1
CHAPTER I
INTRODUCTION
Geographically, a local food system refers to food produced near its point of
consumption in relation to the mainstream food system. Local food production may be
considered a development strategy for rural communities as a way to support their
economies – if consumers buy food produced within a local area instead of imports from
outside the area, more money remains in the local community (Martinez et al., 2010).
Recent national surveys and U.S. Congress reports reflect the strong consumer demand
for local sold directly by farmers at farmers markets (FMs). According to Clark et al.
(2017), much of the interest stems from concerns about the potential impacts of the
global food system, such as increasing food insecurity, reduced farmer incomes, greater
concentration and consolidation in the food system, increased social inequality, and
environmental degradation.
Other concerns are focused on how a food system works in rural and regional
economies. Raja, Ma, and Yadav (2008) and Clark et al. (2017) identified five concerns
in a food system: bottleneck in the industry, winners and losers, perverse incentives,
disparities, and disconnection in the food system. The first concern, bottleneck in the
industry, is related to the unbalanced sales and revenues in the industry (farmers and
ranchers, food processor, manufacturers, wholesalers, distributors, retailers, and
restaurants). Second, winners and losers are related to the profits among farm-share (low)
2
and post-farm share (intermediation - the highest). Third, perverse incentives are related
to the financial support from the government to a specific agricultural industry generating
low costs on low-nutrition products. Fourth, disparities pertain to the household clusters
(often concentrated by race) in neighborhoods with low flow of supermarkets and grocery
stores that offer nutritious food for sale. Finally, disconnection of the food system is
related to the gap between farmers and consumer related to the practices of growing,
processing, selling, cooking, and eating food. These concerns can affect the economic
sustainability of a food system and suggest that improved marketing strategies could be
beneficial to a food system facing such issues.
According to the U.S. Department of Agriculture (USDA) (Low & Vogel, 2011),
the two basic types of local food markets are farmers selling to consumers (direct-to-
consumer) and farmers selling to restaurants, retail stores, and institutions such as
government, hospitals, and schools (direct-to-retail/food service). Venues for the direct-
to-consumer category are typified by FMs, community supported agriculture (CSA), on-
site farm stores or stands, off-site farm stores or stands, online marketplaces, and “pick
your own” operations. As interest in local foods continues to grow and food markets
continue to grow in response, it will be important for farmers to utilize various marketing
strategies to market their products in these types of local food markets.
As a commercial strategy, social media could help to drive sales of agricultural
products and connect sellers and customers at low costs. Thus, the use of social media by
farmers is a potential strategy to promote agricultural products and boost sales. For
instance, social media converts consumers into marketers and advertisers, and consumers
3
can create positive or negative pressure for the small farm business, its products, and its
services (Akar & Topçu, 2011).
Kietzmann, Hermkens, McCarthy, and Silvestre (2011) present a social media
framework of seven building blocks: identity, conversations, sharing, presence,
relationships, reputation, and groups. Identity represents the extent to which users reveal
their identities in a social media setting. Conversations represent the extent to which a
user communicate with other users and facilitates a conversation. Sharing represents the
extent to which users exchange, share, and receive content. Presence represents the extent
to which users might know if other users are accessible. Relationships represent the
extent to which users may be related to other users. Reputation represents the extent to
which users can identify the standings of others. Finally, groups represent the extent to
which users can form communities and subcommunities. These blocks show different
aspects of the experience of social media use and suggest how businesses could engage
with social media and promote their business and products.
Social media introduces substantial and pervasive changes to communication
between organizations, communities, and individuals. However, a gap between
knowledge, perceptions, and attitudes about social media use and acceptance by farmers
and vendors in rural areas might affect connections between farmers and new markets. A
lack of training and marketing support might be barriers to using social media platforms
in these ways. Thus, there may be an opportunity to increase farmers’ awareness of the
use of social media for boosting sales of agricultural products and increasing profits of
their core business.
4
Recognizing the characteristics of the local food system in the U.S., the present
study focused on the “direct-to-consumer” category to assess farmers’ knowledge,
perceptions, and attitudes related to the use of social media to promote and market their
agricultural products. The Unified Theory of Acceptance and Use of Technology
(UTAUT model; Venkatesh, Morris, Davis, & Davis, 2003) was used as the foundation
to identify factors related to farmers’ use and acceptance of social media for marketing
purposes. Another purpose was to measure and compare perceptions and attitudes of the
use of social media as a strategy to promote local competitiveness in farmers’
communities. The growth of local food systems has been driven by farmers, rural
communities, and some governmental institutions who support the local food system. In
some regions, a local food system requires increasing farmers’ knowledge about how to
sell their products in local and external marketplaces using marketing strategies through
the Internet at low costs. Ultimately, this study can contribute to the development of an
Extension program focused on marketing strategies to support farmers’ communities.
Statement of the Problem
Direct-to-consumer marketing, as part of the local food system in rural
communities, provides an advantage in terms of gross sales of locally-marketed food for
developing regional and local economies. Researchers have described the growth of this
marketing category in terms of marketing farm products, sales volumes, and production.
For example, the U.S. farmers market industry shows that this sector continues to
experience brisk growth, but newer FMs might not generate the sales volume enjoyed by
older FMs (Ragland & Tropp, 2009). According to the National Farmers Markets
Managers Survey report, in 2006, the average number of customers at FMs per week
5
declined slightly from 1,055 customers per week, reported in 2000, to 959 customers per
week, reported by managers in 2005 (Ragland & Tropp, 2009). While the U.S. Regional
Government Institutions have concentrated efforts to implement programs to foster
economic development in local food systems, such as the Community Food Project
Grants Program, Community Food Security, Farmers Markets Nutrition Program, Senior
Farmers’ Market Nutrition Program, and the Federal State Marketing Improvement
Program, the U.S. Congress has identified barriers to local food-market entry and
expansion related to distribution, education, research, and training for marketing local
foods (Martinez et al., 2010).
Gillespie, Hilchey, Hinrichs, and Feenstra (2007) suggested that FMs are the
“keystones” for rebuilding local food systems and for supporting economic development
(Brown & Miller, 2008; Gillespie et al., 2007). FMs have a notable impact on local
economies due to the increased retention of local dollars (Henneberry, Whitacre, &
Agustini, 2009). For example, the Crossroad Resource Center published an overview
about the Mississippi farm and food economy explaining that new farmers are emerging
across the state as a response to two concerns: (1) a recognition that the commodity
system is not effectively responding to an emergent local food market and (2) an interest
in learning new skills in a desire to connect with neighbors (Meter & Goldenberg, 2014).
In brief, as part of a local food system, FMs have the potential to improve local
economies and networking among communities.
However, the small farm operation is not being competitive in the agribusiness
industry (Paul, Nehring, Banker, & Somwaru, 2004), and farmers and vendors lack the
expertise and skills in technology platform tools to implement varied marketing strategies
6
(Sumner, 2014). Colasanti, Conner, and Smalley (2010) explained that lack of awareness
of a market existence has been shown to be the most significant deterrent to consumers
shopping at FMs.
It has been assumed that the lack of awareness of FMs among community
residents is largely due to inadequate promotion (Colasanti et al., 2010). Liang (2014)
explains that traditional commercial distribution channels might be less convenient and
less profitable for farm businesses in rural areas due to the lack of information about new
technological marketing formats. A case study of FMs in Oklahoma examined the use of
traditional media (radio, flyers, roadside signs, and television) for promoting agricultural
products and found only 23 percent of customers received information from these
promotional sources (Henneberry et al., 2009). In Florida, the Florida Cooperative
Extension System identified limitations of marketing options centered on location and the
costs of advertising in traditional media channels (Goodwin & Gouldthorpe, 2013). In the
National Farmers Market Manager Survey report, Ragland and Tropp (2009) explained
that traditional media such as signs/banners and newspaper advertising were the two
promotional methods most frequently used to attract customers by 80 percent of farmers
market managers nationwide.
Other marketing case studies, such as Timmons and Wang (2010), found that
direct promotion of agricultural food sales expects to increase sales. Additionally, the
Ragland and Tropp report (2009) explained that the top three FMs’ operational issues that
need improvement are (1) support for market advertising/publicity, (2) strategies for
overcoming low customer attendance, and (3) strategies for boosting vendor sales. Liang
(2014) explained that U.S. small agricultural producers and farmers need to be innovative
7
to survive. Therefore, increasing farmers’ knowledge and skills about the use of social
media to advertise their agricultural products could create opportunities for connecting
farmers with new market niches to support and expand the local food system capabilities
locally, nationally, and even globally.
Social media use in rural communities might enhance marketing strategies among
farmers and consumers to promote agricultural products. In other words, it is possible
that farmers themselves are still using traditional formats of marketing to promote their
brands and products, possibly because they do not know how to use social media as a
marketing tool effectively in their farm business. However, there is no evidence that
confirms that farmers recognize the need to use social media for promoting their
products.
The new information technologies have an influence on all aspects of human
activities (Salim, 2012). Marketing and advertising skills using social media have
received significant attention since social media’s beginnings with the goal of connecting
people. Social media platforms are adopted by businesses primarily to talk with
customers and support multiple communication channels of business-to-consumer,
consumer-to-consumer, and networks of groups (Mandal & Mcqueen, 2012). These
platforms are using innovative tools to integrate marketing skills for developing free or
paid advertisements. For instance, Facebook has tools to develop advertising for
organizations, institutions, and agribusinesses to connect among networks. Generally,
these platforms for marketing purposes are free and simple to use according to the
content generated by the users.
8
Given the popularity of social media, in order to integrate it into the local food
system to promote economic development in rural areas, it is necessary to understand
perceptions and attitudes of farmers related to their acceptance and use of technological
social media platforms (Facebook, Twitter, YouTube etc.) as a marketing tool for
business purposes to promote their products and boost profits. Although social media
research has been conducted in several contexts, there is a gap in the literature on the
usage behaviors of social media by farmers.
One theoretical model that can be used to assess farmers’ knowledge, perceptions,
and attitudes related to social media use for marketing their agricultural products is the
UTAUT model. UTAUT is a compilation of several theories: Diffusion of Innovation
(DOI), Theory of Planned Behavior (TPB), Technology Acceptance Model (TAM),
Technology Acceptance Model 2 (TAM2), and Technology Acceptance Model 3
(TAM3). UTAUT is a popular technology adoption theory that explains almost seventy
percent of variance in adoption behavior (Mandal & Mcqueen, 2012; Venkatesh et al.,
2003). Researchers have applied, integrated, and extended UTAUT across a variety of
settings such as different user types, organization types, types of technologies, tasks,
times, and locations (Venkatesh, Thong, Statistics, Xu, & Acceptance, 2016). The present
study utilized UTAUT to understand farmers’ knowledge, perceptions, and attitudes
related to the use of social media platforms as a marketing strategy for their agricultural
business.
In a context of businesses and organizations, UTAUT offers promise for
improving understanding of technology adoption (Salim, 2012). According to Venkatesh
et al. (2003), four constructs play a significant role as direct determinants of user
9
acceptance and usage behaviors: performance expectancy (PE), effort expectancy (EE),
social influence (SI), and facilitating conditions (FC) which can be affected by
moderators (age, gender, voluntariness, and experience) (see Figure 1).
Figure 1 The Unified Theory of Acceptance and Use of Technology - UTAUT model
Note: Figure extracted from Venkatesh et al. (2003).
PE is defined as the degree to which an individual believes that using the system
will help him or her to attain gains in job performance; EE is defined as the degree of
ease associated with the use and acceptance of the system; SI is defined as the degree to
which an individual perceives how important others believe she/he should use and accept
the new system; and FC is defined as the degree to which an individual believes that an
organizational and technical infrastructure exists to support the use of the system
(Venkatesh et al., 2003). The UTAUT model provides a useful tool for managers needing
to assess the likelihood of success for new technology introductions and helps them
understand the drivers of acceptance in order to proactively design training and marketing
10
interventions (Venkatesh et al., 2003). In summary, the main goal of this study is to
examine factors that affect farmers and vendors’ adoption and use of social media for
marketing their small farm businesses based on the UTAUT model.
Background of the Problem
After World War II, the U.S. food system was structured by local production and
marketing and shifted to a national and global market (Low & Vogel, 2011). According
to a U.S. Congress Report, the land, climate, and technology determined the production
pattern for marketing agricultural products (Martinez et al., 2010). Now, issues related to
environmental awareness, supply chain, technology, and commerce are having important
relevance for government leaders. Today, strong consumer demand for local foods in the
United States is contributing to the agricultural economy due in part to consumer interest
in alternative food systems, including organic production, short supply chains, and direct
marketing involving small farmers. Another issue is U.S. imports of food products. These
imports have grown over the past three decades due to many factors including consumer
demand, the growing U.S. immigrant population, improvements in shipping and
quarantine methods, and the implementation of free-trade agreements (Martinez et al.,
2010). Consumers who are willing to pay more for local production, relative to non-local
production, place importance on product quality, nutritional value, and support of local
farmers.
Consequently, growing interest in local foods is the result of consumer
perceptions regarding local foods and several movements: the environmental movement
(environmental conditions/impact, long distance transportation, challenge to the
dominance of retailers), the slow-food movement (protests against homogeneous, mass-
11
produced food production, and the “fast” nature of people’s lives), and the local food
movement (Gaytán, 2003; Martinez et al., 2010). Encouraging people to make healthy
choices, curbing greenhouse effects, expanding local food access, and supporting local
farmers are important for communities and for government. According to Martinez et al.
(2010), some barriers to market entry and expansion exist, such as capacity limitations
that constrain small local growers, production capacity by lack of infrastructure,
traceback mechanisms (i.e., small farmers combine products with other farmers so
processing and shipping are more economical), and limited farmer expertise and training.
These barriers are relevant issues in the development of a local food system. According
to producers of local foods, one of the major problems was the intermediation demand
with high volumes, time for deliveries, prices, and consistent quality in the agriproducts.
A marketing strategy might be to incorporate more local and regional markets to increase
direct farm sales to local customers, restaurants, and institutions linking with other
partners. Unfortunately, lack of infrastructure in terms of distribution issues is another
barrier to commercializing agricultural products in local and regional markets (Martinez
et al., 2010). According to Gallardo (2016), while an effective online strategy may
increase demand for local foods, a lack of broadband logistical infrastructure to support
such a system could actually undermine the marketing effort. In other words, the interests
of the government for local food systems development around the United States is part of
the agenda to develop regional economies. In summary, barriers were found related to
limited infrastructure, education, and training for members of local food systems; thus,
research could examine marketing channels using innovative technological platforms as a
way to overcome such barriers.
12
The definition of a local food system in the link of a marketing process is
important to generate awareness about economic issues and new market challenges.
Local food market typology has two categories of marketing channels: transactions
between farmers and consumers (direct-to-consumer - DTC: FMs, CSAs, roadside stands,
pick your own, and community gardening), and direct sales by farmers to restaurants,
retail stores, and institutions (direct-to-retail/food service: direct to restaurants,
institutions, regional food aggregators, and retailers) (Martinez et al., 2010). As
mentioned, the study is focused on the DTC category, primarily FMs, CSAs, and on-farm
stores; although findings could be applicable to the direct-to-retail/food service category
as well.
FMs are defined as a common area where several farmers sell quality and fresh
fruits, vegetables, and other farm products to consumers (Low & Vogel, 2011). The most
popular products sold at farmers’ market have been fruits and vegetables followed by
herbs, flowers, honey, nuts, and preserves. A CSA, a replica of the Japanese and
Switzerland model, is defined as a group of people who buy shares for a portion of the
harvest of a farm (Martinez et al., 2010). A typical CSA offers between 8 and 12 types of
produce and herbs per week per shareholder in the production season (Kantor, 2001;
Martinez et al., 2010). Other types of DTC marketing include pick your own (PYO or U-
pick), farm stands, community gardening, and on-farm stores. These types of marketing
depend on the locations and seasonality.
Several studies have found that farmers have a positive impact on local
economies. Marketing of local foods via DTC and intermediated channels grossed $4.8
billion in 2008 (Low & Vogel, 2011). Farmers are selling food commodities through
13
FMs, roadside stands, and other local food sales outlets. The last U.S. Congress Report
explained that in 2012, 7.8 percent of 163,675 farms were marketing food locally and
defined as a DTC; the numbers of farms with DTC sales increased by 17 percent, and
sales increased by 32 percent between 2002 and 2007 (Low et al., 2015). The
Agricultural Resource and Management Survey (ARMS) indicated that local food sales
totaled an estimated $6.1 billion in 2012 (Moss, Kuethe, & Morehart, 2012). Between
2007 and 2012, the number of farmers using DTC channels increased by 5.5 percent,
while the value of DTC sales actually declined by nearly 1 percent when measured in
constant dollars (Low et al., 2015). Table 1 shows a recent census of DTC farms and
sales.
Table 1 Number of direct-to-consumer (DTC) farms and sales, 2002-12
Census year
Item
2002
2007
2012
All farms reporting direct-to-consumer sales
116,733
136,817
144,530
Percent of all farms
5.5
6.2
6.9
Direct-to-consumer sales (millions of dollars)
812
1,211
1,310
Percent of all farm sales
0.4
0.4
0.3
Direct-to-consumer sales (millions of constant dollars:
2012 = 100)
1,002
1,322
1,310
Percentage change from previous census
36.1
31.9
-0.9
Note: Adapted from USDA, National Agricultural Statistics Service, Census of
Agriculture data, various years.
According to Low and Vogel (2011), DTC sales of agricultural products account
for a small but fast-growing segment of U.S. agriculture, increasing by $399 million (49
percent) from 2002 to 2007, and by $660 million (120 percent) from 1997 to 2007.
According to the 2007 Census, 136,800 farms, or 6 percent of all farms in the U.S., sold
$1.2 billion worth of farm products directly to consumers, or 0.4 percent of all
14
agricultural sales if nonedible products are excluded from total agricultural sales
(Diamond & Soto, 2009; Martinez et al., 2010). In brief, the economic indicators about
growth in DTC marketing farms and sales has primarily come from larger operations, and
fruit, vegetable, and beef farms.
On the other hand, there are two main factors for declining sales of the DTC, and
as a consequence the negative growth of the sector in 2012. The first factor relates to the
markets and the new features. Vendors are competing with each other’s consumers, and
they are losing competitiveness against vendors closer to the consumers (Low et al.,
2015). The second factor is the intermediation marketing channel (Low et al., 2015).
Farmers prefer to connect their agricultural products to consumers through retailers and
grocery stores rather than DTC sales primarily due to stock of products, post-harvest
time, and quick distribution to preserve the quality of products. In brief, farms with DTC
sales tended to experience smaller increases in sales than all other farmers.
Basic information about famers and the economic impact of DTC on regions in
Mississippi lacks information in terms of business, financial, marketing, and distribution
channel database. Mississippi has about 90 FMs around the state according to the MS
Government Directory (Mississippi Department of Agriculture & Commerce, n.d.). From
a national point of view, Low and Vogel (2011) mentioned that combining marketing
channels may represent the appropriate market strategy for medium-sized farms and rural
communities to thrive, and that it could be an important topic to study. As a marketing
strategy, these sectors should be able to identify and develop the best advertising format
using social media tools to promote agricultural products. In summary, several issues
about high interest in the economic sector, movements of the local food demand, lack of
15
infrastructure and knowledge to reach new markets are a concern of government and
stakeholders as challenges to promote the DTC sector. Social media as a marketing tool
might support economic growth in rural areas and increase profits in small farmers’
businesses. However, it is important to understand what factors (e.g., attitudes,
perceptions, knowledge) could influence farmers’ decisions to use social media for
marketing their products.
Attitudes are defined as the positions, postures, emotions, and feelings of people
toward any technology item or platform. According to experts, attitude toward an
innovation is a critical intervening variable in the innovation adoption decision (Rogers,
2003). From a marketing view, an attitude can be described as a person’s enduring
favorable or unfavorable evaluation, emotional feeling, and action tendencies toward
some object or idea (Akar & Topçu, 2011; Kotler & Keller, 2006). Attitudes and
perceptions can drive people’s decision making about dislike or like of a social media
platform. For example, the theory of planned behavior (Ajzen, 1991) explained three
conceptually independent determinants of intention that are integrated in UTAUT:
attitude, subjective norms, and perceived behavioral control, which are explained deeply
in the next chapter. UTAUT includes additional components that can affect behavior: PE,
EE, SI, and FC (Venkatesh et al., 2003). At the end, these components might identify
positive or negative influences toward social media use in different amounts (high or
low). The results might be influential in the design of an Extension program about
marketing skills using social media for farmers.
16
Purpose of the Study
The purpose of this quantitative, non-experimental survey research was to
examine factors that affect the acceptance and use of social media by farmers and
vendors in farmers’ communities for marketing their small farm businesses in North
Mississippi using the UTAUT model proposed by Venkatesh et al. (2003). By applying
UTAUT using a multivariate technique to combine factors (PE, EE, SI, and FC), the
study identified relationships among the factors and their influence on the use and
acceptance of social media to promote agricultural products by farmers.
Research Questions
The research questions for this study were based on the UTAUT model
(Venkatesh et al., 2003):
1. How does PE affect BI of farmers and vendors to use social media for
marketing their small farm businesses?
2. How does EE affect BI of farmers and vendors to use social media for
marketing their small farm businesses?
3. How does SI affect BI of farmers and vendors to use social media for
marketing their small farm businesses?
4. How do FC affect farmers and vendors’ USE for marketing their small farm
businesses?
5. How does gender moderate the relationships from PE, EE, and SI to farmers
and vendors’ BI to use social media for marketing their small farm
businesses?
17
6. (a) How does age moderate the relationships from PE, EE, and SI to farmers
and vendors’ BI to use social media for marketing their small farm
businesses? (b) How does age moderate the relationship between FC and
farmers and vendors’ USE for marketing their small farm businesses?
7. (a) How does experience of social media use moderate the relationships from
EE and SI to farmers and vendors’ BI to use social media for marketing their
small farm businesses? (b) How does experience of social media use moderate
the relationship from FC to farmers and vendors’ USE for marketing their
small farm businesses?
8. How does BI affect farmers and vendors’ USE for marketing their small farm
businesses?
9. What are the reasons (barriers) that prevent farmers and vendors from selling
their products through various marketing channels (FM, CSA, retailer,
roadside stand, website, and social media)?
Significance of the Study
The present study is an empirical application of the UTAUT model (Venkatesh et
al., 2016) to the context of social media adoption and use by farmers’ communities in
North Mississippi. This study contributes to the literature on the factors affecting
farmers’ intention to use social media to promote agriproducts to connect new markets.
Additionally, through analyzing factors that affect the acceptance and use of social media
by farmers and vendors to advertise and promote their agriproducts, future benefits from
this work could be the development and implementation of Extension programs on
marketing strategies using social media.
18
Using social media platforms with a low cost (almost free) will help support
farmers and vendors to promote their agricultural products. Networking local economies
will support economic development in the rural communities. Research has demonstrated
that social media platforms have experienced expansion, becoming essential channels for
business and marketing (Herrero, San Martín, & Garcia-De los Salmones, 2017; Warner-
Søderholm et al., 2018), and social media can be a good virtual tool to provide a bridge
between farmers and consumers. In brief, the present study documents factors in the use
and acceptance of social media by farmers, including their knowledge, attitudes, and
perceptions on topics related to social media use for marketing purposes with an ultimate
goal of providing information for the development of Extension programs and generation
of awareness to the local and state government of the value of extending technological
resources for supporting economic development in rural areas.
Extension Programs
Agriculture in Mississippi is of major economic importance. According to the
Agricultural Resource and Management Survey, Extension programs are focused on
meeting expectations and needs of different social and productive, public, and private
sectors in the country (Low et al., 2015). Barriers to local food market entry and
expansion include capacity constraints for small farms and lack of distribution systems
for moving local food into mainstream markets; limited research, education, and training
for marketing local food; and uncertainties related to regulations that may affect local
food production, such as food safety requirements (Martinez et al., 2010). For example,
Mississippi State University (MSU) Extension has as a mission to provide research-based
information, educational programs, and technology transfer according to the needs of
19
rural communities to improve their economic, social, and cultural well-being (Mississippi
State University, n.d.). According to MSU Extension, its main purpose is education to
empower families in rural communities to make decision in terms of their vocations, their
families, and their social and economic environment. One of the goals for an educational
program focused on marketing in rural regions would be to connect the knowledge (the
university) and businesses (agribusiness) to prepare and develop skills and capabilities in
agribusiness and marketing. Outcomes from the present study can support the design of
new Extension programs focused on agribusiness, marketing, and social media use.
Policymakers
Federal, state, and local institutions of the government are implementing
programs to support local food systems in rural areas. For example, government food
assistance programs such as the Women, Infants, and Children Farmers Market Nutrition
Program and the Senior Farmers Market Nutrition Program are programs with a goal to
increase the availability of fresh fruits and vegetables to households at risk for food
insecurity, thus boosting sales of fresh fruits and vegetables (Ragland & Tropp, 2009).
Although there are some federal programs and policies supporting local food systems,
Mississippi could benefit from additional governmental support for small farm businesses
in the region (Meter & Goldenberg, 2014). In addition, the Farmers’ Market Promotion
Program was created in 2002 with the only goal being to expand scope, and the 2014
Farm Bill (federal level) provided support for Extension programs to promote production
and marketing of food locally and regionally. In brief, policymakers may be able to use
findings from this study for decision making related to marketing of local foods and any
associated potential economic growth in rural communities.
20
Social and Economic Issues
The agribusiness field needs to know and support topics related to social and
economy issues. In a local food system, the DTC category has a variety of channels to
promote agricultural products to rural communities. Mississippi consumers spend $7.1
billion buying food each year, including $4.1 billion for home use, and most of this food
is produced outside the state; thus, state consumers spend at least $6.5 billion per year
buying food sourced outside of Mississippi (Meter & Goldenberg, 2014). Social media
creates a linkage between farmers and customers to promote agricultural products and set
up networking in the local food system. Supporting farmers to do this will help them
reach new markets and bring new customers for implementing strategies and innovation
through the use social media. In terms of economic issues, social media will promote
agribusiness as a branch of the economic development in rural areas focusing on the
marketing skills. Using social media to promote agriproducts, farmers might identify new
customers, access new markets, and generate new opportunities. In terms of social issues,
the study will analyze social needs of the farmers, evaluate unemployment rates in rural
areas and communities, and link other rural entities in the local food system such as
schools, government institutions, hospitals, volunteers, and so on. The benefits of the
social media use are to strengthen the local food system as a strong network generating
economic development and growth.
