NEED HELP ASAP EASY ASSIGNMENT

Journal of Current Issues in Media and Telecommunications ISSN: 1935-3588 Volume 6, Number 4 © Nova Science Publishers, Inc.

AN INSTRUMENT FOR MEASURING SOCIAL MEDIA USERS’ INFORMATION PRIVACY

CONCERNS

Babajide Osatuyi University of Texas-Pan American College of Business Administration

Department of Computer Information Systems & Quantitative Methods Texas, USA

ABSTRACT

Privacy concerns associated with the use of social media applications is gaining significant interest both in the academic and organizational communities. Calls have been made to researchers to investigate the theoretical underpinnings that explain concerns for information privacy in the social media context. A review of the information privacy literature reveals that till date, no study has taken a theory-driven approach to understand privacy concerns in the social media context. To fill the gap in the literature, this article evaluates information privacy concern instrument from prior research in the social media context. This Chapter draws on Social Penetration theory and Communication Privacy Management theory to explain social media users’ information privacy concern (SMIPC). Drawing on a sample of 270 avid social media users, this study examines the factor structure of the dimensions of concern for information privacy instrument from prior research. An exploratory factor analysis followed by a confirmatory factor analysis revealed three first-order factor structure of social media users’ concern for information privacy measurement instrument. Further analysis of alternative factor models revealed that a second order factor structure of SMIPC performs better than its first-order factor structure in the social media context.

Keywords: Privacy concerns, social media, instrument development, information privacy, validity, confirmatory factor analysis, exploratory factor analysis

 [email protected].

Babajide Osatuyi 360

INTRODUCTION The prevalence of social media avail individual users and organizations with

unprecedented access to personal information that was once arduous to gather. Undoubtedly, privacy concerns on social media platforms become critically important as vendors can now potentially have access to a large collection of users’ personal information. Unlike other online applications, users voluntarily contribute their personal information on social media platforms [1], thereby increasing the ease of gathering personally identifiable information.

Information privacy in the information systems (IS) literature is conceptualized in terms of personal information gathering, sharing, and usage [2, 3]. However, researchers need to consider taking a broader perspective with the definition of information privacy, especially in the context of social media [1, 4]. Information privacy on social media platforms need to be conceptualized as a multi-agent phenomenon that involves the volunteer of personal information from individuals to a group or collective under the assumption that shared information will be kept confidential among agents—social ties, social media platform, and policy makers [1]. Hence, information privacy is formally conceptualized in this article as the exchange of personal information among members of a social network with the implicit anticipation that members share a responsibility of keeping the shared information private (Osatuyi, forthcoming).

In addition to users’ shared responsibility to protect self-disclosed personal information, social media platforms provide policies, interfaces, and features that structure interaction among users, as well as to third parties on how to provide add-on features and applications to extend the platform’s functionality [5].

Researchers call for “theoretical and operational assumptions underlying the structure of constructs such as concern for information privacy (CFIP) should be re-investigated in light of emerging technology, practice, and research [6, p.37].” Drawing on Communication Privacy Management (CPM) theory [7], this article proposes that social media users’ information privacy concerns can be characterized in terms of access to personal information, errors in storing personal information, and collection of personal information. CPM provides coordination rules for information owners to collectively manage the disclosure of personal information.

In response to calls to study information privacy concerns in the context of social interactions [1], this study seeks to develop an instrument that can predict and explain information disclosure practices on social media platforms. In this study, social media information privacy concern (SMIPC) is defined as concerns about loss of privacy as a result of the disclosure of personal information to known and unknown external agents—including other social media users, social media platforms, and third parties.

This article contributes to the information privacy literature by providing a re- conceptualization of a broader perspective of information privacy on social media platforms, the development and empirical validation of a measurement instrument for studying users’ information privacy concerns on social media platforms.

An Instrument for Measuring Social Media Users’ … 361

THEORETICAL BACKGROUND CURRENT INFORMATION PRIVACY CONCERN MEASUREMENT INSTRUMENTS

One of the major breakthroughs in information privacy research was a study conducted

by Smith et al. [3]. In their work, they found that concern for information privacy was influenced by four fundamental factors based on individuals’ concern in response to organizations’ information practices, namely collection, unauthorized secondary use (internal and external), improper access to personal information, and errors in personal information storage. Smith et al. [3] measured and validated the four factors as first-order constructs in a nomological network.

In a later validation of Smith et al.’s [3] model, Stewart and Segars [6] posited that CFIP is complex and should be measured as a second-order construct. Stewart and Segars [6] then developed, tested, and validated their proposed hypothesis with CFIP as a multi-dimensional construct in a nomological network and found that it mediated the relationship between computer anxiety and behavioral intentions. Since then, researchers have validated the use of CFIP as a second-order construct, in nomological models across different contexts. While the context of CFIP was on direct offline marketing, Malhotra et al. [8] developed a multi- dimensional scale to measure Internet Users’ Information Privacy Concern (IUIPC) in the context of the Internet.

