1
Chapter 1: Introduction to the Study
Introduction
In the period 1950 to 2005, the population of the United States doubled, with the
southern and western regions experiencing the largest population growth (Kenny et al.,
2009). In California, the population grew 137% in the 50 years prior to the 2010 Census
(United States Census Bureau, 1960, 2010a). While population increase on its own is a
strain on a water supply, water pollution, climate change, and periodic drought have also
had a profound impact on California’s supplies (CDWR, 2009a; USEPA, 2002).
In the half-century prior to 2011, for over half of that period (29 years) California
received annual rainfall below the average of 21.85 inches (see Figure 1; NOAA, 2013).
Between 2007 and 2009, the driest period since California’s historic 1986-1992 drought,
the state averaged 15.39 inches of rain (CDWR, 2009a, 2010b; NOAA, 2013; Santa
Barbara County, 2009). In the midst of this 3-year drought, Governor Arnold
Schwarzenegger asked the California legislature to draft water conservation legislation
and, on November 10, 2009, he signed into law the Water Use Efficiency Senate Bill 7
(SBx7-7; California Senate, 2009; CDWR, 2009c).
An important component of SBx7-7 is the requirement that water purveyors in
California reduce their per capita water consumption by 20% by the year 2020 (CDWR,
2009c). According to Rogers (2009), “California now is the first state to set statewide
targets for water conservation. The law also is expected to push cities such as Fresno and
Sacramento—which still don't have water meters on all homes—to do more” (p. 2B).
This innovative legislation encourages the implementation of new water conservation
2
programs and inventive water saving solutions, if only for the purpose of complying with
the new law.
Figure 1. California, climate division 4, precipitation, January-December 1960-2011.
Reprinted from “Temp, Precip, and Drought Time Series” NOAA, 2013. Retrieved
from http://www.ncdc.noaa.gov/cag/time-series/us. Material is not subject to copyright
protection within the United States.
Most commonly, elements of water conservation programs fall into three
categories: restrictions, rates, and rebates. Water conservation restrictions provide
authority to the water purveyor to encourage consumers to reduce their water
consumption during droughts. Encouragement is attempted through the implementation
and enforcement of water budgets. A water budget is an estimate calculated for the
household so that the customer can achieve the most efficient use of water. Should the
household use more water during the billing period than is specified in their water
3
budget, the customer is billed at a punitively higher drought rate for the excess water. For
example, in 2012, customers of Moulton Niguel Water District in southern California
paid $1.38 per unit (100 cubic feet or 1 CCF or 748 gallons) for the amount of water
allotted to their indoor water budget, $1.54 per unit for the water allotted to their outdoor
water budget, and from $2.75 to $11.02 per unit for water used in excess of their water
budget (see Figure 2).
Generally, water conservation rates are either tiered water rates designed to
encourage water conservation year round or drought rates imposed only during declared
drought emergencies. Rather than set an individual water budget for every household, the
water purveyor sets its tiers based on their customers’ cumulative average usage or
industry standard tiers. For example, customers of Vandenberg Village Community
Services District (VVCSD), on California’s central coast, pay $1.25 per unit of water for
the first 10 units, $1.43 per unit of water for the next 7 units, $1.55 per unit of water for
the next 32 units, and $2.49 per unit of water for water usage above 49 units in a billing
cycle (see Figure 3; VVCSD, 2013).
Water conservation rebates encourage the replacement of high-water-use
features—such as toilets, washing machines, dishwashers, and grass—by providing cash
incentives. Landscape rebates, also known as Cash-4-Grass rebates, encourage customers
to reduce the amount of turf grass in their yard and thereby reduce outside watering. In
1994, the United States Environmental Protection Agency introduced the concept of
landscape rebates:
4
Landscape rebate programs pay customers to install low-water-use
landscaping or to convert all or part of their lawn to nonturf landscaping.
Participation in this type of program is predominantly by residential
customers. The amount of rebate paid in a landscape conversion program is
usually based on the amount of land converted to water-efficient landscape.
(Chapter 4, p. 6)
Figure 2. Residential water budget. Reprinted from “Understanding water-budget-
based rates”, by Moulton Niguel Water District, 2012. Retrieved from
http://www.mnwd.com/ customer-service/budget-based-rates.aspx. Copyright 2012 by
Moulton Niguel Water District. Reprinted with permission.
5
However, it was more than a decade before water purveyors embraced the
possibility of reducing water consumption by providing such rebates. In 2007, VVCSD
was the first agency in Santa Barbara County to offer landscape rebates (J. Barget,
personal communication, January 11, 2012). VVCSD provides rebates of $2.00 per
square foot (up to $1,000) to replace grass with low water usage plants, rocks, or
synthetic turf (VVCSD, 2012d).
Figure 3. Inclining block rate structure. Compiled from “Current residential water and
wastewater rates (effective 7/13)” VVCSD, 2013. Retrieved from http://vvcsd.org/
custserv/current.htm. Copyright 2013 by VVCSD. Reprinted with permission.
Developer in-lieu fees fund VVCSD’s water conservation program (VVCSD,
2007). In accordance with the program, developers must retrofit 10 existing homes for
every new home built. Rather than soliciting volunteers and physically retrofitting those
homes, developers pay an in-lieu fee to VVCSD which administers the water
conservation program and coordinates retrofits. Additionally, from 2010 through 2012,
VVCSD and other agencies in the county were awarded more than $160,000 in grant
funds from the United States Bureau of Reclamation to encourage and promote landscape
rebates (Santa Barbara County, 2010). VVCSD received almost $10,000 in
reimbursements (VVCSD, 2012b). As a result, in the period from the program’s
6
inception in November 2007 through November 2012, 46 rebates were issued for the
replacement of 57,570 square feet of lawn (VVCSD, 2012b).
This chapter provides background information on global and local water supply
and demand; it details threats to the water supply, such as pollution, population increase,
climate change, and drought; and it presents solutions to increase supply and reduce
demand, including water conservation. It also introduces the study area of Vandenberg
Village, California, presents the theoretical framework, and includes the problem
statement, purpose and nature of the study, research questions and hypotheses, definitions
for terms, and assumptions, limitations, and delimitations.
This study is an important addition to the body of work on water conservation
because of the current gap in the literature on landscape rebates. Attempting to control
water demand through pricing structures is a very common research topic as are other
types of elements of a water conservation program such as restrictions or indoor rebates
(Gober & Kirkwood, 2010; Grafton & Ward, 2008; Harlan, Yabiku, Larsen, & Brazel,
2009; Kenney, Goemans, Klein, Lowrey, & Reidy, 2008). However, although introduced
close to 2 decades ago, landscape rebates have only recently gained in popularity and
have not yet been targeted for extensive research. Of the 17 water purveyors in Santa
Barbara County, only 11 offer rebates as a part of their water conservation program and
only three of those offer landscape rebates (Santa Barbara County, 2013a). Two of those
programs were implemented in response to SBx7-7 (City of Lompoc, 2009; City of Santa
Barbara, 2009). Only VVCSD began offering landscape rebates before the regulatory
requirement to reduce water consumption was signed into law (VVCSD, 2007). As more
7
water purveyors explore water conservation options, it is anticipated that this study will
allow them to make an informed decision about innovative water conservation
alternatives.
Background
Water Supply
Seventy percent of the earth’s surface is covered in water (United States
Geological Survey, 2012). However, 97.5% of the earth’s water is saltwater. “It is the
freshwater resources, such as the water in streams, rivers, lakes, and groundwater that
provide people (and all life) with most of the water they need every day to live” (United
States Geological Survey, 2012, para. 6). Of the 2.5% that is freshwater, 69% of the
available water supplies are locked up in glaciers, icebergs, and permanent snow (see
Table 1). This means that less than 1% of the earth’s total water supply is drinkable and,
of the available freshwater, almost all of it is located underground. The United States has
over 3.5 million miles of rivers and streams and 40 million acres of lakes (USEPA, 2011,
2012) and many major U.S. cities were built, and thrived, along the banks of these rivers
and lakes. Even though only about 0.3% of the world’s freshwater is located in rivers and
lakes, 80% of the water withdrawals in the United States come from these surface water
sources (Kenny et al., 2009). The Schuylkill River, the largest tributary to the Delaware
River, supplies drinking water to the City of Philadelphia (Philadelphia Water
Department, 2012); New York City receives water for its residents from the Bronx,
Delaware, and Croton Rivers through a series of dams, reservoirs, and aqueducts (Stony
8
Brook University, n.d.); and Lake Michigan provides nearly 60% of the drinking water
used in the State of Illinois (University of Illinois at Urbana-Champaign, 2012).
Table 1
Global water distribution
Water source
Water volume,
in cubic miles
Percent of
freshwater
Percent of
total water
Oceans, seas, and bays
321,000,000
--
96.54
Ice caps, glaciers, & permanent snow
5,773,000
68.6
1.74
Groundwater
5,614,000
--
1.69
Fresh
2,526,000
30.1
0.76
Saline
3,088,000
--
0.93
Soil moisture
3,959
0.05
0.001
Ground ice and permafrost
71,970
0.86
0.022
Lakes
42,320
--
0.013
Fresh
21,830
0.26
0.007
Saline
20,490
--
0.007
Atmosphere
3,095
0.04
0.001
Swamp water
2,752
0.03
0.0008
Rivers
509
0.006
0.0002
Biological water
269
0.003
0.0001
Note. Compiled from Water in crisis: a guide to the world's fresh water resources
(pp. 13-24) by P. H. Gleick. New York, NY: Oxford University Press. Copyright
1993 by Pacific Institute for Studies in Development, Environment, and Security.
Reprinted with permission.
Of the 20% of fresh water drawn from groundwater sources in the United States,
67% is for irrigation (Kenny et al., 2009). Only 18% is drawn for domestic water uses
(Kenny et al., 2009).
9
California is home to more than 37 million people, has more than 800 miles of
coastline, more than 400 groundwater basins, and uses more groundwater than any other
state—nearly one sixth of the withdrawals in the nation (Bachman et al., 2005; CDWR,
2003; United States Census Bureau, 2010a). In California, in an average year, about 30%
of the urban and agricultural water comes from groundwater (CDWR, 2003). During a
drought year, the use of groundwater increases by 40% in some areas and up to 60% in
others (CDWR, 2003).
On California’s central coast, water supplies come from a variety of sources
including groundwater, surface water, water imported from the State Water Project
(SWP), and recycled water (CDWR, 2003). The CDWR (2009d) recognizes that
California’s central coast depends on its 28 groundwater basins for its water supply.
Santa Barbara County, the location of this study, receives about 77%
(approximately 78,000 acre feet per year) of its domestic, commercial, industrial, and
agricultural water from groundwater (Santa Barbara County, 2011). The county also has a
combined allotment of 40,000 acre-foot per year (one acre-foot equals 325,851 gallons)
of water from the SWP (Santa Barbara County, 2011). However, the voters of the city of
Lompoc and the unincorporated areas of Mission Hills and Vandenberg Village rejected
State Water, electing instead to rely solely on their groundwater wells (Santa Barbara
County, 2011).
Threats to the water supply. These limited drinking water supplies are
threatened by pollution, drought, climate change, and population growth (Anderson-
Wilk, 2008; CDWR, 2009a; Harou et al., 2010; Larson, Gustafson, & Hirt, 2009;
10
USEPA, 2002). Although, even in California, a majority of the fresh water withdrawals
are from surface water sources, both surface water and groundwater supplies are
interconnected at the hydrologic level (see Figure 4). Therefore, pollution in the surface
or groundwater supply or an overdraft, where more water is pumped out of the basin than
is replenished (Hanak et al., 2011), can have an impact on a number of different
communities regardless of their primary water source.
Figure 4. Map of the California basins and subbasins illustrating the interconnectivity
of groundwater sources. Reprinted from California’s groundwater (p. 108), by DWR,
1993. Sacramento, CA: State of California. Reprinted with permission.
11
Pollution. Water quality can be threatened by both natural and anthropogenic
(man-made) contaminants (Bachman et al., 2005; Blette, 2008; Kumar, Adak, Gurian, &
Lockwood, 2010; Pricope, 2009; Wilcox, Gotkowitz, Bradbury, & Bahr, 2010).
Constituents occurring naturally include minerals that erode into the water such as
arsenic, asbestos, fluoride, chromium, and cadmium (USEPA, 2009). Coastal
communities are also at risk from seawater intrusion into groundwater basins (Bachman
et al., 2005; CDWR, 2009b, 2009d). Man-made pollutants in the water supply include
pesticides, nitrates, and perchlorates (Bachman et al., 2005; VanDerslice, 2011). The
EPA requires periodic testing of all drinking water in the country to identify harmful
constituents so that they can be removed from the drinking water supply, either during
treatment or at the source, before the problem becomes so severe that it can no longer be
remediated (Bachman et al., 2005; USEPA, 2009).
A contaminated water supply that cannot be remediated must be abandoned as a
water source thereby reducing the available water to the region (Bachman et al., 2005;
Millock & Nauges, 2010). Contamination may also be a threat to neighboring water
sources if the migration of contaminated groundwater is not prevented (Bachman et al.,
2005). Groundwater contamination can also result in higher treatment costs or well
abandonment which would require alternative water sources and added capital costs
(CDWR, 2009b).
Population. The California Department of Finance estimates that the state’s
population could increase by an additional 62% to 59.5 million by the year 2050 (see
Table 2), and experts are concerned that California’s water supplies will remain static or
12
decrease (CDWR, 2009a). Population growth is a concern because it is a major factor for
planning future water use (CDWR, 2009a). The Not in My Backyard (NIMBY) attitude
exhibited by many urban residents demonstrates that some communities are resistant to
reducing their water usage in response to population growth, even with encouragement
from the purveyor (Atwood, Kreutzwiser, & De Loë, 2007; Cockerham & Leinauer,
2011; Nielsen-Pincus et al., 2010). This increase in population would reduce the available
per capita water supply as urban development used more water for turf grass, golf
courses, and backyard pools (Atwood et al., 2007; Balling, Gober, & Jones, 2008; Gober
& Kirkwood, 2010). Although land-use planners make an effort to consider water
supplies when approving new development, legislators will impose building moratoriums
to limit new development in their jurisdiction (Marimow, 2007). However, the courts
have been known to overrule a moratorium when there are housing shortages (Hanak,
2008; "Kawaoka v. the City of Arroyo Grande," 1994). This conflict in priorities
increases the potential for overdraft in the region.
In reviewing the population growth scenarios delineated in Table 2, the CDWR
(2009a) reports that, under the current trend of growth, an increase in population would
result in an increase in urban water demand of 6 million acre-feet per year; slow and
strategic population growth would increase demand by 1.5 million acre-feet per year; and
expansive population growth would increase demand by as much as 10 million acre-feet
per year.
13
Table 2
Scenario factors affecting urban water demand
Scenario factors for
urban water demand Year
2005
Future scenarios – Year 2050
Current
trends
Slow and
strategic
growth
Expansive
growth
Population (millions) 36.7 59.5 44.2 69.8
Single-family housing units (millions) 7.9 13.3 10 14.7
Multiple-family housing units
(millions)
4.3 5.8 4.5 6.6
Commercial employees (millions) 19 36.5 28 40.4
Industrial employees (millions) 1.7 1.9 1.9 1.9
Note. Reprinted from California water plan update 2009 Volume 1 - The Strategic
Plan (p. 5-28), by DWR, 2009. Sacramento, CA: State of California. Reprinted with
permission.
Environmental. Environmental threats, such as climate change and periodic
drought, also reduce the amount of available water and increase the possibility of
overdraft (Gleick, 1993; Hall, 2009; Makki, Stewart, Panuwatwanich, & Beal, 2011;
Polebitski & Palmer, 2010). Currently, during dry years, California’s water supply does
not adequately meet its current level of use (CDWR, 2009b; Hall, 2009). As the
population continues to increase, this trend will only worsen. Additionally, the increased
reliance on groundwater will increase the treatment cost while decreasing the available
water (CDWR, 2009b). Droughts also result in economic harm from fire danger and loss
of crops as well as degraded water quality and species collapse (CDWR, 2009a).
On California’s central coast, one major climate change concern is sea-level rise
due to climate change (Griggs & Russell, 2012). Griggs and Russell project an average
14
sea-level increase of 7 inches by 2030, 14 inches by 2050, and as much as 55 inches by
2100. As a result of sea-level rise, the City of Santa Barbara faces a moderate risk of
seawater flooding and inundation of low-lying coastal areas by the year 2050 and a high
risk by the year 2100 (Griggs & Russell, 2012). Critical water and sanitary sewer
infrastructure are vulnerable to the future sea-level rise and must be protected in order to
maintain a reliable water supply (CDWR, 2008).
Water Consumption
According to the Pacific Institute for Studies in Development Environment and
Security (2012), on average, a human requires a minimum of 1.3 gallons of water per day
for survival. When daily drinking, cooking, bathing, and sanitation requirements are
tallied, the minimum increases to around 13 gallons. In the United States, the average
water use per person is 70 gallons per day, this is a large amount when compared to the
Netherlands where the average use per person is 27 gallons, and in the African nation of
Gambia where the average per person is only 1.17 gallons of water per day. According to
the CDWR (2009c), in California, the average per capita water use is 192 gallons per day.
Unfortunately, a majority of that total, approximately 60% of the state’s treated drinking
water, is used for outdoor irrigation (California Urban Water Conservation Council,
2007).
Hardin’s (1968) concept of the tragedy of the commons can be applied to water
consumption (Bachman et al., 2005; Harlan et al., 2009; Larson, Gustafson, et al., 2009).
“Each party contributes [to the tragedy] by acting in their personal interest and
maximizing their use of the resource” (Bachman et al., 2005, p. 14). Although an
15
individual’s total water consumption may be insignificant by itself, cumulatively, the
water consumption of the community may not be sustainable for the long term (Harlan et
al., 2009; Larson, Casagrande, Harlan, & Yabiku, 2009).
Solutions to the water supply shortage must either increase supply or decrease
demand (CDWR, 2009b). Increasing supply generally involves a large monetary
investment to build dams, reservoirs, pipelines, or desalination plants (CDWR, 2009b). A
desalination plant to produce 300,000 acre-foot of water per year can cost between $1.5
and $2.0 billion (CDWR, 2009b). Some water supply projects, such as the California
State Water Project, the largest state-built water conveyance system in the country,
essentially move the water from one area to another (CDWR, 2010a). Although the net
amount of water in the state is not increased, because this project moves the surplus water
from a less populated area to the population center, it is seen as acceptable solution by
regulatory agencies looking for ways to increase supply. In California, 75% of the
precipitation falls in the northern half of the state while 75% of the population lives in the
southern half (MacDonald, 2007). However, the cost of importing water can be very
expensive. In 2009, a 45-mile pipeline project to transport water from Lake Nacimiento
to communities in southern San Luis Obispo County had a price tag of $140 million
(CDWR, 2009d).
Many factors can influence both indoor and outdoor household water
consumption. The size of the yard is the primary factor in the amount of outdoor water
that is used (Harlan et al., 2009; Willis, Stewart, Panuwatwanich, Capati, & Giurco,
2009) and an increase in temperature or a decrease in rainfall can cause a household to
16
increase their outdoor water use (Harlan et al., 2009; Kenney et al., 2008; Lee, Tansel, &
Balbin, 2011). The number of bathrooms and the age of the home are primary factors in
the amount of indoor water consumption (Harlan et al., 2009; Mansur & Olmstead, 2012;
Polebitski & Palmer, 2010; Polebitski, Palmer, & Waddell, 2011; USEPA, 2008).