Definitions and Key Terms
Agricultural marketing: Commercial functions involved in transferring agricultural
products consisting of farm, horticultural, and other allied products from producer to
consumer (Vadivelu & Kiran, 2013).
21
Community Supported Agriculture – CSA: During the 1960’s, the concept of
community supported agriculture originated in Switzerland and Japan. A group of people
buy shares for a portion of the expected harvest of a farm (Farnsworth, Thompson, Drury,
& Warner, 1996; Martinez et al., 2010).
Confirmatory Factor Analysis - CFA: Use of multivariate technique to test (confirm) a
prespecified relationship (Hair, Black, Babin, & Anderson, 2009). Additionally, it is a
way of testing how well measured variables represent a smaller number of constructs.
Direct-to-Consumer Marketing: The Census of Agriculture, conducted by USDA’s
National Agricultural Statistics Service defines direct sales-to-consumer as the value of
agricultural products sold directly to individuals for human consumption from roadside
stands, farmers’ markets, pick your own operations, etc. It excludes livestock sales. Sales
of agricultural products by vertically integrated, operations through their own processing
and marketing operations are also excluded (Martinez et al., 2010).
Exploratory Factor Analysis - EFA: An interdependence technique to define the
underlying structure among the variables in the analysis where these variables play a key
role in any multivariate analysis (Hair et al., 2009, p. 94).
Farmers’ communities: People who care about protecting farmland and healthy, local
food come together to build and support community farms in their neighborhoods
(Martinez et al., 2010).
Farmers market: A farmers market is a common area where several farmers gather on a
recurring basis to sell a variety of fresh fruit, vegetables, and other farm products directly
to consumers (Martinez et al., 2010).
22
Local food system: Defined in terms of social and supply chain characteristics. Food
produced, processed, and distributed within a particular geography boundary that
consumers associate with their own community. “Local food” or “local food system” are
often used interchangeably to refer to food produced near its point of consumption in
relation to the modern or mainstream food system (Martinez et al., 2010; Peters, Bills,
Wilkins, & Fick, 2008).
Marketing: Identifying and meeting human and social needs. One of the shortest good
definitions of marketing is “meeting needs profitably” (Kotler & Keller, 2014).
Structural Equation Modeling - SEM: Multivariate technique combining aspects of
factor analysis and multiple regression that enables the researcher to simultaneously
examine a series of interrelated dependence relationships among the measured variables
and latent constructs (variates) as well as between several latent constructs (Hair et al.,
2009).
Social media: Defined as a group of internet-based applications that build on the
ideological and technical foundations of Web 2.0, and that allow the creation and
exchange of user generated content (Kaplan & Haenlein, 2010; Whiting & Williams,
2013).
Unified Theory of Acceptance and Use of Technology - UTAUT: Venkatesh et al.
(2003) developed UTAUT as a comprehensive synthesis of prior technology acceptance
research. UTAUT which aims to explain technology acceptance, is based on eight
technology acceptance theories or models (Innovation Diffusion Theory – IDT, Social
Cognitive Theory – SCT, Theory of Planned Behavior - TPB, Theory of Reasoned
Action - TRA, Technology Acceptance Model - TAM). UTAUT uses behavioral
23
intention as a predictor of the technology use behavior and it included predictors of
behavioral intention are based on the components the eight technology adoption models
reviewed (Thomas, Singh, & Gaffar, 2013).
24
CHAPTER II
LITERATURE REVIEW
In this chapter, an overview of local food markets is provided and farmers’ needs
related to marketing and promoting agriproducts, social media marketing, and the
UTAUT model are discussed. In the local food markets section, the concept of the direct-
to-consumer (DTC) category of a local food system (FMs and CSAs), consumer
motivations to buy from farmers’ communities, and the economic impact in rural areas
are covered. Farmers’ community needs related to marketing and promotion skills and
the use of social media as a marketing tool to support and promote agriproducts is
discussed next. Finally, the UTAUT is described in depth to review its concepts and
application to social media.
Local Food Market Overview
Concept and Description
The definition of “local food” is based on the set of marketing channels (as
measured from the farm gate to the consumer) used by farmers (Hand & Martinez, 2010;
Low & Vogel, 2011; Martinez et al., 2010); in other words, it is defined in terms of social
and supply chain characteristics (Martinez et al., 2010). The concept integrates terms
related to geographic proximity, distance, and consumer attributes; but the concept of
“local” refers to food produced near its point of consumption in relation to the
mainstream food system (Peters et al., 2008). Local food system has also been seen as
25
synonymous to the small farm sellers committed to place agricultural products to support
social and economic issues in a community. Likewise, it can be defined as the
characteristics of intermediate stages of the supply chain where people may connect
between food consumers and producers (Martinez et al., 2010).
Local food is categorized into two basic types: those where transactions are
conducted directly between farmers and consumers (direct-to-consumer or DTC) and
direct sales by farmers to restaurants, retailers, and institutions such as government
entities, hospitals, and schools (direct-to-retail/food service). In 2012, the number of
farms with DTC sales increased 17 percent. DTC increased by 32 percent between 2002
and 2007; however, between 2007 and 2012, the number of farms with DTC sales
increased 5.5 percent, with no change in DTC sales (Low et al., 2015). Clearly, consumer
interests in direct-to-retail/food service and the growth in nondirect sales have affected
FM sales. Consumers have more opportunities to purchase food directly from producers,
with 8,268 FMs operating in 2014 in the United States (Low et al., 2015; Martinez et al.,
2010). The most recent data on sales in the FMs industry, for example, explains that total
FMs sales in 2005 were estimated to have slightly exceeded $1 billion, compared with
$888 million in 2000, an average annual growth rate of 2.5 percent (Ragland & Tropp,
2009). For example, in Mississippi, only $9.7 million of food products (0.2 percent of
farm cash receipts and 0.1 percent of the state’s consumer market) are sold by farmers
directly to consumers (Meter & Goldenberg, 2014).
Studies have found that farmers’ communities have a positive impact on local
economies supported by local food system. Marketing of local foods via DTC and
intermediated channels grossed $4.8 billion in 2008 (Low & Vogel, 2011). Farmers
26
selling food commodities through farmers’ communities (e.g., FMs, CSAs, roadside
stands, and other local food sales outlets) depend on transportation, logistics, and
customers. Information about farmers’ communities and their economic impact on each
region in Mississippi is poor in terms of marketing and channels. For example, a USDA
Report stated that the numbers of FMs rose to 5,274 in 2009, up from 2,756 in 1998 and
1,755 in 1994, according to USDA’s Agricultural Marketing Service (Low & Vogel,
2011). At the same time, combining marketing channels may represent the appropriate
market strategy for medium-sized farms to thrive.
Local food sales were estimated at $6.1 billion in 2012 according to the Census of
Agriculture Data; however, farms with DTC tended to experience smaller increases in
sales than all other farms (Low et al., 2015). However, between the year 2000 and the end
of 2005, the number of FMs, specifically, in the United States increased 43 percent, from
2,863 to 4,093, an average growth rate of 8.6 percent a year (Ragland & Tropp, 2009).
This means the interest of the rural population, the government, and the farmers’
communities to promote agriproducts in favor of the healthy food consumption and the
economy can generate new business opportunities.
Low and Vogel (2011) typify a local foods system as a direct-to-consumer (DTC)
outlets and intermediated marketing channels that can include the following channels:
FMs, roadside stands, on farm store, and CSAs. All are considered a local source of fresh,
nutritious products, and typically include vegetables, fruits, herbs, flowers, plants, baked
goods, honey, nuts, meat, and poultry products, as well as dairy products and eggs
(Abelló, Palma, Waller, & Anderson, 2014; Ragland & Tropp, 2009).
27
Farmers Markets - FMs
FMs are the most important marketing channel contributor of a local food system
in rural areas (Henneberry et al., 2009; Hughes, Brown, Miller, & McConnell, 2008;
Sadler, Clark, & Gilliland, 2013). A FM is a gathering area where several farmers sell a
variety of fresh fruits, vegetables, and other farm products directly to consumers
seasonally (Martinez et al., 2010). FMs are a retail outlet in which two or more vendors
sell agricultural products directly to customers through a common marketing channel
(Ragland & Tropp, 2009). The purpose of the FMs as a main marketing channel is for
farmers to promote and sell their agricultural products to local communities. Most
established FMs have hired individuals to oversee the organization, rules and regulations,
and promotions for all growers, and most also charge vendor fees for selling privileges,
including a flat fee as space is available, a membership fee for the entire season, or a fee
based on a percentage of vendor sales (Martinez et al., 2010; Ragland & Tropp, 2009).
They are operated by individual vendors who offer a variety of locally produced,
seasonal, and organic products (Dodds et al., 2014). In other words, it is a place
considered as an important point of sale of agricultural products for a local community.
A Census of Agriculture report explained demographic data related to farmers’
age. The age average of the American farmer is now nearly 60 years old, but there was a
measurable increase in farmers aged 25 to 34 during the 2007 and 2012. However, many
young and beginning farmers may be struggling to stay on the land working for their
small farm business (U.S. Census of Agriculture, 2012). Although the present study is
focused on farmers and vendors in the DTC category of a local food system, farmers’
28
average age is an important variable to be considered in understanding social media
acceptance and use for marketing agricultural products.
According to the USDA Agricultural Marketing Service’s 2006 and National
Farmers’ Market Survey, the most popular product category sold at FMs was fresh fruits
and vegetables, which was sold by nearly 92 percent of farmers’ markets in 2005,
followed by herbs, flowers, honey, nuts, and preserves (Martinez et al., 2010; Ragland &
Tropp, 2009). Curtis, Cowee, Velcherean, and Gatzke (2010) indicated that the majority
of products purchased at FMs are organic. However, not all products sold there are part
of the local food system (Hughes et al., 2008). For example, some vendors may come
from outside the local region, and some local vendors may not sell products that are
produced within the region (Martinez et al., 2010).
Community Supported Agriculture - CSAs
The model of CSA originated from Switzerland and Japan in the 1960s
(Farnsworth et al., 1996; Martinez et al., 2010). The community is described as a group
of people who buy shares for a portion of the expected harvest of a farm (Local Harvest,
2017). A CSA is a marketing strategy where consumers buy “shares” in the farm before
planting begins and then they receive a portion of whatever is available each week of the
growing season (Brown & Miller, 2008). This model is considered an alternative socio-
economic model of agriculture and food distribution. For example, Cooley and Lass
(1998) found local organic retail values of CSA produce to be 1.5 to 2.5 times the values
of customers’ shares for three organic Massachusetts CSA farms (Brown & Miller, 2008;
Cooley & Lass, 1998). Also, a CSA allows the producer and consumer to share risks of
29
farming. In return for subscribing to a harvest, subscribers receive either a weekly or bi-
weekly box of produce or other farm goods.
According to the Economic Research Service of the USDA, by 2005, there were
1,144 CSAs compared to 761 in 2001, an increase of 50 percent (Adam, 2016; Martinez
et al., 2010). An online registry estimates that the number of CSAs exceeds 4,000 (Local
Harvest, 2017). Business organizations for CSA programs include sole proprietorships
(single farm), partnership and farm cooperatives (multiple farms), and limited liability
corporations. The typical CSA offers a mix of between 8 and 12 types of produce and
herbs per week per shareholder throughout the growing season (Kantor, 2001; Martinez
et al., 2010).
FMs and CSAs as Part of the Local Food System
FMs and CSAs have also been described as centerpieces of locally integrated food
systems, which connect producers and consumers, and of civic agriculture, which
(re)embeds market transactions within community relationships that build social capital
(Colasanti et al., 2010). They can provide healthy food as a part of the supply chain
system in local communities. Raja et al. (2008) determined food system elements: (1)
food production, (2) distribution and aggregation, (3) food processing, (4) marketing, (5)
markets and purchasing, (6) preparation and consumption, and (7) resource and waste
recovery. All marketing channels used to promote agriproducts should be incorporated as
strategies to fit into the food system elements.
On the other hand, FMs and CSAs involve small farmers, heterogeneous products,
and short supply chains in which farmers also perform marketing functions, including
storage, packaging, transportation, distribution, and advertising (Martinez et al., 2010).
30
The connections among entities in the communities have an advantage about the supply
chain in the food system. Moreover, two of the most prominent benefits in the system are
obtaining quality products and the local supporting community (Dodds et al., 2014).
Helping the farmers’ communities will generate a strong food system to support local
economies including restaurants, hospitals, schools, businesses, and government
institutions.
In the FMs and CSAs, farmers are encouraged to understand consumer profiles,
provide quality and healthy products, and maintain the important sense of community at
markets by maintaining local vendors with local and organic products (Dodds et al.,
2014). However, farmers should consider the social opportunity of FMs in order to
realize the economic benefits available to successful FMs (Dodds et al., 2014). In other
words, farmers’ communities of FMs and CSAs are an important piece of the local food
system chain to support and sustain quality life and health promotion of any community
in order to promote better social and economic conditions.
Consumer Motivations to Buy in Farmers’ Communities
For the present study, the consumer motivation concept explains how customers
purchase agricultural products according to their needs and preferences. Growing interest
in local foods in the United States is the result of several movements (Guptill & Wilkins,
2002; Martinez et al., 2010). According to several researchers, a movement is an informal
group of people which aims at social change, and in this particular case, several
approaches have been utilized (Martinez et al., 2010). The organic movement emphasizes
soil health, human health, and holistic or ecological farming practices (Delind, 2006;
Dodds et al., 2014). Local food movement is a reflection of the food revolution that is
31
taking place due to rising concerns regarding obesity and other diet-related health
problems (Dodds et al., 2014; Ikerd, 2011). Health issues are the primary concern for
each local and regional government in terms of public politics. An earlier national survey,
collected in 2003, found an interest in healthy, safe, and fresh foods increased the
likelihood of buying locally (Low et al., 2015; Zepeda & Nie, 2012). In other words,
people are becoming more aware of buying and consuming healthy products grown or
produced through organic methods.
The most recent national data suggest that while local food consumers are
demographically diverse, they are very similar in their motivations for buying local
(Martinez et al., 2010). In the nationally representative U.S. Grocery Shopper Trends
Survey, consumers reported their top reasons for buying locally grown foods in grocery
stores –freshness was the most frequent reason (83 percent) cited, and taste was the
second reason (with 56 percent for buying local food (Food Marketing Institute, 2011;
Low et al., 2015). Thus, consumer interests in buying agricultural products leads to the
importance of understanding marketing strategies that can be developed by farmers’
communities. In other studies, the role of demographic characteristics was somewhat
stronger, and consumers who were female, older, more educated, higher income earners,
and members of environmental groups were more likely to buy local food (Brooker &
Eastwood, 1989; Brown, 2003; Eastwood, 1996; Eastwood, Brooker, & Gray, 1999;
Govindasamy & Adelaja, 1998; Martinez et al., 2010). In brief, consumers' behaviors and
motivations explain a farmers’ need to structure new marketing strategies in their
business for promoting and advertising its agriproducts reaching new markets.
32
Other factors influencing consumers might be grouped as product features and
variety, the actual market experience, and socio-demographic characteristics of
consumers as related to the market and products (Abelló et al., 2014). For example, some
consumers who enjoy cooking, growing a food garden, frequenting health food stores,
and purchasing organic food are more likely to buy local food (Martinez et al., 2010).
Recognizing and advertising what the customers are seeking in a FM is a huge
opportunity to structure new marketing strategies based on these customer preferences
using technological platforms such as social media. Consumers seeking quality fruits and
vegetables go to local food outlets such as FMs and CSAs (Adams & Salois, 2010;
Dodds et al., 2014). Additionally, consumers who value high-quality foods produced with
low environmental impact are often willing to pay more for locally produced food. These
consumers place importance on product quality, nutritional value, methods of raising
products and how those methods affect the environment, and support for local farmers
(Martinez et al., 2010). From this standpoint, implementing new marketing strategies
using social media in farmers’ communities could provide a competitive advantage by
reaching those customers online. While grocery stores provide the consumer with
convenience and relatively consistent quality food, farmers’ communities provide
consumers with traceability, verbal quality assurance, nutrition education, and sometimes
negotiable, low-priced, locally grown products (Watson & Gunderson, 2010). Through
identifying customers’ preferences and motivations for agricultural product consumption,
marketing strategies will drive main elements focused on product, price, place, and
promotion. These strategies supported by social media will connect customers and
33
vendors in a local food system to support economic development in farmers’
communities.
Economic Impact in Rural Communities
Economic impact is an important issue for any strategy implemented in rural
communities. Hughes et al. (2008) proposed that FMs help and support local community
enabling a greater portion of local dollars to stay in the local economy. Additionally,
FMs, CSAs, and other direct marketing venues can have a notable impact on local and
regional economies due to the retention of local dollars (Hughes et al., 2008). In terms of
employment, it was also found that DTC category (e.g., FMs) created between one and
five jobs per stall vendor and comprised one-third of vendors’ income on average.
(Dodds et al., 2014). Socioeconomic factors could reduce unemployment rates in rural
communities. On the other hand, price is not regarded as a main motivation to visit these
marketing channels by consumers. Keeping price ranges similar to grocery stores may
encourage more consumers to purchase a larger amount of their grocery needs at FMs
and CSAs and could make them more accessible to those with lower incomes (Dodds et
al., 2014).
On the demand side, past studies have shown that customers are primarily
attracted by three factors: the overall quality of the products offered (freshness, taste, and
food safety), the lower prices compared to those of comparable goods in supermarkets,
and the atmosphere within the FMs and CSAs (Henneberry et al., 2009; Hughes &
Mattson, 1992). As mentioned earlier, these factors are important for implementing
marketing strategies focused on fresh, organic, and quality production using
technological platform to reach new markets. Others factors for enhancing the local
34
economy and reasons for shopping at a local farmers’ community might be focused on
benefiting the environment, conserving resources, building a deeper relationship with the
growers, and providing a meeting place for friends and community members
(Cummings, Kora, & Murray, 1999; Henneberry et al., 2009; Zepeda & Leviten-Reid,
2004). Essentially, FMs and CSAs allow farmers to contribute to the economic life of
local communities by providing goods and services that are not readily available through
formal, mass markets, and they bring producers and consumers together to solidify bonds
of local identity and solidarity (Lyson, Gillespie, & Hilchey, 1995; Watson & Gunderson,
2010). The economic impact generated by FMs and CSAs as gears of a local food system
in rural communities is fundamental for the development of rural regions for promoting
prosperity, economic, and social welfare of their inhabitants.
Some studies have demonstrated the importance of the economic impact in
several regions and communities. For example, a survey of Oklahoma FM vendors found
total gross farmers market sales for the 2001 season of $3.3 million and $7.8 million in
direct and secondary impacts on the state’s economy (Henneberry et al., 2009). Brown
and Miller (2008) estimated that $630,000 spent by consumers in other sectors led to a
total statewide (Oklahoma) impact of $1.9 million, and 795 jobs were directly generated
by farmers’ markets including the largely unpaid labor force of market managers that
sustained an additional 1,145 jobs in related activities. As the number of FMs has grown,
several grassroots initiatives have attempted to provide FMs with tools to determine and
report their economic impacts in order to leverage support from city, county, and state
policymakers (Brown & Miller, 2008). In short, the increase in the sales volume in the
35
FMs generated an increase of vendors, jobs, and resources that positively impacted the
Oklahoma community economy.
In a study of Iowa FMs, Otto and Varner (2005) estimated that each dollar spent
at FMs in Iowa generated 58 cents in indirect and induced sales, while each dollar of
personal income earned at farmers’ markets generated an additional 47 cents in indirect
and induced income. As a result, total sales and the associated economic impact for
Iowa’s farmers’ markets were estimated to be positive. In summary, FMs and CSAs have
a positive impact on local economies and with a strong presence, and marketing
promotion might encourage farmers’ business communities to benefit the local food
system.
Farmers’ Communities Needs
According to several studies, farmers have different needs in terms of marketing
channels, location, advertising, etc., in their business and communities. The most
prominent needs are for signage and promotion, reduced time constraints, convenient
location or adequate facilities, and a welcoming atmosphere (Colasanti et al., 2010).
Other concerns from the farmers’ perspective are marketing risks when selling in local
markets that include low sales volume, price competition from multiple sellers with the
same product and local angle, rejection based on quality requirements, inability to meet
specifications, inability to meet logistical requirements, and buyers backing out of
contracts (Leroux, Schmit, Roth, & Streeter, 2010; Martinez et al., 2010).
In a study in Michigan, focus group participants saw a need for the local farmers
market to better advertise to the community (Colasanti et al., 2010). Recall that
advertising has been identified as a perceived need by farmers in farmers’ communities
36
and FMs as well. The focus group participants also expressed a wide array of ideas about
potential advertising channels, and these ranged from television commercials to mail
pieces; however, the local newspaper was the venue most consistently suggested
(Colasanti et al., 2010). From these perspectives, advertising as a part of marketing plays
an important role in the development of new technological marketing methods
encompassing the traditional media to social media.
Barriers to entry and expansion may hinder progress in local food market
development (Martinez et al., 2010). These barriers can be typified in economic, social,
cultural, and technological needs. Researchers found that consumers who shop at FMs
enjoyed the shopping experience, but others cited access issues such as distance from the
consumer’s home and inconvenience of the location as barriers (Eastwood et al., 1999;
Zepeda & Li, 2006).
Lack of knowledge related to using innovative tools to advertise and promote
events might be barriers to overcome in farmers’ communities. For example, the lack of
FMs’ promotion seems to indicate that more diverse means of market advertising may
help to broaden the customer base (Colasanti et al., 2010). In a recent study, FMs highly
desire assistance with advertising and publicizing the market, improving the operating
budget, increasing federal nutrition benefit sales, improving sales per vendor, and
increasing customer numbers and volunteers (Berry, Moyer, & Oberholtzer, 2013), and it
might be occurring to other communities such as CSAs. In short, barriers for farmers are
focused on marketing and advertising skills but also on the main three legs of any
business: financial, operations, and marketing. Related specifically to marketing, there is
a need to apply innovative tools for advertising and promoting agriproducts.
37
While farmers’ communities are becoming more popular in some environments,
information about the potential, demand, and development of marketing channels in other
regions is lacking (Alonso & O’Neill, 2011). Zepeda (2009) found that while income
does not seem to play a role in influencing purchases at FMs and CSAs, if consumers
perceive cost as the most important attribute of food, the probability of making purchases
would be dramatically reduced (Alonso & O’Neill, 2011; Zepeda, 2009). In other words,
income is not an important role in the decision to buy agricultural products if consumers
perceive the benefits in terms of health, nutrition, and quality of life. Alonso and O'Neill
(2011) also explained that opportunities exist for many farmers to maximize their
revenues, decrease their dependence on wholesalers or ‘middlemen,’ and provide a
service that is of significant benefit to consumers and society. In other words, educating
consumers and raising awareness about local fresh produce are opportunities to present in
advertising. Today, changing eating and lifestyle patterns of individuals are becoming
disengaged, even disenfranchised, from such important aspects of everyday life as locally
grown foods, cooking local dishes and/or the importance of experiencing the educative
side of the visit (Alonso & O’Neill, 2011). In summary, according to Berry et al. (2013),
selling agriproducts through FMs and CSAs might be a good way for farmers to develop
marketing skills, expand their business into the local food system, and build a bridge
between farmers and customers implementing innovative tools supplying farmer’s needs.
Marketing and Promotion
Marketing and promotion are significant activities for agribusiness, as well as
other areas of rural communities. For example, MarketMaker™ is a direct marketing
cutting-edge tool available to anyone whom wants to buy and/or sell agriproducts with
38
access to the Internet. The social network website helps farmers with a food-related
product or service target markets and build a customer base. Suppliers can find buyers,
and buyers can locate suppliers (Barnes, Hood, & Gallardo, 2014). As a promotional tool,
MarketMaker™ links two sectors (buyers and sellers) allowing them to integrate and
advertise farmers’ brands in an online platform. As a part of social media platforms, some
technological tools are taking advantage to support and help farmers in the marketing
process and promotion of agricultural products.
Researchers are focusing on recommendations considering market and policy
interventions on prioritizing attention to marketing and promotion. Schmit and Gómez
(2011) suggest that the success of vendors at FMs and CSAs should not be analyzed only
from sales performance; other outcomes are important as well. In the FMs, for example,
promotional events should be focused on educational activities and festivals (Abelló et
al., 2014); but, this strategy is not enough to promote agriproducts. The goal of the
present study was to identify and analyze factors in the adoption and use of social media
by the farmers and vendors for marketing their products. Limited promotion of FMs and
the use of mostly mainstream advertising channels may encourage proportionately greater
FM and CSA attendance from mainstream cultures (Colasanti et al., 2010). For example,
FMs in Oklahoma undertake a variety of promotional activities using radio, internet,
flyers, roadside signs, and television to attract more customers; however, only 23 percent
of customers indicated they had received their information from these promotional
sources (Henneberry et al., 2009).
Market visitors influence the effects of promotional activities, and it is probably a
reflection of the specific demographic groups associated with each specific market
39
(Abelló et al., 2014). In fact, traditional marketing strategies among CSAs and FMs still
promote agricultural products at high costs. Consumer participation may also be
supported through improved information about the availability of products and benefits
from shopping at the establishment, as well as advertising in the popular media.
However, while pamphlets and websites do a good job of providing information, farmers
are most likely to be found by those actively seeking information on the topic
(Henneberry et al., 2009).
Studies present arguments about the lack of technology tools in the use of
advertising and the presence in rural communities. According to Gallardo (2016), rural
residents, businesses, and governments may not have time, expertise, or resources to
implement a robust online presence strategy, and the author stated that social media can
provide a powerful and effective way to engage. Another study described how the loss of
traditional advertising channels and the cost of advertising are negatively affecting farm
product marketing (Goodwin & Gouldthorpe, 2013).
Another recent concept that has started to describe the marketing strategy in the
agricultural sector is called agricultural marketing. Agricultural marketing might be
defined as the commercial functions involved in transferring agricultural products,
consisting of farm, horticultural, and other allied products, from producer to consumer
(Vadivelu & Kiran, 2013). The concept may be effective if it is looked at from the
collective and integrative efforts from various sectors by addressing farmers’
communities, middlemen (intermediation), researchers, vendors, producers, and
managers.