In response to Stewart and Segar’s [6] call to investigate the shifting dimensions of information privacy concerns in light of emerging technology, practice, and research, Xu et al. [9, p.3] developed a 9-item instrument to measure mobile users’ concerns for information privacy. The research program through which this study is conducted responds to the same call as it investigates measurement scales needed to understand individuals’ concern for information privacy on social media platforms, which is increasingly becoming an important communication medium for users, organizations, and government entities.

COMMUNICATION PRIVACY MANAGEMENT THEORY Privacy concerns about information sharing and collection is particularly germane to the

social media context as users voluntarily generate and contribute their personal information on social media platforms. Users’ perception that their personal information may be accessed and used by other entities (i.e., other users, social media platforms, and third-party vendors), may trigger concern for security and privacy. Web 2.0 technologies such as social media platforms are data-driven requiring regular collection of users’ preferences, conversations, and personal information in order to improve users’ experience on their platform. Users, platform developers, and third party vendors therefore share the ownership and responsibility to ensure that personal information exchanged on social media platforms are kept private [1]. In agreement with Xu et al.’s [9, p.3] notion in their research on mobile users’ concern for information privacy, “concerns about personal information disclosure [on social media platforms] cannot be fully understood without knowing users’ expectations about how their disclosed information will be used and who will have access to the information.”

This article draws on Communication Privacy Management (CPM) theory for its suitability in understanding the privacy consideration for how information is exchanged on

Babajide Osatuyi 362

social media platforms [7]. CPM is especially useful in this context as it recognizes individual and collective ramifications of information exchanged on social media platforms. It also accounts for the need to establish and coordinate boundaries to manage the privacy of information shared among entities (i.e., users, social media platforms, and vendors). Although CPM was developed before the emergence of social media as a widespread communication tool, this study extends it to account for communication beyond a user and other users, but to include the social media platform provider and other third parties.

CPM is positioned as a rule-based theory which posits that individuals develop rules to determine if and how to share information as a function of five criteria [7]: gender, culture, motivations, risk-benefit assessment, and context. Once individuals decide to disclose their personal information, the information moves to a collective domain where collectives (i.e., users, social media platforms, third parties) manage mutually held privacy boundaries. CPM is mainly focused on the collective management of private information. This focus leads to the need for boundary coordination process to collectively control private information by all the agents in the information space i.e., users, social media platforms, and third parties [7]. CPM theory [7] identifies three boundary rules for coordinating information disclosure between agents, including coordinating permeability rules, coordinating ownership rules, and coordinating linkage rules. These coordination rules “illustrate the modes of change for the dialectic of privacy-disclosure as managed in a collective manner” [7, p.127].

Finally, CPM posit that boundary turbulence occurs when co-owners of information are unable to collectively exhibit coordination rules guiding information permeability, ownership, and linkages [7]. The intervention of the Federal Trade Commission requiring Facebook to withdraw proposed privacy changes that would allow the company to use the names, images, and content of Facebook users for advertising without users’ consent is an example of boundary turbulence [10].

Boundary turbulence increases users’ privacy concern, consequently leading to re- coordination of their boundary rules to manage information permeability, ownership, and linkages.

SOCIAL MEDIA INFORMATION PRIVACY CONCERN (SMIPC) Based on the theoretical perspective presented using CPM, four main constructs for

measuring users’ concern for information privacy on social media platforms are introduced, including 1) unauthorized access and secondary used of personal information—information access, 2) information collection—collection, and 3) erroneous storage and representation of personal information—errors.

INFORMATION ACCESS AND USE For research in the information privacy domain, Smith et al. [3] developed two scales

(unauthorized access and secondary use) that relate to the use of users’ personal information without their consent. Secondary use is the use of personal data collected by an organization for a legitimate reason (e.g., shipping information by an e-commerce company), but used for

An Instrument for Measuring Social Media Users’ … 363

a secondary purpose without the consent of the user. Unauthorized access describes the interception of personal information captured from a legitimate transaction by a third party without the consent of either the user or the organization. In the context of social media, secondary use and unauthorized access both describe access and use of personal information without receiving permission from the information owner(s). In accordance to CPM, the unauthorized access and use of users’ personal information triggers the coordination of linkage rules used as response mechanism for “the establishment of mutually agreed-upon privacy rules used to choose others who may be privy to the collectively held information [11, p.70].” Drawing on secondary use and unauthorized access dimensions of CFIP, this article posits information access may be important to characterize SMIPC. Information access is conceptualized in the social media context to relate to the access and use of users’ personal information without the consent of the user.