Although water rate increases and usage restrictions are commonly used to
decrease water consumption (Funk, 2007; Hill & Symmonds, 2011; Kenney et al., 2008;
Mansur & Olmstead, 2012; Willis et al., 2009; Willis, Stewart, Panuwatwanich, Jones, &
Kyriakides, 2010), decreasing demand through water conservation programs is a
relatively low-cost solution that allows the region to subsist on less water (Freeman,
Poghosyan, & Lee, 2008; Nelson, Cismaru, Cismaru, & Ono, 2011). Water use efficiency
can reduce the impact of water shortages on a community by reducing treatment costs as
well as increasing available water supplies (CDWR, 2008). In Southern California, urban
water conservation could increase supply by about 1 million acre-feet per year–25% of
the current annual demand (Freeman et al., 2008). Experts agree that a diverse water
supply portfolio should include a robust water conservation program (CDWR, 2008).
Water Conservation
California has had a long water conservation history. One of the first steps in
water conservation is metering household water consumption (Atwood et al., 2007;
California Assembly, 2004a; Corbella & Pujol, 2009). Awareness by the consumer about
how much water is being used is important to reducing consumption (Coleman, 2009;
Funk, 2007; Olmstead & Stavins, 2009). In 2004, the California Assembly wrote
17
legislature that encouraged water purveyors to use water metering and volumetric pricing
as tools to encourage conservation.
On August 16, 1889, Los Angeles water baron William Mulholland installed the
city’s first water meter in an effort to curb water consumption (Mulholland, 2000).
Because he knew how precious the resource was to Los Angeles, Mulholland believed
that wasting water was an ultimate sin (Mulholland, 2000). Before metering,
“Angelenos” were using 306 gallons per capita per day (Los Angeles Department of
Water and Power, n.d.). In 1902, the Los Angeles Times reported that consumption had
exceeded maximum supply and that Los Angeles was the fourth largest water user in the
United States at that time (Los Angeles Times, 1902b). At the start of Mulholland’s
metering project, only those users whose grounds exceeded three times the floor area of
the structures on the property were required to have their water metered (Los Angeles
Times, 1902a, p. A7). However, in 2 years, Mulholland reduced per capita water usage
by one-third (Kahrl, 1982). By the end of the project, usage in the city was a manageable
200 gallons per capita per day (Los Angeles Department of Water and Power, n.d.).
In 1928, in response to conflicts that were commonly seen in the courts between
riparian and appropriative water rights, the state’s constitution was amended. The
resulting Reasonable Use Doctrine was drafted to settle the conflict between competing
water rights, but also gave water purveyors the authority to enforce water conservation
(Hanak et al., 2011).
Water resources of the State [will] be put to beneficial use to the fullest extent …
capable, and that the waste or unreasonable use … of water be prevented, and that
18
the conservation of such waters is to be exercised with a view to the reasonable and
beneficial use thereof in the interest of the people and for the public welfare.
(California Constitution, Article X, § 2)
It was not until the 1990s that the State of California enacted laws that targeted
end users by requiring the installation of low-flow plumbing devices and appliances in
new construction (Hanak et al., 2011). However, many communities in California
enacted regulations encouraging water conservation as early as 1972 and the most
aggressive water conservation regulations were passed during historic 1987-92 drought
(Renwick & Archibald, 1998; Syme, Nancarrow, & Seligman, 2000). The drought was
considered historic because of its duration and its impact on the entire state. By 1990,
reservoirs were down 60% and storage did not return to normal until two years after the
drought ended (CDWR, 2000). California’s central coast was severely impacted by the
1987-92 drought (Dziegielewski, Garbharran, & Langowski, 1993; Renwick &
Archibald, 1998; Syme et al., 2000). So much so that, in July 1990, for the first time in
the state’s history, Governor George Deukmejian declared a drought emergency for the
City of Santa Barbara (Dziegielewski et al., 1993; Renwick & Archibald, 1998; San
Francisco Chronicle, 1990). In response to the resulting water shortage, the City of Santa
Barbara and its neighbor, Goleta Water District, enacted water conservation policies,
which included restrictions, rates, and rebates, and resulted in statistically significant
water reductions (Renwick & Archibald, 1998).
In the quinquennial California Water Plan, last updated in 2009, the California
Department of Water Resources recommended that water purveyors make water
19
conservation a priority (CDWR, 2009a). Returning to Hardin’s (1968) theory, although a
water user may not believe that their conservation efforts are significant, cooperative
water conservation efforts at the community level have been shown to help sustain a
community’s water supply (San Francisco Chronicle, 1990).
Most water conservation programs contain a combination of price and non-price
conservation elements (Booker, Howitt, Michelsen, & Young, 2012; Coleman, 2009;
Halich & Stephenson, 2009; Nataraj & Hanemann, 2011; Tsai, Cohen, & Vogel, 2011;
Worthington & Hoffman, 2008). Primarily, punitive water rates, consumption
restrictions, and appliance rebates are utilized to encourage water conservation (Osgood,
2011; Pumphrey, Edwards, & Becker, 2008; Rosenberg, Howitt, & Lund, 2008).
Replacing a home’s toilets with a low-flow toilet could save a household between
9,000 and 26,000 gallons of water per year (Vickers, 2001). Although still effective for
homes built prior to 1992 with original fixtures, state and federal regulations are making
toilet retrofit rebates ineffective for new homes. While the federal Energy Policy Act
required all toilets sold, installed, or imported after January 1, 1994 in the United States
be 1.6 gallons per flush (gpf) or less, California law enacted the standard on any house
built after January 1, 1992 (CDWR, 2013; Vickers, 2001). In 2007, California law (AB
715) required that 50% of new homes built after January 1, 2010 install 1.28 gpf high-
efficiency toilets. On January 1, 2014, that number increased to 100% (California
Assembly, 2007).
As a result of these mandatory water reductions, landscape rebates are currently
being added to water conservation programs across Santa Barbara County and are gaining
20
in popularity throughout the state of California (City of Riverside, 2012; City of Santa
Barbara, 2009; Santa Barbara County, 2013b; VVCSD, 2012b). However, although it
makes sense that reducing irrigable yard area should reduce outdoor water consumption,
there is very little empirical evidence to support the hypothesis. This study will fill the
gap in the literature on landscape rebates by evaluating their effectiveness at reducing
water consumption.
Landscape Rebates
Landscape rebates are a water conservation element that rewards residents who
reduce the amount of irrigable lawn with cash reimbursements (St. Hilaire et al., 2008;
USEPA, 1994; Vickers, 2001). Reducing the amount of lawn to be watered has the
potential to save a significant amount of money because native and low-water-use turf
grasses and plants can reduce or eliminate the need for supplemental irrigation
(Anderson-Wilk, 2008; Gober & Kirkwood, 2010; Hanak & Davis, 2006; St. Hilaire et
al., 2008; Vickers, 2001). By removing traditional lawns and replacing them with
drought-tolerant plants a household can substantially reduce water usage (Hanak &
Davis, 2006). However, many residents are afraid to move to a nontraditional lawn
(Harlan et al., 2009; Robbins, 2009; St. Hilaire et al., 2008). The typical American grass
lawn can be traced back to the vast green expanse of the Southern plantation or Medieval
European estate where the landowner was so wealthy that he could use his slaves or serfs
for beautification instead of agriculture (Robbins, 2009).
Grass lawns require a lot of time and money to maintain, and they can be harmful
to the environment. Today, American homeowners spend approximately 40 hours per
21
year maintaining their “perfect” lawns (National Wildlife Federation, n.d.). American use
67 million pounds of pesticides on their lawns annually and these petroleum-based
fertilizers and pesticides add to the pollution in the community’s water supply (Robbins,
2009). Finally, grass lawns add to noise and air pollution through their requirement for
periodic mowing, blowing, edging, and trimming (Cockerham & Leinauer, 2011;
Robbins, 2009).
Many types of grass used in the United States require a significant amount of
water to maintain. Outdoor water use can be substantially reduced by switching from
traditional grass lawns to drought tolerant plants and grasses (Hanak et al., 2011). Native
plants use several times less water than cool-season turf grass (Hanak & Davis, 2006).
Wilson and Livingston (1932) tested the evaporation properties of 17 popular grass
species and found few that were drought tolerant (see Table 3). The test found that certain
varieties of fescue are much more environmentally friendly than the more popular
varieties such as Kentucky bluegrass and Bermuda grass.
22
Table 3
Cubic centimeters of evaporation for 17 turfgrasses
Common name
Evaporation
(cc)
Common name
Evaporation
(cc)
Tall oat grass 100 Kentucky bluegrass 237
Orchard grass 107 Meadow fescue 249
Rhode Island bent grass 124 Chewing’s fescue 252
Velvet bent grass 138 Redtop 275
Brome grass 168 Perennial ryegrass 281
Washington bent grass 178 Canada bluegrass 290
White clover 208 Vermont bent grass 300
Bermuda grass 209 Sheep’s fescue 458+
Alsike clover 210 Red fescue 465+
Timothy 212
Note. The higher the number, the more drought tolerant the grass species. From
“Wilting and withering of grasses in greenhouse cultures as related to water-supplying
power of the soil,” by J. D. Wilson and B. E. Livingston, 1932, Plant Physiology, 7, p.
22. Copyright 1932 by Plant Physiology. Reprinted with permission.
Landscape rebates are a tool that can help water purveyors encourage the
replacement of thirsty lawns (see Table 4). Although gaining in popularity in the 2000s,
the Environmental Protection Agency (EPA) first recommended the benefits of landscape
rebates in 1994 (Hanak & Davis, 2006; USEPA, 1994).
23
Table 4
Turf Conversion Costs and Savings*
ETo
superzone§
Water savings
(gallons/square foot)
Customer years to recoup
investment
Low net conversion costs
($1.00/square foot)
I II III
Coastal
32
23
6
3
Inner coastal
39
17
6
2
Central
42
15
5
2
Desert
51
12
5
2
Costs to utility ($/acre-foot)
High net conversion costs
($1.60/square foot)
Low rebate
($0.40/square
foot)
High rebate
($1.00/square
foot)
I II III
Coastal
363
907
76
10
4
Inner coastal
298
745
38
10
4
Central
276
690
32
9
4
Desert
232
580
23
8
4
* Assumes a retail water price of $678 per acre-foot. Scenario I includes only water savings, scenario II
also includes garden supply savings, and scenario III includes labor cost savings. Both utility and
customer investments are amortized at a rate of 4 percent. Baseline irrigation efficiency is 37.5
percent, with 25 percent of plant water needs met by rainfall (or alternatively, 50% irrigation
efficiency with no rainfall contribution)
§ ETo = evapotranspiration
Note: From Lawns and water demand in California (p. 16), by E. Hanak and M. Davis,
2006. San Francisco, CA: Public Policy Institute of California. Copyright 2006 by
Public Policy Institute of California. Reprinted with permission.
In recent years, the State of California has enacted regulations to help reduce the
amount of water used on landscapes (California Assembly, 2004b). Assembly Bill 2717
(AB 2717), authored by Assemblyman John Laird and signed by Governor Arnold
Schwarzenegger on September 22, 2004, asked the California Urban Water Conservation
Council (CUWCC) to elicit stakeholders who would develop landscape efficiency
24
recommendations (California Assembly, 2004b). The passage of AB 2717 made it
possible for CDWR (2009a) to require local governments by the year 2010 to enact and
enforce landscape water conservation legislation.
Vandenberg Village
Vandenberg Village is an unincorporated area of Santa Barbara County located on
California’s central coast. The bedroom community of 2,700 housing units covers
approximately 5 square miles (United States Census Bureau, 2010c). The community is
bounded on three sides by protected Burton Mesa Ecological Preserve and on the fourth
by Vandenberg Air Force Base and City of Lompoc boundaries (California Department
of Fish and Game, 2009; Santa Barbara County, 2012). As a result, development
expansion is limited and the community is nearing build-out (VVCSD, 2003).
Household level demographics are not available for the area. However, an
analysis of U.S. Census data indicates very little change in area demographics during the
study period. In 2010, U.S. Census data indicates that there were 2.53 persons per
household, 74.1% of residents own their home, and the median household income was
27% above the state median household income at $78,480 (United States Census Bureau,
2010b, 2010c). In 2000, these numbers were 2.49 persons per household, 81.2% owned
their home, and the median household income was 24% above the state median
household income at $58,700 (United States Census Bureau, 2000a, 2000b). Ethnicity
and age demographics are detailed in Figure 5.
25
Figure 5. Ethnicity and age demographics for Vandenberg Village, California.
Compiled from “Vandenberg Village CDP quickfacts”, United States Census Bureau,
2010. Retrieved from http://quickfacts.census.gov/qfd/states/06/0682086. html#.
USFT2ITCnxQ. Material is not subject to copyright protection within the United
States.
The population of 6,497 is provided water service by VVCSD, a California
special district. VVCSD draws all of its water from three groundwater wells (VVCSD,
2012e). The wells draw from the Lompoc Upland aquifer which is hydrologically
connected to the Lompoc Plain aquifer and, in turn, connected to the Santa Ynez River
(Santa Barbara County, 2011; VVCSD, 2012e; West Yost Associates, 2012). On average,
VVCSD produces about 1,500 acre foot of water per year and delivers about 1,275 acre
foot per year (VVCSD, 2012f). The difference between water produced and water
delivered is the expected water loss during the treatment cycle as well as nonrevenue
water (see Figure 6). Nonrevenue water includes authorized unbilled water usage, such as
fire hydrant usage, and water losses from meter inaccuracies, data errors, leaks, breaks,
and reservoir overflows (American Water Works Association, 2009).
26
Figure 6. Average amount of water produced. Compiled from “Water produced and
sold (1988-2011)”, by VVCSD, 2012. Lompoc, CA. Reprinted with permission.
Senate Bill x7-7 requires that urban retail water suppliers that provide potable
water to more than 3,000 end users reduce per capita water usage by 20% by the year
2020 with an interim target of 10% reduction by 2015 (California Senate, 2009).
Although possibly exempt from the SB x7-7 water reduction requirement, the VVCSD
Board of Directors asked staff to calculate the 2015 and 2020 targets under the regulation
(VVCSD, 2011a, 2011b). In accordance with the benchmark calculations delineated in
California Water Code Section 10608.20, the 2015 target was determined to be 205
gallons per person per day and the 2020 target is 182 gallons (VVCSD, 2011a, 2011b). In
2011, VVCSD’s per capita water usage was 199 gallons per day (see Figure 7).
Of the three agencies in Santa Barbara County that offer landscape rebates,
Vandenberg Village was the first to implement a program of its type (Santa Barbara
County, 2013b). VVCSD adopted its program in 2007 while the City of Lompoc and the
27
City of Santa Barbara both adopted their programs in 2009 (City of Lompoc, 2009; City
of Santa Barbara, 2009; VVCSD, 2007).
Figure 7. Vandenberg Village per capita water use (gallons per person per day).
Compiled from “Water produced and sold (1988-2011)”, by VVCSD, 2012. Lompoc,
CA. Reprinted with permission.
VVCSD’s water conservation program, which includes a landscape rebate
program added in 2007, is making it possible for VVCSD to meet the 2015 goal of a 10%
reduction in per capita water usage (California Senate, 2009; VVCSD, 2012d, 2012f).
However, researchers have not studied the effectiveness of landscape rebates; therefore,
there is no empirical evidence that landscape rebates are making a quantifiable
contribution to the reduction in water consumption. This study will not only provide
valuable information to the management staff and elected officials of VVCSD but it will
28
also give future water conservation researchers the tools needed to evaluate other
programs.
Problem Statement
Experts agree that California’s domestic water supplies are threatened by
population growth, drought, pollution, and climate change (Anderson-Wilk, 2008; Harou
et al., 2010; Larson, Gustafson, et al., 2009). In response, the California legislature
concluded that the most economical option to extend water resources is by using less
water per capita (California Assembly, 2004b; CDWR, 2009a). Many different factors
influence household water consumption. Bathrooms are the largest consumer of indoor
water use (USEPA, 2008). Therefore, the age of the toilets, the number of bathrooms and
the age of the home can influence the amount of indoor water consumption (Harlan et al.,
2009; Mansur & Olmstead, 2012; Polebitski & Palmer, 2010; Polebitski et al., 2011;
USEPA, 2008). To encourage household water savings, water conservation programs
generally target indoor water use through toilet and washing machine rebates (USEPA,
2002). The current literature focuses on restrictions, pricing structures, and toilet rebates
(Anderson-Wilk, 2008; Kenney et al., 2008; Mansur & Olmstead, 2012; Olmstead &
Stavins, 2009; St. Hilaire et al., 2008).
However, outdoor water usage accounts for the majority of a household’s water
usage (Gober & Kirkwood, 2010; Harlan et al., 2009; Larson, Gustafson, et al., 2009;
Millock & Nauges, 2010). The size of the yard is the primary factor in the amount of
outdoor water used (Harlan et al., 2009; Willis et al., 2009); an increase in temperature or
a decrease in rainfall can also cause a household to increase its outdoor water use (Harlan
29
et al., 2009; Kenney et al., 2008; Lee et al., 2011). Experts hypothesize that elements
from a water conservation program that target outdoor water use, such as landscape
rebates, can have a significant impact on domestic water usage (Hanak & Davis, 2006;
USEPA, 1994). Nevertheless, to date, water savings in response to landscape rebates are
anecdotal and have not been scientifically tested (Anderson-Wilk, 2008; Funk, 2007;
Goldstein, 2012; Harlan et al., 2009; St. Hilaire et al., 2008). Because of this substantial
gap in the literature regarding landscape rebates, this quantitative study was designed to
evaluate the effectiveness of landscape rebates on residential water consumption in
Vandenberg Village, California. Secondary data has been collected from VVCSD, Santa
Barbara County, and the National Oceanic and Atmospheric Administration.
Purpose of the Study
The purpose of this quantitative study was to determine if the receipt of a
landscape rebate reduced water consumption while controlling for weather, irrigable yard
size, home value, number of bathrooms, year home built, price per unit of water, receipt
of toilet rebates and receipt of washing machine rebates. With the residents of
Vandenberg Village, California as the study population, this research used regression
analysis to evaluate the effectiveness of landscape rebates in reducing water
consumption.
30
Research Questions and Hypotheses
The following research questions were examined in the process of testing the
hypotheses:
1. Do consumers who receive landscape rebates use less water than those
consumers who do not?
H01: There is no significant difference in water consumption between those
who receive a landscape rebate and those who do not receive a
landscape rebate when controlling for receipt of a toilet or washing
machine rebate, size of property, number of bathrooms, value of home,
age of home, and price per unit of water.
Ha1: There is a significant difference in water consumption between those
who receive a landscape rebate and those who do not receive a
landscape rebate when controlling for receipt of a toilet or washing
machine rebate, size of property, number of bathrooms, value of home,
age of home, and price per unit of water.
2. Do landscape rebate recipients use less water after receiving the rebate than
before?
H02: There is no significant difference in water consumption in the 24 months
before and 24 months after receipt of a landscape rebate when
controlling for receipt of a toilet or washing machine rebate, size of
property, number of bathrooms, value of home, age of home, amount of
rainfall, average temperature, and price per unit of water.
31
Ha2: There is a significant difference in water consumption in the 24 months
before and 24 months after receipt of a landscape rebate when
controlling for receipt of a toilet or washing machine rebate, size of
property, number of bathrooms, value of home, age of home, amount of
rainfall, average temperature, and price per unit of water.
Theoretical Perspective
The theoretical framework for this study was Ajzen and Fishbein’s (1972, 1977)
theory of reasoned action (TRA) which attempts to explain the psychological behavior of
consumers by examining the relationship of beliefs and attitudes on behaviors (Liska,
1974; Petty & Cacioppo, 1981). The theory assumes that people make rational decisions
based on the information available to them and that they consider the implications of their
actions before engaging in the behavior (Ajzen, 1980).
The first step in the reasoned action process is to identify the behavior of interest.
Then, the determinant of the action is identified. With this information—since the person
intends to perform (or not perform) the behavior—the theory states that the behavior is
not difficult to predict. “Barring unforeseen events, a person will usually act in
accordance with his or her intention” (Ajzen, 1980, p. 5). TRA is based on a person’s
attitude toward expected behavior (Liska, 1974; Mueller, 1986; Petty & Cacioppo, 1981).