40
Vadivelu and Kiran (2013) stated that it is time to bring out significant strategies
in agricultural marketing with innovative and creative approaches. A study by Timmons
and Wang (2010) stated that promoting direct food sales would be expected to increase
sales as a result of the interest of local food systems in the United States. A study
conducted in Florida found that small farmers conveyed the challenges associated with
marketing, market availability, and competition (Goodwin & Gouldthorpe, 2013).
Another study implemented in Greenwood, Mississippi, showed that for farmers,
Facebook was a valuable marketing tool and an investment in their learning experience
(Barnes & Coatney, 2015).
Costs associated with advertising and accessing advertising channels are also seen
as formidable challenges in agricultural marketing (Barnes & Coatney, 2015). Goodwin
and Gouldthorpe (2013) stated that advertising is expensive in any large form in
newspapers and magazines for agribusiness purposes, and while a small ad for $20 a
month can get a lot of business, it is hard to find an affordable price to advertise FMs.
Although traditional advertising media have been expensive without any visible return on
the investment, social media might be a useful tool to advertise agricultural products
using numerous advertising messages daily through audio, video, and pictures at low
cost. In brief, some studies related to farmers, small farmers, and agribusiness have
identified the concept of agricultural marketing as an important need, especially related to
social media use, in agricultural small businesses due to the perceived limitations of
market availability and marketing opportunities often centered on location and the costs
of advertising.
41
Social Media Marketing
Social Media Overview
Social media is a developing phenomenon in marketing and promotion to connect
consumers with producers. Kaplan and Haenlein (2010) defined social media as a group
of internet-based applications that build on the ideological and technical foundations of
Web 2.0, and that allow the creation and exchange of user generated content. William et
al. (2012) defined social media as a communication mechanism that allows users to
communicate with thousands, and perhaps billions of individuals all over the world
(Whiting & Williams, 2013; Williams, Crittenden, Keo, & Mccarty, 2012).
Marketers are beginning to understand the use of social media as a fundamental
component of competitiveness to structure effective marketing strategies and campaigns
to reach out to customers. Experts and researchers have described social media as a
bridge for communication from producers to consumers and as a channel to promote a
business and its products. Promotions, marketing intelligence, sentiment research, public
relations, marketing communications, and product and consumer management are
subdisciplines of marketing that may use social media (Akar & Topçu, 2011; Tanuri,
2010).
Mangold and Faulds (2009) explained that social media represents a prevalent
source of information and it has changed the tools and strategies companies use to
communicate, highlighting that information control now lies with the customers.
Businesses are taking advantage of introducing social media into their marketing
programs. Roberts and Kraynak (2008) explained that social media converts consumers
into marketers and advertisers, and the consumers might create positive or negative
42
pressure for the business, its products, and its services (Akar & Topçu, 2011; Roberts &
Kraynak, 2008). Social media is an important technological tool that can create positive
expectations about how a business is presented online and how the quality of products
and services is presented to customers.
Social Media - Word-of-Mouth Marketing
Using social media, consumers have generated product reviews, comments,
images, and tags as a valuable source of information for other customers. This online
process has increased with the internet platform and had a strong impact on electronic
commerce. Experts call this process “word-of-mouth” communication which has become
a dominating channel that influences the decision making process of online consumers
(Akar & Topçu, 201; Cheung, Lee, & Thadani, 2009). Consumers might share opinions,
interact with others, and exchange shopping experiences using an online discussion board
or forum through any social media platform. An online consumer review might create
product information (Chen & Xie, 2008) and can be viewed as a new form of electronic
word-of-mouth communication (Akar & Topçu, 2011; Cheung et al., 2009). Hennig-
Thurau, Gwinner, Walsh, and Gremler (2004) concluded that electronic word-of-mouth
communication may present “any positive or negative statement made by potential
customers about a product or a company which is available to a multitude of people and
institutions via the Internet” (Hennig-Thurau et al., 2004). Companies are investing a
significant portion of their annual budgets in social media as a word-of-mouth strategy
which is often managed by experts called community managers.
Researchers need to keep in mind that for the next five years, the use of social
media platforms for businesses will be a strong technological tool. Today, 77 percent of
43
adults ages 30-49 use social media, compared with 8 percent in 2005 (Perrin, 2015). A
more up-to-date figure, some 88 percent of 18- to 29-year-olds indicate that they use any
form of social media and that share falls to 78 percent among those ages 30 to 49, to 64
percent among those ages 50 to 64 and to 37 percent among Americans 65 and older
(Smith & Anderson, 2018). For the next five years, the use of social networks among
Americans will increase substantially due to business and marketing products. Although
these data explain social networking behavior in the United States, the use of social
media around the world was not reported. Social media abroad could be an advantage to
incorporate international marketing strategies (see Figure 2).
Figure 2 Percentage of all Americans adults who use social networking sites by age
Note: Pew Research Center surveys, 2005-2006, 2008-2015. No data are available for
2007. Retrieved from http://www.pewinternet.org/files/2015/10/PI_2015-10-08_Social-
Networking-Usage-2005-2015_FINAL.pdf.
44
For example, Facebook and YouTube dominate this landscape, as notable
majorities of U.S. adults use each of these sites (Smith & Anderson, 2018). On average
between 2012 and 2018, 68 percent and 73 percent of online adults were using Facebook
and YouTube respectively like an alternative social media platform to interconnect
among families, friends, news, and business surpassing other platforms (see Figure 3).
Figure 3 Social media use in 2018
Note: Adapted from Pew Research Center surveys, 2018. Retrieved from
http://www.pewinternet.org/2018/03/01/social-media-use-in-2018/.
In general, social media could be an opportunity to implement marketing
strategies to support agribusiness in rural communities. The interconnected network of
45
social media could be the key to communicate, organize, and diffuse agricultural products
in farmers’ communities in the United States and other countries. According to the Pew
Research Center, Facebook is a potentially valuable tool to use for developing marketing
strategy plans using e-marketing models and has been a bridge between consumers and
buyers; it has increased its languages worldwide and online from 70 to 120 since 2012,
structuring a global social network platform (Myers, 2015). Although social media are
becoming ever more ubiquitous and important for marketing purposes, social media are
substantially different from traditional or other online media due to the network structure
and their egalitarian nature (Peters, Chen, Kaplan, Ognibeni, & Pauwels, 2013).
Factors Influencing Social Media Marketing
Attitudes can shape customers’ minds generating positive or negative ideas about
an object and moving customers toward or away from that object. The behaviors of
customers are influenced by attitudes that make them behave in a quite consistent way
toward similar objects (Kotler & Keller, 2014). Experts of marketing have described
attitudes as a person’s enduring favorable or unfavorable evaluation, emotional feeling,
and action tendencies toward some object or idea (Akar & Topçu, 2011; Kotler & Keller,
2006). This is important to the present study because the better attitude a farmer has
toward social media adoption and use, the more likely the farmer may be to use it for
promotion and advertising purposes. For instance, a case study of Bangladeshi farmers’
use of mobile telephone found that the use of mobile telephones might provide rural
communities access to information and enable them to enhance their quality of life
(Bayes, 2001; Dey, 2013). Dey (2013) also reported that farmers used mobile phones to
find out about the sources and prices of fertilizers and to contact agricultural extension
46
workers; grouping social networks. The farmers’ need to communicate with external
systems makes functional changes in their attitude on the use of technological devices
and gathering information that can be helpful to their daily work. In several cases,
perceptions of the social media platforms are important in bridging producers and
customers. For example, from the standpoint of customers, Cha (2009) found that the
more people perceive virtual shopping services on social media sites as useful and easy to
use, the more favorable they feel toward shopping for real items on those social media
platforms (Cha, 2009). User intention factors such as attitudes, perceptions, expectations,
influences, and conditions related to the use and adoption of social media are explained in
the next section. These factors are key variables that were analyzed in the present study.
Theories Related to the Study
Introduction
Theories related to information technology acceptance and implications about
attitudes and behaviors have been tested, evaluated, and modified by researchers for
decades. Models and theories such as the Theory of Reasoned Action - TRA (Fishbein &
Ajzen, 1975), the Theory of Planned Behavior – TPB (Ajzen, 1991), the Technology
Acceptance Model - TAM (Davis, 1986), an extension of the Technology Acceptance
Model – TAM2 (Venkatesh & Davis, 2000), and the Diffusion of Innovation Theory –
DOI (Rogers, 2003) identify both psychological and macro-environmental factors that
influence users’ behavioral intention to accept and use technological applications (Dey,
2013). These models are explained in brief here. Researchers such as Venkatesh et al.
(2003) suggest using the UTAUT model to integrate adoption models, including the
TRA, TAM, TAM2, TPB, and DOI (Dey, 2013; Venkatesh et al., 2003). Therefore,
47
comprehensive understanding (involving both macro and individual factors) would be
helpful to conceptualize the dynamic and reciprocal interplays between technology,
society, and human actors (Dey, 2013).
In brief, these technology adoption models can provide a partial understanding of
technology use which requires both adoption and adaptation to incorporate it into
practices, called appropriation (Dey, 2013; Isaac, Besseyre Des Horts, & Leclercq, 2006).
According to MacKay and Gillespie (1992), people are not always directed by
technological applications as they may redesign, redefine, change, or decline their use
and adoption (Dey, 2013; Mackay & Gillespie, 2018). Technology adoption is influenced
by macro-environmental factors and individuals’ skills and abilities. UTAUT addresses
some of these factors.
Theories Involved in the UTAUT model
Researchers such as Venkatesh et al. (2003) suggested using UTAUT to integrate
adoption models and theories, including the TRA, TPB, TAM, TAM2, and DOI. Each of
these is discussed briefly below.
The Theory of Reasoned Action – TRA
The Theory of Reasoned Action (TRA) was developed by Fishbein and Ajzen
(Fishbein & Ajzen, 1975) and is one of the most popular theories used to determine
behavioral intention of the person’s attitudes toward that behavior. The authors aimed to
develop a theory that might predict, explain, and influence human behavior. Fishbein and
Ajzen (1975) defined “attitude” as the individual’s evaluation of an object and defined
“belief” as a link between an object and some attribute, and defined “behavior” as a result
48
or intention. A person’s subjective norms of what they perceive their immediate
community’s attitude to certain behavior also influences behavior (Lai, 2017). Ajzen and
Fishbein (1980) explained that the TRA is an appropriate model for the study of the
determinants of user behavior as a theoretical foundation since it predicts behavioral
intention across two important domains: attitudes and subjective norms. The greatest
limitation of the theory stems from the assumption that behavior is under volitional
control (Ajzen, 1985).
Theory of Planned Behavior – TPB
The Theory of Planned Behavior (TPB) has a central factor focused on the
individual’s intention to perform a behavior. According to Ajzen (1991), intentions are
assumed to capture the motivational factors that influence a behavior and they are
indications of how hard people are willing to try. The TPB might explain intentions of
farmers to structure marketing strategies to promote agricultural products using
traditional marketing models on social media platforms. These intentions are related to
considerable variances in the actual behavior of a person at the time of purchase. Ajzen
(1991) explains that intentions to perform behaviors might be predicted with high
accuracy from attitudes toward the behavior, subjective norms, and perceived behavioral
control. Attitudes toward the behavior were used as a variable in the present study to
analyze farmers’ behaviors in the use of social media. Ajzen (1991) found that if these
three factors are strong, the intention will be strong, which will lead an individual to
perform the subsequent behavior. For the present study, attitudes refer to the degree to
which a person has a favorable or unfavorable evaluation or appraisal of the behavior
(Ajzen, 1991). But, TPB has some limitations and the theory does not include
49
demographic variables and assumes that the experience is similar for participants in the
process.
Technology Acceptance Model – TAM
The Technology Acceptance Model (TAM) was introduced by Fred Davis in
1986 (Lai, 2017). The goal of Davis’ (1989) TAM is to explain the general determinants
of computer acceptance that lead to explaining users’ behavior across a broad range of
end-user computing technologies and user populations. This model posits two constructs,
perceived usefulness (U) and perceived ease of use (EOU) which mediate all external
variables to influence an individual’s decision to use any information system. According
to Bajaj and Nidumolu (1998), perceived usefulness is defined as the person’s subjective
probability that using any information system will be of benefit in an organizational
context, and perceived ease of use is defined as the degree to which the person thinks that
using information system will be free of effort. Venkatesh and Davis (2000) cited that the
main limitation for TAM is that the model only provides limited guidance about how to
influence usage through design and implementation. In other words, it does not help
understand acceptance in ways beyond the suggestion that system characteristics impact
the ease of use factor.
According to Porter and Donthu (2006), no previous research has included age,
gender, and level of education in models like TPB and TAM. Additionally, these theories
have not examined the use of social media by farmers for marketing their products even
though these variables are relevant in the context of explaining their social media use.
50
Extension of the Technology Acceptance Model – TAM2
Venkatesh and Davis (2000) proposed the TAM2 by adding concepts covering
social influences (subjective norm, voluntariness, and image) and cognitive instrumental
processes (job relevance, output quality, result demonstrability, and perceived ease to
use). An additional factor in TAM2 is experience. The influence of the subjective norm
on intention to use will decrease and be replaced by experience in using an information
system (Venkatesh & Davis, 2000).
Diffusion of Innovation Theory – DOI
Diffusion of Innovation Theory (DOI) is useful to analyze the process by which
an innovative technology is communicated through certain channels and networks
presented in the present study forward in the Extension services conceptual frame.
Rogers (1995) defines DOI as the process “by which an innovation is communicated
through certain channels over time among the members of social system (Lyytinen &
Damsgaard, 2001; Rogers, 1995). The DOI theory explains how innovations diffuse
through society and how individuals accept new innovations (Rogers, 2003). Rogers’
diffusion of innovation theory contains and innovation-decision process, innovation
characteristics, adopter characteristics, and opinion leader (Rogers, 2003). Roger’s model
(2003) in the Innovation Decision Process describes five stages (knowledge, persuasion,
decision, implementation, and confirmation) which an individual or other decision
making unit must go through during the process of adoption or rejecting an innovation.
Limitations were observed focused in adequate constructs to deal with collective
adoption behaviors including the critical role of standard, critical mass, network
51
externalities, sunk costs, and path dependence among others (Lyytinen & Damsgaard,
2001).
Theoretical Background of UTAUT
UTAUT is a combination of several theories and models just described: Theory of
Reasoned Action (TRA), Theory of Planned Behavior (TPB), Technology Acceptance
Model (TAM), Technology Acceptance Model 2 (TAM2), and Diffusion of Innovation
(DOI). Venkatesh et al. (2003) used the previous models and theories to form the
UTAUT model. UTAUT model has proven to be successful in a wide range of
educational situations, and it brings the benefit of being “a useful tool for managers
needing to assess the likelihood of success for new technology introductions and helps
them understand the drivers of acceptance” (Suki & Suki, 2017, p. 529). Venkatesh et al.
(2003) incorporated four key determinants of technology use in UTAUT: PE, EE, SI, and
FC, as well as four key moderators like gender, age, voluntariness, and experience (Lai,
2017; Venkatesh et al., 2003) (see Figure 1). Researchers have applied, integrated, and
extended UTAUT to study individual technology acceptance and use across a variety of
settings, such as different user types, different organization types, different types of
technologies, different tasks, different times, and different locations (Venkatesh et al.,
2016). For the present study, the variety of settings included groups (farmers), industry
sectors (small business farmers), technology (social media platforms), and tasks
(marketing strategies to reach new customers).
In general, research has repeatedly confirmed the robustness of UTAUT and its
main effects; however, research has scarcely examined the moderating effects of age,
gender, and experiences (Venkatesh et al., 2016). Venkatesh et al. (2016) recommended
52
future research evaluate new context effects along the following four dimensions:
environment, location, organization, and event. For these variables and dimensions, the
present study utilized multivariate techniques combining aspects of factors analysis and
multiple regression to see how four key determinants (four predictors = PE, EE, SI, and
FC) and moderator variables of age, gender, and level of education of farmers and
vendors at FMs, CSA groups, roadside stands, and on-farm stores in North Mississippi
related to their intention to use and actual use of social media for marketing their
agricultural products.
UTAUT – Constructs and Formulation
Four predictors of BI or usage in UTAUT include PE, EE, SI, and FC as direct
determinants of BI and USE (Suki & Suki, 2017; Venkatesh et al., 2003).
Performance Expectancy
PE is defined as “the degree to which an individual believes that using the system
will help him or her to attain gains in job performance” (Venkatesh et al., 2003, p. 447).
PE in UTAUT is derived from a combination of five similar constructs from various
models, including perceived usefulness, extrinsic motivation, job-fit, relative advantage,
and outcome expectation. PE is the strongest predictor of intention and was found
significant at all points for both voluntary and mandatory use settings in Venkatesh et
al.’s (2003) model validation. For the present study, the research question was how PE
affects BI of farmers and vendors to use social media for marketing their small farm
businesses and how gender, age, and level of education moderate this relationship.
According to farmers’ beliefs about using and adopting social media for marketing
53
purposes, farmers could promote agriproducts to attain gains in business performance.
From a theoretical point of view, there is reason to expect that the relationship between
PE and intention will be moderated by gender and age (Venkatesh et al., 2003). Gender
and age differences have been shown to exist in technology adoption contexts (Venkatesh
& Morris, 2000).
Effort Expectancy
EE is defined as “the degree of ease associated with the use of the system”
(Venkatesh et al., 2003, p. 450). EE appropriates the notions of perceived ease of use and
complexity. For the present study, the research question was how EE affects BI of
farmers and vendors to use social media for marketing their small farm businesses.
Farmers could consider social media as a means to use and adopt marketing solutions to
promote their agriproducts or as a barrier due to the complexity of the platforms. Prior
research supports the notion that constructs related to EE are stronger determinants of
individuals’ intention for women and for older workers (Venkatesh & Morris, 2000).
Social Influence
SI is defined as “the degree to which an individual perceives that important others
believe he or she should use the new system” (Venkatesh et al., 2003, p. 451). According
to Venkatesh et al. (2003), SI includes consideration of the person’s perception of the
opinion of others, his or her reference group’s subjective culture, specific interpersonal
agreements with others, and a degree to which use of an innovation is perceived to
enhance one’s image or status in one’s social system. For the present study, the research
question was how SI affects BI of farmers and vendors to use social media for marketing
54
their small farm businesses and how gender, age, and experience of use moderate the
connections from SI to BI of farmers. SI appears to be important only in the early stages
of an individual’s experience with the technology (Venkatesh & Davis, 2000). Gender
effects may be driven by socially-constructed gender roles (Lubinski, Tellegen, &
Butcher, 1983; Venkatesh et al., 2003). Older workers are more likely to place increased
salience on social influences, with the effect declining with experience (Venkatesh &
Morris, 2000). Venkatesh et al. (2003) expected a complex interaction with these
moderating variables simultaneously influencing the social influence-intention
relationship. According to Venkatesh et al. (2003), SI was found to be a complex factor
and was not significant in the voluntary use context, but it became important when its use
was mandated. However, there are not studies about how SI affects the use and adoption
of technology as a marketing strategy to reach new customers for business, which is
essentially voluntary use in nature, but there are arguments that social media is essential
for marketing, and thus, in a sense, could be considered “mandatory” for a business to
survive and thrive.
Facilitating Conditions
As noted by Venkatesh et al. (2003), “the degree to which an individual believes
that organizational and technical infrastructure exist to support use of the system” (p.
453) is an indication of the presence or otherwise of FC. FC represent organizational
support and included perceived behavioral control, facilitating conditions, and
compatibility. Empirical results indicated that FC do have a direct influence on usage
beyond that explained by BI alone (Venkatesh et al., 2003). For the present study, the
research question was how FC affect BI of farmers and vendors to use social media for
55
marketing their small farm businesses and how gender, age, and experience of use
moderate the connections from FC to BI of farmers to use social media. Farmers could
incorporate social media as a marketing strategy in their business, keeping in mind that
the infrastructure, access, costs, internet provider, and connectivity are supported.
However, these characteristics of connectivity are implicit in the factor, and it is
understood that farmers use connectivity platforms via smartphones, personal computers,
and/or tablets. According to Venkatesh et al. (2003), FC have a significant influence on
usage behavior and is moderated by experience and age.
Finally, the theory holds that four key constructs (PE, EE, SI, and FC) are direct
determinants of usage intention and behavior (Venkatesh et al., 2003). Gender, age, and
experience are posited to moderate the impact of the four key constructs on usage
intention and behavior as indicated in Figure 2. Although age, gender, experience, and
voluntariness of use are moderating variables in the original UTAUT (Venkatesh et al.,
2003), these moderators are treated as external constructs in some studies (Dwivedi,
Rana, Chen, & Williams, 2011). For instance, UTAUT has been used to examine the
context of e-government adoption in a developing country (Gupta, Dasgupta, & Gupta,
2008; Venkatesh et al., 2016). Results indicated that performance and EE, SI, and FC all
positively impact the use of the information and communication technologies (ICT), but
they did not find a significant moderating effect of gender in these relationships (Gupta et
al., 2008). Age has received very little attention in the technology acceptance research
literature (Venkatesh et al., 2003). Voluntariness of use is the degree to which use of the
innovation is perceived as being voluntary or through one’s free will (Tan, 2013). The
56
complex nature of the interactions observed, particularly for gender and age, raises
several issues to investigate in future research (Venkatesh et al., 2003).
In the present study, UTAUT applied to farmers and vendors in North Mississippi
was used to examine the relationships between these factors (PE, EE, SI, and FC)
moderated by age, gender, and experience as they influence farmers’ BI and actual use of
social media for marketing their businesses and agriproducts (see Figure 4).
Figure 4 UTAUT for farmers and vendors in FMs and CSAs in Mississippi
Note: Figure extracted from Venkatesh et al. (2003): “User acceptance of information
technology: Toward a unified view.” Elaborated by the author.
UTAUT Tested on Social Media
New information technologies focused on connectivity and structures, or social
networks have influenced human activities especially in the business field related to
marketing. Social media is adopted by small businesses primarily to promote the brand
and connect to new markets and customers. Today, 65 percent of U.S. adults now use
social networking sites – a nearly tenfold jump in the past decade (Perrin, 2015).
57
According to Mandal and Mcqueen (2012), social media applications are simple,
generally free, web based, and dependent on content generated by the user. Network
groups specialize in subjects or share common interests, and therefore have substantial
influence on buying decisions of a consumer (Lee, 2011). Another social media use is
dissemination of information along with advertising and promotion, which can be
broadcast by the business to a select target group of users who have opted to receive this
information (Mandal & Mcqueen, 2012; Mangold & Faulds, 2009; Sepp, Liljander, &
Gummerus, 2011). Social media has been used to disseminate relevant information from
businesses for advertising and branding of its products and services around the world. In
other words, the use of social media has revolutionized small businesses strategies
concentrated on marketing and diffusion.
One of the most rapid growing digital communication mediums at this time is
social media. The primary question that arises from the present study is how farmers can
take advantage of marketing using social media for promoting their agribusinesses. In
reference to social media, limited investigation has been performed on its adoption by
businesses based on UTAUT (Günther, Krasnova, Riehle, & Schöndienst, 2009; Mandal
& Mcqueen, 2012; Meyer & Dibbern, 2010; Schöndienst, Krasnova, Günther, & Riehle,
2011). Literature identifies the scarce resources of small businesses as the primary
problem blocking the use of networking and sharing to undertake business processes
(Qureshil, Kamal, & Wolcott, 2009).
In a case study, UTAUT and technology diffusion theory have been used to
understand social media adoption in a business environment (Günther et al., 2009;
Mandal & Mcqueen, 2012). As a conclusion, small business owners are individuals
58
whose personality may play an important role in adoption of social media (Mandal &
Mcqueen, 2012). Another study of social media use by IBM employees identified that
caring, climbing, and campaigning are the biggest motivators of social media use
(Dimicco et al., 2008; Mandal & Mcqueen, 2012). Unfortunately, existing studies about
organizational adoption of internet technologies (including social media) have mostly
been conducted among large companies and organizations (Li, Troutt, Brandyberry, &
Wang, 2011).
The technology adoption literature on small businesses identified relative
advantage as the main factor in post technology adoption (Li et al., 2011). The factors
that affect small businesses in social media use and adoption remain largely unexplored,
especially variables such as age, gender, and experience. Identifying these factors is
critically important to both business decision makers, particularly if they intend to
stimulate the adoption and continued use of social media as a sales channel among small
businesses (Li et al., 2011). The challenges faced by small business make it even more
difficult for them to adopt information and communication technologies (ICT) for
competitiveness (Qureshil et al., 2009).
The present study explored relationships between the adoption and use of social
media and factors that affect the acceptance of these technologies in small farm
businesses. The UTAUT model identifies key factors in the acceptance of information
and communication technologies as measured by BI to use the technology and actual
usage (Oye, A.Iahad, & Ab.Rahim, 2014). This model is a popular technology adoption
theory that explains almost 70 percent of variance in adoption behavior (Venkatesh et al.,
2003, p. 467). According to Venkatesh et al. (2003), since the model explains quite high
59
variance for behavioral research, further work should attempt to identify and test
additional boundary conditions of the model to provide an even richer understanding of
technology adoption and use behavior. These conditions are related to incorporating and
evaluating different technologies, different user groups, and different organizational
contexts to identify benefits of enhancing the UTAUT model and/or extending the
existing work. Therefore, the present study examined the relationship between attitudes,
perceptions, and knowledge of farmers and vendors and the adoption and use of social
media for marketing their businesses and how those relationships are affected by age,
gender, and level of education.
60
CHAPTER III
METHODS
Introduction
Through an application of the UTAUT model, the purpose of the present study is
to assess farmers and vendors’ knowledge, perceptions, and attitudes related to the use of
social media for marketing their agriproducts. The research method of the present study
was presented to the Institution Review Board (IRB) at Mississippi State University and
its protocol was granted an exemption determination (see Appendix A). The present
chapter describes the research design and methodology, population and sample, sampling
method, variables and measurements, survey instrument, data collection and procedures,
and data analyses.
Research Design and Methodology
The present study implemented a quantitative research methodology based on the
UTAUT model (Venkatesh et al., 2003). The present study followed the original model
(with two differences described later), measurements, and analyses (Venkatesh et al.,
2003) which included correlations, factors analysis, and structural equations modeling
(SEM).