COLLECTION Collection is defined in the information privacy research stream as “concern that

extensive amounts of personally identifiable data are being collected and stored in databases [3, p.172].” In today’s data-driven society, aggressive data collection strategies are used by various organizations to build a better understanding of their customers as well as to maintain competitive advantage. However, as noted by several researchers [3, 8], the practice of data collection, justifiable or not, raises privacy concerns for customers. With minimal integration into one’s daily routine, social media applications may be used to track and gather an individual’s behavioral pattern throughout each day e.g., choice of lunch locations, favorite online radio playlist, running/walking route information sharing with friends etc.

Users may not use social media for the fear that their personal information and private conversations may be collected and stored for future business or intelligence analysis. The act of data collection initiates users’ tendency to activate their coordination of permeability rules, which Petronio [12] describes as the extent to which information within the collectively owned privacy boundary should be disclosed to others. Studies show report that when individuals are provided with a significant control over information disclosure, they create boundary structures that reduce the amount of information collection by others or they establish boundaries with low permeability [12, 13]. Accordingly, this research posits that personal information collection is an important factor that characterizes SMIPC.

ERRORS Smith et al. [3, p.172] described error as individuals’ “concern that protections against

deliberate and accidental errors in personal data are inadequate.” The authors noted that privacy-related concerns are initiated when errors are made in the representation of customers’ information. In the offline context, such errors are not uncommon due to inevitable mistakes with the data entry process. However, in the context of social media where personal information is user-generated, errors may be conceptualized as a deliberate act on the part of the information provider. Studies abound in the social interaction context

Babajide Osatuyi 364

that shows that users deliberately enter incomplete or inaccurate information about themselves on online social platforms [e.g., 14]. Using CPM as a backdrop, the tendency for users to provide erroneous information is consistent with ownership coordination strategy [or rules] used by individuals to ensure that unwanted agents are not privy to their complete personal information. From the perspective of the social media platform, improper update of personal information may also lead to erroneous personal data. Lastly, third party vendors have no way of confirming the integrity of consumers’ information available to them, which may lead to misappropriation of advertisement or other services. Hence, this study posits that errors characterize SMIPC as it can affect the user, the platform provider, and third party vendors.

METHOD

Scale Development A survey instrument was developed to test the hypothesized model in accordance to the

practice in the information privacy domain [3, 15]. Scales from the instruments for measuring concern for information privacy posited Smith et al. [3] were adapted based on prior research. Information access was measured by a combination of items from unauthorized access and secondary use scales from Smith et al. [3]. Collection and error were adapted from Smith et al. [3] and measured with multiple items on five-point Likert scales, anchored with strongly disagree to strongly agree. Behavioral intentions and computer anxiety were adapted from Stewart and Segar’s [6] and measured with three items on five-point Likert scales, anchored with highly likely to not at all likely and strongly disagree to strongly agree respectively.

Experts in both social media and information privacy research fields reviewed the initial versions of the survey questions. Additional feedback was received from a sample of graduate and undergraduate students on the clarity of the questions and options before the final version was finally developed. All measurement items used in this research are included in the Appendix.

Survey Design An online survey was developed and the link was sent to college students in a southern

region of United States. Students received extra-credits toward their final grade for participating in the study. To protect the privacy of participants and in compliance with the IRB regulations of the institution, no personally identifiable information was collected from participants. There were 310 participants in the study, but only 250 of the responses were complete and useful for analysis. Respondents had a median age of 19, and 63.2% of them were female. 91% of the participants reported that they use social media on a daily basis. Based on Pew research reports [16], college students are predominant social media users, hence, choosing college students as the study population is ideal for this research.

An Instrument for Measuring Social Media Users’ … 365

DATA ANALYSIS AND RESULTS Data analysis was conducted in two phases; Phase 1 sought to identify and validate the

factor structure of SMIPC and Phase 2 evaluated SMIPC in a nomological network.

Step 1: Identification: Factor Structure of SMIPC Identifying a proper structure for the SMIPC construct is necessary since it comprises

scales from mature and validated instruments in addition to a newly developed item based on prior literature. As suggested by prior literature [e.g., 8], exploratory factor analysis (EFA) of the various factors of SMIPC was conducted followed by confirmatory factor analysis (CFA).