The more a person is expected to exhibit a particular behavior, the more likely he or she
is to behave in the expected manner.
TRA and its successor, the theory of planned behavior (TPB), have been used
frequently to help explain the motivation to conserve water (Endter-Wada, Kurtzman,
32
Keenan, Kjelgren, & Neale, 2008; Fielding, McDonald, & Louis, 2008; Graymore &
Wallis, 2010; Hurlimann, Dolnicar, & Meyer, 2009; Jorgensen, Graymore, & O'Toole,
2009; Larson, Wutich, White, Muñoz-Erickson, & Harlan, 2011; Marandu, Moeti, &
Joseph, 2010; Willis, Stewart, Panuwatwanich, Williams, & Hollingsworth, 2011). For
example, America has a tradition of lush lawns that has been a symbol of wealth and
power since colonial times (Endter-Wada et al., 2008; Harlan et al., 2009; Robbins,
2009). To reduce outdoor water consumption, consumers need to believe that reducing
their turf is not only acceptable but also desirable (Endter-Wada et al., 2008; Gober &
Kirkwood, 2010; Harlan et al., 2009; Robbins, 2009; St. Hilaire et al., 2008). An increase
in the awareness that lush lawns waste drinking water may create a new expectation that
native lawns are more acceptable (Gober & Kirkwood, 2010; Harlan et al., 2009;
Knutson, 2008; Robbins, 2009; van Putten, Jennings, Louviere, & Burgess, 2011).
Therefore, as more residents take advantage of landscape rebates the acceptance of
nontraditional lawns increases.
Additionally, water conservation messages and rebate announcements during
periods of reduced rainfall may encourage increased participation in water conservation
rebate programs should customers believe it to be the right thing to do.
Nature of the Study
In this quantitative study, secondary data were used to test the effects of several
independent variables on water consumption: monthly maximum daily temperature, total
monthly rainfall, total square footage of irrigable yard, home value, number of toilets in
home, year home built, price per unit of water, receipt of toilet rebates, receipt of washing
33
machine rebates, and receipt of landscape rebates. Customers were divided into groups
where each group comprised the same number of customers with similar rebate receipts
and similar property characteristics, such as yard size, number of bathrooms, and age and
value of home.
Regression analysis was performed on each group to determine if, and to what
extent, certain leading indicators (independent variables) affect the dependent variable.
Regression analysis was selected because (a) it is frequently used for water conservation
studies and (b) because of the number of variables (landscape, washing machine and
toilet rebates, price per unit of water, total monthly rainfall, average monthly temperature,
size of yard, number of toilets, value of home, and age of home) that influence water
consumption (Harlan et al., 2009; House-Peters, Pratt, & Heejun, 2010; Polebitski &
Palmer, 2010). Because the study population targeted publicly available records, the data
collection process required no direct contact with the study participants.
Operational Definitions of Terms
Evapotranspiration: “The loss of water to the atmosphere by the combined
processes of evaporation (from soil and plant surfaces) and transpiration (from plant
tissues)” (CDWR, 2009e, para. 1).
Irrigable yard size: The amount of a property’s outdoor area subject to irrigation.
Calculated by subtracting the square foot of the improvements from the square foot of the
property.
Overdraft: “Overdraft occurs when more groundwater is removed from a basin
than is replaced by recharge over a period of many years” (Bachman et al., 2005, p. 13).
34
Price: The amount per unit of water that the customer pays. In the United States,
domestic water is generally measured in hundred cubic foot (CCF or HCF).
Rebate: A monetary incentive provided by the water purveyor to encourage the
consumer to replace high water using features such as showerheads, toilets, washing
machines and dishwashers, and turfgrass (American Water Works Association, 2006).
Water Consumption: The amount of water consumed by a customer, generally
recorded on a monthly basis.
Water Demand: See water consumption.
Water Purveyor: A water purveyor is an agency or person that supplies water to
end-users (Reclamation, 2009).
Water Rate: See price.
Assumptions
This study assumed that customers who have received a landscape rebate have not
removed the low-water using plants and reintroduced turf grass to their lawns within the
24 months following the receipt of the rebate. Also, because California law requires that
residential customers to request an investigation within 5 days of receiving a disputed bill
(California, 1988), this study assumed that any issues of inaccuracy in the water
consumption data would have a minor impact on the results of the study. Details on
issues of accuracy can be found in the limitations section.
Scope and Delimitations
The delimitation of the study has been the public water agency and its customers
in Vandenberg Village, California, an unincorporated bedroom community in Santa
35
Barbara County north of the City of Lompoc. Vandenberg Village was selected as a
research area because, as the first agency in the county to offer landscape rebates, they
have the longest period of data available for the study. Additionally, Vandenberg
Village’s size makes it an optimal location to include community variables, such as
temperature and precipitation, since there is very little variation in the daily maximum
temperature or the amount of rainfall received at different locations throughout the
community. Finally, a previous study in the area (Allen, 2008) found that landscape
rebates were statistically significant to water conservation but the sample size was too
small to conclude that those savings could be applied to other properties within the study
area.
Limitations
The study was subject to three limitations. First, there were data missing for some
homes during periods of vacancy. Second, household level demographics that can impact
water consumption—such as number of residents per household, their ages, and
income—were not available for the community.
Third, there are normal issues of accuracy inherent to water consumption data. As
meters age, their mechanical parts wear down and their accuracy declines (Jackson, n.d.).
During the end of the study period, VVCSD replaced every meter in the community with
automatic meter reading (AMR) capable meters (VVCSD, 2011c, 2012c). While meters
that stopped recording any water consumption were targeted for early replacement,
before being replaced with a new AMR water meter, the oldest meters may have simply
been under-recording consumption data. Additionally, before the installation of the AMR
36
meters, the employees of VVCSD read meters manually. While most meter readings are
accurate, human error is possible where meter readings were entered into the meter
reading device. Generally, these reading errors self-corrected the following month when
the meter was read again, resulting in 1 month of higher than normal usage and 1 month
of lower than normal usage. But the average was normal. At times, the higher than
normal meter reading was higher than the reading the subsequent month. Per district
policy, higher than normal meter readings that did not “catch up” within 3 months were
adjusted accordingly and the customer service representative edited the customer’s
monthly water consumption data in the customer database so large meter reading errors
were corrected in the data collected.
Significance of the Study
Groundwater and surface water sources are connected across multiple regions
throughout the state of California (United States Geological Survey, 2009). As a result,
the threat of population increase, pollution, drought, and climate change on the water
supply source of one region can have negative effects on water supplies across many
regions (Bachman et al., 2005). Water conservation, including newly regulated landscape
rebates, can play an essential part in protecting water supplies from overdraft and nitrate
pollution (Bachman et al., 2005; CDWR, 2009a; Robbins, 2009). The significance of this
study is that it introduces the positive economic, environmental, and social impacts that
water conservation programs and landscape rebates can have on local, regional, and state
water supplies. As landscape rebates gain in popularity, this study will be beneficial to
37
water professionals around the world who are exploring alternative methods of
encouraging water conservation.
Social Change Implications
Maintaining a sustainable water supply has been an important part of California’s
heritage for over a century (California Constitution; CDWR, 2009a; Mulholland, 2000).
In the past decade, the state legislature recognized that landscape ordinances, through
which landscape rebates can play an important role, need to be enacted to stop wasteful
outdoor water use (California Assembly, 2004b). The success of landscape rebates relies
heavily on the social acceptance of nontraditional lawns (Hanak & Davis, 2006; Robbins,
2009). Therefore, the future of water supply sustainability also relies heavily on the social
acceptance of alternative lawns and landscapes. Currently, although water conservation
has been sufficiently studied, the effectiveness of landscape rebates is an area that is
missing from the literature (Anderson-Wilk, 2008; Funk, 2007; Gober & Kirkwood,
2010; Harlan et al., 2009; Kenney et al., 2008; St. Hilaire et al., 2008). This study is
expected to add to the body of water conservation studies while filling a significant gap
in the literature. It is the intent of this study to provide water professionals with empirical
data to assist in their goal of providing safe, clean, dependable water to California’s
residents in perpetuity.
Summary
Chapter 1 presented an in-depth introduction to water supply, water demand,
water conservation, and landscape rebates and includes a comprehensive background;
purpose statement; research questions and hypotheses; operational terms and acronyms;
38
assumptions, limitations, and delimitations; significance of the study; and social change
implications.
This chapter explained that water supplies in California are stretched to critical
levels as a result of population growth, periodic drought, and climate change and that the
California legislature recognized that the best way to increase supply is to decrease
demand. The Water Use Efficiency Senate Bill 7 (SBx7-7) was signed into law in 2009
which requires water purveyors to reduce per capita water usage by 20% by the year
2020. To comply, water purveyors are searching for innovative ways to increase water
conservation and, as a result, Cash-4-Grass rebates are increasing in popularity. Although
experts believe that Cash-4-Grass rebates can significantly decrease water consumption,
there is no scientific evidence to substantiate the theory.
The chapter goes on to explain that the purpose of this study was to fill the gap in
the literature and to determine if the receipt of a landscape rebate reduces water
consumption in Vandenberg Village, California. The theoretical frameworks for this
study were Ajzen and Fishbein’s TRA and its successor, TPB, which explains the
motivation to conserve water. Secondary data were statistically analyzed to determine to
what extent landscape, washing machine and toilet rebates, price per unit of water, total
monthly rainfall, average monthly temperature, size of yard, number of toilets, value of
home, and age of home influence water consumption. The significance of this study is the
positive economic, environmental, and social impacts that water conservation programs
and landscape rebates can have on the local, regional, and state water supply.
39
Finally, this chapter explains the social change implications in that this study is
expected to add to the body of water conservation studies while filling a significant gap
in the literature which will provide water professionals with empirical data to assist in
their goal of providing safe, clean, dependable water to California’s residents in
perpetuity.
Chapter 2 will present the review of the existing literature for the study. Chapter 3
will present the research methodology that has been used for this project. Chapter 4 will
present the results of the statistical analysis. Chapter 5 will present the interpretation of
the findings, limitations of the study, recommendations, and implications.
40
Chapter 2: Literature Review
Introduction
California’s domestic water supplies are threatened by population growth,
drought, pollution, and climate change. In recent years, the California legislature
concluded that the most economical option to extend the resources is by using less water
per capita. To encourage household water savings, most water conservation programs
utilize restrictions, rates, and rebates. Prior research has focused on water restrictions,
pricing structures, and toilet rebates. Since outdoor water usage accounts for the majority
of a household’s water usage, elements of a water conservation program that target
outdoor water use, such as landscape rebates, can have a significant impact on domestic
water usage. To date, however, water savings in response to landscape rebates are
anecdotal and have not been scientifically tested.
The purpose of this study was to evaluate the effectiveness of landscape rebates in
Vandenberg Village, California. This chapter outlines the literature on water supply,
water demand, water conservation, landscape rebates, and the theoretical perspective of
water conservation. The available literature on landscape rebates is sparse and primarily
focuses on the theoretical; no studies were found that examine the effectiveness of
landscape rebates (Anderson-Wilk, 2008; Funk, 2007; Gober & Kirkwood, 2010; Harlan
et al., 2009; Kenney et al., 2008; St. Hilaire et al., 2008). Other types of water
conservation measures—such as incentive programs, public information campaigns, and
pricing structures—have been thoroughly researched in California for many years. That
research can be used to make some assumptions for research design (Atwood et al., 2007;
41
Booker et al., 2012; Coleman, 2009; Graymore & Wallis, 2010; Halich & Stephenson,
2009; Kenney et al., 2008; Mansur & Olmstead, 2012; Millock & Nauges, 2010; Nataraj
& Hanemann, 2011; Nelson et al., 2011; Olmstead & Stavins, 2009; Tsai et al., 2011;
Worthington & Hoffman, 2008).
This literature review was based on academic journals published between 2007
and 2012. The primary databases used were ABI/INFORM Complete, Academic Search
Elite, Business Source Complete, Elsevier ScienceDirect Social Sciences, Free E-
Journals, Highwire Press, JSTOR, ProQuest Central, ProQuest Science Journals,
PsycARTICLES, SAGE Journals Online, and SSRN eLibrary. Some keyword phrases
included: water conservation AND California, water conservation AND theory of
planned behavior, rebate OR incentive AND water conservation, and water conservation
AND drought.
Theoretical Perspective
Because the key to a successful Cash-4-Grass program is changing the perception
of alternative landscapes, the theoretical framework used for this study was Ajzen and
Fishbein’s (1980, 1985; 1972, 1977; 1975) theory of reasoned action (TRA) and their
theory of planned behavior (TPB). TRA attempts to explain the psychological behavior
of consumers by examining the relationship of beliefs and attitudes on behaviors (Ajzen,
1980; Ajzen & Fishbein, 1972, 1977; Fishbein & Ajzen, 1975; Liska, 1974; Petty &
Cacioppo, 1981). TPB adds the concept of perceived conscious choice to the previously
explained TRA (Ajzen, 1985, 2011).
42
TRA and TPB have been used frequently to attempt to explain the motivation
behind water conservation (Endter-Wada et al., 2008; Fielding et al., 2008; Graymore &
Wallis, 2010; Hurlimann et al., 2009; Jorgensen et al., 2009; Larson et al., 2011;
Marandu et al., 2010; Willis et al., 2011). Larson et al. (2011) use a modified version of
the TPB to explain why more long-term residents of Phoenix, Arizona tend to prefer lush,
well-watered landscapes to the native landscapes embraced by the area’s newer residents.
“Long-term residents … appear acculturated to the status quo of well-watered landscapes
and few regulations in the Phoenix oasis, given heightened opposition to water-use
restrictions among them compared to newcomers” (Larson et al., 2011, p. 85). Graymore
and Wallis (2010) and Jorgensen et al. (2009) found that public trust in the water
authority and trust in the water-using community increased the positive attitude of water
conservation. Marandu et al. (2010) found that the TRA was a statistically significant
indicator of water conservation behavior and that changing the attitude of influential
members of the community would allow the process to gain critical mass and promote
widespread attitudinal and behavioral changes.
Review of the Literature
Water Supply
Water supplies in the western United States are being threatened by population
increase, climate change, water pollution, and periodic drought (Anderson-Wilk, 2008;
Harou et al., 2010; Larson, Gustafson, et al., 2009; Larson et al., 2011; Olmstead &
Stavins, 2009; Ward, Michelsen, & DeMouche, 2007; Willis et al., 2011). “The most
significant determinants for increases in water demand are population growth, climate
43
change, and the type of urban development that occurs” (House-Peters et al., 2010, p.
461). To help address a majority of these water supply concerns, the California
legislature recently enacted Senate Bill 7 (SBx7-7) to require every water purveyor in the
state to reduce their per capita water usage by 10% by 2015 and 20% by 2020 (California
Senate, 2009). Working with another legislative act, which increases plumbing
regulations for new homes (Millock & Nauges, 2010), the law encourages options to
decrease demand such as water conservation opportunities including water restrictions,
punitive water rates, and conservation rebates.
Population Increase. The rate of population growth is an important factor for
planning future water needs. Funk (2007) believes that population growth alone “poses
challenges beyond the capacity of current water resources” (p. 171) and Corbella and
Pujol (2009) suggest that the rate of population growth is more important than population
size in predicting future water consumption. Gleick, Christian-Smith, and Cooley (2011)
report that water consumption in the United States has remained relatively level since
1980 despite substantial population growth. The authors feel that this is because of
increased water conservation awareness and improved regulations. However, Harlan et
al. (2009) report that global per capita water consumption has increased several times
faster than the population due to increased water consumption by more prosperous
nations. The authors believe that water conservation should focus on “reducing high-end
urban household consumption” (Harlan et al., 2009, p. 692).
Climate change. Climate change increases the frequency and severity of droughts
and causes substantial changes in rainfall patterns (Hall, 2009; Makki et al., 2011;
44
Polebitski & Palmer, 2010; Willis et al., 2011). As a result of these changes, experts
believe that climate change will significantly reduce surface water availability, increasing
dependence on groundwater supplies (Booker et al., 2012; Hall, 2009; Larson, Gustafson,
et al., 2009; Polebitski & Palmer, 2010; Ward et al., 2007; Willis et al., 2010). Climate
change can also have an alarming impact on water quality as contaminants are
concentrated and waterborne diseases increase in the diminishing water supplies (Binder,
2011; Nelson et al., 2011). Knutson (2008) recommends that water managers look
beyond the accepted statistical and managerial solutions and understand how climate
change will impact future water supplies. Gober and Kirkwood (2010) advocate for
informed climate change planning. The authors believe that designing a system for the
worst case scenario would be cost prohibitive, if even feasible, but that planning for the
best case scenario would leave the area “vulnerable to water shortage with little time to
adapt” (Gober & Kirkwood, 2010, p. 21298).
Drought. Although lack of rain is a primary factor in drought, it is not the only
cause (Corbella & Pujol, 2009). The Palmer Hydrological Drought Index (PHDI) utilizes
precipitation, temperature, evapotranspiration, soil runoff, and soil recharge to categorize
drought severity (see Figure 8; Balling, Gober, & Jones, 2008). Zero is normal, plus
numbers indicate an increasing severity of rainfall, and minus numbers indicate an
increasing severity of drought.
45
Figure 8. Monthly Palmer Hydrological Drought Index (PHDI) values for Phoenix,
Arizona. Reprinted from “Sensitivity of residential water consumption to variations in
climate: an intraurban analysis of Phoenix, Arizona,” by R. C. Balling, P. Gober, and
N. Jones, 2008, Water Resources Research, 44, p. 5. Copyright 2008 by the American
Geophysical Union. Reprinted with permission.
Periodic drought creates a natural competition for diminishing water supplies
(Harlan et al., 2009) and, while it is expected that water usage increase during drought
conditions (Balling et al., 2008; Bennear, Taylor, & Lee, 2011; Funk, 2007; Gober &
Kirkwood, 2010; Polebitski et al., 2011), during these periods of drought, households
may overwater their outdoor vegetation due to their perception of what the plant requires
rather than the plant’s actual requirements (Endter-Wada et al., 2008). Temperature,
precipitation, and humidity are major factors in outdoor water usage (Harlan et al., 2009).
Additionally, House-Peters et al. (2010) report that properties in Hillsboro, Oregon with
large outdoor landscapes are the most susceptible to drought with some census tracts
“consuming up to 1.85 times more water for external purposes during a drought summer
than an average summer” (p. 468). Polebitski, Palmer, and Waddell (2011) found that
46
even in Seattle, a city with the nation’s lowest outdoor watering rates, summer usage is
1.4 times higher than winter usage.
Water Consumption
Many of the factors that influence water consumption are beyond the control of
the water purveyor. Factors such as household income, temperature and rainfall, number
of bathrooms in the home, and the size, age, and value of the home and property can all
impact the amount of water used by the household. Although outside their influence,
prudent water planners must strategize for every variable that may influence residential
water consumption. Those factors that the water purveyor can control, such as
restrictions, rates, or rebates, are generally the most popular elements of water
conservation programs.
Temperature. Ambient temperature can have a profound impact on outdoor
water consumption (Lee et al., 2011). Higher outdoor temperatures can increase
evapotranspiration, the combination of evaporation and plant transpiration, which
increases the need for outdoor watering (Balling et al., 2008; Hall, 2009). As a result,
many researchers utilize temperature to evaluate water usage patterns (Balling et al.,
2008; Hall, 2009; Harlan et al., 2009; Kenney et al., 2008). Kenney et al. (2008) utilized
daily weather data from NOAA to construct average maximum daily temperature. The
authors discovered that, although the irrigation season saw an increase in water usage
regardless of actual temperature, the average temperature can be a valuable tool for
planning and management (Kenney et al., 2008).