61
The present study is classified into a UTAUT application category due to being an
empirical study that applied at least part of UTAUT as a research model (Venkatesh et
al., 2016). The independent variables in the present study included:
Performance Expectancy (PE) - the degree to which individuals believe that
using system will help them improve their job performance.
Effort Expectancy (EE) - the degree of ease associated with the use of the
system.
Social Influence (SI) – the degree to which peers influence the use of the
system, whether positive or negative.
Facilitating Conditions (FC) - the degree to which an individual believes that
an organizational and technical infrastructure exists to support the system.
The dependent variables in the present study included:
Behavioral Intention (BI) – a person’s subjective possibility that he or she will
perform the behavior in question.
Use Behavior of Social Media (USE) - the actual use (USE) of specific system
(Ong, Day, Chen, & Hsu, 2008). According to Ajzen and Fishbein (1980), the
actual USE is dominated by BI. In UTAUT, the direct influence of behavior
intention on USE was tested and validated during the development of UTAUT
(Venkatesh et al., 2003).
The moderator variables in the present study included:
Age - Age is theorized to play a moderating role. Research on job-related
attitudes (e.g., Hall & Mansfield, 1975; Porter, 1963; Venkatesh et al., 2003)
62
suggests that younger workers may place more importance on extrinsic
rewards.
Gender - Gender schema theory suggests that such differences stem from
gender roles and socialization processes reinforced from birth rather than
biological gender per se (Bem, 1981; Bem & Allen, 1974; Kirchmeyer &
Bullin, 1997; Lubinski et al., 1983; Lynott & McCandless, 2000; Motowidlo,
1982; Venkatesh et al., 2003).
Experience – Experience using social media for business purposes.
The model for this study is presented in Figure 5.
Figure 5 Research model tested based on UTAUT
Note: Figure extracted from Venkatesh et al. (2003): “User acceptance of information
technology: Toward a unified view.” Elaborated by the author.
The model tested in the present study is structured following UTAUT presented
by Venkatesh et al. (2003) with some modifications. First, the independent variables
63
(main constructs) are similar to those in UTAUT with the addition of an FC question
referring to internet connectivity, access, or cost as limitations in farmers’ ability to
promote their business using social media. According to Gallardo (2016), broadband
technology is the backbone of the digital age, and this connectivity is considered the new
essential utility; however, a digital divide, measured by internet access, exists between
urban and rural areas placing rural FMs at a potential disadvantage. Thus, in rural areas,
connectivity and access are major concerns related to Internet use.
Second, in the present study, voluntariness of use as a potential moderator was
omitted. Voluntariness refers to the extent to which potential adopters perceive the
adoption decision to be non-mandated (Agarwal & Prasad, 1997). Because small farm
businesses are not part of a larger organization where use of technology (i.e., social
media) could be mandated, the voluntariness variable is not relevant.
Research Hypotheses
The following hypotheses tested are based on the UTAUT model (Venkatesh et
al., 2003) and are displayed in Figure 5:
H1: PE has a positive effect on farmers and vendors’ BI to use social media for
marketing their small farm businesses.
H2: EE has a positive effect on farmers and vendors’ BI to use social media for
marketing their small farm businesses.
H3: SI has a positive effect on farmers and vendors’ BI to use social media for
marketing their small farm businesses.
H4: FC have a positive effect on farmers and vendors’ social media USE for
marketing their small farm businesses.
64
H5a-c: Gender has a moderating effect on the relationships between PE, EE, and
SI and farmers and vendors’ BI to use social media for marketing their small farm
businesses, with females being more likely to intend to use social media.
H6a-d: Age has a moderating effect on the relationships between PE, EE, and SI
on farmers and vendors’ BI to use social media for marketing their small farm
businesses, as well as on the relationship between FCs and farmers and vendors’
social media USE for marketing their small farm businesses, with younger
farmers being more likely to intend to use or actually use social media.
H7a-c: Experience in the use of social media has a moderating effect on the
relationships between EE and SI on farmers and vendors’ BI to use social media
for marketing their small farm businesses, as well as on the relationships between
FCs and farmers and vendors’ social media USE for marketing their small farm
businesses, with those who have had a positive experience being more likely to
intend to use or actually use social media.
H8: BI has a positive effect on farmers and vendors’ social media USE for
marketing their small farm businesses.
Population and Sample
The population in the present study includes farmers’ communities in North
Mississippi who sell agricultural products through marketing channels where transactions
are conducted directly between farmers and consumers DTC which includes venues such
as FMs, CSAs, farm stands/on farm sales, and “pick your own” operations (Martinez et
al., 2010).
65
A sample in a research study is the group on which information is obtained
(Fraenkel, Wallen, & Hyun, 2014). The sample selected for the present study is farmers’
communities in North Mississippi, specifically FMs, CSAs, roadside stand, and on-farm
sales basically.
Sampling Method
Sampling refers to the process of selecting the individuals who will participate in
a research study (Fraenkel et al., 2014). The sampling method used for the present study
of farmers’ communities in North Mississippi is a convenience sample method (Fraenkel
et al., 2014). This method was selected due to the locations of the FMs (North Mississippi
rural areas), distances, transportation costs, and visit approval by the managers.
The sample size in factor analysis is an important issue in terms of the theory and
the statistical model applied. While different opinions about minimum sample size
requirements exist (Maccallum, Widaman, Preacher, & Hong, 2001), Fraenkel et al.
(2014) recommend a minimum sample size of 150 for a model containing seven or fewer
constructs, with each construct having more than three items.
Method for Locating Participants
Potential participants were located geographically using three sources from U.S.
Government and State Government databases on the Internet:
United States Department of Agriculture (USDA), Local Food Directories:
National Farmers Market Directory (https://www.ams.usda.gov/local-food-
directories/farmersmarkets).
66
Government of Mississippi, Farmers’ Markets Directory
(http://www.ms.gov/content/Pages/FarmersMarkets.aspx).
Mississippi Department of Agriculture & Commerce, FMs in Mississippi
(https://www.mdac.ms.gov/bureaus-departments/farmers-market/markets-
mississippi/).
From information obtained at these websites, 37 FMs (see Table 2) and 8 CSAs
(see Table 3) were identified as potential locations for conducting the study in North
Mississippi. After these locations were identified, the researcher contacted each manager
to obtain permission to conduct the study at that specific setting.
67
Table 2 FMs in north Mississippi
#
FMID
Market Name
City
County
1
1003348
Green Market
Corinth
Alcorn
2
1003998
Corinth Farmers Market
Corinth
Alcorn
3
1006670
Cleveland Farmers Market
Cleveland
Bolivar
4
1006671
Cleveland Farmers Market
Cleveland
Bolivar
5
1003599
Calhoun City Farmers Market
Calhoun City
Calhoun
6
1001146
Vaiden Farmers Market
Vaiden
Carroll
7
1009755
Choctaw County FM
Ackerman
Choctaw
8
1003679
West Point Farmers Market
West Point
Clay
9
1006285
Hernando Farmers Market
Hernando
DeSoto
10
1007251
Southaven Farmers Market
Southaven
DeSoto
11
1000545
CJ's Vegetable Market
Grenada
Grenada
12
1009513
Grenada, MS Farmers Market
Grenada
Grenada
13
1010827
The Grenada Farmers' Market
Grenada
Grenada
14
1004698
Volume Vegetable FM
Fulton
Itawamba
15
1004854
Itawamba Farmers Market
Fulton
Itawamba
16
1003637
Tupelo Farmers Depot
Tupelo
Lee
17
1000407
Hitching Lot Farmers Market
Columbus
Lowndes
18
1000201
Holly Springs Farmers' Market
Holly Springs
Marshall
19
1006153
Downtown Amory FM
Amory
Monroe
20
1000155
Starkville Community Market
Starkville
Oktibbeha
21
1003786
Henry's Farmers Market
Starkville
Oktibbeha
22
1003588
Square Market
Batesville
Panola
23
1001230
Prentiss County Farmers Market
Booneville
Prentiss
24
1011747
Delta Regional Mule Train
Marks
Quitman
25
1006302
Indianola Open Air Market
Indianola
Sunflower
26
1006280
Tate County Farmers Market
Senatobia
Tate
27
1001106
Main Street Farmers Market
Ripley
Tippah
28
1009334
Bear Creek Farmers Market
Belmont
Tishomingo
29
1009754
Tishomingo County FM
Iuka
Tishomingo
30
1011692
Biscuits and Jam FM
New Albany
Union
31
1001099
Webster County FM
Eupora
Webster
32
1003326
Water Valley main Street FM
Water Valley
Yalobusha
33
1000021
Mid-Town Farmers Market
Oxford
Lafayette
34
1009922
Oxford City Market
Oxford
Lafayette
36
1003229
Noxubee County FM
Macon
Noxubee
37
N/REG
Greenwood FM
Greenwood
Leflore
Note: Information obtained from government agency websites (United States Department
of Agriculture - USDA, Government of Mississippi, Farmers’ Markets Directory, and
Mississippi Department of Agriculture & Commerce, FMs in Mississippi).
68
Table 3 CSAs in north Mississippi
#
City
County
Name
Address
Operates
Contact
1
Starkville
Oktibbeha
Bountiful
Harvest Farms
1420 Pat Station
rd.
April-
November
Sam and
Isabel
McLemore
2
West Point
Clay
Double D Farms
No address
May-
August
Dallas
O'Bryant
3
Columbus
Lowndes
Merry Heart
Farm
7924 Hwy 50 E
Columbus
January
through
January
Amanda
Fondren
4
Louisville
Winston
Winston County
Self Help
Cooperative
5226 Old West
Point
May-
August
Orlando
Trainer
5
Ashland
Benton
Tubby Creek
Farm
1394 Gray
Academy Rd
April -
September
Randy
Alexander
6
Plantersville
Lee
City Point
Produce
3432 Highway 6
April -
August
Candace Tate
7
Mt Vernon
Lee
Isis Gardens
955 Mt. Vernon Rd
April-
December
Genevieve
Yeakel
8
Belden
Lee
Poultry Peeps
Farm
1196 Chesterville
Rd
Year
around
Benji &
Robin
Boydstun
Note: Information obtained from government agency websites (United States Department
of Agriculture - USDA, Government of Mississippi, Farmers’ Markets Directory, and
Mississippi Department of Agriculture & Commerce, FMs in Mississippi).
Participants
The final sample for the present study included FMs and CSAs in the following
North Mississippi counties: Oktibbeha, Marshall, Lee, Lowndes, Lafayette, Choctaw,
Clay, Monroe, Yalobusha, Noxubee, Union, Leflore, Grenada, and Desoto. Out of the 37
FMs and 8 CSAs contacted, 16 responded to the initial contact, and 14 agreed to allow
their vendors to participate in the study. A total of 67 zip codes were identified by survey
respondents. The zip code was used to identify and map rural areas and towns’
population who are using FMs as a marketing channel (see Appendix B). Overall, an
average of 12 farmers and vendors per FM (14 FMs visited) were selling agricultural
products in the summer season 2018.
69
Figure 6 shows the 14 FMs and on-farm stores that participated in the present
study (CSA members included in the study sold their products at a FM). FMs were
visited during summer season in open hours mainly on Tuesdays, Thursdays, and
Saturdays.
Figure 6 Map of participating FMs
Note: Elaborated by the author using Arc GIS software.
The survey was completed by 169 farmers and vendors at FMs, CSAs, and on-
farm stores in northern Mississippi. Table 4 displays the number of respondents at each
FM.
70
Table 4 FMs visited and number of respondents
Farmers Market - Town
Frequency
(%)
Starkville FM
11
6.5%
On-farm stores
3
0.6%
Holly Springs FM
2
1.2%
Columbus FM
20
11.8%
Tupelo FM
11
0.6%
Oxford Mid-Town FM
15
6.5%
Louisville FM
9
8.9%
West Point FM
8
0.6%
Aberdeen FM
2
5.3%
Water Valley FM
5
4.7%
Oxford FM
19
1.2%
New Albany FM
13
3.0%
Greenwood FM
17
11.2%
Grenada FM
7
7.7%
Hernando FM
27
10.1%
Total
169
100.0%
Data Collection and Procedures
The present study used a questionnaire to collect data from farmers and vendors at
FMs, farmers of CSAs, and farmers with on-farm stores. Data collection included the
following procedures.
The researcher visited each establishment and talked individually with each
farmer and vendor about the purpose of the study. The researcher explained to the farmer
the sequence and the time needed (approximately 8 minutes) to complete the
questionnaire. This process ensured that farmers and vendors could receive assistance
when completing the questionnaire to reduce the likelihood of skipping questions due to
misunderstanding, thus reducing the amount of missing data. Once the study was
explained, each farmer was asked to complete the questionnaire. Finally, the researcher
71
asked the farmer for questions, doubts, suggestions, and comments related to the broader
study and the questionnaire.
Visits to FMs occurred from May through August 2018 during open hours of the
FMs. Most of these FMs were open to the public Tuesday, Thursday, and Saturday of
every week during the spring and summer seasons. The best time to request farmers’
participation was prior to the market opening. Because it is possible that the same farmer
or vendor sold at multiple FMs, the researcher documented the name of the participating
farmer or vendor to ensure multiple surveys were not collected from the same individual.
Description of Questionnaire
For the present study, the questionnaire contained 36 questions distributed across
three general domains: demographic information, UTAUT constructs, and marketing
channel preferences (see Appendix C – Measurable Concepts from the Research
Questions and Appendix D – Study Questionnaire).
Demographic Questions
Four questions were used to collect demographic information: zip code, gender
(Male/Female), age (categories), and level of education (categories). One question
assessed general computer experience using a five-point Likert-type scale ranging from 1
to 5 with response options of Very Poor, Poor, Moderate, Good, and Very Good.
UTAUT Questions
Questions related to UTAUT were based on questionnaires used by Venkatesh et
al. (2003) in studies of the model. Participants are asked to indicate the extent to which
they either agree or disagree with a series of statements about the use of social media for
72
business purposes. Each series is grouped by the different constructs in UTAUT as
described below.
PE measured farmers’ perceptions of using social media for marketing purposes
in terms of how useful it was in promoting their agricultural business or products,
enabling them to complete tasks related to promoting their business or products more
quickly, increasing their productivity to promote their business or products, and
increasing their chances of getting more sales in business. EE measured farmers’
perceptions of the ease-of-use of social media for marketing purposes related to their
interaction with social media platforms being clear and understandable, being easy to
become skillful at using social media for their business, finding social media easy to use
for their business, and being easy to learn to operate social media for their business. SI
measured farmers’ perception of how people who are important to them think about
farmers’ use of social media to promote their business, how people who influence
farmers’ behavior think farmers should use social media to promote their business, how
people whose opinions farmers value think that farmers should use social media to
promote their business, and how other farmers and producers who use social media to
promote their businesses affect farmers’ decision to use social media in their business. FC
measured farmers’ perception of being able to access required resources necessary to use
social media to promote farmers’ business, having the knowledge necessary to use social
media to promote farmers’ business, social media’s incompatibility with other ways that
farmers promote their business, knowing who can assist if farmers have technical
problems when using social media to promote their business, and how internet
connectivity, access, or cost limit farmers’ ability to promote their business using social
73
media. BI measured farmers’ perception of their plan to use social media to promote their
business or products in the next 12 months, their prediction that they will use social
media to promote their business or products in the next 12 months, and their intention to
use social media in the next 12 months. Each of the items within these constructs used
seven-point Likert-type scales with response options of Strongly Disagree, Disagree,
More or Less Disagree, Undecided, More or Less Agree, Agree, and Strongly Agree),
where 1 being the negative end of the scale and 7 being the positive end of the scale
(Venkatesh et al. 2003, p. 438).
USE was assessed by eight questions. Four questions focused on use of social
media for personal purposes. (Yes/No initial response). If the participant answered No,
he/she was asked to skip to the next section of questions. However, if the participant
indicated Yes, he/she used social media for personal purposes, he/she was then asked to
“mark all that apply” to indicate which social media platforms (e.g., Facebook, Twitter,
Instagram, Pinterest, Snapchat, YouTube, LinkedIn, Wikis, Blogs, and Other) were used
for personal purposes. The next question asked participants to describe their social media
experience for personal purposes using a five-point Likert-type scale ranging from 1 to 5
with response options of Very Poor, Poor, Moderate, Good, and Very Good. Two
questions were also included about how long and how often farmers use social media for
personal purposes. Participants indicated how long they had been using social media for
personal purposes by choosing from four responses coded from 1 to 4, respectively: less
than 1 year, 1-2 years, 3-4 years, or more than 4 years. Participants also indicated how
often they used social media per day for personal purposes by choosing from four
74
responses coded from 1 to 4, respectively: less than 1 hour, 1-2 hours, 3-4 hours, more
than 4 hours.
Four questions assessed use of social media for business purposes. The questions
and response options were identical to those asked in the questions related to use of social
media for personal purposes, with the phrase “for personal purposes” in each question
being replaced by the phrase “for business purposes.”
In preliminary tests of UTAUT (Venkatesh et al., 2003), internal consistency
reliability for each of the model components was high. Based on three datasets, internal
consistency reliability ranged from .91 to .92 for PE, .90 to .94 for EE, .88 to .94 for SI,
and .83 to .87 for FC. According to Venkatesh et al. (2003), in the preliminary test of the
UTUAT model, the square roots of the shared variance between the constructs and their
measures were higher than the correlations across constructs, supporting convergent and
discriminant validity.
For the present study, reliability and validity were determined in various ways. A
scale reliability analysis was performed to assess the internal consistency and item-total
correlations to demonstrate consistency and accuracy of the model constructs. Internal
consistency reliability is a frequently used type of reliability in the information system
domain and it refers to the degree to which responses are consistent across the items
(variables) within a single measurement scale (Kline, 2011; Sekaran, 2003). Items-total
correlation refers to the correlation of a variable, with the composite score of all variables
forming the measure of the construct (Lu, Lai, & Cheng, 2007). Finally, Confirmatory
Factor Analysis (CFA) was conducted to assess and examine the convergent and
discriminant validity. The results of these analyses are presented in Chapter 4.
75
Marketing Channel Preference Questions
The last group of questions assessed marketing channel preferences and interest in
participating in an MSU Extension program about marketing agricultural products using
social media. One question was related to marketing channel preferences of farmers for
selling their agricultural products by marking all that apply in a list that includes farmers
market, retailer, CSA, roadside stand, website, social media, and other. The next question
asked farmers to mark all of the reasons that would prevent them from selling their
products through those different marketing channels. Reasons included expensive fees,
low flow of customers, high logistic costs. Too many food regulations, limited hours of
customer service, and other.
The final question asked farmer to indicate their interest (Yes/No) in participating
in an MSU Extension program about marketing agricultural products using social media.
If farmers answered Yes, they were asked to provide an email address for receiving
information on such programs in the future.
Data Analysis
The database was compiled and organized using Excel software. SPSS statistical
software was the statistical tool used to analyze the database. The Statistical Package for
the Social Sciences - SPPS® (Version 24.0) software and its module, Analysis of a
Moment Structures - AMOS® (Version 22.0), were used to analyze the data. SPSS
AMOS was the most suitable tool for analyzing the data collected for the present study
due to its ability to model latent variables for data screening and data analysis. Analyses
included an Exploratory Factor Analysis (EFA) to identify the underlying data structure
for each construct, a CFA for testing how well measured variables represent a smaller
76
number of constructs (Hair et al., 2009), and Structural Equations Modeling (SEM) to
examine a series of interrelated dependence relationships among the measured variables
and latent constructs (variates) as well as between several latent constructs (Hair et al.,
2009).
Factor analysis was used to define the underlying structure among the variables in
the analysis (Hair et al., 2009). Factor analysis allows researcher to determine if many
variables can be described by a few factors (Fraenkel et al., 2014). This analysis provides
the tools for analyzing the structure of the interrelationships (correlations) among a large
number of variables (questionnaire responses) by defining set of variables that are highly
interrelated (Hair et al., 2009). These variables are known as a factors or constructs; in
UTAUT, they include PE, EE, SI, FC, BI, and USE.
Two exploratory factor analysis procedures, EFA and CFA, were used to identify
the underlying data structure for each construct. The EFA analysis provided information
about how many factors were needed to best represent the data (Hair et al., 2009). EFA
was used to examine a single relationship at a time and explore the construct validity of
the questionnaire subscales (Hair et al., 2009). CFA tested how well the measured
variables represented a smaller number of constructs (Hair et al., 2009). CFA was used to
assess the multidimensionality and the factorial validity of the constructs of the model
(Byrne, 2009). The CFA analysis allowed the variables to be analyzed as a cluster in
UTAUT. The CFA statistics explained how well the theoretical specification of the
factors matched reality or the actual data. According to Hair et al. (2009), CFA is used to
provide a confirmatory test of the measurement theory. In the present study, the
measurement theory is UTAUT.
77
Finally, the SEM analysis examined a series of interrelated dependence
relationships among the measured variables and latent constructs (variates) as well as
between several latent constructs (Hair et al., 2009). According to Tabachnick and Fidell
(2007), SEM combines factor analysis, canonical correlation, and multiple regression.
The technique evaluates whether a model provides a reasonable fit to the data and how
each of the IVs contributes to explaining the DVs. In other words, SEM is a collection of
statistical techniques that allows a set of relationships between one or more IVs and one
or more DVs to be examined (either continuous or discrete variables). The term SEM
conveys two important aspects of the procedures: (a) that the casual processes under
study are represented by a series of structural equations and (b) that these structural
relations can be modeled pictorially to enable a clear conceptualization of the theory
under study (Byrne, 2010). Gefen, Straub, and Boudreau (2000) recommend the use of
SEM in behavioral sciences research and for topics such as IT/IS research.
Several aspects of SEM offer significant benefits over other multivariate
techniques. According to Byrne (2010), SEM set it apart from older generations of
multivariate procedures in four characteristics:
(1) SEM takes a confirmatory rather than an exploratory approach;
(2) SEM provides explicit estimates of errors and variances parameters;
(3) SEM procedures can incorporate both unobserved and observed variables; and
(4) SEM has become a popular methodology for non-experimental research.
Researchers are often interested in studying theoretical constructs that cannot be
observed directly in the behavioral science. SEM is designed to maximize, then test, the
78
degree of consistency between the theoretical model, and actual data (Kline, 2011). For
the present study, SEM was employed using the AMOS 24.0 package version.
In the present study, SEM examined constructs in UTAUT (PE, EE, SI, FC, and
BI). In the moderator analysis (testing influence of age, gender, and experience on the
constructs), a multi-group analysis method was used. The data were split into relevant
groups for analysis. The gender moderator was divided between males and females, the
age moderator was divided into young and old categories, and the social media
experience moderator was divided between high and low experience. The experience
moderator was analyzed for both personal and business purposes for social media use.
79
CHAPTER IV
RESULTS
Introduction
In this chapter, findings are presented. Three important processes are described:
examination of the data, descriptive statistics analysis of the data, and testing of the
UTAUT model.
The pre-analysis data screening was conducted on the raw data before starting the
multivariate data analysis. The major advantage of descriptive statistics is that they
permit researchers to describe the information contained in many scores with just a few
indices (Fraenkel et al., 2014). Structural Equations Modeling assessed how well UTAUT
fits reality as represented by data (Hair et al., 2009). Therefore, Hair et al. (2009),
described six stages SEM process: (1) defining individual constructs, (2) developing the
overall measurement model, (3) designing a study to produce empirical results, (4)
assessing the measurement model validity, (5) specifying the structural model, and (6)
assessing structural model validity. At the end, a measurement theory specifies how
measured variables logically and systematically represent constructs involved in UTAUT
(Hair et al., 2009). While a descriptive statistics analysis was used to avoid incorrect
findings and results (Field, 2009), a multivariate data analysis (CFA and SEM) is used to
provide a confirmatory test of the measurement theory (Hair et al., 2009).
80
Data Examination
A data examination was conducted on the raw data before reporting the
descriptive statistical analyses and conducting the multivariate analysis. In this process,
the main goal was how to identify and overcome pitfalls resulting from the research
design and data collection practices. According to Hair et al. (2009), missing data,
outliers, and the statistical characteristics of the data are the main analyses to conduct
before applying any multivariate techniques to (1) attain a basic understanding of the data
and relationships between variables and (2) ensure that the data underlying the analysis
meet all of the requirements for a multivariate analysis (Hair et al., 2009).
Missing Data
According to Hair et al. (2009, p. 36), the concept of missing data refers to
information not available for a subject (or case) about whom other information is
available, and it occurs when a respondent fails to answer one or more questions in a
survey. Missing data is one of the common barriers in data analysis within social research
(Kline, 2011). According to The Measurable Concepts from the Research Questions
(Appendix C), the missing data analysis process for the present study divided data into
the three domains: demographic information, UTAUT constructs, and marketing channel
preference and Extension program questions. The present study applied a Missing Value
Analysis using SPSS software.
The first domain examined was the demographic data. Results showed no missing
data related to demographic information. These results were presented successfully due to
the face-to-face previous brief interview to the farmers explaining them the importance of
the research goals in terms to evaluate the perceptions of use social media for business
81
purposes. Missing data analysis results for the demographic information domain is
presented in Table 5.
Table 5 Missing demographic data
Zip Code
Gender
Age
Education Level
Valid
169
169
169
169
Missing
0
0
0
0
For the second domain, surveys with any missing data related to UTAUT model
(constructs or variables) were discarded. According to Arbuckle (1995), missing data will
cause several problem to the Goodness-of-Fit-Index (GIF) in Structural Equations
Modeling using AMOS.
As mentioned, the researcher was present when respondents completed the
questionnaire to help reduce misunderstanding of questions and thus missing data.
Therefore, all data were complete; there were no missing data.
Finally, the third domain examined questions related to prefer marketing channels
and Extension programs. Question number 34 and 35 were evaluated about marketing
channels preferences for selling agriproducts and question number 36 about the farmers
and vendors’ interest to be part of an Extension program. All data was complete; there
was no missing data. On the other hand, question number 35 was evaluated carefully
since some of vendors did not want to answer to avoid conflicts with the FMs’ managers
or coordinators. Hair et al. (2009) describe this as censored data.