Two sets of EFA were conducted with IBM SPSS software using Principal Components Analysis technique with VARIMAX rotation and Kaiser Normalization. The first EFA included only the four dimensions of CFIP posited by Smith et al. [3]. As shown in Table 1, three components were revealed and all the items loaded cleanly on their respective constructs with no cross loadings. The second EFA included the newly developed scale, privacy policy transparency, in addition to the four dimensions of CFIP posited by Smith et al. [3]. As shown in Table 1, three components were also revealed with the newly developed scale (privacy policy transparency) and two of the dimensions from prior research (unauthorized access and secondary use) converging on the same factor component.

Cronbach Alpha was used to assess the reliability of the factors revealed from the EFAs [17]. Cronbach Alpha for the three factors in both EFAs exceed the 0.70 threshold recommended by Nunnally [18], indicating convergent validity. Since the privacy policy transparency construct loads on the same factor as the composite construct that comprise secondary use and unauthorized access constructs, it is plausible to propose two factor structures for SMIPC. The first SMIPC structure (SMIPC Structure I) can be divided into three dimensions (unauthorized access and secondary use, collection, and error) and the second structure (SMIPC Structure II) into three dimensions (privacy policy transparency, collection, and error).

Table 1. Principal Component Analysis Using VARIMAX Rotation

Factor

Items

Component I

1 2 3 CA

Collection

COL1 .783 0.82

COL2 .747

COL3 .809

COL4 .740

Error

ERR1 .852 0.92

ERR2 .812

ERR3 .873

UAC1 .743

Babajide Osatuyi 366

Information Access = Unauthorized Access + Secondary Use of Personal Information

UAC2 .766 0.93

UAC3 .819

UAC4 .660

SUS1 .805

SUS2 .765

SUS3 .805

Rotation Sums of Squared Loadings

Total 4.655 2.894 2.720

% Variance 33.251 20.671 19.426

Cumulative Variance

33.251 53.921 73.347

CA-Cronbach’s Alpha. Consistent with prior study [3], CFA is used to assess the efficacy of the model structure

suggested from the EFA, which contains 14 items. This research follows the procedures used by Stewart and Segars [6] to conduct CFA, which compared covariance matrices based on observed data and the hypothesized models. In the case of this research there are two observed covariance is a 14 × 14 matrix of the measures adopted to measure SMIPC. According to Stewart and Segars [6], CFA allows a researcher to specify, estimate, and re- specify multiple and interrelated dependence relationships as well as unobserved constructs. CFA assesses a hypothesized model by comparing its observed covariance matrix with the implied covariance matrix. The implied matrix is a set of covariances (14 × 14) generated through maximum likelihood estimation as a result of the specified model [6]. The closer the two models are, the better the model fit indicating that the specified model is indicative of the data collected. Goodness-of-fit indices are then used to report the result of the comparison between the observed and implied matrices. In accordance with prior studies [19, 20], multiple fit indices were used for assessing the fit between the observed and implied models in this study. The following section presents four1 hypothesized models to assess the factorial nature of SMIPC.

Model 1 hypothesizes that all items of SMIPC form into one first-order factor accounting for all the common variance among the 14 items. Prior research [3] measured privacy concern as though it were a unidimensional construct indicating that one first-order factor can be used to explain the underlying data structure. If this model is accepted, then it is appropriate to consider SMIPC as a single dimension that governs similarities in variation among all 14 items.

Model 2 hypothesizes that all 14 items of SMIPC form into two first-order factors: secondary use and unauthorized access is loaded onto one factor and error and collection are loaded onto the second factor. This model proposes that individuals’ concern for information privacy on social media sites is divided into two areas: 1) individuals’ concern for privacy may be triggered by their perception of other people’s access to their personal information and misuse of such information, and 2) users’ concern as a result of erroneous collection of personal information on social media sites.

1 Although only four models are discussed in this article, the four-structure CFIP model proposed in prior research

was specified and tested with data collected in this study. The fit indices for the models with four independent measures were not as strong as the 3-factor structure revealed in the context of social media.

An Instrument for Measuring Social Media Users’ … 367

Model 3 hypothesizes that three first-order factors account for the covariance among the 14 items as unauthorized access and secondary use, collection, and error. This model suggests scaling SMIPC as an average of the subscale scores to calculate an overall score. As noted by Stewart and Segars [6], the assumption of this model is that all the items are equally important in computing each factor and each factor is equally vital to computing the overall score for the SMIPC construct.

Model 4 hypothesizes that all items form into three first-order factors, which are then measured by a second-order factor SMIPC. According to Stewart and Segars [6], in such a model, “the inter-correlations among first-order factors form a system of interdependence (or covariation) that is itself important in measuring the construct. Conceptually, each factor and the second-order factor are necessary in capturing the nature of the construct domain [6, p.39].” SMIPC can therefore be defined as three distinct factors as well as the structure of interrelationships among these factors.