47
Rainfall. A lack of rainfall can impact water storage by reducing recharge
(Larson, Gustafson, et al., 2009; Rosenberg et al., 2008; Willis et al., 2010) but
precipitation can also have varying impacts on outdoor water usage (Graymore & Wallis,
2010; Harlan et al., 2009; Kenney et al., 2008; Lee et al., 2011; Tsai et al., 2011; Willis et
al., 2009). In Southwest Australia, Graymore and Wallis (2010) found no relationship
between water usage and rainfall but Willis et al. (2009) discovered reduced irrigation
during increased rainfall in Australia’s Gold Coast region. In the United States, Harlan et
al. (2009) found that water usage was sensitive to rainfall in Phoenix, Arizona and Lee et
al. (2011) discovered significant seasonal effects on water usage in Miami-Dade County,
Florida. A prior study in Vandenberg Village (Allen, 2008) found that only rainfall had a
significant impact on water consumption. The study concluded that, in the model that
evaluated aggregated water consumption for all customers, for every inch of rainfall,
VVCSD customers conserved 2,655 CCF of water (Allen, 2008).
Yard size. The size of the household’s yard is one of the primary factors in
outdoor water consumption, along with landscape composition and irrigation technology
(Harlan et al., 2009; Willis et al., 2009). Harlan et al. (2009) report that irrigable lot size
had a significant effect on water usage. Many researchers (Balling et al., 2008; Hall,
2009; Harlan et al., 2009; Larson, Gustafson, et al., 2009; Mansur & Olmstead, 2012;
Rosenberg et al., 2008) found that square footage of the structures subtracted from the lot
size is an estimator of irrigable yard size. When yard size is not available, Mansur and
Olmstead (2012) believe that lot size is a valuable proxy for outdoor water consumption
48
preferences (e.g., lawns, gardens, and pools). Polebitski and Palmer (2010) and Polebitski
et al. (2011) found that lot size was significant in explaining water demand.
Home value. Home value, land value, and property value have been used
interchangeably with income as a predictor of water use (Dolnicar, Hurlimann, & Grün,
2012; Endter-Wada et al., 2008; Smith & Wang, 2008). Harlan et al. (2009) reported
“appraised house value increased water consumption more than any other socioeconomic
variable” (p. 692). House-Peters et al. (2010) found that homes with higher property
values used more outdoor water and Harlan et al. (2009) attribute lower home values with
lower income owners and higher home values with more affluent neighborhoods.
Number of Bathrooms. The United States Environmental Protection Agency
(EPA) reports that the bathroom is the largest consumer of indoor water (USEPA, 2008).
Because of the additional housing cost of multiple bathrooms, Mansur and Olmstead
(2012) and Harlan et al. (2009) opine that the number of bathrooms can be used as a
proxy for style of living that utilizes a higher water usage as well as a proxy for indoor
water consumption.
Year built. The age of a home has been used frequently to predict the water
efficiency of the home (Harlan et al., 2009; House-Peters et al., 2010; Jorgensen et al.,
2009; Kenney et al., 2008; Mansur & Olmstead, 2012; Nataraj & Hanemann, 2011;
Polebitski & Palmer, 2010; Polebitski et al., 2011). Mansur and Olmstead (2012) found
that old and new homes use less water than ‘middle aged’ homes. The explanation is that
“old homes may have smaller connections to water systems and fewer water-using
appliances, such as dishwashers and hot tubs, than newer homes. The newest homes in
49
the sample may have been constructed with water-conserving toilets and showerheads”
(Mansur & Olmstead, 2012, p. 336).
Water Conservation
Water conservation programs generally consist of price and non-price elements in
the form of rates, restrictions, and rebates. There is an ample supply of literature
espousing the benefits of both price and non-price approaches (Atwood et al., 2007;
Booker et al., 2012; Coleman, 2009; Halich & Stephenson, 2009; Kenney et al., 2008;
Millock & Nauges, 2010; Nataraj & Hanemann, 2011; Olmstead & Stavins, 2009; Tsai et
al., 2011; Worthington & Hoffman, 2008). Most experts believe that a comprehensive
conservation program employs a balance of both.
Water rates. Price elasticity is a commonly researched measure of demand (Hall,
2009; Rosenberg et al., 2008). “Elasticity is the measure of the responsiveness of quantity
demanded to the price level” (Munger, 2000, p. 232). Were water demand to follow a
typical demand model, as the price of a commodity increased, the demand for that item
would decrease. However, many researchers have concluded that, because the demand
for water does not generally change in response to changes in price, which indicates the
apparent inelasticity of water, the effect of water pricing on water consumption can be
unpredictable (Graymore & Wallis, 2010; Olmstead & Stavins, 2009; Osgood, 2011;
Pumphrey et al., 2008; St. Hilaire et al., 2008; Tsai et al., 2011). Munger (2000) has
coined this the “diamonds and water” paradox:
Diamonds face an elastic demand curve, and therefore involve only a
modest amount of consumer surplus. Water, on the other hand, faces a
50
highly inelastic demand, with enormous consumer surplus. Consequently,
though the prices may seem backward, the values the market system places
on diamonds, and on water, are exactly in line with what one would expect:
water is far more valuable. (p. 215)
Nonetheless, Funk (2007) found that increasing block-rate pricing structures can
encourage water consumption and Mansur and Olmstead (2012) found that outdoor
demand was more price elastic than indoor water usage. Additionally, Kenney et al.
(2008) concluded “residential water demand is largely a function of price, the impact of
nonprice demand management programs, and weather and climate” (p. 204).
Restrictions. Water restrictions include individual household water budgets and
regional limitations (Anderson-Wilk, 2008; Kenney et al., 2008; Knutson, 2008; Larson,
Gustafson, et al., 2009; Lee et al., 2011; Makki et al., 2011; Mansur & Olmstead, 2012;
Olmstead & Stavins, 2009; Pumphrey et al., 2008; Rosenberg et al., 2008; Willis et al.,
2010). On Australia’s Gold Coast, water restrictions are commonly enforced during times
of drought. Those restrictions “dictate a total outdoor watering ban and encourage
residents to consume 140 L/p/d” (Willis et al., 2009, p. 2000). While restrictions have
been an effective conservation tool, because VVCSD has implemented no policy
restricting water usage, this water conservation element has not been tested in this study.
Monetary incentives. Monetary incentives, generally in the form of rebates, are a
component of most water conservation programs (Anderson-Wilk, 2008; Bennear et al.,
2011; Booker et al., 2012; Funk, 2007; Hall, 2009; Harlan et al., 2009; Harou et al., 2010;
Kenney et al., 2008; Knutson, 2008; Larson, Gustafson, et al., 2009; Lee et al., 2011;
51
Makki et al., 2011; Mansur & Olmstead, 2012; Millock & Nauges, 2010; Olmstead &
Stavins, 2009; Osgood, 2011; Rosenberg et al., 2008; St. Hilaire et al., 2008; Tsai et al.,
2011; van Putten et al., 2011; Ward et al., 2007; Willis et al., 2009; Willis et al., 2010).
Monetary incentives are frequently utilized to encourage consumers to replace
high water-using features—such as showerheads, toilets, washing machines and
dishwashers, and turfgrass—or to install water saving products—such as climate
controlled irrigation controllers and rainwater tanks. The EPA lists over 100 rebate
programs across the nation (Bennear et al., 2011). Mansur and Olmstead (2012) are of the
opinion that rebates “could make everyone better off” (p. 333) and Tsai et al. (2011)
found that in the first four years of the water conservation program in Reading,
Massachusetts, the town realized “an overall average savings of approximately 3,950 m3 /
quarter” (p. 700). Lee et al. (2011) discovered similar results in Miami-Dade County,
Florida. However, Funk (2007) notes that rebates and retrofits do not typically “create
enough incentive to maximize end-user participation” (p. 178) and van Putten et al.
(2011) found that incentive programs with higher monetary rewards were more attractive
to consumers. Prudent conservation planners need to design programs that “allow
flexibility in terms of the legal arrangements, land use options, and other program
attributes” (van Putten et al., 2011, p. 2653).
Landscape Rebates
The primary goal of this research is to answer the question Can cash-4-grass
programs save water? Cash-4-Grass programs offer monetary incentives, in the form of
landscape rebates, in exchange for the conversion of grass to low-water-using landscapes
52
(Anderson-Wilk, 2008; Funk, 2007; Goldstein, 2012; Harlan et al., 2009; St. Hilaire et
al., 2008). Although “aesthetically pleasing landscapes and water-efficient landscapes are
not mutually exclusive concepts….homeowners consistently show a preference for
traditional, nonwater-conserving landscapes” (St. Hilaire et al., 2008, p. 2089). Larson,
Casagrande, Harlan, and Yabiku (2009) found that cultural norms reinforcing the
traditional lawn created a barrier against converting turfgrass to alternative landscapes. In
Australia, the driest inhabited continent on Earth, Head (2007) discovered that, in
Sydney, water was conserved inside the house so that it could be used outside to maintain
the landscape. Harlan et al. (2009) reports that, even though ill-suited for arid climates
and potentially damaging to the environment, water intensive turf grass remains the most
popular residential landscape choice in the United States and Canada. By 2005, turfgrass
had surpassed the area of any irrigated crop in the United States by 300% (Larson, Cook,
Strawhacker, & Hall, 2010).
Because outdoor water usage typically accounts for more than half of a
household’s water consumption, landscape rebates have the potential to significantly
reduce overall household water use (Gober & Kirkwood, 2010; Harlan et al., 2009;
Larson, Gustafson, et al., 2009; Millock & Nauges, 2010). Knutson (2008) predicts water
savings of 15 to 100% when landscape conservation measures are employed.
Some researchers believe that restrictions on outdoor watering is a desirable
solution (Atwood et al., 2007; Brennan, Tapsuwan, & Ingram, 2007; Halich &
Stephenson, 2009). However, most experts acknowledge that the public is resistant to
mandatory limits on water use and that the solution is not economically efficient
53
(Brennan et al., 2007; Grafton & Ward, 2008; Halich & Stephenson, 2009; Jones,
Evangelinos, Gaganis, & Polyzou, 2011; Kallis, Ray, Fulton, & McMahon, 2010;
Worthington & Hoffman, 2008). Another group of experts believe that reducing the
amount of irrigable lawn through a rebate program is a more efficient method of long-
term water conservation (Funk, 2007; Harlan et al., 2009; Kenney et al., 2008; Larson et
al., 2011; St. Hilaire et al., 2008).
Existing Literature and the Literature Gap
There is a significant gap in the literature when focusing on the effectiveness of
landscape rebates on water consumption. While it is possible to make some assumptions
with the existing water conservation rebate research on other water saving features and
the theoretical assumptions made by other researchers on the effectiveness of Cash-4-
Grass programs, to date, water savings in response to landscape rebates are anecdotal and
have not been scientifically tested (Anderson-Wilk, 2008; Funk, 2007; Gober &
Kirkwood, 2010; Harlan et al., 2009; Kenney et al., 2008; St. Hilaire et al., 2008).
Because landscape rebates can result in a significant monetary outlay for the agency,
VVCSD pays up to $1,000 per household, empirical evidence of their effectiveness is
important to the continuation of the programs (VVCSD, 2012d).
In response to this substantial gap in the literature regarding landscape rebates,
this quantitative study has been designed to evaluate the effectiveness of landscape
rebates on residential water consumption in Vandenberg Village, California. This paper
contributes to the literature by providing empirical data on the effectiveness of landscape
54
rebates that can be used by water professionals to expand their programs and by future
researchers to expand on the subject of water conservation.
Summary
Chapter 2 presented a review of the existing literature on water supply, water
demand, water conservation, landscape rebates, the effect of temperature, rainfall, yard
size, home value, number of bathrooms, year home was built, price of water, restrictions,
and rebates have on water consumption, the theoretical perspective of water conservation,
and the literature gap.
This chapter reiterated that California’s domestic water supplies are threatened by
population growth, drought, pollution, and climate change and explained that a review of
the literature has shown that many factors influence water consumption. The chapter
further explains the theoretical frameworks for this study, Ajzen and Fishbein’s TRA and
its successor, TPB, and that TRA was found to be a statistically significant indicator of
water conservation behavior. Temperature, rainfall, yard size, home value, number of
bathrooms, and age of the home were also found to be significant influences on water
consumption. Water rates, restrictions, and monetary incentives are primary components
of water conservation programs. Because outdoor landscaping accounts for a majority of
a household’s water consumption, traditional grass lawns have increasingly been targeted
for conservation measures by municipalities. However, scientific studies focusing on
Cash-4-Grass rebates are missing from the literature. In closing, this chapter explains that
this chapter will contribute to the literature by filling this gap.
55
Chapter 1 presented an in-depth introduction to water supply, water demand,
water conservation, and landscape rebates. Chapter 3 will present the research
methodology that has been used for this study. Chapter 4 will present the results of the
statistical analysis. Chapter 5 will present the interpretation of the findings, limitations of
the study, recommendations, and implications.
56
Chapter 3: Research Method
Introduction
This research was designed to measure the effectiveness of landscape rebates on
water consumption in the bedroom community of Vandenberg Village, California.
Analysis was performed on the public data requested from VVCSD, Santa Barbara
County (SBC), and the National Oceanic and Atmospheric Administration (NOAA).
This chapter outlines the methodology used to collect and analyze the data to
address the study’s research questions and hypotheses. The chapter covers the following
topics: a restatement of the problem; the research questions and hypotheses; research
design and approach; target population; setting and sample; instrumentation and
materials; data collection; protection of participants’ rights; data analysis; and dependent,
control, and independent variables.
Restatement of the Problem
Experts agree that California’s domestic water supplies are threatened by
population growth, drought, pollution, and climate change. In response, the California
legislature concluded that the most economical option to extend water resources was to
use less water per capita. Many different factors influence household water consumption
but bathrooms are the largest consumer of indoor water use. Thus, the age of the toilets,
the number of bathrooms, and the age of the home can influence the amount of indoor
water consumption. To encourage household savings, water conservation programs
generally target indoor water use through toilet and washing machine rebates. The current
literature focuses on restrictions, pricing structures, and toilet rebates.
57
However, outdoor water usage accounts for the majority of a household’s water
usage. The size of the yard is the primary factor, but an increase in temperature or a
decrease in rainfall can also cause a household to increase its outdoor water use. Experts
hypothesize that elements of a water conservation program that target outdoor water use,
such as landscape rebates, can have a significant impact on domestic water usage.
Nevertheless, to date, water savings in response to landscape rebates are anecdotal and
have not been scientifically tested. Because of this substantial gap in the literature, this
quantitative study was designed to evaluate the effectiveness of landscape rebates on
residential water consumption in Vandenberg Village, California. Secondary data was
collected from VVCSD, SBC, and the NOAA.
Research Questions and Hypotheses
The following research questions have been examined in the process of testing the
hypotheses:
1. Do consumers who receive landscape rebates use less water than those
consumers who do not?
H01: There is no significant difference in water consumption between those
who receive a landscape rebate and those who do not receive a
landscape rebate when controlling for receipt of a toilet or washing
machine rebate, size of property, number of bathrooms, value of home,
age of home, and price per unit of water.
Ha1: There is a significant difference in water consumption between those
who receive a landscape rebate and those who do not receive a
58
landscape rebate when controlling for receipt of a toilet or washing
machine rebate, size of property, number of bathrooms, value of home,
age of home, and price per unit of water.
2. Do landscape rebate recipients use less water after receiving the rebate than
before?
H02: There is no significant difference in water consumption in the 24 months
before and 24 months after receipt of a landscape rebate when
controlling for receipt of a toilet or washing machine rebate, size of
property, number of bathrooms, value of home, age of home, amount of
rainfall, average temperature, and price per unit of water.
Ha2: There is a significant difference in water consumption in the 24 months
before and 24 months after receipt of a landscape rebate when
controlling for receipt of a toilet or washing machine rebate, size of
property, number of bathrooms, value of home, age of home, amount of
rainfall, average temperature, and price per unit of water.
Research Design and Approach
The purpose of this study was to investigate the relationship between the
dependent variable of monthly residential water usage and the independent variables of
receipt of a toilet, washer, and/or landscape rebate, size of irrigable property, number of
toilets, value of home, age of home, amount of monthly rainfall, average monthly high
temperature, and price per unit of water. The primary goal of this research is to answer
the question Can cash-4-grass programs save water? To answer the question this study
59
has utilized a multiple time-series quasi-experimental research design to evaluate the
effectiveness of landscape rebates on residential water consumption in Vandenberg
Village, California.
This type of research design has been selected because participants cannot easily
be randomly assigned for this type of study, as it would be cost-prohibitive to randomly
select residential properties for turf reduction. Additionally, there is a significant amount,
a minimum of ten years, of aggregated time-series and water conservation monthly rebate
data available. Rebate recipients can also be easily isolated from the customer base for
separate testing.
A time series “is a set of observations obtained by measuring a single variable
regularly over a period of time” (IBM Corporation, 2011, p. 1). An aggregated time series
is time series data whose values are combined over equal intervals of time. For this study,
the aggregated time series observations are monthly water meter readings obtained for the
customers of Vandenberg Village, California. The water meter reading represents the
continuous water usage for the previous billing period. VVCSD Ordinance §2.9.1 et seq.
allows for the water meter readings to be collected monthly at 25 to 35 day intervals
provided the annual average is 30 days (VVCSD, 2010).
A time-series design “is the presence of a periodic measurement process on some
group or individual and the introduction of an experimental change into this time series of
measurements” (Campbell & Stanley, 1963, p. 37). According to Campbell and Stanley
(1963), a multiple time-series design is the best of the quasi-experimental research
designs. Also known as intervention analysis, this technique evaluates the impact of an
60
event, the introduction of the experimental change, on the time series observations
(Balkin & Ord, 2001; Harvey, 1989). The experimental change, or intervention, in this
study is the introduction of landscape rebates. The customers initiate the intervention by
replacing their turfgrass and requesting a rebate.
Autocorrelation is a common issue with time-series data. To decrease the
likelihood of autocorrelation, the data for this research question Do consumers who
receive landscape rebates use less water than those consumers who do not? has been
split into separate models consisting of 12 months of data. Additionally, the data for the
research question Do landscape rebate recipients use less water after receiving the rebate
than before? has been limited to the data 24 months before and 24 months after the
receipt of a landscape rebate.
Target Population, Setting, and Sample
Target Population
In water conservation research, a population is generally the customers serviced
by the water purveyor or a random sampling of customers, if the service area is too large
to reliably test each individual. The population for this study are the customers serviced
by VVCSD. Ten years of historical data for its water customers has been gathered via a
public records request. This study has been approved and sponsored by the general
manager of VVCSD. A letter of cooperation can be found in Appendix A.
Setting and Sample
The setting for this research is Vandenberg Village, California. Vandenberg
Village is a bedroom community of 6,497 (United States Census Bureau, 2010c) located
61
in an unincorporated area of Santa Barbara County north of the city of Lompoc.
Domestic water is supplied by VVCSD. Due to the water usage collection method
utilized by VVCSD, analysis of usage at the individual resident level is not feasible.
Therefore, this study has analyzed the monthly household water use. The population for
this study has been limited to residential connections in Vandenberg Village. As of
June 30, 2012, the number of residential connections served by VVCSD was 2,577
(VVCSD, 2012a).
Yanovitzky and VanLear (2008) state that time series data analysis “performs best
when the number of sequential observations available for analysis is 50 or greater, the
unit of time is consistent among all variables measured, and each time series is uniform
and unbroken” (p. 108). A nonprobability convenience sample has been utilized for this
research study because those customers who have received water conservation rebates
can be easily isolated from those customers that have received no rebate. The available
data includes monthly observations for a minimum of 10 years times the number of
customers selected for each regression model. Therefore, all three conditions suggested
by Yanovitzky and VanLear have been met.
Instrumentation and Materials
This quantitative research study has utilized available data that has been collected
through public records requests to VVCSD and SBC as well as Internet searches. Table 5
details the variables and their measurements.