Univariate Normality – Skewness and Kurtosis
According to Hair et al. (2009), the most fundamental assumption in multivariate
analysis is normality. It refers to the shape of the data distribution for an individual metric
82
variable and its correspondence to the normal distribution (Hair et al., 2009). Multivariate
normality is more difficult to test (Gnanadeskian, 1997), but the statistical techniques for
testing univariate normality are Pearson’s skewness parameter and graphical analysis of
normality (Hair et al., 2009). In the present study, visual examination of the database
distributed in a histogram was mainly used to test the univariate normality. Thus, visual
assessment of the histogram with the data distributions for all constructs indicated that
the shapes of all the univariate distributions were reasonably usual and acceptable. Table
6 indicates all values of the constructs were within the accepted range of skewness and
kurtosis (i.e. -2.58 → +2.58) (Hair et al., 2009, p. 73). If the z value exceeds the specific
critical value, then the distribution is considered non-normal in terms of that
characteristic.
83
Table 6 Skewness and kurtosis statistics for the study variables (n = 169)
Scale
PE
Skewness
Kurtosis
* Q#14_PE_1
-1.340
1.671
* Q#15_PE_2
-1.101
1.000
* Q#16_PE_3
-0.955
0.609
* Q#17_PE_4
-1.121
0.991
EE
Skewness
Kurtosis
* Q#18_EE_1
-1.027
1.335
* Q#19_EE_2
-0.822
0.326
* Q#20_EE_3
-0.773
0.127
* Q#21_EE_4
-1.025
0.756
SI
Skewness
Kurtosis
* Q#22_SI_1
-0.803
0.081
* Q#23_SI_2
-0.758
0.221
* Q#24_SI_3
-0.743
0.243
* Q#25_SI_4
-0.419
0.975
FC
Skewness
Kurtosis
* Q#26_FC_1
-1.293
1.244
* Q#27_FC_2
-0.891
-0.003
* Q#28_FC_3
0.509
-0.760
* Q#29_FC_4
-0.744
-0.516
* Q#30_FC_5
0.376
-1.170
BI
Skewness
Kurtosis
* Q#31_BI_1
-1.065
0.329
* Q#32_BI_2
-1.260
1.021
* Q#33_BI_3
-1.366
1.188
USE
Skewness
Kurtosis
* Q#11_USE_1
-0.589
-1.094
* Q#12_USE_2
-0.126
-1.351
* Q#13_USE_3
0.672
0.207
Outliers
Outliers are observations with a unique combination of characteristics identifiable
as distinctly different from the other observation and typically judged to be an unusually
high or low value on a variable (Hair et al., 2009). It is fundamental to screen the data to
84
identify outliers since they can affect the results. According to Hair et al. (2009), outliers
can be classified through three types of detection: univariate (examines the distribution of
observations for each variable in the analysis), bivariate (pairs of variables can be
assessed jointly through a scatter plot), or multivariate (analyses involving more than two
variables) perspective based on the number of variables (characteristics) considered. In
the present study, univariate and multivariate detection method were examined using
residual analysis (Tabachnick & Fidell, 2014). SPSS was used to detect outliers through
the following steps.
The first step was to identify univariate outliers using mean composites created
for each variable (PE, EE, SI, FC, BI, and USE) and then standardized. According to
Tabachnick and Fidell (2007), cases whose standardized values exceeded the absolute
value of 3.29 were considered potential outliers (p < .001, two-tailed test). Table 7 shows
that there were no univariate outlier cases with residuals above 3.29.
85
Table 7 Standardized values per each variable – maximum and minimum
PE
N
Maximum
Minimum
Q#14_PE_1
169
0.971
-3.501
Q#15_PE_2
169
1.109
-3.425
Q#16_PE_3
169
1.140
-2.745
Q#17_PE_4
169
1.031
-2.930
EE
N
Maximum
Minimum
Q#18_EE_1
169
1.194
-3.788
Q#19_EE_2
169
1.166
-3.583
Q#20_EE_3
169
1.155
-3.457
Q#21_EE_4
169
1.098
-2.967
SI
N
Maximum
Minimum
Q#22_SI_1
169
1.142
-3.193
Q#23_SI_2
169
1.227
-2.921
Q#24_SI_3
169
1.241
-3.114
Q#25_SI_4
169
1.445
-1.939
FC
N
Maximum
Minimum
Q#26_FC_1
169
0.932
-3.196
Q#27_FC_2
169
1.042
-2.900
Q#28_FC_3
169
2.157
-1.239
Q#29_FC_4
169
1.118
-2.149
Q#30_FC_5
169
1.932
-1.279
BI
N
Maximum
Minimum
Q#31_BI_1
169
0.945
-2.857
Q#32_BI_2
169
0.880
-2.983
Q#33_BI_3
169
0.817
-2.983
USE
N
Maximum
Minimum
Q#11_USE_1
169
1.230
-1.579
Q#12_USE_2
169
1.363
-1.347
Q#13_USE_3
169
2.731
-1.205
86
The second step is to detect multivariate outliers using the Mahalanobis D²
measure (Hair et al., 2009). According to Hair et al. (2009), this analysis evaluates the
position of each observation compared with the center of all observation on a set of
variables. In the present study, the calculation of the D²/df value (df = 23) allows for
identification of outliers through an approximate test of statistical significance. Because
the sample had only 169 observations, a threshold value of 3.5 was used rather than the
value of 4.0 used in large samples (Hair et al., 2009). Table 8 shows the detection of 9
observations (166, 50, 129, 100, 54, 37, 74, 127, and 35) identified as significantly
different.
Table 8 Multivariate outlier detection results
Observation
number
Mahalanobis D-
squared (D²)
p1
p2
Cases with values
of D²/df greater
than >3.5
166
71.406
0
0
4.760
50
70.433
0
0
4.696
129
69.644
0
0
4.643
100
69.434
0
0
4.629
54
61.83
0
0
4.122
37
56.144
0
0
3.743
74
55.45
0
0
3.697
127
54.589
0
0
3.639
35
53.106
0
0
3.540
Note: Results extracted from the AMOS output – Observations farthest from the centroid
(Mahalanobis distance).
These outliers were not seen in earlier univariate analysis but appear only in the
multivariate tests. These results are probably not unique on a single variable but instead
are unique in combination (Hair et al., 2009). In conclusion, no observations are extreme
on a sufficient number of variables to be considered unrepresentative of the population.
In all instances, the observations designated as outliers, even with the multivariate tests,
87
seem similar enough to the remaining observations to be retained in the multivariate
analysis. However, the researcher should always examine the results of each specific
multivariate technique to identify observations that may become outliers. For the present
study, outliers are retained and further verified in the assessment model.
Descriptive Statistics Analysis
The survey was completed by 169 respondents (farmers and vendors) at FMs,
CSAs, and roadside stands in north Mississippi. The following sections present
descriptive statistics for the questionnaire items.
Demographic Analysis of Farmers at the Mississippi Northeast Region
A total of 169 participants were surveyed at the FMs visited by the researcher.
Table 9 provides a general overview of the farmers and vendors of the different
participating FMs in North Mississippi in terms of demographic information such as
gender, age, education level, and computer experience.
88
Table 9 Descriptive data analysis demographic information (n = 169)
Variable
Frequency
(%)
Gender
Female
93
55.0
Male
76
45.0
Age
Between 18 and 24 years
13
7.7
Between 25 and 34 years
23
13.6
Between 35 and 44 years
25
14.8
Between 45 and 54 years
33
19.5
Between 55 and 74 years
65
38.5
75 years or older
10
5.9
Education
Less than a high school diploma
4
2.4
High school degree or equivalent (e.g. GED)
27
16.0
Some college
54
32.0
Bachelor’s degree (e.g., BA, BS)
51
30.2
Some Graduate School
13
7.7
Graduate degree (e.g., MA, MS, PhD, JD,
MD, EdD)
20
11.8
Computer
Experience
Very Poor
8
4.7
Poor
11
6.5
Moderate
36
21.3
Good
58
34.3
Very Good
56
33.1
Descriptive Statistics for UTAUT
A descriptive statistical analysis is described in this section to provide an
understanding of the farmer perceptions in terms of PE, EE, SI, FC, BI, and USE
(personal and business purposes). Table 10 summarizes the frequencies and
corresponding percentages for farmers and vendors’ perceptions related to PE questions.
Over two-thirds of respondents (69.3 percent) agreed that they would find social media
useful to promote their agricultural business or products. The majority agreed that use of
social media would enable them to complete tasks related to promoting their business or
89
products more quickly (61.6 percent), that use of social media would increase their
productivity to promote their business or products (57.6 percent), and that use of social
media would increase their chances of getting more sales in their business (65.7 percent).
Table 10 Descriptive statistics for PE
Questionnaire Item
Strongly
Disagree
Disagree
More or
Less
Disagree
Undecided
More or
Less
Agree
Agree
Strongly
Agree
Mean
Std.
Dev.
PE-Q14-1: I would find
Social Media useful to
promote my agricultural
business or products.
2
(1.2%)
5
(3.0 %)
5
(3.0 %)
17
(10.1%)
23
(13.6%)
66
(39.1%)
51
(30.2%)
5.70
1.340
PE-Q15-2: Using Social
Media would enable me to
complete tasks related to
promoting my business or
products more quickly.
1
(0.6%)
7
(4.1%)
5
(3.0%)
19
(11.2%)
33
(19.5%)
64
(37.9%)
40
(23.7%)
5.53
1.323
PE-Q16-3: Using Social
Media would increase my
productivity to promote my
business or products.
0
(0.0%)
7
(4.1%)
6
(3.6%)
17
(10.1%)
40
(23.7%)
58
(34.3%)
41
(24.3%)
5.53
1.287
PE-Q17-4: Using Social
Media would increase my
chances of getting more sales
in my business.
0
(0.0%)
6
(3.6%)
5
(3.0%)
15
(8.9%)
32
(18.9%)
61
(36.1%)
50
(29.6%)
5.70
1.262
Note: The highest percentage for each item is bolded.
Table 11 summarizes the frequencies and corresponding percentages for farmers
and vendors’ perceptions related to EE questions. Again, the majority of respondents
agreed (or strongly agreed) that their interaction with social media for promoting their
business or products would be clear and understandable (59.8 percent), that it would be
easy for them to become skillful using social media in their business (60.4 percent), that
they would find social media easy to use for promoting for their business (58.0 percent),
and that learning to operate social media for their business would be easy for them (56.8
percent).
90
Table 11 Descriptive statistics for EE
Questionnaire Item
Strongly
Disagree
Disagree
More or
Less
Disagree
Undecided
More or
Less
Agree
Agree
Strongly
Agree
Mean
Std.
Dev.
EE-Q18-1: My interaction
with Social Media for
promoting my business or
products would be clear and
understandable.
1
(0.6%)
4
(2.4%)
3
(1.8%)
21
(12.4%)
39
(23.1%)
64
(37.9%)
37
(21.9%)
5.56
1.204
EE-Q19-2: It would be easy
for me to become skillful at
using Social Media in my
business.
1
(0.6%)
2
(1.2%)
8
(4.7%)
27
(16.0%)
29
(17.2%)
62
(36.7%)
40
(23.7%)
5.53
1.263
EE-Q20-3: I would find Social
Media easy to use for my
business.
1
(0.6%)
2
(1.2%)
11
(6.5%)
24
(14.2%)
33
(19.5%)
56
(33.1%)
42
(24.9%)
5.50
1.301
EE-Q21-4: Learning to
operate Social Media for my
business would be easy for me.
5
(3.0%)
3
(1.8%)
10
(5.9%)
24
(14.2%)
31
(18.3%)
55
(32.5%)
41
(24.3%)
5.38
1.476
Note: The highest percentage for each item is bolded.
Table 12 summarizes the frequencies and corresponding percentages for farmers
and vendors’ perceptions related to SI questions. Approximately half of the respondents
agreed (or strongly agreed) that people who are important to them think they should use
social media to promote their business (57.4 percent), that that people who influence their
behavior think they should use social media to promote their business, and that people
whose opinions they value think they should use social media to promote their business
(52.1 percent). However, slightly more than one-third of respondents agreed that other
farmers and producers who use social media to promote their businesses would affect
their decision to use social media in their business (37.3 percent).
91
Table 12 Descriptive statistics for SI
Questionnaire Item
Strongly
Disagree
Disagree
More or
Less
Disagree
Undecided
More or
Less
Agree
Agree
Strongly
Agree
Mean
Std.
Dev.
SI-Q22-1: People who are
important to me think that I
should use Social Media to
promote my business.
1
(0.6%)
6
(3.6%)
8
(4.7%)
29
(17.2%)
28
(16.6%)
56
(33.1%)
41
(24.3%)
5.42
1.383
SI-Q23-2: People who
influence my behavior think
that I should use Social Media
to promote my business.
3
(1.8%)
7
(4.1%)
7
(4.1%)
33
(19.5%)
36
(21.3%)
48
(28.4%)
35
(20.7%)
5.22
1.446
SI-Q24-3: People whose
opinions I value think that I
should use Social Media to
promote my business.
2
(1.2%)
6
(3.6%)
5
(3.0%)
37
(21.9%)
31
(18.3%)
54
(32.0%)
34
(20.1%)
5.29
1.378
SI-Q25-4: Other farmers and
producers who use Social
Media to promote their
businesses will affect my
decision to use Social Media in
my business.
10
(5.9%)
27
(16.0%)
9
(5.3%)
34
(20.1%)
26
(15.4%)
48
(28.4%)
15
(8.9%)
4.44
1.772
Note: The highest percentage for each item is bolded.
Table 13 summarizes the frequencies and corresponding percentages for farmers
and vendors’ perceptions related to FC questions. Approximately two-thirds of
respondents agreed (or strongly agreed) that they have the resources necessary to use
social media to promote their business (67.4 percent), while 59.2 percent agreed that they
have the knowledge necessary to use social media to promote their business. Nearly half
agreed that someone could help them if they had technical problems when using social
media to promote their business (49.1 percent), On the other hand, 46.1 percent of
respondents disagreed (or strongly disagreed) that social media is not compatible with
other ways that they promote their business. Finally, 45 percent of respondents disagreed
that Internet connectivity, access, or costs limit their ability to promote their business
using social media.
92
Table 13 Descriptive statistics for FC
Questionnaire Item
Strongly
Disagree
Disagree
More or
Less
Disagree
Undecided
More or
Less
Agree
Agree
Strongly
Agree
Mean
Std.
Dev.
FC-Q26-1: I have the
resources necessary to use
Social Media to promote my
business.
3
(1.8%)
6
(3.6%)
7
(4.1%)
16
(9.5%)
23
(13.6%)
59
(34.9%)
55
(32.5%)
5.64
1.453
FC-Q27-2: I have the
knowledge necessary to use
Social Media to promote my
business.
2
(1.2%)
8
(4.7%)
12
(7.1%)
22
(13.0%)
25
(14.8%)
52
(30.8%)
48
(28.4%)
5.41
1.522
FC-Q28-3: Social Media is not
compatible with other ways
that I promote my business.
32
(18.9%)
46
(27.2%)
17
(10.1%)
37
(21.9%)
14
(8.3%)
15
(8.9%)
8
(4.7%)
3.19
1.766
FC-Q29-4: If I have technical
problems when using Social
Media to promote my
business, someone could help
me.
11
(6.5%)
15
(8.9%)
9
(5.3%)
23
(16.6%)
28
(16.6%)
45
(26.6%)
38
(22.5%)
4.95
1.836
FC-Q30-5: I feel that internet
connectivity, access, or costs
limit my ability to promote my
business using Social Media.
27
(16.0%)
49
(29.0%)
18
(10.7%)
22
(13.0%)
20
(11.8%)
25
(14.8%)
8
(4.7%)
3.39
1.868
Note: The highest percentage for each item is bolded.
Table 14 summarizes the frequencies and corresponding percentages for farmers
and vendors’ perceptions related to BI to use social media. Approximately two-thirds of
respondents agreed (or strongly agreed) that they plan to use social media to promote
their business or products in the next 12 months (63.9 percent), that they predict they will
use social media to promote their business or products in the next 12 months (66.9
percent), and that they intend to use social media in the next 12 months (69.8 percent).
Table 14 Descriptive statistics for BI
Questionnaire Item
Strongly
Disagree
Disagree
More or
Less
Disagree
Undecided
More or
Less
Agree
Agree
Strongly
Agree
Mean
Std.
Dev.
BI-Q31-1: I plan to use Social
Media to promote my business
or products in the next 12
months.
3
(1.8%)
10
(5.9%)
6
(3.6%)
24
(14.2%)
18
(10.7%)
52
(30.8%)
56
(33.1%)
5.51
1.578
BI-Q32-2: I predict I will use
Social Media to promote my
business or products in the
next 12 months.
5
(3.0%)
5
(3.0%)
6
(3.6%)
22
(13.0%)
18
(10.7%)
50
(29.6%)
63
(37.3%)
5.63
1.553
BI-Q33-3: I intend to use
Social Media in the next 12
months.
5
(3.0%)
6
(3.6%)
6
(3.6%)
18
(10.7%)
16
(9.5%)
48
(28.4%)
70
(41.4%)
5.71
1.579
Note: The highest percentage for each item is bolded.
93
Table 15 presents information about the use of social media for personal
purposes. Results indicate that 80.5 percent of respondents use social media for personal
purposes, while 19.5 percent do not use social media for personal purposes. Facebook is
the main social media platform and is used by 91.2 percent of respondents. YouTube is
the second most used social media platform; it is used by 49.3 percent of respondents.
Table 15 Social media use for personal purposes (n = 169)
Variable
Frequency
(%)
Social Media Use PERSONAL Purposes
136
80.5
Facebook
124
91.2
Twitter
38
27.9
Instagram
61
44.9
Pinterest
46
33.8
Snapchat
28
20.6
YouTube
67
49.3
LinkedIn
23
16.9
Wikis
7
5.1
Blogs
13
9.1
Table 16 provides information about USE for business purposes. Over two-thirds
of the respondents (69.2 percent) use social media for business purposes, while 92.3
percent of respondents use Facebook as the main social media platform, and 39.3 percent
use Instagram as a second main social media platform.
94
Table 16 Social media use for business purposes (n = 169)
Variable
Frequency
(%)
Social Media Use BUSINESS Purposes
117
69.2
Facebook
108
92.3
Twitter
23
19.7
Instagram
46
39.3
Pinterest
12
10.3
Snapchat
4
3.4
YouTube
17
14.5
LinkedIn
13
11.1
Wikis
1
0.8
Blogs
4
3.4
Other
2
1.7
Table 17 summarizes the frequencies and corresponding percentages for farmers
and vendors’ social media use. Approximately one-third of respondents described a good
experience with the use of social media (32.5 percent), while 23.7 percent did not have
experience. Although 27.2 percent of respondents have not used social media for business
purposes, nearly one-fourth of respondents had been using it between 1 - 2 years for
business purposes (23.7 percent). Over one-third reported using social media less than
one hour per day for business purposes (36.1 percent).
95
Table 17 Descriptive statistics for USE
Questionnaire Item
No
experience
Very
Poor
Poor
Moderate
Good
Very
Good
Mean
Std.
Dev.
USE-Q11-1: How would you describe
your Social Media experience for
business purposes?
40
(23.7%)
6
(3.6%)
7
(4.1%)
36
(21.3%)
55
(32.5%)
25
(14.8%)
2.80
1.791
No Use SM
for Business
Purposes
Less than
1 year
Between
1 – 2
years
Between
3 – 4
years
More than
4 years
Mean
Std.
Dev.
USE-Q12-2: How long have you been
using Social Media for business
purposes?
46
(27.2%)
13
(7.7%)
40
(23.7%)
37
(21.9%)
33
(19.5%)
1.99
1.476
No Use SM
for Business
Purposes
Less than
1 Hour
Between
1 – 2
hours
Between
3 – 4
hours
More than
4 hours
Mean
Std.
Dev.
USE-Q13-3: How often do you use
Social Media per day for business
purposes?
46
(27.2%)
61
(36.1%)
47
(27.8%)
9
(5.3%)
6
(3.6%)
1.22
1.020
Note: The highest percentage for each item is bolded.
Table 18 presents summary and central tendency statistics calculated considering
the average of the items for each construct.
Table 18 Descriptive statistics summary by constructs
Constructs
N
Min
Max
Mean
Std. Dev.
Var.
PE
169
1.50
7.00
5.62
1.303
1.698
EE
169
1.00
7.00
5.49
1.311
1.729
SI
169
1.00
7.00
5.90
1.495
2.260
FC
169
1.00
7.00
4.52
1.689
2.881
BI
169
1.00
7.00
5.62
1.570
2.465
USE
169
0.0
5.00
2.00
1.429
2.143
Valid N (list wise)
169
Note: Elaborated by the author.
Descriptive Statistics for Preferred Marketing Channels and Extension Programs
Tables 19, 20, and 21 provide general information about preferred marketing
channels and Extension program interest. Table 19 shows 98.2 percent of respondents
prefer to use FMs as a marketing channel to promote their products, and 47.3 percent of
respondents prefer social media.
96
Table 19 Marketing channels farmers and vendors prefer to use to promote their
agricultural products
Variable
Frequency
(%)
FMs
166
98.2
Retailer
58
34.3
CSAs
34
20.1
Roadside Stand
34
20.1
Website
41
24.1
Social Media
80
47.3
Other
21
12.4
Table 20 provides general information about reasons that would prevent farmers
and vendors from selling their products through various marketing channels. In this
question, 72.8 percent of the respondents answered this question, while 27.2 percent were
not willing to answer to avoid conflict with the farmers market’s manager.
“Low Flow of Customers” was the most commonly selected reason to prevent
them from selling their agricultural products at a FM (26.6 percent of respondents), in
CSAs (10.7 percent of respondents), or at roadside stands (23.7 percent of respondents).
“Too Many Food Regulations” was most commonly selected (17.2 percent of
respondents) as a reason to prevent them from selling their products in a retail setting.
Finally, 10.1 percent of respondents selected “Expensive Fees” as a reason to prevent
them from selling their products on websites.
97
Table 20 Reasons to prevent sell agricultural products through marketing channels
Valid
Cases Missing
Total
N
(%)
N
(%)
N
(%)
Respondents
123
72.8
46
27.2
169
100
Note: Dichotomous group tabulation.
Expensive
Fees (%)
Low Flow
of
Customers
(%)
High
Logistic
Costs (%)
Too Many
Food
Regulations
(%)
Limited
Hours of
Customer
Service (%)
Other
Reasons
(%)
FMs
9.5
26.6
7.7
8.9
14.8
6.5
Retailer
14.8
5.3
11.8
17.2
2.4
4.7
CSAs
4.1
10.7
7.1
4.7
7.1
3.6
Roadside Stand
3.0
23.7
6.5
3.6
7.7
4.7
Website
10.1
9.5
8.3
3.6
4.1
5.3
Social Media
7.1
7.1
6.5
1.8
2.4
4.1
Other Channels
0.0
0.0
0.6
0.0
0.0
1.8
Note: The highest percentage for each item is bolded.
Respondents were also asked if they were interested in participating in an MSU
Extension program about marketing agricultural products using Social Media. Table 21
shows that 42.6 percent of respondents are interested in participating in this type of
Extension program.
Table 21 Farmers interested in participating in an MSU Extension program
Variable
Frequency
(%)
Interest in Extension program participation
No
97
57.4
Yes
72
42.6
Measurement Model Analysis
Results from a series of analyses needed to prepare for testing the UTAUT model
as applied to farmers’ use of social media for marketing their agricultural products are
described in the following sections. Factor analysis (FA) was conducted to study and
98
confirm the validity of the factor structures that represent each individual model
construct. Scale reliability is reported through an assessment of internal consistency.
Additionally, results of the EFA and the CFA are presented, both of which are employed
to confirm and refine the identified structure of each model construct to ensure its validity
and unidimensionality.
Reliability
The reliability of a measure refers to the degree to which the instrument is free of
random error and it is concerned with the consistency and stability of the measurement
(Hair et al., 2009). In the present study, four independent scales and two dependent scales
were used to measure the constructs of the UTAUT model. The independent scales were
PE, EE, SI, and FC. The dependent scales were BI to use social media (BI) and actual
USE. A scale reliability analysis was performed to assess the internal consistency and
item-total correlations to demonstrate the model constructs’ consistency and accuracy.
Internal Consistency Reliability
Internal consistency reliability refers to the degree to which responses are
consistent across the items (variables) within a single measurement scale (Kline, 2011;
Sekaran, 2003). The type of reliability coefficient reported most often in the literature is
coefficient alpha also called Cronbach’s alpha. In the present study, Cronbach’s
coefficient alphas, which are calculated based on average inter-item correlations, were
used to measure internal consistency. If internal consistency is low, then the content of
the items may be so heterogeneous that the total score is not the best possible unit of
analysis for the measure. As stated by Straub (1989, p. 151), “high correlations between
99
alternative measures or large Cronbach’s alphas are usually signs that the measures are
reliable.” Hair et al. (2006) recommended that construct reliability should be 0.7 or
higher to demonstrate convergence or internal consistency (Hair et al., 2009, p. 687).
However, Hinton, McMurray, and Brownlow, 2014 propose four levels of reliability:
excellent (0.90 and above); high (0.70 to 0.89); high moderate (0.50 to 0.69); and low
(0.49 and below). A scale reliability analysis was performed to assess the internal
consistency to show that those scales satisfied the model constructs consistently and
accurately. Table 22 shows the Cronbach’s alpha (α) value for each construct.
Table 22 Cronbach’s alpha reliability results
Constructs
Items
Cronbach’s
Alpha (α)
Comment
PE
4
0.929
Excellent Reliability
EE
4
0.925
Excellent Reliability
SI
4
0.816
High Reliability
FC
5
0.490
High Moderate Reliability
BI
3
0.968
Excellent Reliability
USE
3
0.866
High Reliability
The results in the analysis show all the constructs to have high reliability of more
than 0.70, except the FC construct with a high moderate reliability (approximately 0.50).
However, the FC construct will be redefined in the model assessment and model re-
specification processes.
Item-total Correlations
The study of correlations demonstrates the relationships between the research
constructs. It also provides comparisons with the existing sample data. Items-total
correlation refers to the correlation of an individual variable, with the composite score of
100
all variables forming the measure of the construct (Lu et al., 2007). These relationships
provide a check for how well the proposed model captures important properties of the
study sample. In this study, a corrected item correlation analysis was performed for all
constructs of the proposed model. According to Hair et al. (2009), a value of the
corrected item-total correlation of less than 0.30 indicates that the variable is measuring
something different from the construct as a whole. The results of item correlations are
presented in Tables 23 to 46.