Table 2 presents the results of testing all four models for SMIPC. As in prior research [6, 9], fit indices are examined in terms of NFI, GFI, AGFI, CFI, NNFI, Std. RMR, and RMSEA. As shown in the results, Model 1 and Model 2 have poor goodness of fit indices. Models 3 and 4 have acceptable goodness fit indices, but Model 4 (2nd-order factor) indicates better fit to the data.

This result supports Stewart and Segar’s [6] thesis that information privacy concern models are better structured as a higher order factor models rather than 1st-order factor models.

Since Model 3 and Model 4 exhibit stronger measures of fit compared to other alternative models, the convergent and discriminant validity of both models are further examined.

Table 2. Measures of Model Fit: Confirmatory Factor Analysis Results for Alternative

Factor Structures

Fit Indices Recommended Indices

Alternative SMIPC Factor Structures

Model 1: One 1st- Order Factor

Model 2: Two 1st- Order Factors

Model 3: Three 1st- Order Factors

Model 4: 2nd-Order Factor

χ2 387.70* 306.58* 114.35* 82.65

df 64 63 69 64

χ2/(df) < 3.00 6.06 4.87 1.66 1.29

NFI > 0.90 0.87 0.90 0.98 0.97

GFI > 0.90 0.84 0.85 0.95 0.96

AGFI > 0.80 0.73 0.75 0.92 0.94

CFI > 0.90 0.89 0.92 0.98 0.99

NNFI > 0.90 0.84 0.88 0.98 0.99

Std.RMR < 0.05 0.08 0.07 0.03 0.03

RMSEA < 0.06 0.13 0.11 0.05 0.03

*p<0.05; Std. RMR-Standardized RMR.

Table 3. Convergent Validity for Model 3 and Model 4

Construct

Item

Model 3

Model 4

Std. Loading t-value AVE CR Std. Loading t-value AVE CR

Collection

COL1 0.90 8.26 0.76

0.93

0.97 7.79 0.95

0.99 COL2 0.84 7.43 0.95 7.74

COL3 0.94 10.78 0.99 9.44

COL4 0.81 7.44 0.99 8.37

Error

ERR1 0.97 16.24

0.86 0.95

0.89 15.89

0.90 0.96 ERR2 0.91 15.92 0.99 16.14

ERR3 0.90 16.03 0.95 16.66

Information Access and Use = Unauthorized Access + Secondary Use of Personal Information

UAC1 0.88

16.66 0.87

0.98

UAC2 0.99 20.48

UAC3 0.98 16.89

UAC4 0.94 17.00

SUS1 0.90 16.65

SUS2 0.90 15.89

SUS3 0.92 17.53

AVE—Average Variance Extracted; CR—Composite Reliability.

An Instrument for Measuring Social Media Users’ … 369

As shown in Table 3 the t-values for all the construct items indicate significant factor loadings and provide evidence to support the convergent validity of the items measured [21]. The composite reliability results, shown in Table 3, measure internal consistency of the scales, each exceeding the recommended 0.70 threshold [22], indicating satisfactory reliability for all the factors. The average variance extracted (AVE) for each scale exceeds the recommended 0.50 threshold [22]. Put together, both models demonstrate strong properties of convergent validity.

Although the psychometric properties of Model 3 and Model 4 are strong and satisfactorily fit the data, Model 4 represents the structure of SMIPC more parsimoniously than Model 3.

To assess the discriminant validity of the SMIPC model, correlations among the latent variables are examined as shown in Table 4. Discriminant validity is assessed if the square root of the AVE is larger than correlation coefficients [22]. The result shows that correlations among the latent variables are less than the square root of the AVEs along the diagonals of Table 4. Hence, the measurement model is shown to exhibit strong discriminant validity.

Table 4. Correlations among Latent Variables

IAC COL ERR Information Access (IAC) 0.96 Collection (COL) 0.58 0.97 Error (ERR) 0.68 0.44 0.95

Step 2: Validation: SMIPC within a Nomological Network Consistent with prior work [6], the construct for SMIPC is tested for nomological

validity. In accordance with the approach recommended by Chin [23], establishing the efficacy of a second-order model involves its assessment with other constructs within a nomological network. In accordance to the procedure established by Stewart and Segars [6], a second-order factor is expected to act as a significant mediator when embedded within a network of predictor and consequent variables. Accordingly, SMIPC is placed between a predictor variable (computer anxiety) and a consequent variable (behavioral intention). Stewart and Segars [6] validated CFIP as a mediator between individuals’ frustration with the use of computers (i.e., computer anxiety) and their behavioral intention to use it in future. Individuals that are apprehensive or fearful about current or future use of computers are found to have stronger levels of privacy concerns [3, 6]. As such, this article argues that SMIPC will act as a consequent of computer anxiety. Individuals that are anxious about the use of computers are highly likely to be concerned about how their personal information is collected and used on social media platforms.