62
Table 5
Research variables
Variable Source Collection Method
Water use VVCSD Public records request
Rebate VVCSD Public records request
Price VVCSD Public records request
Temperature NOAA Internet search
Rainfall SBC Internet search
Yard Size SBC Public records request
Number of Toilets SBC Public records request
Age of Home SBC Public records request
Value of Home SBC Public records request
Data Collection
The following time-series data, at the household level, has been requested from
VVCSD via a public records request:
• Domestic water use
• Rebates received
• Price/rate per CCF of water
The following information, at the assessor’s parcel number (APN) level, has been
requested from SBC clerk, recorder, and assessor via a public records request:
• Acreage of property
• Square footage of structures
• Number of bathrooms
• Value of land
63
• Value of structures
• Year built
The monthly rainfall for Vandenberg Village has been obtained from the SBC
Flood Control District website (Santa Barbara County, 2008) and the monthly
temperature for Vandenberg Village has been obtained from the NOAA website (NOAA,
n.d.).
Protection of Participants’ Rights
The confidentiality of the data collected through public records for this research
has been protected at all times. The only individual identifier in the raw data is the APN.
This identifier is assigned by the SBC clerk, recorder, and assessor’s office and is unique
to every house. The APN has been used to combine the data from VVCSD and SBC into
a single database to be used in the IBM SPSS® Statistics software package for statistical
analysis. Only the researcher has had access to the APN and it has not been referenced in
the written report. The raw data and the combined database file has been stored
electronically on a password-protected portable hard drive and will be kept in a fireproof
safe for a minimum of 5 years. At that time, the data on the hard drive will be destroyed.
Walden Institutional Review Board approval #09-13-13-0277917 was obtained prior to
data collection.
Reliability and Validity
Reliability and validity refers to the credibility of the research measurement
(McNabb, 2008). Measurement reliability is the extent that the research is reproducible
and validity is the extent that the research measures what it is intended to measure
64
(Frankfort-Nachmias & Nachmias, 2008; Rossi, Lipsey, & Freeman, 2004; Trochim,
2006b). One threat to internal validity can be the passage of time. Water usage is
measured constantly but the water purveyor generally reads the meter just once a month.
Therefore, in order to collect enough data to make an informed analysis, many months,
ideally years, of data must be analyzed. Additionally, during that passage of time, events
occur that impact water usage. Drought increases the annual average water consumption
while a rainy year decreases the average. To increase validity in this study, the regression
equations have analyzed many factors that impact water consumption and each variable
selected has been thoroughly researched to verify that it is contributing to the analysis.
All of the data has been obtained through public records and, because California law
requires that residential customers to request an investigation within 5 days of receiving a
disputed bill (California, 1988), this study assumes that the data is reliable and accurate.
Measures
Dependent Variables
Domestic water use. Domestic water use [WATER] is a ratio measurement of
time series data that represents the amount of water consumed by a residential customer
on a monthly basis. Residential water usage in Vandenberg Village is recorded on a
monthly basis by electronically reading the meter attached to the residential service line.
Units of water are measured in increments of one hundred cubic foot (CCF).
Lagged Dependent Variables
Domestic water use – lagged. Lagged domestic water use [WATERt1] is a ratio
measurement of time series data that represents the amount of water consumed by a
65
residential customer on a monthly basis lagged one month. The lagged variable has been
used to help predict the nonlagged version of the same dependent variable (Vogt &
Johnson, 2011). Ruijs, Zimmermann, and van den Berg (2008) used this technique to
correct for autocorrelation in their regression models.
Independent Variables
Landscape rebates. Landscape rebates [GRASS] is a nominal level dummy
variable that focuses on the receipt of a rebate for the replacement of turf grass and is
used to determine the impact of rebates on water consumption. A “1” has been used for
the months following the receipt of the rebate. A “0” has been used for those months
before the rebate was received. Because rebates are reimbursements there is a delay
between the date that the eligible item was installed and the rebate received. Therefore,
rebates are considered leading indicators and, subsequently, the variable has been lead by
one month to account for water saved before the rebate was requested. A negative
coefficient was expected on this variable because those customers who receive a rebate
should reduce their total water consumption by actively reducing their water usage.
Control Variables
Rebates. Rebates [TOILET] [WASHER] are nominal level dummy variables that
focus on the receipt of a rebate for the replacement of a toilet and/or washing machine
and are used to determine the impact of rebates on water consumption. As mentioned
above, a “1” has been used for the months following the receipt of the rebate. A “0” has
been used for those months before the rebate was received. Again, because rebates are
reimbursements there is a delay between the date that the eligible item was installed and
66
the rebate received. Therefore, rebates are considered leading indicators and,
subsequently, the variable has been lead by one month to account for water saved before
the rebate was requested. A negative coefficient was expected on this variable because
those customers who receive a rebate should reduce their total water consumption by
actively reducing their water usage.
Price/rate. Price or rate [PRICE] is a ratio measurement that represents the
amount per CCF of water that the customer pays. This variable is used to determine the
impact of price on water consumed. VVCSD utilizes tiered water rates in an effort to
encourage water conservation. As more water is used, the rate per unit increases until the
customer reaches the punitive top tier. This study assumes that the highest tier is a
deterrent against water waste. Therefore, this study has utilized the price at the highest
tier. Because there is a delay between the month that the price change goes into effect and
when a consumer takes notice and begins conserving, price is considered a lagging
indicator and, subsequently, this variable has been lagged one month to account for this
delay, in accordance with the Taylor-Nordin specification (Nordin, 1976; Taylor, 1975).
As a basic human necessity, water is somewhat insensitive to price. Therefore, it was not
expected that price increases would significantly impact water consumption. However, a
negative coefficient would indicate a reduction in water consumption in response to the
price change.
Yard size. Yard size [SIZE] is an ordinal measurement that represents the area
that has the potential for outdoor water use. This variable is calculated by subtracting the
square footage of the structures from the total square footage of the property. A positive
67
coefficient is expected on this variable because water use increases as more outdoor area
is landscaped with turf grass.
Number of bathrooms. Number of bathrooms [BATH] is an ordinal
measurement that represents the number of bathrooms in a home. A negative coefficient
is expected on this variable because as increased amounts of older toilets are replaced
with low-flow toilets, water usage decreases. A positive coefficient is not unexpected on
this variable because as the number of toilets in a home increases, water usage may also
increase.
Age of home. Age of home [AGE] is a nominal level dummy variable that
represents the year the home was built. A “1” has been used for homes built before 1992.
A “0” has been used for homes built after 1991. A positive coefficient is expected on this
variable for homes built in California before January 1992 because the older Plumbing
Code results in a higher indoor water usage for the same tasks.
Rainfall. Rainfall [RAIN] is a ratio measurement that represents the monthly
amount of rainfall. This variable is used to determine the impact that rainfall has on the
amount of monthly residential water consumed. A negative coefficient is expected on this
variable because water use increases as rainfall decreases.
Temperature. Temperature [TEMP] is an interval measurement that represents
the monthly average high temperature. Because the average high temperature in
Vandenberg Village ranges between 60° F and 80° F (The Weather Channel, 2012),
temperature is not expected to be statistically significant to water consumption.
68
Value of home. Value of home [VALUE] is an ordinal measurement that
represents the 2012 Assessed Value of the land and structures. This variable acts as a
representative for income and has been adjusted from the 2013 SBC assessed value
utilizing the ENR 20 Cities Construction Cost Index (Engineering News-Record, 2012).
A positive coefficient is expected on this variable because water usage increases as the
value of the home increases (Harlan et al., 2009).
Data Analysis
IBM SPSS® Statistics version 21 has been used to generate descriptive statistics
and perform a linear regression analysis (IBM Corporation, 2012). Regression analysis is
commonly utilized to measure direct effects and has been used in this study to predict the
amount of water consumed as a function of landscape rebates while controlling for
washing machine and toilet rebates, price per unit, total monthly rainfall, average
monthly temperature, size of yard, number of toilets, value of home, and age of home.
The statistical significance of the relationship between the amount of water consumed
each month and landscape rebates has been used as an indicator of the effectiveness of
VVCSD landscape rebates.
Descriptive Statistics Analysis
Descriptive statistics has been used to describe the basic features of the data in
this study (Trochim, 2006a). The median, standard deviation, and correlation has been
calculated for study variables where appropriate. The level of correlation has been
examined for the relationship between the control variables and dependent variable.
69
Direct Effects Using Regression Analysis
The hypotheses assume that landscape rebates are positively associated with water
conservation. Linear regression analysis has been utilized to test for the statistical
significance of the hypothesized relationship.
Regression Analysis Equations
A review of the literature was used to identify factors that have a significant
impact on water consumption. Some of these factors have been selected as regression
analysis variables, detailed in a separate section. Additionally, because of the different
assumptions for different types of customers, separate water demand regression models
have been created. Table 6 details each model and its corresponding regression analysis
equation.
The models have analyzed similar variables for two different sets of customers.
Model RQ1 (All Customers) has analyzed the data for the landscape rebate customers as
well as a comparable number of non-landscape rebate customers with similar property
characteristics. The goal of this model is to determine the extent to which landscape
rebate customers use less water than non-landscape rebate customers. Model RQ2
(Landscape Rebate Customers) has analyzed the data for landscape recipient customers to
determine the extent to which their water consumption reduced as a result of receiving a
landscape rebate.
70
Table 6
Regression Analysis Equations
Model
Equation
Rebate/Non-Rebate
Customers
(Model RQ1)
WATER = β
0
+ β
1
WATER
t-1
+ β
2
PRICE + β
3
SIZE +
β4BATH + β5VALUE + β6AGE + β7TOILET
+ β8WASHER + β9GRASS + εt
Landscape Rebate Customers
(Model RQ2)
WATER = β
0
+ β
1
WATER
t-1
+ β
2
PRICE + β
3
SIZE +
β4BATH + β5VALUE + β6AGE + β7TOILET
+ β8WASHER + β9GRASS + β10RAIN + β11TEMP +
εt
Summary
Chapter 3 presents the research methodology for this study and outlines the
method that has been used to collect and analyze the data to evaluate the research
questions and hypotheses developed for this study. It includes: a restatement of the
problem; the research questions and hypotheses; research design and approach; target
population; setting and sample; instrumentation and materials; data collection; protection
of participants’ rights; data analysis; and dependent, control, and independent variables.
This chapter explains that the purpose of this study is to investigate the
relationship between the dependent variable of monthly residential water usage and the
independent variables of receipt of a toilet, washer, and/or landscape rebate, size of
irrigable property, number of toilets, value of home, age of home, amount of monthly
rainfall, average monthly high temperature, and price per unit of water. The target
population for this study was the customers of VVCSD. Secondary data were collected
from VVCSD, SBC, and NOAA. The primary goal of this research is to answer the
71
question Can cash-4-grass programs save water? through research questions which
focused on the extent that water conservation rebates reduce water consumption.
Chapter 1 presented an in-depth introduction to water supply, water demand,
water conservation, and landscape rebates. Chapter 2 presented the review of the existing
literature for the study. Chapter 4 will present the results of the statistical analysis.
Chapter 5 will present the interpretation of the findings, limitations of the study,
recommendations, and implications.
72
Chapter 4: Results
Introduction
This research was designed to measure the effectiveness of landscape rebates on
water consumption in the bedroom community of Vandenberg Village, California. The
analysis was performed on public data obtained from VVCSD, SBC, and NOAA. This
chapter presents the results of the statistical analysis. It includes: a description of the data
collection process; an explanation of the data transformation performed; a description of
the data analysis; a testing of the hypotheses; and a summary of the findings.
Data Collection
Data collection was consistent with the process outlined in Chapter 3 and per the
terms of Walden IRB. The data were collected through public records requests to
VVCSD and SBC and through data downloads from the NOAA and SBC websites.
VVCSD and SBC responded to the request with electronic data. As a result, all of the
data required for analysis were received in electronic format. Therefore, none of the data
required manual transcription, which reduced the possibility of data entry errors.
VVCSD provided 15 separate data files including a separate water usage file for
each fiscal year from 2002 through 2013 sorted by customer identification, one file
containing all rebates by address, one file containing water rates from 1990 through
present, and one file containing the key to convert the customer identification to APN.
The SBC assessor’s office provides to the public an ASCII text file containing for
every property in the county: ownership, mailing and situs address, assessment, Tax Rate
Area (TRA), and acreage information. By special request, SBC provided to the researcher
73
a data file, which contained the standard information as well as property characteristic
such as the square footage of the home, the year the home was built, and the number of
bedrooms, bathrooms, and fireplaces.
On its website, the SBC Water Agency publishes daily, monthly, and annual
rainfall data collected from rainfall stations placed throughout the county. The historic
monthly and yearly rainfall records were downloaded for Station 205 at the Burton Mesa
Fire Station #51 located within the service area of VVCSD.
On its website, NOAA provides access to the data collected at the weather
stations within their land-based Automated Weather Observing System (AWOS). The
historic hourly temperature data was downloaded for the AWOS stations near Lompoc
and Vandenberg Air Force Base.
Data Transformation
Electronic data from VVCSD, SBC, and NOAA were transformed as needed and
merged into single databases for each dataset. The data for water consumption (WATER)
was provided by VVCSD separated by month, year, and APN. This format was used for
the remainder of the data.
• WATERt1 – this variable constructed by lagging water consumption for each
APN by one month.
• PRICE – no transformation was required.
• TOILET/WASHER/GRASS – month and year of rebate was matched with the
corresponding APN in database containing all VVCSD Assessor Parcel
Numbers by utilizing Microsoft Excel® 2013 IF function (Microsoft
74
Corporation, 2012). Further transformation included constructing dummy
variables by converting date of rebate and subsequent dates to “1” and all
dates prior to rebate to “0” in database separating all data by month, year, and
APN. The final transformation included leading variable for each APN by one
month.
• SIZE – this variable was constructed by subtracting size of structures (square
footage) from size of lot (square footage).
• BATH – the number of bathrooms was rounded up to account for the number
of toilets per home (e.g., 2½ bath = 3 toilets).
• VALUE – this variable was constructed by adding the land value and the
improvement value for September 2013 and multiplying the sum by the ENR
20 Cities Construction Cost Index.
• AGE – this dummy variable was constructed by converting year built for
homes built before 1992 to “1” and all dates after 1991 to “0” in database
separating all data by month, year, and APN.
• RAIN – no transformation was required.
• TEMP – no transformation was required. However, the high temperature for
each month was selected from the hourly temperature readings by utilizing
Microsoft Excel® 2013 MAX function (Microsoft Corporation, 2012).
The Microsoft Excel® 2013 VLOOKUP function was utilized to combine all of
the electronic files into single files for each dataset by matching month, year, and APN
(Microsoft Corporation, 2012).
75
Model Formation
Autocorrelation is a common issue with time-series data. To decrease the
likelihood of autocorrelation, annual models were formed so that the data could be
analyzed by year rather than aggregated over 10 years. Individual datasets were created
for RQ1 for the calendar years 2007 through 2012 and analyzed separately. Additionally,
the data for RQ2 has been limited to the data 24 months before and 24 months after the
receipt of a landscape rebate. Finally, the average number of bathrooms, yard size, value
of home, and toilet and washing machine rebates was compared for each model against
the database as a whole with the goal of being representative of the population in its
entirety.
Data Analysis
IBM SPSS® Statistics version 21 was utilized to generate descriptive statistics
and perform a linear regression analysis (IBM Corporation, 2012). Descriptive statistics
include means, standard deviations, correlations, and frequencies and percentages.
Regression analysis is commonly utilized to measure direct effects and was used in this
study to predict the amount of water consumed as a function of landscape rebates while
controlling for washing machine and toilet rebates, price per unit, total monthly rainfall,
average monthly temperature, size of yard, number of bathrooms, value of home, and age
of home. The statistical significance of the relationship between the amount of water
consumed each month and landscape rebates has been used as an indicator of the
effectiveness of VVCSD landscape rebates.
76
Hypothesis Testing
Research Question 1 and Hypothesis
Do consumers who receive landscape rebates use less water than those consumers
who do not?
H01: There is no significant difference between the receipt of landscape
rebates and no receipt of landscape rebates on water consumption
when controlling for receipt of a toilet or washing machine rebate,
size of property, number of bathrooms, value of home, age of home,
and price per unit of water.
Ha1: There is a significant difference between the receipt of landscape
rebates and no receipt of landscape rebates on water consumption
when controlling for receipt of a toilet or washing machine rebate,
size of property, number of bathrooms, value of home, age of home,
and price per unit of water.
Descriptive statistics. For easier evaluation, each year’s descriptive statistics
results have been presented in two tables. The complete statistical results can be found in
Appendix C.
Calendar Year 2007. Although washing machine and Cash-4-Grass rebates were
added to the VVCSD water conservation program in July 2007, no rebates were issued
that year. As a result, the variables for washing machine rebate and Cash-4-Grass rebate
were constant at 0. Therefore, the mean and standard deviation are 0.00 and the
77
correlation coefficients cannot be computed. The means, standard deviations, and
correlations are presented in this section for reference only.
The mean water consumption was 19.98 (SD = 20.012), mean price was $1.33
(SD = $0.08), mean size was 10,584.20 sq. ft. (SD = 8,423.158), mean number of
bathrooms was 2.41 (SD = .814), mean value of home was $180,352.60 (SD =
$116,387.78), mean age was .98 (SD = .137), and mean toilet rebate was .29 (SD = .455).
The correlation coefficients for number of bathrooms (.641), value of home (.583), size
of yard (.559), and toilet rebate (.218) are statistically significant at the one percent level
on a one-tailed test and indicate that those variables have a direct relationship with water
consumption. The positive correlation suggests that as the amount of the variable
increases, the amount of water consumption also increases. The size of the correlation
coefficient for number of bathrooms indicates that the variable has a strong relationship
with water consumption, the size of the correlation coefficients for value of home and
size of yard indicate moderate relationships, and the coefficient for toilet rebate indicates
a weak relationship.
Calendar Year 2008. The mean, standard deviation, and correlations between the
variables for calendar year 2008 are shown in Table 7 and Table 8.
78
Table 7
Means, Standard Deviations, and Intercorrelations for Water Usage and Water
Consumption Predictor Variables for Calendar Year 2008
Variable
M
SD
WATER
1
3
4
5
WATER
19.68
22.052
–
.803**
.458**
.619**
.526**
Predictor variable
1.WATERt1
20.05
22.096
.803**
–
.469**
.622**
.538**
2.PRICE
$1.48
$0.06
.008
.112**
-.004
-.004
.037
3.SIZE
11132.04
9001.029
.458**
.469**
–
.615**
.600**
4.BATH
2.42
.806
.619**
.622**
.615**
–
.736**
5.VALUE
$193526.40
$123435.10
.526**
.538**
.600**
.736**
–
6.AGE
.98
.135
-.006
-.007
.081*
.072
-.030
7.TOILET
.33
.470
.199**
.203**
.073
.117**
.031
8.WASHER
.03
.168
.023
.012
.122**
.124**
.124**
9.GRASS
.02
.140
-.095*
-.086*
-.060
-.075
-.082*
* p < 0.05. ** p < 0.01.
Table 8
Means, Standard Deviations, and Intercorrelations for Water Usage and Water
Consumption Predictor Variables for Calendar Year 2008
Variable
M
SD
2
6
7
8
9
WATER
19.68
22.052
.008
-.006
.199**
.023
-.095*
Predictor variable
1.WATERt1
20.05
22.096
.112**
-.007
.203**
.012
-.086*
2.PRICE
$1.48
$0.06
–
.001
.031
.037
.168**
3.SIZE
11132.04
9001.029
-.004
.081*
.073
.122**
-.060
4.BATH
2.42
.806
-.004
.072
.117**
.124**
-.075
5.VALUE
$193526.40
$123435.10
.037
-.030
.031
.124**
-.082*
6.AGE
.98
.135
.001
–
.096*
.024
.020
7.TOILET
.33
.470
.031
.096*
–
.247**
-.007
8.WASHER
.03
.168
.037
.024
.247**
–
-.025
9.GRASS
.02
.140
.168**
.020
-.007
-.025
–
* p < 0.05. ** p < 0.01.