Validity
Construct validity is referred to as the extent to which a set of measured variables
actually represents the theoretical latent construct that those variables are designed to
measure (Hair et al., 2009). It demonstrates that the research instrument truly measures
what it is intended to be measured (Netemeyer, Bearden, & Sharma, 2003). One of the
primary objectives of CFA/SEM is to assess the construct validity of a proposed
measurement theory (Hair et al., 2009). In this study, the validity of the scales was
assessed by using EFA and examining the correlation coefficients for all of the
instrument scales (constructs). The most two widely accepted forms of validity are
convergent and discriminant. According to Hair et al. (2009), convergent validity
assesses the degree to which two measures of the same concept are correlated and
discriminant validity assesses the degree to which conceptually similar concepts are
distinct. In addition, convergent and discriminant validity of the measurement scales were
also assessed using CFA.
101
Exploratory Factor Analysis (EFA)
Exploratory factor analysis (EFA) might be defined as an orderly simplification
and/or organized of interrelated measures. According to Byrne (2010), EFA is designed
for the situation where links between the observed and latent variables are unknown or
uncertain. The primary purpose of EFA is to define the underlying structure among the
variables in the analysis (Hair et al., 2009). Additionally, EFA is used to explore data to
determine the number or nature of factors that account for the covariation between
variables. This covariation process is used when a researcher does not have enough
evidence to form a hypothesis about the number of factors underlying the data. In brief,
EFA is useful in assessing the relationships among variables and exploring the construct
validity of test scale. In this step, SPSS was used to conduct the EFA and the scales were
analyzed one by one in this process.
Analysis of PE Scale
Table 23 shows the description of the PE variable with its four items, variable
codes, and questionnaire statements respectively. The correlation coefficient matrix for
the four items in the PE scale is shown in Table 24. The results revealed that the
correlation coefficients between items are generally greater than 0.3. A value for a
corrected item-correlation of less than 0.30 indicates that the variable is measuring
something different from the construct as a whole (Pallant, 2007). However, for a matrix
correlation in a factor analysis, sampling adequacy was assessed by examining the
Kaiser-Meyer-Olkin (KMO) statistic. According to Hinton et al. (2014), the KMO test is
a helpful measure of whether the data are suitable for a factor analysis. If the test comes
out at 0.5 or higher, factor analysis can proceed reflecting that the data is suitable for the
102
analysis. Another test included in factor analysis is the Bartlett’s Test of Sphericity.
According to Hinton et al. (2014), this test demonstrates whether there is a relationship
between the variables. If no relationship is found, then there is no need to proceed with
the factor analysis. A p-value < .05 indicates a reason to continue with the factor analysis,
and if the analysis has found p < .001, it can be concluded that there are relationships
between the variables (Hinton et al., 2014, p. 347). As Table 25 shows, The KMO
statistic is 0.821, which is above the minimum acceptable level of 0.5 (Hinton et al.,
2014). The KMO test is indicating sampling adequacy. Additionally, Bartlett’s test of
Sphericity was Chi-square = 558.666, which was highly significant at (p < 0.001)
indicating that there were adequate relationships between the variables of the factor (PE)
included in the analysis. In brief, the tests indicated that the data were appropriate for the
factor analysis.
Table 23 Coding of PE construct
Construct
Variable
Code
Questionnaire Statement
PE
Q#14_PE_1
I would find Social Media useful to promote my
agricultural business or products.
Q#15_PE_2
Using Social Media would enable me to complete tasks
related to promoting my business or products more
quickly.
Q#16_PE_3
Using Social Media would increase my productivity to
promote my business or products.
Q#17_PE_4
Using Social Media would increase my chances of getting
more sales in my business.
103
Table 24 Correlation matrix for PE scale
Q#14_PE_1
Q#15_PE_2
Q#16_PE_3
Q#17_PE_4
Correlation
Q#14_PE_1
1.000
0.766
0.712
0.724
Q#15_PE_2
0.766
1.000
0.798
0.749
Q#16_PE_3
0.712
0.798
1.000
0.851
Q#17_PE_4
0.724
0.749
0.851
1.000
Note: **. Correlation is significant at the 0.01 level (2-tailed)
Table 25 KMO and Bartlett’s test for PE
KMO and Bartlett’s Test
Kaiser-Meyer-Olkin Measure of Sampling Adequacy
0.821
Bartlett’s Test of Sphericity
Approx. Chi-Square
558.666
df
6
Sig.
.000
The next step to be evaluated is the factor loading of scale items. These loadings
contain both unique variance between variables as well as factors and the correlations
among factors (Hair et al., 2009). The factor loading analysis is usually used in factor
interpretation unless data reduction is the sole objective. According to Hair et al. (2009,
p. 118), values greater than +/- 0.50 are generally considered necessary for practical
significance. For the present study, the cut-off factor loading of 0.5 was used to ensure
that all variables had practical significance for the confirmatory factor analysis. A larger
loading is needed given a factor solution with a large number of factors, especially in
evaluating the loadings on later factors (Hair et al., 2009; p. 118). Table 26 shows the
loading values for the PE factor; all four items exceed the cut-off level of 0.5. In brief, the
four items are unidimensional which means that this set of variables (items) represents
the present construct.
104
Table 26 Factor loading for PE
Component Matrix
Component
1
PE_1
0.878
PE_2
0.912
PE_3
0.927
PE_4
0.916
Note: Extraction Method: Principal Component Analysis
a. 1 component extracted
Analysis of EE Scale
Table 27 shows four questionnaire statements and codes for the EE scale which
measures the degree of ease of use social media. The correlation matrix (see Table 28)
indicates that correlation coefficients are generally greater than 0.3. On the other hand, in
Table 29, the KMO analysis shows a highly significant result (0.822) and Bartlett’s test
(Chi-square = 583.059) is highly significant (p < 0.001). Table 30 shows the factor
loadings of the items are highly than the cut-off level (+/- 0.5) indicating that the four
items for EE are unidimensional.
Table 27 Coding of EE construct
Construct
Variable
Code
Questionnaire Statement
EE
Q#18_EE_1
My interaction with Social Media for promoting my
business or products would be clear and understandable.
Q#19_EE_2
It would be easy for me to become skillful at using Social
Media in my business.
Q#20_EE_3
I would find Social Media easy to use for my business.
Q#21_EE_4
Learning to operate Social Media for my business would
be easy for me.
105
Table 28 Correlation matrix for EE scale
EE_1
EE_2
EE_3
EE_4
Correlation
EE_1
1.000
0.646
0.676
0.720
EE_2
0.646
1.000
0.887
0.796
EE_3
0.676
0.887
1.000
0.822
EE_4
0.720
0.796
0.822
1.000
Table 29 KMO and Bartllet’s test for EE
KMO and Bartlett’s Test
Kaiser-Meyer-Olkin Measure of Sampling Adequacy
0.822
Bartlett’s Test of Sphericity
Approx. Chi-Square
583.059
df
6
Sig.
.000
Table 30 Factor loading for EE
Component Matrix
Component
1
EE_1
0.832
EE_2
0.923
EE_3
0.939
EE_4
0.924
Note: Extraction method: principal component analysis
a. 1 component extracted
Analysis of SI Scale
Table 31 shows four questionnaire statements and codes in the SI scale that
measures the degree to which a farmer or vendor at the FM perceives that others who are
important to them believe he or she should use social media. The correlation matrix (see
Table 32) indicates that correlation coefficients are generally greater than 0.3. However,
the item SI_4 indicates a low correlation with other items. Also, the KMO analysis
(0.778) is highly significant) and the Bartlett’s test (Chi-square = 440.538) is highly
significant (p < 0.001) as presented in Table 33. Table 34 shows that the factor loading of
106
items are higher than the cut-off level (0.50). However, the item SI_4 indicates a low
factor loading, but it meets the minimum cut-off required (0.5). In conclusion, the four
items for SI are unidimensional.
Table 31 Coding of SI construct
Construct
Variable
Code
Questionnaire Statement
SI
Q#22_SI_1
People who are important to me think that I should use
Social Media to promote my business.
Q#23_SI_2
People who influence my behavior think that I should use
Social Media to promote my business.
Q#24_SI_3
People whose opinions I value think that I should use
Social Media to promote my business.
Q#25_SI_4
Other farmers and producers who use Social Media to
promote their businesses will affect my decision to use
Social Media in my business.
Table 32 Correlation matrix for SI scale
SI_1
SI_2
SI_3
SI_4
Correlation
SI_1
1.000
0.804
0.861
0.308
SI_2
0.804
1.000
0.804
0.328
SI_3
0.861
0.804
1.000
0.277
SI_4
0.308
0.328
0.277
1.000
Table 33 KMO and Bartlett’s test for SI
KMO and Bartlett’s Test
Kaiser-Meyer-Olkin Measure of Sampling Adequacy
0.778
Bartlett’s Test of Sphericity
Approx. Chi-Square
440.538
df
6
Sig.
.000
107
Table 34 Factor loading for SI
Component Matrix
Component
1
SI_1
0.935
SI_2
0.918
SI_3
0.929
SI_4
0.470
Note: Extraction method: principal component analysis
a. 1 component extracted
Analysis of FC Scale
Table 35 shows the five questionnaire statements which were used to measure
how an individual believes that technical infrastructure, connectivity, and resources exist
to support the use of social media. The correlation matrix for the scales FC_1, FC_2, and
FC_4 indicated that the correlation coefficients are greater than 0.3 as presented in Table
36. However, items FC_3 and FC_5 indicated low correlation to other items (below
0.30). Also, the KMO analysis (0.585) was high moderate significant and Bartlett’s test
(Chi-square = 182.441) was highly significant (p < 0.001) as presented in Table 37.
Finally, the factor loadings of items FC_1, FC_2, and FC_4 are higher than the cut-off
level (0.50) as shown in Table 38. While FC_3 and FC_5 indicate a low factor loading
among items, both still meet the minimum cut-off. In conclusion, three of the five items
measuring SI (FC_1, FC_2, and FC_4) are unidimensional.
108
Table 35 Coding of FC construct
Construct
Variable
Code
Questionnaire Statement
FC
Q#26_FC_1
I have the resources necessary to use Social Media to
promote my business.
Q#27_FC_2
I have the knowledge necessary to use Social Media to
promote my business.
Q#28_FC_3
Social Media is not compatible with other ways that I
promote my business.
Q#29_FC_4
If I have technical problems when using Social Media to
promote my business, there is someone who could help me.
Q#30_FC_5
I feel that internet connectivity, access, or cost limit my
ability to promote my business using Social Media.
Table 36 Correlation matrix for FC scale
FC1
FC2
FC3
FC4
FC5
Correlation
FC1
1.000
0.718
-0.078
0.399
-0.088
FC2
0.718
1.000
-0.038
0.398
0.022
FC3
-0.078
-0.038
1.000
0.115
0.331
FC4
0.399
0.398
0.115
1.000
-0.030
FC5
-0.087
0.022
0.331
-0.030
1.000
Table 37 KMO and Bartlett’s test for FC
KMO and Bartlett’s Test
Kaiser-Meyer-Olkin Measure of Sampling Adequacy
0.585
Bartlett’s Test of Sphericity
Approx. Chi-Square
182.441
df
10
Sig.
.000
109
Table 38 Factor loading for FC
Component Matrix
Component
Component
1
2
FC_1
0.887
-0.062
FC_2
0.879
0.049
FC_4
0.679
0.190
FC_3
-0.054
0.825
FC_5
-0.093
0.790
Note: Extraction method: principal component analysis
a. 1 component extracted
Analysis of BI Scale
Table 39 shows the three questionnaire statements used to measure the influence
of BI on the use of social media by respondents. The correlation matrix (see Table 40)
indicates that the correlation coefficients are generally greater than 0.3. Also, the KMO
analysis (0.773) is highly significant) and the Bartlett’s test (Chi-square = 637.548) is
highly significant (p < 0.001) as presented in Table 41. Finally, the factor loadings of
items BI_1, BI_2, and BI_3 presented in Table 42 are higher than the cut-off level (0.50).
In brief, the three items measuring BI are unidimensional.
Table 39 Coding of BI construct
Construct
Variable
Code
Questionnaire Statement
BI
Q#31_BI_1
I plan to use Social Media to promote my business or
products in the next 12 months.
Q#32_BI_2
I predict I will use Social Media to promote my business or
products in the next 12 months.
Q#33_BI_3
I intend to use Social Media in the next 12 months.
110
Table 40 Correlation matrix for BI scale
BI_1
BI_2
BI_3
Correlation
BI_1
1.000
0.929
0.889
BI_2
0.929
1.000
0.910
BI_3
0.889
0.910
1.000
Table 41 KMO and Bartlett’s test for BI
KMO and Bartlett’s Test
Kaiser-Meyer-Olkin Measure of Sampling Adequacy
0.773
Bartlett’s Test of Sphericity
Approx. Chi-Square
637.548
df
3
Sig.
.000
Table 42 Factor loading for BI
Component Matrix
Component
1
BI_1
0.969
BI_2
0.977
BI_3
0.962
Note: Extraction method: principal component analysis
a. 1 component extracted
Analysis of USE Scale
Table 43 shows the three questionnaire statements that were used to assess the
actual use of social media by respondents to promote their business. The correlation
matrix (see Table 44) for the three scale items of use of behavior (USE) of social media
indicated that the correlations coefficients are generally greater than 0.3. Also, the KMO
analysis (0.723) is highly significant, and the Bartlett’s test (Chi-square = 292.201) is
highly significant (p < 0.001) as presented in Table 45. Finally, the factor loadings of
111
items USE_1, USE _2, and USE _3 presented in Table 46 are higher than the cut-off
level (0.50). In conclusion, the three items measuring USE are unidimensional.
Table 43 Coding of USE construct
Construct
Variable Code
Questionnaire Statement
USE
Q#11_USE_1
How would you describe your Social Media
experience for business purposes?
Q#12_USE_2
How long have you been using Social Media for
business purposes?
Q#13_USE_3
How often do you use Social Media per day for
business purposes?
Table 44 Correlation matrix for USE scale
USE1
USE2
USE3
Correlation
USE1
1.000
0.780
0.737
USE2
0.780
1.000
0.654
USE3
0.737
0.654
1.000
Table 45 KMO and Bartlett’s test for USE
KMO and Bartlett’s Test
Kaiser-Meyer-Olkin Measure of Sampling Adequacy
0.723
Bartlett’s Test of Sphericity
Approx. Chi-Square
292.201
df
3
Sig.
.000
Table 46 Factor loading for USE
Component Matrix
Component
1
USE_1
0.932
USE_2
0.899
USE_3
0.880
Note: Extraction method: principal component analysis
a. 1 component extracted
112
Confirmatory Factor Analysis (CFA)
After the EFA, the next step was to apply a CFA. This analysis indicates the
extent to which the hypothesized model fits or adequately describes the data (Byrne,
2010). Additionally, CFA is used to determine the goodness-of-fit between a model
already obtained (the UTAUT model) and the collected data. In other words, CFA is used
to provide a confirmatory test of the present study’s measurement theory.
Assessment of Construct Validity
The main objective of CFA is to assess the construct validity of the proposed
measurement (Hair et al., 2009). In the present study, confirmatory factor analysis was
conducted to assess and examine the convergent and discriminant validity. According to
Hair et al. (2009), construct validity is the extent to which a set of measured items
actually reflects the theoretical latent construct those items are designed to measure.
Discriminant and convergent validity are both part of the process analysis. The evaluation
of both (discriminant and convergent validity) was done using AMOS Version 24.0
(Analysis of Moment Structures).
Convergent Validity
Convergent validity is evidenced when items thought to reflect a construct
converge, or show significant, high correlations with one another (Straub, Boudreau, &
Gefen, 2004). According to Hair et al. (2009), the items that are indicators of a specific
construct should converge or share a high proportion of variance in common, known as
convergent validity. In the present study, two methods were used to evaluate convergent
validity: average variance extracted (AVE) and construct reliability (CR).
113
AVE is calculated as the mean variance extracted for the items loading on a
construct (Hair et al., 2009). In other words, AVE is mainly used to calculate the
explanatory power of all variables of the dimension to the average variations. This is
calculated using standardized loading:
AVE = ∑𝐿𝑖²
𝑛
𝑖=1
𝑛
Where: n = total number of items; and Li represents the standardized factor loading.
According to Hair et al. (2009), an AVE of 0.5 or higher suggests adequate
converge, and an AVE less than 0.5 indicates that more errors remain in the items than
variance explained. In other words, the higher the AVE, the higher the reliability and
convergent validity of the dimension. AVE should be computed for each latent construct
in a measurement model.
On the other hand, construct reliability (CR) is also a method to evaluate
convergent validity and another measure of reliability and internal consistency of the
measured variables. It is computed from the following formula:
CR =(∑𝐿𝑖
𝑛
𝑖=1 )²
(∑ 𝐿𝑖
𝑛
𝑖=1 )² + (∑𝑒𝑖
𝑛
𝑖=1 )
Where: n = total number of items; and Li represents the standardize factor loading; and ei
= error variance term.
According to Hair et al. (2009), construct reliability estimates of 0.7 or higher
suggest good reliability, while between 0.6 and 0.7 may be acceptable. In other words,
high construct reliability indicates that internal consistency exists meaning that all
measures represent the same latent construct. Using AMOS software, the standardized
regression weights were extracted. Table 47 shows the AVE and CR indicators for the
114
present study. In summary, all results support the instrument’s convergent validity as
adequate. However, the square root of the AVE for FC is less than one of the correlations
with another factor (0.332). All the other results confirm adequate convergent validity.
Discriminant Validity
Discriminant validity is the extent to which a construct is truly distinct from other
constructs both in terms of how much it correlates with other constructs and how
distinctly measured variables represent only this single construct (Hair et al., 2009).
According to Straub et al. (2004), discriminant validity is defined as an instrument’s
ability that the scales reflect their suggested construct differently from the relationship
with all the other scales in the research model. In the present study, discriminant validity
was tested and assessed through inter-factor correlations by comparing the absolute value
of the correlations between the constructs and the square root of average variance
extracted (AVE) by a construct. Using AMOS software, standardized regression weights
were extracted (Group number 1 Default model). When the correlations are lower than
the square root of the average variance extracted (AVE) by a construct, constructs are
said to have discriminant validity (Fornell & Larcker, 1981). Table 47 shows that all
square roots of the AVE (diagonal cells) are higher than the correlations between
constructs. However, the square root of the AVE for FC is less than one of the
correlations with another factor (0.576).
115
Table 47 Convergent validity for the constructs AVE and CR
Construct
(AVE)
(CR)
1
2
3
4
5
6
1. (PE)
0.767
0.929
0.876
2. (EE)
0.771
0.930
0.790
0.878
3. (SI)
0.671
0.881
0.737
0.680
0.819
4. (FC)
0.332**
0.567
0.638
0.848
0.557
0.576*
5. (BI)
0.910
0.968
0.680
0.717
0.624
0.716
0.954
6. (USE)
0.728
0.889
0.512
0.573
0.489
0.564
0.634
0.853
Note: Calculations prepared by the authors using AMOS and Excel software.
*Discriminant Validity: the square root of the AVE for FC is less than one of the
correlations with another factor (0.576).
**Convergent Validity: the AVE for FC is less than 0.50.
Model Assessment
The next step presents the process of assessing the conceptual UTAUT model and
includes testing the structural and measurement models. In the present study, the main
aim is to test the hypotheses related to the proposed UTAUT model.
Measurement Model Assessment Criteria
Measurement model validity depends on 1) finding specific evidence of construct
validity (step made previously), and 2) establishing acceptable level of goodness-of-fit
for the measurement model (Hair et al., 2009). The measurement model was assessed
using goodness-of-fit (GOF) tests. According to Hair et al. (2009), GOF indicates how
well the specific model reproduces the observed covariance matrix among the indicator
items. Once a specified model is estimated, model fit compares the theory to reality by
assessing the similarity of the estimated covariance matrix (theory) to reality (the
observed covariance matrix) (Hair et al., 2009). In other words, the researcher’ theory
were perfect, the observed and estimated covariance matrices will be the same. A number
of alternative GOF measures are available to the researcher. The basic index of this test is
116
Chi-square (χ²) statistics, degree of freedom (df), and significance level (p-value).
Additionally, Comparative Fix Index (CFI), the Root Mean Square Error of
Approximation (RMSEA), goodness-of-fit index (GFI), Tucker-Lewis index (TLI),
incremental-fit index (IFI), and relative Chi-square (χ²/ df) test were used to evaluate the
measurement model. According to Hair et al. (2009), reporting Chi-square χ² value,
degrees of freedom (df), CFI, TLI, IFI, and RMSEA will usually provide sufficient
unique information to evaluate a model. The main assessment criteria of the model fit are
summarized in Table 48.
Table 48 Measurement model assessment criteria
GOF Test
Requirement
References
Chi-square (χ²)
χ² < df
Hu & Bentler (1999)
Hair et al. (2009)
Byrne (2010)
Kline (2011)
Degree of Freedom (df)
> 0
Relative Chi-square (χ²/df)
< 3
Goodness-of-Fit Index (GFI)
> 0.90
Tucker-Lewis Index (TLI)
> 0.90
Incremental-Fit Index (IFI)
> 0.90
Comparative Fix Index (CFI)
> 0.90
Root Mean Square Error of Approx. (RMSEA)
< 0.08
Factor Loading
> 0.50
A model is considered to fit the data when the CFI value is above 0.95 (Hu &
Bentler, 1999). A model with a RMSEA value less than 0.05 has good fit, one with a
value of 0.08 has reasonable fit, and a model with RMSEA less than 0.1 has poor fit
(Browne & Cudeck, 1993). A small χ² value relative to the degree of freedom indicates a
good model fit (Hu & Bentler, 1999). Factor loading should be larger than 0.50 (Hair et
al., 2009).
117
Measurement Model Results
The measurement model results related to the testing of the hypothesized model
(AMOS graphical output) are presented in Figure 7 with the calculation estimates. This
measurement model was drawn using AMOS Version 24.0.
Figure 7 Model assessment (Model A) - path diagram for hypothesized model
(AMOS graphical output)
118
In Table 49, the fit indices are presented. The model presented an acceptable level
of fit (χ² = 451.489, df = 215, χ²/df = 2.100, GFI = 0.822, TLI = 0.920, CFI = 0.932, IFI =
0.933, and RMSEA = 0.081). All factors had significant loadings greater than 0.50 (p <
0.001) on each construct.
Table 49 Measurement model results
Construct/Factor
Loading
Composite
Reliability
AVE
Correlation Between
Constructs
PE
0.929
0.767
PE EE = 0.79
PE SI = 0.74
PE FC = 0.64
PE BI = 0.68
PE USE = 0.51
EE SI = 0.68
EE FC = 0.85
EE BI = 0.72
EE USE = 0.54
SI FC = 0.56
SI BI = 0.62
SI USE = 0.49
FC BI = 0.72
FC USE = 0.56
BI USE = 0.63
Q#14_PE_1
0.81**
Q#15_PE_2
0.86**
Q#16_PE_3
0.92**
Q#17_PE_4
0.91**
EE
0.930
0.771
Q#18_EE_1
0.78**
Q#19_EE_2
0.89**
Q#20_EE_3
0.92**
Q#21_EE_4
0.91**
SI
0.881
0.671
Q#22_SI_1
0.93**
Q#23_SI_2
0.87**
Q#24_SI_3
0.92**
Q#25_SI_4
0.34**
FC
0.567
0.332
Q#26_FC_1
0.81**
Q#27_FC_2
0.89**
Q#28_FC_3
-0.54*
Q#29_FC_4
0.46**
Q#30_FC_5
-0.14*
BI
0.968
0.910
Q#31_BI_1
0.96**
Q#32_BI_2
0.97**
Q#33_BI_3
0.94**
USE
0.889
0.728
Q#11_USE_1
0.95**
Q#12_USE_2
0.82**
Q#13_USE_3
0.78**
χ² = 451.489, df = 215, χ²/df = 2.100, GFI = 0.822, TLI = 0.920, CFI = 0.932, IFI = 0.933,
RMSEA = 0.081
Note: *not significant, **p < 0.000.
119
Measurement Model Refinement Procedure
Based on measurement model results, a researcher may respecify an originally
hypothesized model if the model provided information related both to model fit and
possible areas of model misspecification. Model refinement is a method known as a
model correction or model updating. According to Minas and Inman (1990), the goal of
this method is to improve the initial model to better fit the study data by following some
guidelines.
For model refinement, inadequate fit of the hypothesized model to the sample
data must be present and at least two parameters in the model must be misspecified
(Byrne, 2010). Model respecification must be supported by strong substantive and/or
empirical rationale (Byrne, 2010; Jöreskog, 1993). Based on Byrne (2010), given
apparent item content overlap and replication of same error covariance, respecification is
of the initial UTAUT model is justified.
Therefore, based on the estimates and modification indices, focus was placed on
the items with low correlation indices (factor loadings) and covariates of those errors
with high estimation index among the same construct respectively. The first step was to
verify the items with low factor loadings (low correlations). Table 50 shows the items to
take out of the model (Q#28_FC_3: it is part of the UTAUT model initially and
Q#30_FC_5: it is not part of the UTAUT model initially). These items are related to be
incorporated for indications of the researchers.
120
Table 50 Items to be removed from the initial model
Items removed
Factor Lodging
p-value
Q#28_FC_3
Social Media is not compatible with
other ways that I promote my
business.
-0.54
.506
Q#30_FC_5
I feel that internet connectivity,
access, or cost limit my ability to
promote my business using Social
Media.
-0.14
.864
Note: Extracted from the scalar estimates (Group number 1 – Default Model) – maximum
likelihood estimates - regression weights outputs.
On the other hand, the respecification of error variances are presented in Table 51
applied for the Model A fixed using AMOS software. The modification indices (MI)
reveal some evidence of misfit in the model (Byrne, 2010). In other words, if a parameter
with a high MI was to be freely estimated in a subsequent model, the overall χ² value
would drop by at least this MI’s amount. The parameter change value (Par Change)
represents the approximate value that the newly estimated parameter would assume
(Byrne, 2010). In summary, the flow of interest for a researcher to adjust modification
indices is from right to left on the outputs.
Table 51 Selected AMOS output for Model A: modification indices
Construct
Covariance
M.I.
Par Change
PE
err3 err1
5.450
-0.087
err2 err1
11.828
0.159
err4 err2
5.392
-0.078
EE
err6 err5
7.564
-0.103
err7 err5
9.032
-0.104
err7 err6
33.140
0.156
err8 err6
4.457
0.068
err8 err5
5.616
-0.089
Note: Extracted from the modification indices (Group number 1 – Default Model) –
covariance outputs. MI = modification indices. Par Change = parameter change.