As for the predictor variable of SMIPC, individuals with higher levels of privacy concerns are more likely in the future to refuse to disclose their personal information, and refuse to use a technology that demands data collection with online merchants. Prior research provides evidence for a negative relationship between privacy concern and behavioral intention [24]. Therefore, a negative relationship is expected between SMIPC and users’ behavioral intentions.

Babajide Osatuyi 370

Using previously defined scales for computer anxiety [3] and an adaptation of behavioral intentions [6], the analysis of SMIPC is expanded using a 20 × 20 covariance matrix consisting of a 14-item SMIPC scale, a 3-item computer anxiety scale, and a 3-item behavioral intentions scale. Table 5 presents the structural model for Model 5 (1st-order factor model) and Model 6 (2nd-order factor model).

Table 5. Measures of Model Fit: SMIPC within a Nomological Network

Fit Indices Recommended Indices Model 5: 1st-Order Factor

Model 6: 2nd-Order Factor

χ2 158.11 105.17

df 145 101

χ2/(df) < 3.00 1.10 1.04

NFI > 0.90 0.95 0.97

GFI > 0.90 0.95 0.96

AGFI > 0.80 0.93 0.94

CFI > 0.90 0.99 0.99

NNFI > 0.90 0.99 0.99

Std.RMR < 0.05 0.04 0.04

RMSEA <0.06 0.02 0.01

Following recommendations from prior research, fit indices in terms of Normed χ2, NFI,

GFI, AGFI, CFI, NNFI, and RMSEA indicate good model fit for Model 5 and Model 6. Based on the NFI, GFI, AGFI, and RMSEA indices, Model 61 demonstrates stronger fit compared to Model 5. Figure 1 demonstrates both models (Model 5 and Model 6) and associated estimates of SMIPC mediating the relationship between computer anxiety and behavioral intentions. The paths are all significant and consistent with the theoretical prediction. When compared to Model 4 (1st-order factor model), Model 6 (2nd-order factor model) appears to have a better fit.

DISCUSSION AND CONCLUSION The research program through which this study was conducted seeks to respond to the

call [1, 4] for a better understanding of information privacy concerns with interpersonal social interactions in the context of social media. As noted in the call [1], information privacy on social media platforms needs to be conceptualized as a multi-agent phenomenon that involves the volunteer of personal information from individuals to a group or collective under the assumption that shared information will be kept confidential among agents, including social ties members, social media platforms, and third party vendors [1].

1 Discriminant validity was also verified using nested model method recommended by Bagozzi et al. [25], which

showed significant differences among all the models, showing that the constructs are distinct from one another. Common method bias analysis was conducted to rule out the variance attributable to the measurement method rather than to the constructs, and none was observed in the data.

The Use of Social Media for Job Placement in Career Centres… 371

It is plausible to expect that privacy issues in the context of social media will become important as consumers’ personal information is more readily accessible by other users, vendors, and social media platforms. Although research in the area of information privacy is matured and extensive, the social media context of information privacy research is in its infancy, hence the call for research in this domain [1]. Drawing on Communication Privacy theory [7], this study empirically developed and measured SMIPC based on the three dimensions including, collection of personal information, and errors with the storage and representation of personal information. The three-factor structure of SMIPC was revealed in an exploratory factor analysis (EFA), which was further confirmed through confirmatory factor analysis. Three factors emerged from the EFA with secondary use and unauthorized access loading on one factor, errors loaded on the second factor, and collection loaded on the third factor. Further analysis revealed that the 2nd-order factor model of SMIPC outperformed the 1st-order factor model of SMIPC based on better fit indices. The better fit indices for the second-order model of SMIPC does not simply imply that social media users are concerned about these issues, but it also suggest that interdependencies among these issues are vital to measuring individuals’ information privacy concerns on social media platforms (SMIPC).

Figure 1. SMIPC as a Second-Order Model within its Nomological Network.

The results of this study provide interesting insights into the dimensionality of SMIPC construct in the context of social interactions. This study therefore contributes to the literature by developing and validating an instrument for measuring individuals’ privacy concern on social media platforms. As noted by Stewart and Segars [6], with this validated instrument, further research can now be conducted into the relationships among antecedents and consequences of information privacy concerns on social media platforms. Additionally, researchers are invited to use the instrument developed in this study with confidence due to the strong reliability and validity of the constructs indicated in the results. The SMIPC instrument can now be used as a standardized measure of individuals’ concern for information privacy on social media platforms.