In calendar year 2008, the mean water consumption was 19.68 (SD = 22.052),
mean price was $1.48 (SD = $0.06), mean size was 11,132.04 sq. ft. (SD = 9,001.029),
mean number of bathrooms was 2.42 (SD = .806), mean value of home was $193,526.40
79
(SD = $123,435.10), mean age was .98 (SD = .135), mean toilet rebate was .33 (SD =
.470), mean washing machine rebate was .03 (SD = .168), and mean Cash-4-Grass rebate
was .02 (SD = .140).
The correlation coefficients for number of bathrooms (.619), value of home
(.526), size of yard (.458), and toilet rebate (.199) are statistically significant at the one
percent level on a one-tailed test and indicate that those variables have a direct
relationship with water consumption. The positive correlation suggests that as the amount
of the variable increases, the amount of water consumption also increases. The
correlation coefficient for Cash-4-Grass rebate (-.095) is statistically significant at the
five percent level on a one-tailed test and indicates that the variable has an indirect
relationship with water consumption. The negative correlation suggests that as the
amount of the variable increases, the amount of water consumption decreases. The size of
the correlation coefficient for number of bathrooms indicates that the variable has a
strong relationship with water consumption, the size of the correlation coefficients for
value of home and size of yard indicate moderate relationships, and the coefficients for
toilet rebate and Cash-4-Grass rebate indicate a weak relationship.
Calendar Year 2009. The mean, standard deviation, and correlations between the
variables for calendar year 2009 are shown in Table 9 and Table 10.
80
Table 9
Means, Standard Deviations, and Intercorrelations for Water Usage and Water
Consumption Predictor Variables for Calendar Year 2009
Variable
M
SD
WATER
1
3
4
5
WATER
18.40
22.746
–
.855**
.400**
.646**
.543**
Predictor variable
1.WATERt1
18.24
22.705
.855**
–
.405**
.643**
.537**
2.PRICE
$1.90
$0.41
.037
.133**
.000
.000
.004
3.SIZE
11066.65
8965.708
.400**
.405**
–
.616**
.599**
4.BATH
2.42
.803
.646**
.643**
.616**
–
.735**
5.VALUE
$199685.46
$126391.14
.543**
.537**
.599**
.735**
–
6.AGE
.98
.134
.010
.012
.080*
.071
-.030
7.TOILET
.38
.486
.072
.081*
.016
.059
-.069
8.WASHER
.05
.208
.037
.055
.105**
.158**
.082*
9.GRASS
.14
.345
-.169**
-.161**
-.028
-.093*
-.081*
* p < 0.05. ** p < 0.01.
Table 10
Means, Standard Deviations, and Intercorrelations for Water Usage and Water
Consumption Predictor Variables for Calendar Year 2009
Variable
M
SD
2
6
7
8
9
WATER
18.40
22.746
.037
.010
.072
.037
-
.169
**
Predictor variable
1.WATERt1
18.24
22.705
.133**
.012
.081*
.055
-
.161
**
2.PRICE
$1.90
$0.41
–
.000
.003
.037
.090*
3.SIZE
11066.65
8965.708
.000
.080*
.016
.105**
-.028
4.BATH
2.42
.803
.000
.071
.059
.158**
-.093*
5.VALUE
$199685.46
$126391.14
2
.004
-.030
-.069
.082*
-.081*
6.AGE
.98
.134
.000
–
.107**
.030
.054
7.TOILET
.38
.486
.003
.107**
–
.279**
.058
8.WASHER
.05
.208
.037
.030
.279**
–
-.087*
9.GRASS
.14
.345
.090*
.054
.058
-.087*
–
* p < 0.05. ** p < 0.01.
In calendar year 2009, the mean water consumption was 18.40 (SD = 22.746),
mean price was $1.90 (SD = $0.41), mean size was 11,066.65 sq. ft. (SD = 8,965.708),
81
mean number of bathrooms was 2.42 (SD = .803), mean value of home was $199,685.46
(SD = $126,391.14), mean age was .98 (SD = .134), mean toilet rebate was .38 (SD =
.486), mean washing machine rebate was .05 (SD = .208), and mean Cash-4-Grass rebate
was .14 (SD = .345).
The correlation coefficients for number of bathrooms (.646), value of home
(.543), and size of yard (.400) are statistically significant at the one percent level on a
one-tailed test and indicate that those variables have a direct relationship with water
consumption. The positive correlation suggests that as the amount of the variable
increases, the amount of water consumption also increases. The correlation coefficient for
Cash-4-Grass rebate (-.169) is statistically significant at the one percent level on a one-
tailed test and indicates that the variable has an indirect relationship with water
consumption. The negative correlation suggests that as the amount of the variable
increases, the amount of water consumption decreases. The size of the correlation
coefficient for number of bathrooms indicates that the variable has a strong relationship
with water consumption, the size of the correlation coefficients for value of home and
size of yard indicate moderate relationships, and the coefficient for Cash-4-Grass rebate
indicates a weak relationship.
Calendar Year 2010. The mean, standard deviation, and correlations between the
variables for calendar year 2010 are shown in Table 11 and Table 12.
82
Table 11
Means, Standard Deviations, and Intercorrelations for Water Usage and Water
Consumption Predictor Variables for Calendar Year 2010
Variable
M
SD
WATER
1
3
4
5
WATER
15.58
16.225
–
.794**
.458**
.522**
.471**
Predictor variable
1.WATERt1
15.82
16.453
.794**
–
.450**
.541**
.493**
2.PRICE
$2.38
$0.00
c
c
c
c
c
3.SIZE
11001.66
8905.403
.458**
.450**
–
.616**
.596**
4.BATH
2.41
.797
.522**
.541**
.616**
–
.722**
5.VALUE
$206161.60
$129577.88
.471**
.493**
.596**
.722**
–
6.AGE
.98
.132
-.003
-.001
.079*
.069
-.028
7.TOILET
.41
.493
.060
.061
-.011
.022
-
.104
**
8.WASHER
.08
.278
.105**
.118**
.122**
.272**
.072
9.GRASS
.22
.412
-.184**
-.182**
-.010
-.080*
-.079*
* p < 0.05. ** p < 0.01. c. variable is constant.
Table 12
Means, Standard Deviations, and Intercorrelations for Water Usage and Water
Consumption Predictor Variables for Calendar Year 2010
Variable
M
SD
2
6
7
8
9
WATER
15.58
16.225
c
-.003
.060
.105**
-
.184
**
Predictor variable
1.WATERt1
15.82
16.453
c
-.001
.061
.118**
-
.182
**
2.PRICE
$2.38
$0.00
–
c
c
c
c
3.SIZE
11001.66
8905.403
c
.079*
-.011
.122**
-.010
4.BATH
2.41
.797
c
.069
.022
.272**
-.080*
5.VALUE
$206161.60
$129577.88
c
-.028
-.104**
.072
-.079*
6.AGE
.98
.132
c
–
.113**
.041
.071
7.TOILET
.41
.493
c
.113**
–
.362**
.128**
8.WASHER
.08
.278
c
.041
.362**
–
.047
9.GRASS
.22
.412
c
.071
.128**
.047
–
* p < 0.05. ** p < 0.01. c. variable is constant.
In calendar year 2010, the mean water consumption was 15.58 (SD = 16.225),
mean size was 11,001.66 sq. ft. (SD = 8,905.403), mean number of bathrooms was 2.41
83
(SD = .797), mean value of home was $206,161.60 (SD = $129,577.88), mean age was
.98 (SD = .132), mean toilet rebate was .41 (SD = .493), mean washing machine rebate
was .08 (SD = .278), and mean Cash-4-Grass rebate was .22 (SD = .412). The variable
for price was constant at $2.38. Therefore, the mean was $2.38 and the standard deviation
is 0.00.
The correlation coefficients for number of bathrooms (.522), value of home
(.471), size of yard (.458), and washing machine rebate (.105) are statistically significant
at the one percent level on a one-tailed test and indicate that those variables have a direct
relationship with water consumption. The positive correlation suggests that as the amount
of the variable increases, the amount of water consumption also increases. The
correlation coefficient for Cash-4-Grass rebate (-.184) is statistically significant at the one
percent level on a one-tailed test and indicates that the variable has an indirect
relationship with water consumption (see Figure 9). The negative correlation suggests
that as the amount of the variable increases, the amount of water consumption decreases.
The size of the correlation coefficients for number of bathrooms, value of home, and size
of yard indicate moderate relationships and the coefficients for washing machine rebate
and Cash-4-Grass rebate indicate a weak relationship. The variable for price was constant
during the analysis period. Therefore, the correlation coefficient cannot be computed for
this variable.
85
Table 13
Means, Standard Deviations, and Intercorrelations for Water Usage and Water
Consumption Predictor Variables for Calendar Year 2011
Variable
M
SD
WATER
1
3
4
5
WATER
15.83
18.175
–
.768**
.487**
.600**
.446**
Predictor variable
1.WATERt1
15.66
18.050
.768**
–
.475**
.582**
.433**
2.PRICE
$2.38
$0.00
c
c
c
c
c
3.SIZE
10928.41
8811.361
.487**
.475**
–
.614**
.594**
4.BATH
2.40
.787
.600**
.582**
.614**
–
.700**
5.VALUE
$214669.16
$133483.71
.446**
.433**
.594**
.700**
–
6.AGE
.98
.130
-.005
-.012
.077*
.067
-.026
7.TOILET
.46
.499
.000
.003
-.053
-.028
-
8.WASHER
.13
.334
.232**
.227**
.148**
.327**
.121**
9.GRASS
.37
.484
-.111**
-.094*
-.027
-.015
-.071
* p < 0.05. ** p < 0.01. c. variable is constant.
Table 14
Means, Standard Deviations, and Intercorrelations for Water Usage and Water
Consumption Predictor Variables for Calendar Year 2011
Variable
M
SD
2
6
7
8
9
WATER
15.83
18.175
c
-.005
.000
.232**
-
Predictor variable
1.WATERt1
15.66
18.050
c
-.012
.003
.227**
-.094*
2.PRICE
$2.38
$0.00
–
c
c
c
c
3.SIZE
10928.41
8811.361
c
.077*
-.053
.148**
-.027
4.BATH
2.40
.787
c
.067
-.028
.327**
-.015
5.VALUE
$214669.16
$133483.71
c
-.026
-.118**
.121**
-.071
6.AGE
.98
.130
c
–
.123**
.051
.102**
7.TOILET
.46
.499
c
.123**
–
.318**
.213**
8.WASHER
.13
.334
c
.051
.318**
–
.087*
9.GRASS
.37
.484
c
.102**
.213**
.087*
–
* p < 0.05. ** p < 0.01. c. variable is constant.
In calendar year 2011, the mean water consumption was 15.83 (SD = 18.175),
mean size was 10,928.41 sq. ft. (SD = 8,811.361), mean number of bathrooms was 2.40
(SD = .787), mean value of home was $214,669.16 (SD = $133,483.71), mean age was
86
.98 (SD = .130), mean toilet rebate was .46 (SD = .499), mean washing machine rebate
was .13 (SD = .334), and mean Cash-4-Grass rebate was .37 (SD = .484). The variable
for price was constant at $2.38. Therefore, the mean was $2.38 and the standard deviation
is 0.00.
The correlation coefficients for number of bathrooms (.600), size of yard (.487),
value of home (.446), and washing machine rebate (.232) are statistically significant at
the one percent level on a one-tailed test and indicate that those variables have a direct
relationship with water consumption. The positive correlation suggests that as the amount
of the variable increases, the amount of water consumption also increases. The
correlation coefficient for Cash-4-Grass rebate (-.111) is statistically significant at the one
percent level on a one-tailed test and indicates that the variable has an indirect
relationship with water consumption. The negative correlation suggests that as the
amount of the variable increases, the amount of water consumption decreases. The size of
the correlation coefficients for number of bathrooms, value of home, and size of yard
indicate moderate relationships and the coefficients for washing machine rebate and
Cash-4-Grass rebate indicate a weak relationship. The variable for price was constant
during the analysis period. Therefore, the correlation coefficient cannot be computed for
this variable.
Calendar Year 2012. The mean, standard deviation, and correlations between the
variables for calendar year 2012 are shown in Table 15 and Table 16.
87
Table 15
Means, Standard Deviations, and Intercorrelations for Water Usage and Water
Consumption Predictor Variables for Calendar Year 2012
Variable
M
SD
WATER
1
3
4
5
WATER
18.38
19.279
–
.874**
.505**
.689**
.498**
Predictor variable
1.WATERt1
18.33
19.202
.874**
–
.513**
.687**
.508**
2.PRICE
$2.38
$0.00
c
c
c
c
c
3.SIZE
10928.41
8811.361
.505**
.513**
–
.614**
.594**
4.BATH
2.40
.787
.689**
.687**
.614**
–
.700**
5.VALUE
$220310.16
$136990.75
.498**
.508**
.594**
.700**
–
6.AGE
.98
.130
.030
.036
.077*
.067
-.026
7.TOILET
.47
.499
.015
.015
-.057
-.031
-
8.WASHER
.14
.351
.190**
.187**
.123**
.293**
.117**
9.GRASS
.49
.500
-.021
-.025
.000
.031
-.036
* p < 0.05. ** p < 0.01. c. variable is constant.
Table 16
Means, Standard Deviations, and Intercorrelations for Water Usage and Water
Consumption Predictor Variables for Calendar Year 2012
Variable
M
SD
2
6
7
8
9
WATER
18.38
19.279
c
.030
.015
.190**
-.021
Predictor variable
1.WATERt1
18.33
19.202
c
.036
.015
.187**
-.025
2.PRICE
$2.38
$0.00
–
c
c
c
c
3.SIZE
10928.41
8811.361
c
.077*
-.057
.123**
.000
4.BATH
2.40
.787
c
.067
-.031
.293**
.031
5.VALUE
$220310.16
$136990.75
c
-.026
-.121**
.117**
-.036
6.AGE
.98
.130
c
–
.124**
.054
.129**
7.TOILET
.47
.499
c
.124**
–
.340**
.289**
8.WASHER
.14
.351
c
.054
.340**
–
.117**
9.GRASS
.49
.500
c
.129**
.289**
.117**
–
* p < 0.05. ** p < 0.01. c. variable is constant.
In calendar year 2012, the mean water consumption was 18.38 (SD = 19.279),
mean size was 109,28.41 sq. ft. (SD = 8,811.361), mean number of bathrooms was 2.40
(SD = .787), mean value of home was $220,310.16 (SD = $136,990.75), mean age was
88
.98 (SD = .130), mean toilet rebate was .47 (SD = .499), mean washing machine rebate
was .14 (SD = .351), and mean Cash-4-Grass rebate was .49 (SD = .500). The variable
for price was constant at $2.38. Therefore, the mean was $2.38 and the standard deviation
is 0.00.
The correlation coefficients for number of bathrooms (.689), size of yard (.505),
value of home (.498), and washing machine rebate (.190) are statistically significant at
the one percent level on a one-tailed test and indicate that those variables have a direct
relationship with water consumption. The positive correlation suggests that as the amount
of the variable increases, the amount of water consumption also increases. The size of the
correlation coefficient for number of bathrooms indicates that the variable has a strong
relationship with water consumption, the size of the correlation coefficients for value of
home and size of yard indicate moderate relationships, and the coefficient for washing
machine rebate indicates a weak relationship. The variable for price was constant during
the analysis period. Therefore, the correlation coefficient cannot be computed for this
variable.
Summary
A summary of significant descriptive statistics can be found in Table 17. Washing
machine rebates and Cash-4-Grass rebates were added to the VVCSD water conservation
program in 2007. Therefore, the values for those variables were constant at 0 for calendar
year 2007 and are noted in the results. By comparing the means for years 2008 through
2012, it is apparent that both washing machine rebates and Cash-4-Grass rebates gained
89
in popularity during the research period and, excluding year 2012, Cash-4-Grass rebates
exhibited a statistically significant correlation with water consumption.
Table 17
Summary of Descriptive Statistics for Water Usage and Water Consumption Predictor
Variables
Means
Intercorrelations
Year Water Toilet Washer Grass
Water/
Bath
Water/
Grass
Water/
Toilet
2007
19.98
0.29
c
c
.641**
c
.218**
2008
19.68
0.33
0.03
0.02
.619**
-.095*
.199**
2009
18.40
0.38
0.05
0.14
.646**
-.169**
.072
2010
15.58
0.41
0.08
0.22
.522**
-.184**
.060
2011
15.83
0.46
0.13
0.37
.600**
-.111**
.000
2012
18.38
0.47
0.14
0.49
.689**
-.021
.015
* p < 0.05. ** p < 0.01. c. variable is constant.
Regression Analysis
Ordinary Least Squares linear regression analysis has been utilized to test for the
statistical significance of the hypothesized relationship. The results for calendar year
2008 can be found in Table 18, results for year 2009 in Table 19, results for year 2010 in
Table 20, results for year 2011 in Table 21, and results for year 2012 in Table 22. The
complete statistical results can be found in Appendix C.
Calendar Year 2007
Although washing machine and Cash-4-Grass rebates were added to the VVCSD
water conservation program in July 2007, no rebates were issued that year. As a result,
the variables for washing machine rebate and Cash-4-Grass rebate were constant at 0 and
were deleted from the analysis equation. The analysis results presented in this section are
for reference only.
90
For the year 2007, four of the eight coefficients have the expected signs and three
are less than .05, which means they are statistically significant at the five percent level on
a one-tailed test with a critical t-value of 1.645. The coefficient on price was negative,
which, although not predicted, is not unexpected. Because the water consumption is
somewhat insensitive to price, the research plan did not indicate an expected sign. The
negative coefficient on price indicates that as the price of water increases, water
consumption decreases.
Overall, the F-statistic of 249.294 is more than 2.21 which means that the model,
as a whole, is statistically significant at the five percent level of significance and the
adjusted R-squared is .735 which means that this set of variables explains almost 74
percent of residential water consumption in Vandenberg Village.
Calendar Year 2008
Table 18
Regression Analysis Summary for Water Consumption Variables Predicting Water
Usage for Calendar Year 2008
Variable
B
SE B
β
t
p
Constant
33.141
13.350
–
2.482
.013
WATERt1
.670
.030
.671
22.051
.000
PRICE
-25.098
8.601
-.067
-2.918
.004
SIZE
.000
.000
.024
.805
.421
BATH
4.249
1.037
.155
4.098
.000
VALUE
.000
.000
.038
1.084
.279
AGE
-2.740
3.752
-.017
-.730
.466
TOILET
2.365
1.125
.050
2.102
.036
WASHER
-2.795
3.099
-.021
-.902
.367
GRASS
-1.482
3.614
-.009
-.410
.682
Note. R2 = .67 (N = 651, p < .001).
91
For the year 2008, six of the eight coefficients have the expected signs and two
are less than .05, which means they are statistically significant at the five percent level on
a one-tailed test with a critical t-value of 1.645. Again, the negative coefficient on price
indicates that as the price of water increases, water consumption decreases. The positive
coefficient for number of bathrooms indicates that as the number increases, water
consumption also increases. The variables for washing machine rebates and Cash-4-Grass
rebates displayed the expected negative coefficient, which indicates that as rebates are
received, water consumption decreases. However, the coefficients were not statistically
significant with a p-value of 0.367 and 0.682 respectively.
Overall, the F-statistic of 149.340 is more than 2.21 which means that the model,
as a whole, is statistically significant at the five percent level of significance and the
adjusted R-squared is .672 which means that this set of variables explains almost 68
percent of residential water consumption in Vandenberg Village.