121
Figure 8 shows Model A adjusted with refinement procedures for a new model fit.
The corrected Model A presents an acceptable level of fit (χ² = 260.438, df = 166, χ²/df =
1.569, GFI = 0.880, TLI = 0.965, CFI = 0.972, IFI = 0.973, and RMSEA = 0.058). All
factors had significant loadings greater than 0.50 (p < 0.001) on each construct. In brief,
the corrected Model A is ready to be tested as the UTAUT model in the present study.
Figure 8 Corrected Model A
122
Structural Model Assessment
The next step is to assess the structural model in order to test the hypotheses of
the theoretical model and the relationships between constructs. This step involves
specifying the structural model by assigning relationships from one construct to another
based on the proposed theoretical model (Unified Theory of Acceptance and Use of
Technology - UTAUT). This analysis addresses research questions #1 through #4. The
goal in the process is that the researcher may identify the dependence relationships that
are hypothesized to exist among the constructs and each hypothesis represents a specific
relationship that must be specified (Byrne, 2010). The structural model differs from the
measurement model in that the emphasis moves from the relationships between
constructs and measured variables to the importance and significance of the relationships
between constructs (Hair et al., 2009). The structural model was designed in AMOS
Version 24.0 replacing the relationships between constructs (all double-headed arrows)
with single-headed (causal) arrows. Figure 9 shows the UTAUT conceptual model
incorporating the factor structures and the hypothesized relationships. The structural
model assessment includes an analysis of model fit indices and the standardized path
coefficients to explore which hypothesized relationships are supported and which are not.
For the hypothesized relationships to be supported, the standardized path coefficients are
required to be greater than 0.30 and significant at the p > 0.05 level to be considered
meaningful (Byrne, 2010).
123
Figure 9 The structural model
Structural Model Results
Overall, the model showed good fit indices: χ² = 272.443, df = 170, χ²/df = 1.603,
GFI = 0.875, TLI = 0.963, CFI = 0.970, IFI = 0.970, RMSEA = 0.060. Findings show
that five of five path coefficients (hypotheses) were statistically significant and were
considered meaningful (ranging from 0.22 to 0.59). A theoretical model is considered
valid to the extent that the parameter estimates are statistically significant and in the
predicted direction (Hair et al., 2009). Findings revel that PE construct did not
significantly predict BI (0.02, n.s.); thus, H1 was not supported. Second, EE positively
predicted the BI construct (0.59, p < 0.001); thus, H2 was supported. Third, SI positively
predicted the BI construct (0.22, p < 0.001); thus, H3 was supported. Fourth, FC
positively predicted the social media use (USE) construct (0.24, p < 0.05); thus, H4 was
supported. Finally, BI positively predicted the USE construct (0.48, p < 0.001); thus, H8
was supported. In Table 52 and Figure 10, the assessment of the model fit and results are
summarized and illustrated.
124
Table 52 Structural model results
Path (Hypothesis)
Standardized Path
Coefficients (Beta)
Hypothesis Testing
Result
PE BI (H1)
0.017 n.s.
Not supported
EE BI (H2)
0.585***
Supported
SI BI (H3)
0.219*
Supported
FC USE (H4)
0.241*
Supported
BI USE (H8)
0.475***
Supported
Note (1): Model fit indices: χ² = 272.443, df = 170, χ²/df = 1.603, GFI = 0.875, TLI =
0.963, CFI = 0.970, IFI = 0.970, RMSEA = 0.060.
Note (2): *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
Figure 10 Initial structural model a with standardized path coefficients
Note: *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
The Effects of Moderators
The present study also assessed how the moderators of gender, age, and
experience influence BI to use or actual USE of social media. Figure 11 explains the
structure of each moderator in the model. According to Baron and Kenny (1986), a
moderator is a qualitative (e.g., sex, race, class) variable that affects the direction and/or
strength of the relation between an independent or predictor variable and a dependent or
125
criterion variable. A technique called multiple-group analysis can be used to evaluate
how moderators affect other variables. In the present study, variables of gender, age, and
social media experience were examined to determine their influence on other variables
and relationships in the UTAUT model.
Figure 11 Moderators structure in the model
Note: Elaborated by the author
Gender Moderator Results – Multi-group Analysis
First, research question #5 is focused on how gender moderates the connections
from PE, EE, and SI to BI of farmers and vendors to use social media for promoting their
businesses. This analysis tested three hypotheses: H5a, H5b, and H5c. Results are
presented in Tables 53 and 54 and Figures 12 and 13. Changes in Chi-square from the
baseline to the constrained model were not statistically significant (∆χ² (17) = 25.822, p =
0.078). Therefore, not all paths were invariant across gender. In other words, groups
126
(female and male) were different at the model level. Specific hypotheses with the gender
moderator were tested as follows:
H5a: Gender has a moderating effect on the relationships between PE and
farmers and vendors’ BI to use social media for marketing their small farm businesses,
with females being more likely to intend to use social media.
Analysis revealed that the relationship between PE and BI of farmers and vendors
to use social media for marketing their businesses varied significantly across males and
females (∆χ² (1) = 4.428, p < 0.05). In other words, groups (female and male) were
different. An examination of the slopes indicated that the relationship between PE and BI
was stronger in the female group (β = 0.54, p < 0.05) than in the male group (β = -0.02,
not significant). Thus, this hypothesis was supported.
H5b: Gender has a moderating effect on the relationships between EE and
farmers and vendors’ BI to use social media for marketing their small farm businesses,
with females being more likely to intend to use social media.
Analysis revealed that the relationship between EE and BI of farmers to use social
media for marketing their businesses was not significant across male and female groups
(∆χ² (1) = 2.147, not significant). In other words, groups (female and male) were not
different so slopes were not further examined. Thus, this hypothesis was not supported.
H5c: Gender has a moderating effect on the relationships between SI and
farmers and vendors’ BI to use social media for marketing their small farm businesses,
with females being more likely to intend to use social media.
Analysis revealed that the relationship between SI and BI of farmers to use social
media for marketing their business was significant across male and female groups (∆χ²
127
(1) = 6.014, p < 0.05). In other words, groups (female and male) were different.
Examination of slopes indicated that the relationship between SI and BI was stronger in
the male group (β = 0.44, p = 0.001) than the female group (β = 0.08, not significant).
Thus, because the relationship was stronger for males, this hypothesis was not supported.
Table 53 Multi-group analysis by gender groups
Model
χ²
df
CFI
∆χ²
∆df
Females only
178.940
120
0.970
Males only
157.177
120
0.969
Unconstrained model (baseline)
336.132
240
0.970
Fully constrained model
361.953
257
0.967
25.822*
17
PE to BI path constrained
340.560
241
0.969
4.428*
1
EE to BI path constrained
338.279
241
0.969
2.147 n.s.
1
SI to BI path constrained
342.145
241
0.968
6.014*
1
Note: *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
Model Fit Indices: χ² = 336.132, df = 240, χ²/df = 1.40, GFI = 0.826, TLI = 0.961, CFI =
0.967, IFI = 0.967, RMSEA = 0.049.
Table 54 Multi-group analysis for gender groups by pathways (standardized
estimates)
Pathways
Predictor Outcome
Female
Male
Std. β
Std. β
PE BI
.537**
-.021 n.s.
EE BI
.195 n.s.
.391*
SI BI
.077 n.s.
.442***
BI USE
.685***
.547***
Note: *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
128
Figure 12 Standardized coefficients for female group
Note: *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
Figure 13 Standardized coefficients for male group
Note: *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
Age Moderator Results – Multi-group Analysis
Research question #6 focused on how age moderates the connections from PE,
EE, SI, and FC to BI of farmers to use social media for promoting their small farm
businesses. The multi-group analysis to evaluate age as a moderator divided respondents
in two groups: young and old (Yarlagadda, Murthy, & Krishna Prasad, 2015).
Respondents between 18 and 44 years old were categorized as young farmers, and those
between 45 and 75 or older were categorized as old farmers. This analysis tested four
129
moderating hypotheses: H6a, H6b, H6c, and H6d. Results are presented in Table 55 and
56 and Figures 14 and 15. Changes in Chi-square from the baseline to the constrained
model were not statistically significant (∆χ² (20) = 32.461, p = 0.039). Therefore, not all
paths were invariant across age. In other words, age groups (young and old farmers) were
different at the model level. Specific hypotheses with the age moderator were tested as
follows:
H6a: Age has a moderating effect on the relationship between PE and farmers
and vendors’ BI to use social media for marketing their small farm businesses, with
younger farmers being more likely to intend to use social media.
Analysis revealed that the relationship between PE and BI of farmers to use social
media for marketing their businesses was not significant across younger and older
farmers (∆χ² (1) = 2.557, p < 0.110). In other words, age groups (younger and older
farmers) were not different, so slopes were not further examined. Thus, this hypothesis
was not supported.
H6b: Age has a moderating effect on the relationship between EE and farmers
and vendors’ BI to use social media for marketing their small farm businesses, with
younger farmers being more likely to intend to use social media.
Analysis revealed that the relationship between EE and BI of farmers to use social
media for marketing their businesses varied significantly across younger and older farmer
groups (∆χ² (1) = 6.465, p < 0.05). In other words, age groups (younger and older
farmers) were different. The relationship between EE and BI was stronger in the older
farmers group (β = 0.78, p < 0.000) than it was in the younger farmers group (β = 0.09,
130
not significant). Thus, because the relationship was stronger for older farmers, this
hypothesis was not supported.
H6c: Age has a moderating effect on the relationship between SI and farmers
and vendors’ BI to use social media for marketing their small farm businesses, with
younger farmers being more likely to intend to use social media.
Analysis revealed that the relationship between SI and BI of farmers to use social
media for marketing their businesses was not significant across younger and older farmer
groups (∆χ² (1) = 0.730, p = 0.393). In other words, age groups (younger and older
farmers) were not different, so slopes were not further examined. Thus, this hypothesis
was not supported.
H6d: Age has a moderating effect on the relationship between FC and farmers
and vendors’ social media USE for marketing their small farm businesses, with younger
farmers being more likely to actually use social media.
Analysis revealed that the relationship between FC and USE was not significant
across younger and older farmers groups (∆χ² (1) = 0.256, p = 0.613). In other words,
groups (younger and older farmers) were not different, so slopes were not further
examined. Thus, this hypothesis was not supported.
131
Table 55 Multi-group analysis by age groups
Model
χ²
df
CFI
∆χ²
∆df
Young F&V only
325.897
170
0.835
Old F&V only
242.435
170
0.970
Unconstrained model (baseline)
569.212
340
0.931
Fully constrained model
601.673
360
0.928
32.461*
20
PE to BI path constrained
571.769
341
0.931
2.557 n.s.
1
EE to BI path constrained
575.677
341
0.930
6.465**
1
SI to BI path constrained
569.942
341
0.931
0.730 n.s.
1
FC to USE path constrained
569.468
341
0.932
0.256 n.s.
1
Note: *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
Model Fit Indices: χ² = 569.212, df = 340, χ²/df = 1.67, GFI = 0.789, TLI = 0.915, CFI =
0.931, IFI = 0.933, RMSEA = 0.064.
Table 56 Multi-group analysis for age groups by pathways (standardized estimates)
Pathways
Predictor Outcome
Young F&V
Old F&V
Std. β
Std. β
PE BI
.239 n.s.
-.173 n.s.
EE BI
.085 n.s.
.783***
SI BI
.406**
.210*
FC USE
.133 n.s.
.321*
BI USE
.598***
.371**
Note: *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
132
Figure 14 Standardized coefficients for younger group
Note: *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
Figure 15 Standardized coefficients for the older group
Note: *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
133
Social Media Experience Moderator Results – Multi-group Analysis
Finally, research question #7 focused on how social media experience moderated
the connections from EE and SI to BI of farmers to use social media for promoting their
small farm businesses, as well as how experience moderated the relationship from FC to
farmers and vendors’ actual social media use for marketing their small farm businesses.
For this moderator, the analysis was evaluated in the social media experience for personal
(Per.) and business (Bus.) purposes.
Respondents were classified into one of two groups based on their reported social
media experience for personal purposes. Those who reported “no experience,” “very
poor,” “poor,” and “moderate” were categorized into a low experience group.
Respondents who reported “good” and “very good” were categorized into a high
experience group. This analysis tested three hypotheses: H7a, H7b, and H7c. The results
are presented in Tables 57 and 58 and Figures 16 and 17. The change in Chi-square from
the baseline to the constrained model was not statistically significant (∆χ² (16) = 40.041,
p < 0.001). Therefore, not all paths were invariant across social media experience for
personal purposes. In other words, groups (low and high experience) were different at the
model level. The hypotheses for the personal purposes social media experience moderator
were tested as follows:
H7a(Per.): Experience in the use of social media has a moderating effect on the
relationship between EE and farmers and vendors’ BI to use social media for marketing
their small farm businesses, with those who have had a positive (high) experience being
more likely to intend to use social media.
134
The analysis revealed that the relationship between EE and BI was not significant
across farmers and vendors with high and low experience (∆χ² (1) = 2.043, p = 0.153). In
other words, groups (high and low experience) are not different, so slopes were not
further examined. Thus, this hypothesis was not supported.
H7b(Per.): Experience in the use of social media has a moderating effect on the
relationship between SI and farmers and vendors’ BI to use social media for marketing
their small farm businesses, with those who have had a positive (high) experience being
more likely to intend to use social media.
The analysis revealed that the relationship between SI and BI was not significant
across farmers and vendors with low and high experience (∆χ² (1) = 0.480, p = 0.488). In
other words, groups (high and low experience) were not different, so slopes were not
further examined. Thus, this hypothesis was not supported.
H7c(Per.): Experience in the use of social media has a moderating effect on
the relationship between FC and farmers and vendors’ social media USE for marketing
their small farm businesses, with those who have had a positive (high) experience being
more likely to actually use social media.
The analysis revealed that the relationship between FC and USE was not
significant for farmers and vendors with low and high experience (∆χ² (1) = 0.001, p =
0.981). In other words, groups (high and low experience) were not different, so slopes
were not further examined. Thus, this hypothesis was not supported.
135
Table 57 Multi-group analysis by social media experience groups (personal purposes)
Model
χ²
df
CFI
∆χ²
∆df
High-experience only (Per.)
166.005
107
0.951
Low-experience only (Per.)
138.691
107
0.969
Unconstrained model (baseline)
304.750
214
0.959
Fully constrained model
344.791
230
0.948
40.041****
16
EE to BI path constrained
306.793
215
0.959
2.043 n.s.
1
SI to BI path constrained
305.230
215
0.959
0.408 n.s.
1
FC to USE path constrained
304.750
215
0.960
0.001 n.s.
1
Note: *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
Model Fit Indices: χ² = 304.750, df = 214, χ²/df = 1.424, GFI = 0.836, TLI = 0.948, CFI =
0.959, IFI = 0.960, RMSEA = 0.050.
Table 58 Multi-group analysis social media experience by pathways standardized
estimates (personal purposes)
Pathways
Predictor Outcome
High-exp.
Low-exp.
Std. β
Std. β
EE BI
.430****
.666****
SI BI
.282**
.140 n.s.
FC USE
.068 n.s.
.083 n.s.
BI USE
.691****
.356**
Note: *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
Figure 16 Standardized coefficients for high-experience group (personal purposes)
Note: *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
136
Figure 17 Standardized coefficients for low-experience group (personal purposes)
Note: *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
Respondents were classified into one of two groups based on their reported use of
social media for business purposes. Those who reported “no experience,” “very poor,”
“poor,” and “moderate” were categorized into a low experience group. Respondents
who reported “good” and “very good” were categorized into a high experience group.
This analysis is performed by testing three moderating hypotheses which are: H7a, H7b,
and H7c. Results are presented in Tables 59 and 60 and Figures 18 and 19. Change in
Chi-square from the baseline to the constrained model was not statistically significant
(∆χ² (16) = 59.440, p < 0.000). Therefore, not all paths were invariant across social media
experience for business purposes. In other words, groups (low and high experience) were
different at the model level. The hypotheses for the social media experience moderator
for business purposes were tested as follows:
H7a(Bus.): Experience in the use of social media has a moderating effect on the
relationships between EE and farmers and vendors’ BI to use social media for marketing
137
their small farm businesses, with those who have had a positive (high) experience being
more likely to intend to use social media.
Analysis revealed that the relationship between EE and BI was not significant for
those with high and low experience (∆χ² (1) = 0.379, p = 0.538). In other words, groups
(high and low experience) were not different, so slopes were not further examined. Thus,
this hypothesis was not supported.
H7b(Bus.): Experience in the use of social media has a moderating effect on the
relationship between SI and farmers and vendors’ BI to use social media for marketing
their small farm businesses, with those who have had a positive (high) experience being
more likely to intend to use social media.
Analysis revealed that the relationship between SI and BI was significant with 90
percent confidence across low and high experience groups (∆χ² (1) = 2.878, p < 0.100).
In other words, groups (high and low experience) were different. An examination of
slopes indicated that the relationship between SI and BI was stronger in the low
experience group (β = 0.520, p < .050) than the high experience group (β = -0.051, not
significant). Thus, since the relationship was stronger in the low experience group, this
hypothesis was not supported.
H7c(Bus.): Experience in the use of social media has a moderating effect on the
relationship between FC and farmers and vendors’ social media USE for marketing their
small farm businesses, with those who have had a positive (high) experience being more
likely to actually use social media.
Analysis revealed that the relationship between FC and USE was significant with
95 percent confidence across farmers and vendors with low and high experience (∆χ² (1)
138
= 4.280, p < 0.050). In other words, groups (high and low experience) were different.
Examination of slopes indicated that the relationship between FC and USE was stronger
in farmers and vendors with high experience (β = 0.487, p < 0.000) than it was in farmers
and vendors with low experience (β = -0.26, not significant). Thus, this hypothesis was
supported.
Table 59 Multi-group analysis by social media experience groups (business purposes)
Model
χ²
df
CFI
∆χ²
∆df
High-experience only (Bus.)
146.590
107
0.944
Low-experience only (Bus.)
145.632
107
0.970
Unconstrained model (baseline)
292.233
214
0.961
Fully constrained model
351.673
230
0.939
59.440****
16
EE to BI path constrained
292.612
215
0.961
0.379 n.s.
1
SI to BI path constrained
295.111
215
0.960
2.878*
1
FC to USE path constrained
296.512
215
0.959
4.280**
1
Note: *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
Model Fit Indices: χ² = 292.233, df = 214, χ²/df = 1.366, GFI = 0.841, TLI = 0.950, CFI =
0.961, IFI = 0.962, RMSEA = 0.047.
Table 60 Multi-group analysis social media experience by pathways standardized
estimates (business purposes)
Pathways
Predictor Outcome
High-exp.
Low-exp.
Std. β
Std. β
EE BI
.589***
.520****
SI BI
-.051 n.s.
.256**
FC USE
.487****
.083 n.s.
BI USE
-.054 n.s.
-.026 n.s.
Note: *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
139
Figure 18 Standardized coefficients for high-experience group (business purposes)
Note: *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
Figure 19 Standardized coefficients for low-experience group (business purposes)
Note: *p < 0.100, ** p < 0.05, *** p < 0.01, **** p < 0.001, n.s. = not significant.
Data Analysis Summary
This chapter presented results of the analysis for applying the UTAUT model to
the use of social media by North Mississippi farmers for marketing their small farm
businesses. The first step was to present an introduction of the UTAUT model and the
methods to evaluate the model (CFA, SEM, and EFA). Then, a data examination
described missing data, univariate normality, and outliers. Descriptive statistics were
140
presented for survey items. The measurement scale analysis was presented to demonstrate
reliability (incorporating internal consistency and items correlations methods) and
validity (EFA) of each UTAUT construct. A CFA evaluated outcomes in the model. An
assessment of construct validity demonstrated convergent and discriminant validity. The
SEM tested and revised the measurement model and produced the final structural model.
Finally, moderators (gender, age, and social media experience) in the UTAUT model
were tested using multi-group analysis. Table 61 summarizes results of the hypothesis
testing and Table 62 summarizes the moderator relationships for each construct.
Table 61 Summary of the hypotheses
Hypothesis
Affecting Construct
Affected
Construct
Moderator
Hypothesis
testing result
H1
PE
BI
Not Supported
H2
EE
Supported
H3
SI
Supported
H4
FC
USE
Supported
H5a
PE
BI
Gender
Supported
H5b
EE
Not Supported
H5c
SI
Not Supported
H6a
PE
BI
Age
Not Supported
H6b
EE
Not Supported
H6c
SI
Not Supported
H6d
FC
USE
Not Supported
H7a
EE
BI
Experience
(Per.)
Not Supported
H7b
SI
Not Supported
H7c
FC
USE
Not Supported
H7a
EE
BI
Experience
(Bus.)
Not Supported
H7b
SI
Not Supported
H7c
FC
USE
Supported
H8
BI
USE
Supported
141
Table 62 Summary of moderator relationships on each construct
No Hypo.
Ind. Var.
Dep. Var.
Moderator
Explanation
H5a
H6a
PE
BI
Gender &
Age
Effect stronger for females
H5b
H6b
H7a
EE
BI
Gender,
Age, &
Experience
Effect stronger for old farmers and
vendors
H5c
H6c
H7b
SI
BI
Gender,
Age, &
Experience
Effect stronger for male farmers
and vendors and low experience in the
social media use
H6d
H7c
FC
USE
Age &
Experience
Effect stronger for high experience
in the use social media
142
CHAPTER V
DISCUSSION, CONCLUSIONS, AND RECOMMENDATIONS
Introduction
The chapter contains a discussion of answers to the research questions related to
the UTAUT model and to farmers’ preferences for different marketing channels. The
discussion is organized based on the study’s research questions. Further discussion and
conclusions related to UTAUT are included. Then, the contribution of the study’s
findings to the body of knowledge, limitations, and recommendations for future research
are presented.
Answering the Research Questions
The purpose of the study was to determine factors that affect the acceptance and
use of social media platforms to advertise and promote agriproducts in farmers’
communities among FMs, CSAs, and on-farm stores in North Mississippi based on
concepts proposed in the UTAUT model. A questionnaire was completed by 169
participants who were farmers and vendors at FMs, CSAs, and on-farm stores in North
Mississippi. Research questions 1 through 5 and their related hypotheses were based on
the UTAUT (Venkatesh et al., 2003) and are discussed below.
143
Research Question 1: How does PE affect BI of farmers and vendors to use social
media for marketing their small farm businesses?
Hypothesis 1 (H1): PE has a positive effect on farmers and vendors’ BI to use
social media for marketing their small farm businesses.
Discussion: PE is the degree to which an individual believes that using social
media will help him or her to attain gains in job performance. According to the theory,
the PE construct is the strongest predictor of intention and remains significant at all
points of measurement (Venkatesh et al., 2003). Some studies have supported this
strength of PE in terms of acceptance of social media platforms by followers and users
(Salim, 2012) and in business (Mandal & Mcqueen, 2012).
However, in the present study, the PE construct did not significantly predict BI of
the use social media by farmers and vendors for marketing their small farm businesses.
This finding indicates that social media is not considered by farmers and vendors of FMs
to be a useful tool for promoting their small farm businesses. It could be that farmers and
vendors at FMs and CSAs in North Mississippi are not aware of the benefits of social
media platforms as a tool to promote their small farm businesses. If an individual is
unaware of social media’s benefits, it would not be surprising that they would not
perceive social media as being able to improve their job performance or enhance their
small farm business. However, from a theoretical point of view, those two constructs can
be moderated by gender and age (Venkatesh et al., 2003). A discussion of how
moderators impacted the present study’s findings is presented with research questions 5
through 7 below. In conclusion, farmers and vendors at FMs in North Mississippi who
participated in this study believe that social media will not be a beneficial tool for
marketing their small farm business and products.
144
Research Question 2: How does EE affect BI of farmers and vendors to use social
media for marketing their small farm businesses?
Hypothesis 2 (H2): EE has a positive effect on farmers and vendors’ BI to use
social media for marketing their small farm businesses.
Discussion: EE is the degree of ease associated with the use of social media.
According to UTAUT, EE is a strong determinant of individual intention (Venkatesh et
al., 2003), yet study findings have been mixed. Some studies support this idea, such as
that reported by Mandal and Mcqueen (2012) related to the use of social media adoption
by microbusinesses, while others do not, such as those that studied social networking site
adoption by young people in Africa (Kaba & Touré, 2014) or mobile banking services for
banks (Baptista & Oliveira, 2015).
In the present study, EE positively predicted the BI of farmers and vendors to use
social media for marketing their small farm businesses. From a theoretical point of view,
this construct can also moderated by gender, age, and experience (Venkatesh et al.,
2003); this will be discussed with research questions 5 through 7 below. In conclusion,
farmers and vendors at FMs in North Mississippi who participated in this study think that
social media is an easy-to-use tool for marketing their small farm business and products.
Research Question 3: How does SI affect BI of farmers and vendors to use social
media for marketing their small farm businesses?
Hypothesis 3 (H3): SI has a positive effect on farmers and vendors’ BI to use
social media for marketing their small farm businesses.
Discussion: SI is the degree to which an individual perceives that others who are
important to him or her believe he or she should use social media. According to the
theory, SI in voluntary contexts operates by influencing perceptions about the
145
technology—the mechanisms at play here are internalization and identification
(Venkatesh et al., 2003). However, the role of SI in technology acceptance decisions is
complex and subject to a wide range of influences. Again, previous studies have mixed
results. For example, this idea was not supported in a study related to the use of social
media for a social networking site adoption by young people in Africa (Kaba & Touré,
2014) but was moderately supported in a study of social media adoption by
microbusinesses (Mandal & Mcqueen, 2012).
In the present study, SI did affect BI of farmers and vendors to use social media
for marketing their small farm businesses. The UTAUT model suggests that SI has an
impact on individual behavior through three mechanisms: compliance, internalization,
and identification (Venkatesh & Davis, 2000). Internalization and identification alter an
individual’s belief structure and/or cause an individual to respond to potential social
status gains, while compliance leads an individual to simply alter his or her intention in
response to the social pressure generated, such that the individual intends to comply with
the SI (Venkatesh et al., 2003). In conclusion, farmers and vendors at FMs in North
Mississippi who participated in this study perceive it is important to use social media if
others believe they should use social media for marketing their small farm business and
products.