Although the results presented are insightful, there are limitations associated with the conduct of this study. As noted in prior research [6, p.45], results of confirmatory factor

Babajide Osatuyi 372

analysis must be interpreted cautiously, since the “criteria for comparing [competing] models and assessing goodness-of-fit indices are relative and not absolute.” This suggests that a model is a good representation of reality when it can be replicated in subsequent studies. As such, researchers the authors call on researchers to further confirm the validity of SMIPC in other contexts.

CONTRIBUTION Drawing on CPM theory, this study seeks to define the nature of privacy concerns in the

social media context based on three boundary coordination rules [7], including (a) information access (rooted in linkage and permeability coordination rules), (b) error in storing and representation of personal information (rooted in ownership coordination rule), and (c) collection of personal information (rooted in permeability coordination rule). In the social media context, this article argues that users are likely to become concerned about the safety of their personal information shared on social media platforms with the knowledge that other users, the platform, and third party vendors may gain access to their personal information and misappropriate it. In particular, social media users’ concern over information access and information misuse can be triggered by the violation of linkage and permeability rules i.e., when linkage to personal data occurs and vendors have access to users’ personal information without expressed consent from the user. With regards to error, users’ perception of error can become salient with the violation of ownership rules i.e., when vendors and social media platforms are able to make decisions about the possession of users’ personal information. Finally, users’ perception of collection of their personal information can become salient when permeability rules are violated i.e., when social media platforms and their vendors can readily access users’ personal information without consent. Table 6 presents a summary of information privacy measurement instruments with their theoretical foundations.

APPENDIX: CONCERN FOR INFORMATION PRIVACY MEASUREMENT SCALES

Computer Anxiety (CAX) [Source: [6]] 1. I am sometimes frustrated by increasing automation in my home 2. I am sometimes afraid that I will delete all my important files 3. I am sometimes afraid that I will mistakenly send personal information to the wrong

recipient Concern for Information Privacy [Source: [3]] Collection (COL) 1. It usually bothers me when social media sites ask me for personal information 2. It usually bothers me when companies ask me to like or follow their social media

sites during transactions 3. It bothers me to give personal information to so many people on the social media

sites I am registered with

The Use of Social Media for Job Placement in Career Centres… 373

4. I am concerned that companies are collecting too much personal information about me through the social media sites I am registered on

Unauthorized Access (UAC) 1. Computer databases that contain personal information should be protected from

unauthorized access—no matter how much it costs 2. Social media sites should take more steps to make sure that unauthorized people

cannot access personal information in their computers 3. Databases that contain personal information should be stored in a highly secured

location 4. Social media sites should delete a user’s account for illegally accessing other users’

personal information Errors (ERR) 1. Companies should take more steps to make sure that personal information in their

files is accurate 2. Companies should have better procedures to correct errors in personal information 3. Companies should devote more time and effort to verifying the accuracy of the

personal information in their databases Secondary Use (SUS) 1. Social media sites and companies should not use personal information for any

purpose unless it has been authorized by the individuals who provide the information 2. When people give personal information to a social media site or company for some

reason, the company should never use the information for any other purpose 3. Social media sites or companies should never share personal information with other

companies unless it has been authorized but the individual who provided the information

Behavioral Intentions (BIN) [Source: [6]] How likely are you, within the next three years to… 1. Create an online account with an organization or company to complete a transaction? 2. Create an online account with an organization or company you frequently do

business with? 3. Register online with a company or organization that you frequently do business with?

REFERENCES

[1] Xu, H., and BéLanger, F., "Information Systems Journal Special Issue on: Reframing Privacy in a Networked World," Information Systems Journal (23, 2013, 371-375.

[2] Bélanger, F., and Crossler, R.E., "Privacy in the digital age: a review of information privacy research in information systems," MIS Quarterly (35:4), 2011, 1017-1042.

[3] Smith, H.J., Milberg, S.J., and Burke, S.J., "Information privacy: measuring individuals' concerns about organizational practices," MIS Quarterly, 1996, 167-196.

Babajide Osatuyi 374

[4] Lipford, R.H., Wisniewski, J.P., Lampe, C., Kisselburgh, L., and Caine, K., "Workshop on Reconciling Privacy with Social Media", The ACM Conference on Computer Supported Cooperative Work, 2012

[5] Aral, S., Dellarocas, C., and Godes, D., "Introduction to the Special Issue-Social Media and Business Transformation: A Framework for Research," Information Systems Research (24:1), 2013, 3-13.

[6] Stewart, K.A., and Segars, A.H., "An empirical examination of the concern for information privacy instrument," INFORMATION SYSTEMS RESEARCH (13:1), 2002, 36-49.

[7] Petronio, S., Boundaries of privacy: Dialectics of disclosure, State University of New York Press, New York, 2002.