92
Calendar Year 2009
Table 19
Regression Analysis Summary for Water Consumption Variables Predicting Water
Usage for Calendar Year 2009
Variable
B
SE B
β
t
p
Constant
.888
4.175
–
.213
.832
WATERt1
.749
.027
.747
28.100
.000
PRICE
-3.270
1.109
-.059
-2.948
.003
SIZE
.000
.000
-.017
-.644
.520
BATH
4.036
.968
.142
4.171
.000
VALUE
.000
.000
.049
1.576
.116
AGE
-.877
3.377
-.005
-.260
.795
TOILET
.849
.975
.018
.870
.384
WASHER
-3.736
2.264
-.034
-1.650
.099
GRASS
-1.995
1.325
-.030
-1.505
.133
Note. R2 = .75 (N = 659, p < .001).
For the year 2009, five of the eight coefficients have the expected signs and two
are less than .05, which means it is statistically significant at the five percent level on a
one-tailed test with a critical t-value of 1.645. Again, the negative coefficient for price
indicates that as the price of water increases, water consumption decreases. The variables
for washing machine rebates and Cash-4-Grass rebates displayed the expected negative
coefficient, which indicates that as rebates are received, water consumption decreases.
However, the coefficients were not statistically significant with a p-value of 0.099 and
0.133 respectively.
Overall, the F-statistic of 220.508 is more than 2.21 which means that the model,
as a whole, is statistically significant at the five percent level of significance and the
adjusted R-squared is .750, which means that this set of variables explains approximately
75 percent of residential water consumption in Vandenberg Village.
93
Calendar Year 2010
Table 20
Regression Analysis Summary for Water Consumption Variables Predicting Water
Usage for Calendar Year 2010
Variable
B
SE B
β
t
p
Constant
.267
2.989
.089
.929
WATERt1
.682
.028
.691
24.180
.000
PRICE
c
c
c
c
c
SIZE
.000
.000
.089
2.892
.004
BATH
1.678
.779
.082
2.154
.032
VALUE
.000
.000
.018
.505
.614
AGE
-1.725
2.866
-.014
-.602
.547
TOILET
1.144
.832
.035
1.374
.170
WASHER
-1.222
1.510
-.021
-.809
.419
GRASS
-2.027
.932
-.052
-2.175
.030
Note. R2 = .65 (N = 673, p < .001). c. variables are constants or have missing
correlations
For the year 2010, five of the eight coefficients have the expected signs and three
are less than .05, which means they are statistically significant at the five percent level on
a one-tailed test with a critical t-value of 1.645. The variables for price were constant and
were deleted from the analysis equation. The variables for washing machine rebates and
Cash-4-Grass rebates displayed the expected negative coefficient, which indicates that as
rebates are received, water consumption decreases. The coefficient for Cash-4-Grass
rebates was statistically significant. However, the coefficient for washing machine
rebates was not statistically significant with a p-value of 0.419.
Overall, the F-statistic of 154.737 is more than 2.21 which means that the model,
as a whole, is statistically significant at the five percent level of significance and the
adjusted R-squared is .646, which means that this set of variables explains almost 65
percent of residential water consumption in Vandenberg Village.
94
Calendar Year 2011
Table 21
Regression Analysis Summary for Water Consumption Variables Predicting Water
Usage for Calendar Year 2011
Variable
B
SE B
β
t
p
Constant
-3.743
3.445
-1.086
.278
WATERt1
.614
.029
.610
20.838
.000
PRICE
c
c
c
c
c
SIZE
.000
.000
.079
2.528
.012
BATH
4.499
.903
.195
4.983
.000
VALUE
.000
.000
-.007
-.189
.850
AGE
-2.085
3.304
-.015
-.631
.528
TOILET
.489
.924
.013
.529
.597
WASHER
1.087
1.434
.020
.758
.449
GRASS
-1.977
.899
-.053
-2.198
.028
Note. R2 = .63 (N = 695, p < .001). c. variables are constants or have missing
correlations
For the year 2011, three of the eight coefficients have the expected signs and all
three are less than .05, which means they are statistically significant at the five percent
level on a one-tailed test with a critical t-value of 1.645. The variables for price were
constant and were deleted from the analysis equation. The variable for Cash-4-Grass
rebates displayed the expected negative coefficient, which indicates that as rebates are
received, water consumption decreases. The coefficient for Cash-4-Grass rebates was
statistically significant.
Overall, the F-statistic of 147.245 is more than 2.21 which means that the model,
as a whole, is statistically significant at the five percent level of significance and the
adjusted R-squared is .627, which means that this set of variables explains almost 63
percent of residential water consumption in Vandenberg Village.
95
Calendar Year 2012
Table 22
Regression Analysis Summary for Water Consumption Variables Predicting Water
Usage for Calendar Year 2012
Variable
B
SE B
β
t
p
Constant
-3.988
2.856
-1.397
.163
WATERt1
.756
.025
.753
29.997
.000
PRICE
c
c
c
c
c
SIZE
.000
.000
.027
1.128
.260
BATH
4.219
.782
.172
5.399
.000
VALUE
.000
.000
-.019
-.712
.477
AGE
-1.704
2.707
-.012
-.629
.529
TOILET
.566
.780
.015
.726
.468
WASHER
-.295
1.111
-.005
-.266
.791
GRASS
-.408
.727
-.011
-.561
.575
Note. R2 = .78 (N = 695, p < .001). c. variables are constants or have missing
correlations
For the year 2012, four of the eight coefficients have the expected signs but only
number of bathrooms has a p-value less than .05. Again, the variables for price were
constant and were deleted from the analysis equation. The variables for washing machine
rebates and Cash-4-Grass rebates displayed the expected negative coefficient, which
indicates that as rebates are received, water consumption decreases. However, the
coefficients were not statistically significant with a p-value of 0.791 and 0.575
respectively.
Overall, the F-statistic of 304.369 is more than 2.21 which means that the model,
as a whole, is statistically significant at the five percent level of significance and the
adjusted R-squared is .777, which means that this set of variables explains almost 78
percent of residential water consumption in Vandenberg Village.
96
Summary
A summary of the regression analysis model results for the years 2007 through
2012 can be found in Table 23. All six models were statistically significant at the five
percent level of significance and the variables selected explain between 63 percent and 78
percent of residential water consumption in Vandenberg Village.
Table 23
Regression Analysis Summary for Water Consumption Models
Model
R2
SE
F
2007
.735
10.310
249.294**
2008
.672
12.625
149.340**
2009
.750
11.376
220.508**
2010
.646
9.649
154.737**
2011
.627
11.095
147.245**
2012
.777
9.096
304.369**
*p < .05. **p < .01.
As previously stated, in 2007, the variables for Cash-4-Grass rebates were
constant and were deleted from the analysis equation. For the remaining 5 years, the
Cash-4-Grass rebates exhibited the expected negative sign and the variable was
statistically significant at the five percent level of significance for 2 of the 5 years
analyzed (see Table 24).
97
Table 24
Regression Analysis Summary for Cash-4-Grass Rebates
Model
B
SE B
β
t
p
2007
c
c
c
c
c
2008
-1.482
3.614
-.009
-.410
.682
2009
-1.995
1.325
-.030
-1.505
.133
2010
-2.027
.932
-.052
-2.175
.030*
2011
-1.977
.899
-.053
-2.198
.028*
2012
-.408
.727
-.011
-.561
.575
*p < .05. c. variables are constants or have missing correlations
The results of the regression analyses indicate that receipt of a Cash-4-Grass
rebate, size of property, number of bathrooms, and price per unit of water have varying
degrees of statistical significance on water consumption. Additionally, the correlation
between Cash-4-Grass rebates and water consumption was statistically significant for
four of the five years analyzed. Although this does not imply causation, it is an indicator
of a relationship between the variables.
The lagged domestic water use [WATERt1] has been used in these models to help
predict the nonlagged version of the same dependent variable and to correct for
autocorrelation (Ruijs et al., 2008; Vogt & Johnson, 2011). The lagged domestic water
use also acts as a proxy for factors that are not able to be measured directly (e.g., number
of household residents).
t test
An independent samples t test was used to compare the differences in customers
who receive a Cash-4-Grass rebate and those who do not. The results for calendar year
2008 can be found in Table 25, results for year 2009 in Table 26, results for year 2010 in
Table 27, results for year 2011 in Table 28, and results for year 2012 in Table 29
98
Calendar Year 2008
Table 25
Group Differences for Water Consumption Between Groups That Did or Did Not Receive
Cash
-4-Grass Rebates for Calendar Year 2008
No Grass Rebate
Grass Rebate
M
SD
M
SD
df
t
p
WATER
19.98
22.172
5.00
2.739
650
2.433
.015
PRICE
1.48
0.06
1.55
0.00
650
-4.332
.000
SIZE
11209.10
9074.49
7344.23
1145.77
650
1.534
.125
BATH
2.43
0.81
2.00
0.00
650
1.916
.056
VALUE
194970.39
124062.87
122548.38
51736.48
14.968
4.776
.000
AGE
.98
.136
1.00
.000
638
-3.494
.001
TOILET
.33
.47
.31
.48
12.473
.167
.870
WASHER
.03
.17
.00
.00
638
4.422
.000
In 2008, there was a significant difference in the scores for water consumption,
price per unit of water, value of home, and age of home in customers who did not receive
a Cash-4-Grass rebate and those that did. Washing machine rebate in customers who did
not receive a Cash-4-Grass rebate was significant. However, none of the Cash-4-Grass
rebate recipients had also received a washing machine rebate. Size of yard, number of
bathrooms, and toilet rebate were not significant.
99
Calendar Year 2009
Table 26
G
roup Differences for
Water Consumption Between Groups That Did or Did Not Receive
Cash
-4-Grass Rebates for Calendar Year 2009
No Grass Rebate
Grass Rebate
M
SD
M
SD
df
t
p
WATER
19.93
23.912
8.80
8.476
658
4.394
.000
PRICE
1.88
0.41
1.99
0.42
119.092
-2.275
.025
SIZE
11166.39
8873.11
10443.01
9552.25
116.203
.677
.500
BATH
2.45
0.84
2.23
0.42
658
2.408
.016
VALUE
203779.08
128569.38
174089.10
108990.11
133.469
2.35
.020
AGE
.98
.144
1.00
.000
658
-1.398
.163
TOILET
.37
.48
.45
.50
658
-1.487
.138
WASHER
.05
.22
.00
.00
658
2.247
.025
In 2009, there was a significant difference in the scores for water consumption,
price per unit of water, number of bathrooms, and value of home in customers who did
not receive a Cash-4-Grass rebate and those that did. Washing machine rebate in
customers who did not receive a Cash-4-Grass rebate was significant. However, none of
the Cash-4-Grass rebate recipients had also received a washing machine rebate. Size of
yard, age of home, and toilet rebate were not significant.
100
Calendar Year 2010
Table 27
Group Differences for Water Consumption Between Groups That Did or Did Not Receive
Cash
-4-Grass Rebates for Calendar Year 2010
No Grass Rebate
Grass Rebate
M
SD
M
SD
df
t
p
WATER
17.15
16.89
9.92
11.991
672
4.847
.000
SIZE
11049.38
8757.196
10829.08
9451.398
218.624
.253
.800
BATH
2.44
.865
2.29
.454
672
2.093
.037
VALUE
211565.50
134242.111
186618.73
109284.213
277.929
2.317
.021
AGE
.98
.149
1.00
.000
672
1.840
.066
TOILET
.38
.486
.53
.501
672
-3.357
.001
WASHER
.08
.268
.11
.313
672
-1.227
.220
In 2010, there was a significant difference in the scores for water consumption
and toilet rebate in customers who did not receive a Cash-4-Grass rebate and those that
did. Size of yard, number of bathrooms, value of home, age of home, and washing
machine rebate in customers who did not receive a Cash-4-Grass rebate were not
significant. Price per unit of water was constant.
101
Calendar Year 2011
Table 28
G
roup
Differences for Water Consumption Between Groups That Did or Did Not Receive
Cash
-4-Grass Rebates for Calendar Year 2011
No Grass Rebate
Grass Rebate
M
SD
M
SD
df
t
p
WATER
17.39
17.703
13.21
18.682
521.599
2.912
.004
SIZE
11108.94
8900.998
10625.68
8667.550
556.218
.704
.481
BATH
2.41
.781
2.38
.799
534.742
.406
.685
VALUE
221941.04
138406.177
202474.76
124099.001
591.848
1.916
.056
AGE
.97
.164
1.00
.000
694
-2.709
.007
TOILET
.38
.486
.60
.491
540.499
-5.722
.000
WASHER
.11
.308
.17
.372
694
-2.294
.022
In 2011, there was a significant difference in the scores for water consumption,
age of home, toilet rebate, and washing machine rebate in customers who did not receive
a Cash-4-Grass rebate and those that did. Size of yard, number of bathrooms, and value
of home in customers who did not receive a Cash-4-Grass rebate and those that did were
not significant. Price per unit of water was constant.
102
Calendar Year 2012
Table 29
G
roup Differences for Water Consumption
Between Groups That Did or Did Not Receive
Cash
-4-Grass Rebates for Calendar Year 2012
No Grass Rebate
Grass Rebate
M
SD
M
SD
df
t
p
WATER
18.78
15.637
17.97
22.503
694
.553
.581
SIZE
10929.43
8955.952
10927.35
8669.716
693.741
.003
.998
BATH
2.37
.710
2.42
.861
655.746
-.821
.412
VALUE
225175.13
143479.667
215186.87
129822.171
692.398
.964
.335
AGE
.97
.180
1.00
.000
694
-3.429
.001
TOILET
.32
.469
.61
.488
694
-7.960
.000
WASHER
.10
.305
.19
.390
694
-3.107
.002
In 2012, there was a significant difference in the scores for age of home, toilet
rebate, and for washing machine rebate in customers who did not receive a Cash-4-Grass
rebate and those that did. Water consumption in customers, size of yard, number of
bathrooms, and value of home in customers who did not receive a Cash-4-Grass rebate
and those that did were not significant. Price per unit of water was constant.
Summary
Water consumption and washing machine rebates were statistically significant for
4 out of the 5 years tested; value of the home, age of the home, and toilet rebate for 3 out
of the 5 years; and price per unit of water and number of bathrooms for 2 out of the 5
years. Size of yard was not significant in any of the years selected.
Research Question 1 Results
The null hypothesis (H01) states that there is no significant difference between the
receipt of landscape rebates and no receipt of landscape rebates on water consumption
when controlling for receipt of a toilet or washing machine rebate, size of property,
103
number of bathrooms, value of home, age of home, and price per unit of water. The
research hypothesis (Ha1) assumes that there is a significant difference between the
receipt of landscape rebates and no receipt of landscape rebates on water consumption
when controlling for receipt of a toilet or washing machine rebate, size of property,
number of bathrooms, value of home, age of home, and price per unit of water. Based on
the statistical results, the null hypothesis (H01), which states that there is no significant
difference between the receipt of landscape rebates and no receipt of landscape rebates on
water consumption, is rejected and the research hypothesis (Ha1) is accepted.
104
Research Question 2 and Hypothesis
Do consumers use less water after receiving a landscape rebate than before?
H02: There is no significant difference in water consumption in the 24
months before and 24 months after receipt of a landscape rebate
when controlling for receipt of a toilet or washing machine rebate,
size of property, number of bathrooms, value of home, age of home,
amount of rainfall, average temperature, and price per unit of water.
Ha2: There is a significant difference in water consumption in the 24
months before and 24 months after receipt of a landscape rebate
when controlling for receipt of a toilet or washing machine rebate,
size of property, number of bathrooms, value of home, age of home,
amount of rainfall, average temperature, and price per unit of water.
Descriptive Statistics. Frequencies and percentages were calculated for the
household characteristics and the results summarized in Table 30. The analyzed data
consists of a total of 48 months of records for 21 landscape recipients. The frequency for
variables that remained constant for the data collection period (e.g., number of
bathrooms, year home was built, size of yard) were divided by 48 to accurately represent
the number of homes in the study. The complete statistical results can be found in
Appendix D.
105
Table 30
Frequencies and Percentages for Household Characteristics
Variable
Frequency
Percent
Bathrooms
2
14
66.7
3
5
23.8
4
1
4.8
6
1
4.8
Rebates Received
Toilet
569
56.4
Washing Machine
136
13.5
Grass
504
50.0
Age of Home
Post 1991
21
100.0
Value of Home
$50,000-99,999
288
28.6
$100,000-149,999
132
13.1
$150,000-199,999
172
17.1
$200,000-249,999
192
19.0
$250,000-299,999
64
6.3
$300,000-349,999
40
4.0
$350,000-399,999
24
2.4
$400,000-449,999
32
3.2
$450,000-499,999
16
1.6
$500,000-549,999
25
2.5
$550,000-599,000
23
2.3
Size of Yard
4,000-4,999 sq. ft.
4
19.0
5,000-5,999
2
9.6
6,000-6,999
7
33.3
7,000-7,999
1
4.8
8,000-8,999
2
9.5
9,000-9,999
0
0.0
10,000-19,999
2
9.5
20,000-29,999
3
14.3
Most of the homes in the study (90.5%) have either two or three bathrooms. A
majority (66.7%) of the homes have two bathrooms and 23.8% have three bathrooms.
More than 56% have retrofitted their bathrooms with new toilets. Size of the yard is a
106
primary factor in how much water is used for irrigation. In this dataset, 28.6% have yards
under 6,000 square feet, 33.3% have yards between 6,000 and 7,000 square feet, and
38.1% have yards larger than 7,000 square feet.
The means, standard deviations, and intercorrelations are presented in Table 31
and Table 32. In this dataset, the mean water consumption was 16.83 (SD = 22.262),
mean size was 10,072.29 sq. ft. (SD = 7,217.827), mean number of bathrooms was 2.52
(SD = .958), mean value of home was $194,409.56 (SD = $126,397.419), mean price was
$2.01 (SD = $0.46), mean toilet rebate was .56 (SD = .496), mean washing machine
rebate was .13 (SD = .342), mean Cash-4-Grass rebate was .50 (SD = .500), mean
temperature was 66.64 (SD = 3.68), and mean rainfall was 1.29 (SD = 2.15). The variable
for age was constant at 1. Therefore, the mean was 1.0 and the standard deviation is 0.00.
Table 31
Means, Standard Deviations, and Intercorrelations for Water Usage and Water
Consumption Predictor Variables
Variable
M
SD
WATER
1
2
10
11
WATER
16.83
22.262
–
.841**
-.077*
-.141**
.118**
Predictor variable
1.WATERt1
16.95
22.270
.841**
–
-.071*
-.097**
.140**
2.PRICE
$2.01
$0.46
-.077*
-.071*
–
.154**
-.042
3.SIZE
10072.29
7217.827
.620**
.621**
-.080*
-.013
.015
4.BATH
2.52
.958
.759**
.759**
.025
.002
-.013
5.VALUE
194409
126397
.681**
.682**
.041
.001
-.009
6.AGE
1.00
.000
c
c
c
c
c
7.TOILET
.56
.496
.136**
.136**
.257**
.028
-.049
8.WASHER
.13
.342
.303**
.299**
.215**
.007
-.045
9.GRASS
.50
.500
-.054
-.062*
.534**
.026
-.081**
10.RAIN
1.2872
2.15346
-.141**
-.097**
.154**
–
-.372**
11.TEMP
66.64
3.680
.118**
.140**
-.042
-.372**
–
* p < 0.05. ** p < 0.01. c. variable is constant.
107
Table 32
Means, Standard Deviations, and Intercorrelations for Water Usage and Water
Consumption Predictor Variables
Variable
M
SD
3
4
5
7
8
9
WATER
16.83
22.262
.620**
.759**
.681**
.136**
.303**
-.054
Predictor variable
1.WATERt1
16.95
22.270
.621**
.759**
.682**
.136**
.299**
-.062*
2.PRICE
$2.01
$0.46
-.080*
.025
.041
.257**
.215**
.534**
3.SIZE
10072.29
7217.827
–
.661**
.744**
-.023
.353**
.000
4.BATH
2.52
.958
.661**
–
.827**
.280**
.500**
.000
5.VALUE
194409
126397
.744**
.827**
–
.155**
.472**
.046
6.AGE
1.00
.000
c
c
c
c
c
c
7.TOILET
.56
.496
-.023
.280**
.155**
–
.347**
.122**
8.WASHER
.13
.342
.353**
.500**
.472**
.347**
–
.052
9.GRASS
.50
.500
.000
.000
.046
.122**
.052
–
10.RAIN
1.2872
2.15346
-.013
.002
.001
.028
.007
.026
11.TEMP
66.64
3.680
.015
-.013
-.009
-.049
-.045
-.081**
* p < 0.05. ** p < 0.01. c. variable is constant.