Research Question 4: How do FC affect the farmers and vendors’ USE for
marketing their small farm businesses?
Hypothesis 4 (H4): FC have a positive effect on farmers and vendors’ USE for
marketing their small farm businesses.
146
Discussion: FC refer to the degree to which an individual believes that his and her
business (as an organization) and technical infrastructure (service provider) exists to
support use of social media platforms. According to UTAUT, when both PE constructs
and EE constructs are present, FC become not significant in predicting intention;
however, empirical results indicate that FC do have a direct influence on use beyond that
explained by BI alone (Venkatesh et al., 2003).
Additionally, when moderated by experience and age, FC can have a significant
influence on USE (Venkatesh & Morris, 2000). As with the other constructs’ influence
on BI and USE, study results related to the relationship between FC and BI is mixed.
Dwivedi et al.’s (2011) work did not support this relationship; thus, they argued that it
needs further research consideration. However, other studies did find the relationship,
such as Mandal and Mcqueen’s (2012) examination of the decision to adopt social media
for microbusiness (Mandal & Mcqueen, 2012). A discussion of how moderators impacted
findings is presented with research questions 5 through 7 below.
In the present study, FC positively predicted the social media use (USE)
construct; thus, farmers and vendors at FMs in North Mississippi who participated in this
study believe that resources and support are available for using social media to market
their small farm business and products.
Research Question 5: How does gender moderate the relationships from PE, EE,
and SI to farmers and vendors’ BI to use social media for marketing their small
farm businesses?
Hypothesis 5 (H5a-c): Gender has a moderating effect on the relationships
between PE, EE, and SI and farmers and vendors’ BI to use social media for marketing
their small farm businesses, with females being more likely to intend to use social media.
147
Discussion: Gender scheme theory suggests that gender differences stem from
gender roles and socialization processes reinforced from birth rather than being due only
to biological gender (Bem, 1981; Bem & Allen, 1974; Kirchmeyer & Bullin, 1997;
Lubinski et al., 1983; Lynott & McCandless, 2000; Motowidlo, 1982; Venkatesh et al.,
2003). Studies on gender differences indicated that men tend to be highly task-oriented
compared to women (Minton & Schneider, 1980; Venkatesh et al., 2003).
In the present study, gender was confirmed as a moderator variable in the PE and
BI relationship in the hypothesized direction. In other words, women were more likely to
intend to use social media for marketing their small farm business than men because they
believed that using social media would help them improve their job performance.
However, the relationship between EE and BI was not moderated by gender; thus, there
were no differences between males and females in how they perceived the degree of ease
of social media use. Finally, the relationship between SI and BI was moderated by
gender, with a stronger effect for men. This suggests that males were more likely than
females to indicate that peers would influence their intention to use social media for
marketing their small farm business.
Research Question 6: (a) How does age moderate the relationships from PE, EE,
and SI to farmers and vendors’ BI to use social media for marketing their small
farm businesses? (b) How does age moderate the relationship between FC and
farmers and vendors’ USE for marketing their small farm businesses?
Hypothesis 6 (H6a-d): Age has a moderating effect on the relationships between
PE, EE, and SI on farmers and vendors’ BI to use social media for marketing their small
farm businesses, as well as on the relationship between FC and farmers and vendors’
148
USE for marketing their small farm businesses, with younger farmers being more likely
to intend to use or actually use social media.
Discussion: Venkatesh and Morris (2000) reported that attitude was more salient
for younger workers while perceived behavioral control was more salient for older
workers. First, in the present study, analysis revealed that age did not moderate the
relationship between PE and BI; thus, there were no differences by age in the relationship
between beliefs that using social media would improve their job performance and
intention to use social media for marketing agribusinesses. However, age was a
moderator in the relationship between EE and BI of farmers and vendors to use social
media for marketing their businesses, but not in the anticipated direction. Old farmers and
vendors were more likely to report that the ease of use associated with social media
would influence their intention to use it for marketing. Third, the relationship between SI
and BI on farmers and vendors’ use of social media for marketing their businesses was
not moderated by age, indicating no differences in how peers influence intention to use
social media. Finally, the relationship between FC and farmers and vendors’ USE was
not moderated by age. In other words, there was no relationship between farmers and
vendors belief that an organizational and technical infrastructure exists to support their
use of social media and their actual use of it.
Research Question 7: (a) How does experience of social media use moderate the
relationships from EE and SI to farmers and vendors’ BI to use social media for
marketing their small farm businesses? (b) How does experience of social media use
moderate the relationship from FC to farmers and vendors’ USE for marketing
their small farm businesses?
Hypothesis 7 (H7a-c): Experience in the use of social media has a moderating
effect on the relationships between EE and SI on farmers and vendors’ BI to use social
149
media for marketing their small farm businesses, as well as on the relationships between
FC and farmers and vendors’ USE for marketing their small farm businesses, with those
who have had a positive experience being more likely to intend to use or actually use
social media.
Discussion: Empirical evidence has demonstrated that experience moderates the
relationship between subjective norms and BI, such that subjective norms become less
important with increasing levels of experience (Venkatesh & Morris, 2000). In other
words, if experience level increases, subjective norms decrease and become less salient in
influencing behavior. In the present study, experience was tested as a moderating variable
in the relationships between EE and BI, SI and BI, and FC and USE. Experience as a
moderator related to intent to use and actual use of social media for personal purposes
(Per.) as well as for business purposes (Bus.) were assessed.
(Per.): First, analysis revealed that the relationships between PE and BI and
between EE and BI were not moderated by the experience of using social media for
personal purposes. Additionally, the relationship between FC and USE was not
moderated by experience. These are contrary to what the UTAUT model proposed.
(Bus.): Analysis revealed that the relationship between EE and BI of farmers and
vendors to use social media for marketing their agribusinesses was not moderated by
experience using social media for business purposes. However, analysis indicated that
both the relationship between SI and BI and the relationship between FC and USE were
moderated by experience. According to Venkatesh et al. (2003), the effect of experience
moderator should be stronger particularly at early stages of experience between EE and
BI; it should be stronger particularly in mandatory settings in the early stages of
150
experience between SI and BI; and it should have a significant influence on FC and USE.
Interestingly, the moderating role was different in these two relationships. Experience as
a moderator in the SI – BI relationship was present for farmers and vendors who reported
low experience. On the other hand, in the FC – USE relationship, the moderator role was
present in farmer and vendors with high experience of using social media for business
purposes.
Research Question 8: How does BI affect farmers and vendors’ USE for
marketing their small farm businesses?
Hypothesis (H8): BI has a positive effect on farmers and vendors’ social media
USE for marketing their small farm businesses.
Discussion: According to Venkatesh et al. (2003), BI has a strong and significant
positive influence on technology usage. For the present study, result showed that BI
positively predicted USE; therefore, BI did have a significant positive influence on use of
social media platforms by farmers and vendors in North Mississippi to promote their
small farm business and products.
Overall Fit of the UTAUT Model in the Present Study
Venkatesh et al. (2003) identified four key factors and three moderators related to
predicting BI to use a technology and actual technology use primarily in organizational
contexts. According to UTAUT, PE, EE, and SI were theorized and found to influence BI
to use a technology, while BI and FC determined actual technology use. In the present
study, results provided strong support for two direct determinants of farmers and vendors
intention to use (EE and SI) social media to market their small farm business and two
direct determinants of usage behavior (BI and FC).
151
First, the degree to which farmers and vendors believe that using social media will
help them to attain gains in their business (PE) is minimal and does not affect their BI to
use social media for marketing their small farm businesses and products. Age did not
have a moderating effect on this relationship. However, as hypothesized, gender did
moderate the relationship, with female farmers and vendors being more likely to report
that using social media would enhance their businesses performance. Pew Research
Center has reported that 73 percent of U.S. females use at least one social media site,
compared to 65 percent of U.S. males (Pew Research Center, 2017). Additionally,
previous research (Minton & Schneider, 1980; Venkatesh et al., 2003) has indicated that
females are more likely to buy local food than males; thus, they may be more likely to see
the value in using social media to promote their businesses and agriproducts. It could be
concluded that among females, social media is perceived to be instrumental in achieving
valued outcomes, social media is perceived with a relative advantage, and social media
contains performance (business-related) and personal (individual goals) expectations.
Second, the degree of ease associated with the use of the social media by vendors
and farmers (EE) affects their BI to use social media for marketing their small farm
businesses and products. In essence, farmers and vendors believe that using social media
would be free of effort and relatively easy to understand and use. Age was a moderator in
this relationship, with old farmers and vendors reporting that social media was easy to
use and not a complex tool; however, this was opposite of the hypothesis indicating that
young farmers and vendors would be more likely to perceive social media as being easy
to use. Previous research has indicated that older adults are more likely to buy local food
than younger adults (Venkatesh & Morris, 2000); thus, older adults may be more likely to
152
see the value of using social media to promote their businesses and agriproducts, as long
as it is perceived as easy to use. However, this finding is still interesting in light of
research that indicates greater social media use by younger adults. Pew Research Center
(2018) reported that 88 percent of U.S. adults aged 18 to 29 years use at least one social
media site, compared to 64 percent of those age 50 to 64 years and 37 percent of those
age 65 and over.
Third, the degree to which farmers and vendors perceive that others who are
important to them believe they should use social media (SI) affected their BI to use social
media for marketing their small farm businesses and products. In essence, the farmers
and vendors' perceptions in terms of subjective norms, social factors, and business image
are important for marketing their small farm businesses and products. Moreover, gender
and experience using social media for business purposes moderated this relationship, with
males and those with less experience being more strongly influenced by how they believe
others will view them (positive or negative) as a result of having used the technology.
These results also contradicted the hypotheses for these moderators.
Fourth, the degree to which farmers and vendors believe that resources and
support exist for social media use (FC) affected farmers and vendors’ USE for marketing
their small farm businesses and products. In essence, the farmers and vendors reflected
perceptions of internal and external constraints on behavior, resource-FC, and
technology-FC in their USE. The experience using social media for business purposes did
moderate this relationship, with those having more experience reporting that FC
influenced USE. It is anticipated that this effect would increase with experience as users
of technology find multiple avenues for help and support when needed.
153
Research Question 9: What are the reasons (barriers) that prevent farmers and
vendors from selling their products through various marketing channels (FM, CSA,
retailer, roadside stand, website, and social media)?
When farmers and vendors were asked about the channels where they preferred to
market their small farm business and products, 98 percent indicated a preference for
selling their products through FMs, 47 percent through social media, and 34 percent
through retailers. Thus, farmers and vendors view FMs as a primary and effective
resource to commercialize their products and obtain profits. However, they are still
looking for alternative marketing channels focusing on digital channels such as social
media and websites.
Farmers and vendors in the present study reported barriers to selling their
products through different marketing channels (i.e., farmers market, retailer, CSA,
roadside stand, website, social media, and other channel). The most commonly
mentioned barriers related to FMs were a low flow of customers and limited hours of
customer service. These barriers help explain why farmers and vendors desire alternative
marketing channels using digital platforms (Low & Vogel, 2011). Although respondents
identified expensive fees for websites and social media as barriers to selling and
promoting their products, this could be due to the farmers and vendors’ lack of
knowledge about the advantages of social media use, its potential effectiveness for their
businesses, and its free or low cost use (Cha, 2009). In summary, farmers and vendors
have expressed barriers in different marketing channels for selling their products and
support identifying alternatives for promoting and selling their products using digital
platforms.
154
As a format to encourage farmers and vendors to promote and sell their products
through social media, programs that provide education on the benefits of social media use
in marketing as well as “how to” manage and implement marketing strategies using
digital platforms, like social media, could be implemented. However, 57 percent of
respondents did not indicate having interest in such a program. Although farmers and
vendors did not provide the reasons for their interest or lack thereof, the demographic
characteristics of respondents may help explain this finding. Increased age has been
shown to be associated with difficulty in processing complex stimuli and allocating
attention to information on the job (Plude & Hoyer, 1985; Venkatesh et al., 2003). Many
respondents (44.4 percent) were age 55 or older, and half reported their highest education
level as “some college” (32 percent) or less (18.4 percent). Research has indicated that
individuals with at least some college experience have been consistently more likely than
those with a high school degree or less to use social media over the past decade (Perrin,
2015). Knowing that nearly half of farmers and vendors at different FMs in North
Mississippi reported a willingness to participate and acquire knowledge through the
Mississippi State Extension is a good indicator of trust for building an Extension program
on these topics.
Theoretical and Methodological Contributions of the Present Study
This study makes an important theoretical contribution in terms of the UTAUT
model applications (Venkatesh et al., 2016). It is the first empirical study that applied the
UTAUT model to the social media adoption for promoting small farm business and
agricultural products using the model varied across a variety of settings (e.g., different
user types, different organization types, different technology types, different tasks and
155
different locations). Rural farmers’ communities, such as FMs, CSAs, roadside stands,
and on-farm stands were settings for the research rather than a more traditional
organizational setting. Additionally, social media as a digital platform (IT system) was
the tool studied related to its use for promoting and selling agricultural products from
small farm businesses. By visiting locations such as FMs in North Mississippi, the study
demonstrated different attitudes about the use social media by these farmers with various
demographic characteristics.
The study tested the UTAUT constructs and moderators (except for voluntariness)
delineated by Venkatesh et al. (2003). Given the internet connectivity issues in the state
of Mississippi, an item was incorporated into the UTAUT model related to availability
and cost of internet in respondents’ communities. Although the item was ultimately
excluded from the model, 29 percent of the respondents disagreed that the Internet
connectivity, access, or costs limited their ability to promote their business using social
media, suggested that this was not a potential barrier to social media use for many
respondents.
The UTAUT model has been expanded and refined over time (Venkatesh et al.,
2016). Since its original publication, the UTAUT model has served as a baseline model
and has been applied to the study of a variety of technologies in both organizational and
non-organizational settings (Venkatesh, Thong, & Xu, 2012). However, UTAUT
components and moderating variables have both been supported by and contradicted in
these studies. That is the case in the present study as well. This study tested the original
model in a new organizational scenario: farmers’ communities in rural areas. It is
possible that characteristics of this study scenario are different from previously studied
156
scenarios and that is why several of the moderating relationships were not supported as
hypothesized. Therefore, future research should include additional variables as
moderators that may be unique and more relevant to farmers’ communities or rural
settings when compared to larger urban businesses or organizations (e.g., size of
organization, frequency of occurrence, number of farmers and vendors, types of products
sold – agricultural produce vs. crafted items). Although Internet accessibility was
considered in this study with no effect, there may be other variables, such as culture,
adopters vs non-adopters, locations, IT service providers, user groups, and social media
platforms that could be important. Future research could also examine sales, ads vs.
profits, number of customers vs. profits, customer flows, adaptation services, and
economic development.
Practical Application
Findings from the present study also have a practical application. Information
about farmers and vendors’ perceptions and attitudes related to the use of social media to
promote and sell their agricultural products can be useful to organizations, such as MSU
Extension, for creating educational programs focused on marketing for FM managers,
CSA leaders, and farmers and vendors. Extension organizations may be particularly
appropriate for providing such programming. For instance, ukur, Karaturhan, and
Budak (2013) explained that agricultural Extension acts as a bridge between research
institutions and farmers, transfers farmers’ problems to the research institutions and
accomplishes an important service by transferring new technologies to the farmers.
Specifically related to Extension marketing programs, ukur et al. (2013) stated that such
programs should include the following topics: production planning, finding markets for
157
farmers, training and giving advice about important marketing practices, and establishing
and running rural markets. These topics are important to be diffused to farmers’
communities to support their economy. In a recent study, Balderrama, García, Fíoriega,
and Preciado Rodríguez (2012) identified the greatest effects of agribusiness programs
included improving product quality, improving production processes, accessing to new
markets, and increasing sales describing and evaluating the needs of the small
agricultural businesses.
Mississippi State University Extension has some existing programs focused on
marketing and social media. For instance, Bricks-to-Clicks (B2C) is an entrepreneurial
effort aimed at helping small business owners use social media, blogs, websites, and e-
commerce solutions to grow their businesses (Barnes & Coatney, 2015). Additionally,
B2C trains entrepreneurs to understand and implement best management practices when
using social media in business and in how using social media can translate into economic
value. B2C is focused on the participant’s experience in the use of social media and
specific implementation of Facebook marketing strategies designed to help each business
owner advertise and grow its business. Mississippi MarketMaker™ is an MSU Extension
database that serves as a social network to connect food-related businesses at all
components of the production and distribution chain (e.g., buyers, fisheries, farmers,
wineries, restaurants, FMs, processors) (Posadas, 2017).
However, more than half of the people who participated in the study did not wish
to participate in an Extension program focused on marketing. Unfortunately, no questions
were included to allow participants to indicate why they were not interested in such
programs. It is likely that reasons such as lack of time, lack of clarity regarding how
158
useful and relevant such information would be, and doubt about benefits of a marketing
program could be considered barriers to their attendance. Additionally, while Extension
is generally well-known for its work related to agriculture, its work related to marketing
may be less known. The results of this study provide information that can be used to
tailor marketing programs and services to different audiences. Therefore, MSU Extension
should further promote its marketing educational programs and services to expand its
clientele.
First of all, results of this study indicated that females were more likely than
males to believe that using social media platforms will help them to attain gains in their
small farm business. While other organizations (e.g., Small Business Administration, The
Women’s Business Enterprise National Council) provide education and networking
opportunities focused specifically on female-owned businesses, Extension could rely on
its strong relationships in rural areas to provide education to women with small farm
businesses on topics related to the characteristics of advertising design, marketing
content, customer service, promotion, and sharing strategies using social media. Since
females in this study already seem to understand the value of social media for their
businesses, a program targeted to this audience would not need spend time describing the
benefits of social media and could emphasize building relevant social media advertising
and marketing strategies and skills.
Second, results of this study indicated that old farmers and vendors were more
likely than young farmers and vendors to consider that social media was an easy-to-use
tool for their small farm business promotion. This finding is encouraging for the
development of Extension programs for this specific audience given that research shows
159
the percentage of U.S. adults who use at least one social media site decreases as age
increases (Pew Research Center, 2018). While it might have been assumed that older
farmers and vendors would need a program that emphasized basic skills for using social
media, this study suggests that an educational program may not have to devote time to
building skills and could move directly to social media advertising, marketing, and
promotion strategies. Incorporating success cases in the use of social media as examples
can demonstrate the effectiveness of the tool for promoting agriproducts to this audience.
Additionally, younger adults could serve as guest speakers in an educational program to
share suggestions for how to market small farm businesses to younger audiences who are
typically more skilled in social media use.
Other older adults may be curious about using social media as a marketing tool,
but they may not know how effective and functional it could be for their business. To
incorporate a social media mindset in older farmers and vendors who are not comfortable
with social media, young volunteers may provide a powerful and effective way to engage
these older adults. The best way to engage older adults unfamiliar with the use of social
media would be incorporating success stories as examples along with information about
the benefits of and “how-to” for using social media.
Third, the results indicate that males and farmers and vendors with low
experience in the use of social media for business purposes intention to use social media
were influenced by other people who were important to them. Based on Rogers (2003)
Diffusion of Innovation Theory, the findings of this study suggest that an Extension
program could be developed that is directed to early adopters of social media and early
majority groups, such as leaders of farmers’ communities (Rogers, 2003). CSA leaders
160
and FM managers should be considered as important diffusion points for disseminating
the idea of social media as a tool to promote agriproducts. When the researcher discussed
the study with FM managers, they expressed an interest in the study for the purpose of
identifying and addressing the needs of farmers and vendors at each FM. Therefore, FM
managers and CSA leaders’ involvement in structuring and promoting an Extension
program related to social media use by small farm businesses would be important.
Additionally, since the relationships between SI and BI to use social media and
FC and actual use were moderated by experience using social media for business
purposes, having individuals who have high experience share their stories during
educational programs for all audiences may have a positive effect on participants’
decision to adopt social media for marketing their small farm businesses and products.
Ultimately, a diverse offering of programs would be beneficial in helping farmers and
vendors develop and use new marketing strategies to support farmers’ communities in
Mississippi through economic development, as well as increase individual farmers and
vendors’ small business profits. It will be important to evaluate these programs to
determine their effectiveness and identify any modifications or additions to content that
may be needed.
Limitations
Three limitations were identified in the present study: sample as related to size,
participant characteristics, and nonresponse bias; locations related to timing and setting,
as well as weather conditions; and researcher characteristics in terms of language and
ethnicity. These limitations were considered in advance of data collection and efforts
were made to reduce their impacts on data collection and the findings of the study.
161
Sample
Sample size is a potential limitation of this study. While 169 farmers and vendors
completed the survey and only 150 (minimum) were required based on the number of
constructs included in the Structural Equations Modeling analysis, a larger number of
respondents may have resulted in stronger findings and more consistent findings. To
obtain a sample, contact was made with managers of 45 FMs and CSAs in North
Mississippi. However, only 16 responded to the initial contact, and 14 agreed to allow
their vendors to participate in the study. Thus, nonresponse bias likely occurred related to
the settings/locations in which the study was conducted. However, it was essential to
have such prior approval; without approval from the manager, farmers and vendors did
not trust and support the data collection process.
This nonresponse bias could also mean that the respondents may have differed
from one to another in unintended ways that are related to the variables studied (Fraenkel
et al., 2014). While potential respondent characteristics that could have affected the
results of the present study include age, attitude towards the completion of the survey,
and socioeconomic status, frequency distributions indicated a relatively proportional
distribution across the different categories of characteristics that were assessed. During
data collection, to minimize threats related to respondent characteristics, the researcher
explained the research project goals and expected results in front of the farmers and
vendors, was present when the questionnaires were completed to answer any questions
could be answered and listened to respondents comments beyond the questionnaire.
162
Locations and Weather Conditions
Location can be a threat whenever data are collected in different settings.
According to Fraenkel et al. (2014), this threat could be involved when the location in
which data are collected may create alternative explanations for results. Given that data
were collected on-site at outdoor FMs and CSAs, it was not uncommon to encounter
differences testing conditions. Additionally, the FMs and CSAs were located at distances
from the researcher’s home ranging from 20 to 160 miles, meaning that the researcher
only could visit one farmers market per week if it was open on Thursday, Tuesday,
and/or Saturday. The majority of the FMs’ open hours for customers on Saturdays were
strict and short (only in the mornings or afternoons). On Tuesdays and Thursdays, open
hours were often reduced by approximately 3 hours. This limited the amount of time
available to collect data in each of the different settings since data were collected face-to-
face. Respondents were most willing to complete the questionnaire before the FMs were
open to the public or after the FMs were closed. This enabled them to attend to the data
collection process while avoiding high density of customer flows during open hours.
Weather was also a limiting factor in this study. Summer 2018 was a rainy season,
and this affected some scheduled visits. The majority of the FMs visited were outdoors
and the weather affected the attendance of farmers and vendors and also the opening of
the market place itself. The researcher cancelled two scheduled visits and lost two
scheduled visits to FMs due to weather conditions. However, summer (May, June, and
July) was the busiest season for FMs and CSAs which made it the best time to conduct
the study to get the largest number of participating locations and farmers and vendors.
163
Researcher Characteristics
Researcher characteristics can affect the data collection process and the results in
terms of explanation, communication, reliability, and credibility. Gender, age, ethnicity,
and language patterns of the individual who collects the data are important to consider
(Fraenkel et al., 2014). During most data collection efforts, the researcher was
accompanied by the social leader of the CSA or cooperative and/or the manager of the
farmers market. Because this individual was immersed within the farmers’ community,
he or she was able to understand any challenges faced by the researcher in that setting
and provide any support needed for the data collection process.
Summary and Recommendations
This study examined factors that affect the acceptance and use of social media
platforms to advertise and promote agriproducts in farmers’ communities among FMs,
CSAs, and on-farm stores in North Mississippi using the UTAUT model proposed by
Venkatesh et al. (2003). UTAUT served as the framework for this study. Results from
169 respondents who completed questionnaires indicated that PE, EE, SI, and FC (key
constructs in UTAUT) did affect farmers and vendors’ BI to use social media for
marketing their small farm businesses and agriproducts. Some respondents’
characteristics (e.g., age, gender, social media experience) moderated these relationships
in different ways as described in Chapters 4 and 5. However, in general, the present study
provided additional empirical support for UTAUT. Additional results indicated that
farmers and vendors prefer to market their products through FMs and CSAs, but note
barriers present in different marketing channels. However, responses to questions that
164
assessed constructs in UTAUT suggest that respondents are open to social media as a
marketing tool for their small farm businesses.
Marketing, outreach, promotion, and social media are relevant topics to be
considered for future work. Many Extension programs exist across the U.S. on broad
topics such as marketing, promotion, entrepreneurship, business, food production, and
guidelines for selling food at FMs. However, results of the present study suggest that
owners of small farm businesses could benefit from training and programming in these
topics with the goal of helping such farmers and vendors attract more customers and
increase sales. Using digital technologies as tools (e.g., social media, broadband, Wi-Fi
spots, advertising platforms) to incorporate these topics into programming for rural
farmers’ communities can play a key role in connecting and facilitating communication
between customers and farmers. The challenge will be to find the best approach to
engage farmers and vendors so that they might adopt these technologies to support their
agribusinesses. Marketing and promoting innovations for FM vendors and managers
related to topics such as healthy food, fresh food, local economy, networking, local food
system, etc., could also be an important way to support these establishments around the
state.
One issue faced in Mississippi specifically is internet connectivity. In many
locations across the state, broadband accessibility is limited (Barnes et al., 2014).
Therefore, an item was included in the present study’s questionnaire to see if this was a
barrier for farmers and vendors related to marketing their small farm business.
Quantitative findings indicated that it was not an issue; however, during discussions with
farmers and vendors while data were being collected, they stated a need/desire to be
165
connected through a free Internet Wi-Fi or hotspot connection to promote their business
and the FMs as well. While some of the farmers and vendors were using paid Internet
through their own service provider, they described barriers such as Internet speed, the
need to be connected and have that fee included in the standard farmers market fee, and
limited support from the local chambers of commerce. Farmers and vendors are seeing
the benefits of connectivity and realize broadband has made it easier for people to
connect across the world and transfer information, farmers do perceive barriers to its use
and could benefit from additional information and support related to how they can use it
regularly to communicate about, promote, and market their small farm businesses and
agriproducts more easily and without much expense.