[8] Malhotra, N.K., Kim, S.S., and Agarwal, J., "Internet users' information privacy concerns (IUIPC): the construct, the scale, and a causal model," Information Systems Research (15:4), 2004, 336-355.

[9] Xu, H., Gupta, S., Rosson, M.B., and Carroll, J.M., "Measuring Mobile Users' Concerns for Information Privacy", Proceedings of 33rd Annual International Conference on Information Systems (ICIS 2012), 2012

[10] Epic, "Pressure Mounts on Facebook to Withdraw Proposed Changes, New Scrutiny of "Faceprints"", in (Editor, 'ed.'^'eds.'): Book Pressure Mounts on Facebook to Withdraw Proposed Changes, New Scrutiny of "Faceprints", Electronic Privacy Information Center, 2013

[11] Jin, S.-a.A., "“To disclose or not to disclose, that is the question”: A structural equation modeling approach to communication privacy management in e-health," Computers in Human Behavior (28:1), 2012, 69-77.

[12] Petronio, S., "Communication privacy management theory: What do we know about family privacy regulation?," Journal of Family Theory & Review (2:3), 2010, 175-196.

[13] Child, J.T., Pearson, J.C., and Petronio, S., "Blogging, communication, and privacy management: Development of the blogging privacy management measure," Journal of the American Society for Information Science and Technology (60:10), 2009, 2079- 2094.

[14] Dwyer, C., Hiltz, S., and Passerini, K., "Trust and Privacy: A Comparison of Facebook and MySpace," Americas Conference on Information Systems, 2007,

[15] Son, J.-Y., and Kim, S.S., "Internet users' information privacy-protective responses: A taxonomy and a nomological model," MIS Quarterly (32:3), 2008, 503-529.

[16] Smith, A., Raine, L., and Zickuhr, K., "College students and technology", in (Editor, 'ed.'^'eds.'): Book College students and technology, Pew Research Center, Washington, D.C. (July 19, 2011), 2011

[17] Bagozzi, R.P., Casual Methods in Marketing, John Wiley and Sons, New York, 1980. [18] Nunnally, J., "Psychometric Theory", McGraw-Hill, New York, NY, 1978 [19] Bentler, P.M., and Bonett, D.G., "Significance tests and goodness of fit in the analysis

of covariance structures," Psychological Bulletin (88:3), 1980, 588. [20] Marsh, H.W., and Hocevar, D., "A new, more powerful approach to multitrait-

multimethod analyses: Application of second-order confirmatory factor analysis," Journal of Applied Psychology (73:1), 1988, 107.

The Use of Social Media for Job Placement in Career Centres… 375

[21] Anderson, J.C., and Gerbing, D.W., "Structural Equation Modeling in Practice: A Review and Recommendation Two-Step Approach," Psychological Bulletin (103:3), 1988, 411-423.

[22] Fornell, C., and Larcker, D.F., "Evaluating structural equation models with unobservable variables and measurement error," Journal of Marketing Research, 1981, 39-50.

[23] Chin, W.W., "Issues and Opinion on Structural Equation Modeling," MIS Quarterly (22:1), 1998, VII-XVI.

[24] Xu, H., and Teo, H.-H., "Alleviating Consumers' Privacy Concerns in Location-Based Services: A Psychological Control Perspective", Proceedings of the Twenty-Fifth Annual International Conference on Information Systems (ICIS 2004), 2004, pp. 793- 806.

[25] Bagozzi, R.P., Yi, Y., and Phillips, L.W., "Assessing construct validity in organizational research," Administrative Science Quarterly (36:3), 1991.

[26] Podsakoff, P.M., Mackenzie, S.B., Lee, J.-Y., and Podsakoff, N.P., "Common method biases in behavioral research: a critical review of the literature and recommended remedies," Journal of Applied Psychology (88:5), 2003, 879.

[27] Bansal, G., Zahedi, F., and Gefen, D., "The moderating influence of privacy concern on the efficacy of privacy assurance mechanisms for building trust: A multiple-context investigation", Proceedings of 29th Annual International Conference on Information Systems (ICIS 2008), 2008.

[28] http://www.ftc.gov/reports/privacy2000/privacy2000.pdf, accessed April 20, 2014. [29] Loeffler, C., "Privacy issues in social media," IP Litigator, 2012, 12-18. [30] Preibusch, S., Hoser, B., Gürses, S., and Berendt, B., "Ubiquitous social networks –

opportunities and challenges for privacy-aware user modelling", in (Editor, 'ed.'^'eds.'): Book Ubiquitous social networks – opportunities and challenges for privacy-aware user modelling, 2007.

Copyright of Journal of Current Issues in Media & Telecommunications is the property of Nova Science Publishers, Inc. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.