The correlation coefficients for temperature (.118), size of yard (.620), number of
bathrooms (.759), value of home (.681), toilet rebate (.136), and washing machine rebate
(.303) are statistically significant at the one percent level on a one-tailed test and indicate
that those variables have a direct relationship with water consumption. The positive
correlation suggests that as the amount of the variable increases, the amount of water
consumption also increases. The correlation coefficient for rain (-.141) is statistically
significant at the five percent level on a one-tailed test and indicates that the variable has
an indirect relationship with water consumption. The negative correlation suggests that as
the amount of the variable increases, the amount of water consumption decreases. The
correlation coefficient for price (-.077) is statistically significant at the five percent level
on a one-tailed test and indicates that the variable has an indirect relationship with water
consumption. The negative correlation suggests that as the amount of the variable
108
increases, the amount of water consumption decreases. The size of the correlation
coefficients for size of yard, number of bathrooms, and home value indicate that the
variables have a strong relationship with water consumption. The coefficients for price,
rain, temperature, toilet rebates, and washing machine rebates indicate a weak
relationship. The correlation coefficient for Cash-4-Grass rebate was not statistically
significant when compared to water consumption.
Regression Analysis
Ordinary Least Squares linear regression analysis has been utilized to test for the
statistical significance of the hypothesized relationship. The results of the regression
analysis can be found in Table 33. The complete statistical results can be found in
Appendix D.
Table 33
Regression Analysis Summary for Water Consumption Variables Predicting Water
Usage
Variable
B
SE B
β
t
p
Constant
-12.919
7.402
–
-1.745
.081
WATERt1
.565
.026
.566
21.680
.000
PRICE
-.608
.975
-.012
-.624
.533
SIZE
.000
.000
.078
3.090
.002
BATH
6.319
.794
.272
7.954
.000
VALUE
.000
.000
.033
1.021
.307
TOILET
.058
.813
.001
.071
.944
WASHER
-2.705
1.279
-.042
-2.114
.035
GRASS
-.415
.843
-.009
-.492
.623
RAIN
-.828
.179
-.080
-4.631
.000
TEMP
.053
.105
.009
.500
.617
Note. R2 = .75 (N = 1007, p < .001)
109
In this dataset, eight of the nine coefficients have the expected signs and four are
less than .05, which means they are statistically significant at the five percent level on a
one-tailed test with a critical t-value of 1.645.
WATERt1: The lagged domestic water use has been used in this model to help
predict the nonlagged version of the same dependent variable and to correct for
autocorrelation. The lagged domestic water use also acts as a proxy for factors that were
not able to be measured directly (e.g., number of household residents).
PRICE: The coefficient on price was negative, which, although not predicted, is
not unexpected. Because the water consumption is somewhat insensitive to price, the
research plan did not indicate an expected sign. The negative coefficient on price
indicates that as the price of water increases, water consumption decreases. However, the
coefficient was not statistically significant with a p-value of 0.533.
SIZE: The coefficient on yard size displayed the expected positive sign, which
indicates that as yard size increases, water consumption increases. This coefficient was
statistically significant at the five percent level on a one-tailed test with a critical t-value
of 1.645.
BATH: The coefficient on bathrooms was positive, which, although not predicted,
is not unexpected. The research plan predicted that the coefficient would be negative
because as increased amounts of older toilets are replaced with low-flow toilets, water
usage decreases. However, this assumption was dependent on the home being built before
1992. Because all of the homes in the dataset were built after 1991, they were required to
be built with low-flow toilets. Therefore, as the number of bathrooms in a home increase,
110
water consumption also increases resulting in a positive coefficient. The coefficient was
statistically significant at the five percent level on a one-tailed test with a critical t-value
of 1.645.
TOILET: The coefficient on toilet rebates did not display the expected negative
sign. This was not unexpected. Because all of the homes in the dataset were built after
1991 and, therefore, homes were required to be sold with low-flow toilets, the receipt of a
toilet rebate would not significantly reduce water consumption.
WASHER: The coefficient on washer rebates displayed the expected negative
sign, which indicates that as clothes washer rebates are received, water consumption
decreases. This coefficient was statistically significant at the five percent level on a one-
tailed test with a critical t-value of 1.645.
GRASS: The coefficient on Cash-4-Grass rebates displayed the expected negative
sign, which indicates that as Cash-4-Grass rebates are received, water consumption
decreases. However, the coefficient was not statistically significant with a p-value of
0.623.
RAIN: The coefficient on rainfall displayed the expected negative sign, which
indicates that as rainfall increases, water consumption decreases. This coefficient was
statistically significant at the five percent level on a one-tailed test with a critical t-value
of 1.645.
TEMP: The coefficient on temperature displayed the expected positive sign,
which indicates that as the ambient temperature increases, water consumption also
111
increases. However, the coefficient was not statistically significant with a p-value of
0.617.
VALUE: The coefficient on home value displayed the expected positive sign,
which indicates that as the value increases, water consumption also increases. However,
the coefficient was not statistically significant with a p-value of 0.307.
AGE: The variable in the dataset is constant and was deleted from the analysis
equation.
Overall, the F-statistic of 305.259 is more than 2.21 which means that the model,
as a whole, is statistically significant at the five percent level of significance and the
adjusted R-squared is .751 which means that this set of variables explains 75 percent of
residential water consumption in Vandenberg Village.
t test
An independent samples t test was used to compare the differences in water
consumption before and after the receipt of a Cash-4-Grass rebate. The results are
presented in Table 34. Comparison of water consumption for Cash-4-Grass rebate
recipients (M = 15.62, SD = 21.77) and those customers not receiving Cash-4-Grass
rebates (M = 18.03, SD = 22.701) showed a reduction in water consumption in average
but revealed no significant differences between the groups t(1006) = 1.721, ns.
Table 34
Group Differences for Water Consumption Between Groups That Did or Did Not Receive
Cash
-4-Grass Rebates
No Grass Rebate
Grass Rebate
M
SD
M
SD
df
t
p
WATER
18.03
22.701
15.62
21.770
1004.240
1.721
.086
112
Research Question 2 Results
The null hypothesis (H01) states that there is no significant difference in water
consumption in the 24 months before and 24 months after receipt of a landscape rebate
when controlling for receipt of a toilet or washing machine rebate, size of property,
number of bathrooms, value of home, age of home, amount of rainfall, average
temperature, and price per unit of water. The research hypothesis (Ha1) assumes that there
is a significant difference in water consumption in the 24 months before and 24 months
after receipt of a landscape rebate when controlling for receipt of a toilet or washing
machine rebate, size of property, number of bathrooms, value of home, age of home,
amount of rainfall, average temperature, and price per unit of water.
The descriptive statistics are presented in in Table 30, Table 31, and Table 32.
The correlation coefficients for size of yard, number of bathrooms, and value of homes
indicate that the variables have a strong relationship with water consumption. These
coefficients are statistically significant at the one percent level on a one-tailed test.
The results of the regression analysis can be found in Table 33 and indicate that
receipt of a washing machine rebate, size of property, number of bathrooms, and rainfall
have a statistical significance on water consumption. The model was statistically
significant at the five percent level of significance and the variables selected explain 75
percent of residential water consumption in Vandenberg Village. Because the coefficient
on landscape rebates displayed the expected negative sign but was not statistically
significant with a p-value of 0.623, a t test was used to compare the differences in water
consumption before and after the receipt of a Cash-4-Grass rebate (see Table 34).
113
Although not statistically significant, the results indicate a reduction in water
consumption after the receipt of a Cash-4-Grass rebate (Difference = 2.41).
Based on these statistical results, the null hypothesis (H02), which states that there
is no significant difference in water consumption in the 24 months before and 24 months
after receipt of a landscape rebate, cannot be rejected.
Findings
The purpose of this study is to investigate the relationship between the dependent
variable of monthly residential water usage and the independent variables of receipt of a
toilet, washer, and/or landscape rebate, size of irrigable property, number of bathrooms,
value of home, age of home, amount of monthly rainfall, average monthly high
temperature, and price per unit of water. The primary goal of this research is to answer
the question Can cash-4-grass programs save water?
Research Question 1
Based on the descriptive statistics and regression analyses conducted on the
datasets, the null hypothesis (H01), which states that there is no significant difference
between the receipt of landscape rebates and no receipt of landscape rebates on water
consumption, is rejected and the research hypothesis (Ha1) is accepted.
Research Question 2
Based on the descriptive statistics and regression analyses conducted on the
datasets, the null hypothesis (H02), which states that there is no significant difference in
water consumption in the 24 months before and 24 months after receipt of a landscape
rebate, cannot be rejected.
114
Summary
Chapter 4 presented the results of the statistical analysis. It includes: a description
of the data collection process; an explanation of the data transformation performed; a
description of the data analysis; a testing of the hypotheses; and a summary of the
findings. This chapter explained that data collected from VVCSD, SBC, and NOAA were
transformed and combined into seven separate datasets for statistical analysis. IBM
SPSS® Statistics version 21 was used to generate descriptive statistics and perform a
linear regression analysis. The statistical results were mixed. However, according to
hypothesis testing, a robust water conservation program, including Cash-4-Grass rebates,
can have a significant impact on water consumption.
Chapter 1 presented an in-depth introduction to water supply, water demand,
water conservation, and landscape rebates. Chapter 2 presented the review of the existing
literature for the study. Chapter 3 presented the research methodology that was used for
this project. Chapter 5 will present the interpretation of the findings, limitations of the
study, recommendations, and implications.
115
Chapter 5: Discussion, Conclusions, and Recommendations
Introduction
California’s domestic water supplies are threatened by population growth,
drought, pollution, and climate change. In recent years, the California legislature
concluded that the most economical option to extend the resources was to use less water
per capita. To encourage household water savings, most water conservation programs use
restrictions, rates, and rebates. Prior research has focused on water restrictions, pricing
structures, and toilet rebates. Since outdoor water usage accounts for the majority of a
household’s water usage, elements of a water conservation program that target outdoor
water use, such as landscape rebates, can have a significant impact on domestic water
usage.
This research was designed to measure the effectiveness of landscape rebates on
water consumption in the bedroom community of Vandenberg Village, California. The
analysis was performed on public data obtained from VVCSD, SBC, and NOAA. The
purpose of this study was to investigate the relationship between the dependent variable
of monthly residential water usage and the independent variables of receipt of a toilet,
washer, and/or landscape rebate, size of irrigable property, number of bathrooms, value
of home, age of home, amount of monthly rainfall, average monthly high temperature,
and price per unit of water.
The primary goal of this research was to answer the question Can cash-4-grass
programs save water? To answer this question the study used a multiple time-series
quasi-experimental research design to evaluate the effectiveness of landscape rebates on
116
residential water consumption in Vandenberg Village, California. Using descriptive
statistics and regression analyses, the study addressed two research questions to evaluate
the effectiveness of landscape rebates:
1. Do consumers who receive landscape rebates use less water than those
consumers who do not?
2. Do consumers use less water after receiving a landscape rebate than before?
The findings for this study were mixed. Overall, the results show that Cash-4-
Grass rebates do reduce water consumption. However, this reduction in water use was not
consistently statistically significant. Therefore, the research hypothesis could not be
accepted for both research questions. This chapter presents the interpretation of the
findings, limitations of the study, recommendations for further research, and a
conclusion.
Interpretation of the Findings
As explained in Chapter 1, TRA is based on a person’s attitude toward expected
behavior. The more a person is expected to exhibit a particular behavior, the more likely
he or she will behave in the expected manner. During California’s periodic dry periods,
customers are inundated with messages to conserve water. Most recently, on January 17,
2014, Governor Jerry Brown officially declared a drought emergency for California and
asked its residents to reduce their water consumption by 20% (York, 2014). Figure 10
illustrates that, except for anomalous calendar year 2012, water conservation expectations
during dry periods were realized by an increase in Cash-4-Grass rebates during those
117
years. An increase in rainfall in calendar year 2010 resulted in a decrease of Cash-4-
Grass rebate requests.
Calendar year 2012 may have exhibited anomalous findings because of the heavy
rainfall during the winter of 2011. Customers may have increased their irrigation regime
in an effort to duplicate the lush, green lawns that occurred naturally during the previous
year (Endter-Wada et al., 2008).
Figure 10. Comparison of the number of Cash-4-Grass rebates and annual rainfall.
Although not statistically significant, the results of the RQ2 t test demonstrate that
Cash-4-Grass rebate recipients reduced their water consumption, on average, by 13.4%.
Similarly, the results of the RQ1 t tests demonstrate that Cash-4-Grass rebate recipients
used between 4% and 75% less water than their non-rebate counterparts (see Figure 11).
118
These savings are consistent with the predictions outlined by the literature review in
Chapter 2.
Figure 11. Comparison of water consumption for customers that have and have not
received Cash-4-Grass rebates.
Research Question 1
The null hypothesis (H01) states that there is no significant difference between the
receipt of landscape rebates and no receipt of landscape rebates on water consumption
when controlling for receipt of a toilet or washing machine rebate, size of property,
number of bathrooms, value of home, age of home, and price per unit of water. The
research hypothesis (Ha1) assumes that there is a significant difference between the
receipt of landscape rebates and no receipt of landscape rebates on water consumption
119
when controlling for receipt of a toilet or washing machine rebate, size of property,
number of bathrooms, value of home, age of home, and price per unit of water.
The results of the regression analyses indicate that receipt of a Cash-4-Grass
rebate, size of property, number of bathrooms, and price per unit of water have varying
degrees of statistical significance on water consumption. Additionally, the correlation
between Cash-4-Grass rebates and water consumption was statistically significant for
four of the five years analyzed. Based on these results, the null hypothesis (H01) is
rejected and the research hypothesis (Ha1) is accepted.
Research Question 2
The null hypothesis (H01) states that there is no significant difference in water
consumption in the 24 months before and 24 months after receipt of a landscape rebate
when controlling for receipt of a toilet or washing machine rebate, size of property,
number of bathrooms, value of home, age of home, amount of rainfall, average
temperature, and price per unit of water. The research hypothesis (Ha1) assumes that there
is a significant difference in water consumption in the 24 months before and 24 months
after receipt of a landscape rebate when controlling for receipt of a toilet or washing
machine rebate, size of property, number of bathrooms, value of home, age of home,
amount of rainfall, average temperature, and price per unit of water.
Although the regression analysis model for the research question was statistically
significant and the coefficient on Cash-4-Grass rebates displayed the expected negative
sign, the Cash-4-Grass rebates variable was not statistically significant with a p-value of
0.623. Additionally, the correlation coefficient between Cash-4-Grass rebates and water
120
consumption was not statistically significant. Based on these results, the null hypothesis
(H02) cannot be rejected.
Limitations of the Study
Upon a detailed review of the data received from VVCSD, it was discovered that
some Cash-4-Grass rebate recipients may have replaced their lawns in an attempt to
increase the home’s curb appeal. Because the customer moved out of the home within a
few months of receiving the rebate, and a new customer with a different family
demographic benefitted from the reduction in irrigable lawn, the before and after water
usage numbers may not have been measuring the same impacts to water consumption. A
limitation identified in Chapter 1 was the impact of missing data for homes during
periods of vacancy. Both of these limitations were addressed during the model formation
stage and homes with periods of zero usage for 2 or more subsequent months were
eliminated from the model.
Another limitation identified in Chapter 1 was meter reading accuracy. During the
study period, VVCSD replaced every meter within their community with AMR capable
meters (VVCSD, 2011c, 2012c). The maximum usage fluctuations detailed in Table 35
may be a result of normal household usage fluctuations and meter reading corrections
brought on by automation because the usage does not gradually reduce as meters were
replaced.
A final limitation that was discovered when the data from VVCSD was divided
into models is that water consumption for some Cash-4-Grass rebate recipients began to
decline months before the rebate was requested. The work plan attempted to take into
121
account the time lag between the time the landscape was replaced and the rebate was
received by leading the variable by one month. However, it seems as though some
recipients prepared to remove their turf grass by turning off their irrigation systems and
allowing their grass to die. This process could have taken many months.
Table 35
Number, Maximum, and Mean for Water Usage Variables
N
Max
Mean
RQ1
2007
629
158
19.98
2008
652
200
19.68
2009
660
225
18.40
2010
674
105
15.58
2011
696
169
15.83
2012
696
171
18.38
RQ2
1008
225
16.83
Recommendations for Further Research
When before and after consumption was tested by the RQ2 model, Cash-4-Grass
rebate recipients did not exhibit a statistically significant water consumption reduction.
However, when rebate and non-rebate consumption was tested in the RQ1 models, there
was a statistically significant reduction. This may be explained by an untested
“conservation attitude” (Dyckman, 2005) and that the Cash-4-Grass rebate was a
monetary manifestation of their conservation efforts. Conversely, the inability to pinpoint
the actual month of lawn replacement may have contributed to the lack of significance in
the RQ2 model through imprecise before and after rebate data points.
It is my opinion that further research should focus on an in-depth analysis of
select Cash-4-Grass rebate recipients. By focusing on specific recipients, additional
122
elements, such as household demographics, can be included in the analysis and the rebate
recipients can be surveyed to analyze their conservation attitude. Finally, the content of
their water consumption history and rebate receipts can be individually analyzed to
determine when the changes to the landscape were implemented. A companion study
should be an in-depth analysis of non-rebate recipients. An evaluation of their
conservation attitude may help determine the barriers to participation in the Cash-4-Grass
rebate program.
Finally, to expand the knowledge regarding the effectiveness of landscape rebates,
a similar study should be performed in other areas that offer rebates for lawn removal.
Since 2009, the City of Los Angeles has paid for the removal of more than 1 million
square feet of grass for a total of $1.4 million in rebates (Lovett, 2013). City officials
expect to save 47 million gallons of water per year (Lovett, 2013). Since 2003, the Las
Vegas Valley Water District has paid more than $200 million for the removal of 165.6
million square feet of grass (Lovett, 2013). District officials report that they have saved
9.2 billion gallons of water in the decade since the program’s implementation (Lovett,
2013). Studies in these areas will greatly expand the academic knowledge regarding
water conservation and landscape rebates.
Implications
A comparison of all of the data collected revealed that in the decade from 2002 to
2012, overall, VVCSD customers reduced their average summer water consumption by
20%. While this consumption reduction cannot be attributed to Cash-4-Grass rebates
alone, in this study, all of the water conservation rebates were statistically significant to
123
varying degrees. Therefore, a robust water conservation program, including Cash-4-Grass
rebates, can have a significant impact on water consumption.
This study has the potential to positively impact a multitude of water customers
not only in California but around the world. While droughts get most of the press
coverage, water conservation is a topic that water professionals deal with year round,
during the rainy season as well as the droughts. Ideas for new and improved methods for
encouraging customers to conserve water are continually being sought. This study adds
new information to the topic of Cash-4-Grass rebates that may assist water conservation
coordinators in broaching the subject with their elected officials in the future. The
addition of Cash-4-Grass rebates to a water conservation program can help reduce water
consumption locally and increase water supplies globally.
Conclusion
In 2014, California enters yet another drought and water conservation is once
again in the local and national news. As Governor Brown is asking California residents to
reduce their water consumption by 20%, the temperatures on the central coast in February
are an unseasonably hot 80 degrees and residents throughout the area are turning on their
sprinklers in an attempt to revive their thirsty lawns. This study has shown that reducing
the amount of irrigable yards can significantly reduce water consumption. However, the
results have also shown that, as theorized by Ajzen and Fishbein (1972, 1977), the person
has to believe that the behavior is expected of them.
