A Social Cognitive Perspective on Mobile Communication Technology Use and Adoption

http://ssc.sagepub.com/Social Science Computer Review

http://ssc.sagepub.com/content/27/1/76The online version of this article can be found at:

DOI: 10.1177/0894439308322594

2009 27: 76 originally published online 13 August 2008Social Science Computer ReviewOscar PetersAdoption

A Social Cognitive Perspective on Mobile Communication Technology Use and

Published by:

http://www.sagepublications.com

can be found at:Social Science Computer ReviewAdditional services and information for

http://ssc.sagepub.com/cgi/alertsEmail Alerts:

http://ssc.sagepub.com/subscriptionsSubscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://ssc.sagepub.com/content/27/1/76.refs.htmlCitations:

What is This?

- Aug 13, 2008 OnlineFirst Version of Record

- Dec 29, 2008Version of Record >>

at TEXAS SOUTHERN UNIVERSITY on October 18, 2014ssc.sagepub.comDownloaded from at TEXAS SOUTHERN UNIVERSITY on October 18, 2014ssc.sagepub.comDownloaded from

http://ssc.sagepub.com/

http://ssc.sagepub.com/content/27/1/76

http://www.sagepublications.com

http://ssc.sagepub.com/cgi/alerts

http://ssc.sagepub.com/subscriptions

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsPermissions.nav

http://ssc.sagepub.com/content/27/1/76.refs.html

http://ssc.sagepub.com/content/27/1/76.full.pdf

http://ssc.sagepub.com/content/early/2008/08/13/0894439308322594.full.pdf

http://online.sagepub.com/site/sphelp/vorhelp.xhtml



76

Social Science Computer ReviewVolume 27 Number 1February 2009 76-95

© 2009 Sage Publications10.1177/0894439308322594

http://ssc.sagepub.comhosted at

http://online.sagepub.com

A Social Cognitive Perspective onMobile Communication TechnologyUse and AdoptionOscar PetersUniversity of Twente, The Netherlands

This study examined the triadic relationship between expected outcomes, habit strength, andmobile communication technology use and adoption within the model of media attendance(LaRose & Eastin, 2004). Mobile phone users (N = 644) were divided into two groups usinga stratified random sampling method. Respondents from Group 1 (n = 334) were surveyed formobile phone use and respondents from Group 2 (n = 310) were surveyed for the intention toadopt mobile video telephony. On the basis of structural equation analysis for the model ofmedia attendance, the results of this study support the assumptions that (a) mobile phone useis more likely to be explained by habit strength and (b) the intention to adopt mobile videotelephony is more likely to be predicted by outcome expectations.

Keywords: model of media attendance; mobile communication technology; social cognitivetheory; media adoption; media use; habit strength; expected outcomes

The present research further extends the study of media use behavior by examining therole of habituation in explaining and predicting mobile communication technology use

and adoption from a social cognitive perspective. Social cognitive theory is a broad theoryof human behavior (Bandura, 1986) that stems from social learning theory, which has a richhistorical background dating back to the late 1800s. The early foundations of social learn-ing theory were suggested by behavioral and social psychologists. Social cognitive theory,however, has also been applied in the context of media use behavior (e.g., Bandura, 2002).The comprehensiveness and complexity of social cognitive theory make it somewhat diffi-cult to operationalize, and many applications of social cognitive theory focus on just one ortwo constructs (e.g., self-efficacy in Hofstetter, Zuniga, & Dozier, 2001), while ignoringothers. Inspired by Bandura’s social cognitive theory, LaRose and Eastin (2004) proposedand tested a new model of media attendance (see Figure 1) that integrates several key con-structs of social cognitive theory into one causal model.

Within the model of media attendance, media usage is defined as overt media consumptionbehavior. It is determined by the expected outcomes that follow from media consumption.According to LaRose, Lin, and Eastin (2003), outcome expectations reflect current beliefsabout the outcomes of prospective future behavior but are predicated on comparisonsbetween incentives expected and incentives attained in the past. The model was able to

Author’s Note: The author wishes to thank Jan A.G.M. van Dijk, Ard Heuvelman, Robert LaRose, ArunVishwanath, and the anonymous reviewers for their constructive criticism and valuable feedback regarding thispaper. Please address correspondence to Oscar Peters at [email protected].

at TEXAS SOUTHERN UNIVERSITY on October 18, 2014ssc.sagepub.comDownloaded from


Peters / Social Cognitive Perspective on Mobile Communication Technology 77

explain media consumption to an unprecedented degree. In the context of Internet use, forexample, studies that have employed prospective measures have consistently doubled,tripled, or quadrupled the amount of variance explained in Internet attendance behaviorcompared to conventional approaches (LaRose, Mastro, & Eastin, 2001).

Validation studies for the model of media attendance (Peters, Rickes, Jöckel, Von Criegern,& Van Deursen, 2006) suggested that once media use was strongly habitualized, furtherincrease of habitualization would lead to a decreasing consciousness of expected outcomes.Therefore, once media use is strongly determined by habituation, the effect of outcome expec-tations in determining media use may no longer have much influence. This occurs becausepeople no longer consciously evaluate expected outcomes; the adoption decision has alreadybeen made, or people no longer have expectations because the outcomes are already known.

In the next section, I begin with a description of both the relevant components of the modelof media attendance and the validation of the model by previous studies. I then present themethod and report the findings of a study to test the merits of the model for explaining mobilephone use as well as predicting the intention to adopt mobile video telephony. With mobilevideo telephony, people can not only talk to, but can also see each other on the phone’s screen.

Outcome Expectations

Within social cognitive theory, human behavior is defined as a triadic, dynamic, andreciprocal interaction of personal factors, behavior, and the environment (Bandura, 1986).This triadic causal mechanism is mediated by symbolizing capabilities that transform sen-sory experiences into cognitive models that guide actions. Within social cognitive theory,LaRose and Eastin (2004) suggest that behavior is an observable act. Furthermore, theperformance of behavior is determined in large part by the expected outcomes of behavior.

Figure 1The Model of Media Attendance (LaRose & Eastin, 2004)

ExpectedOutcomes

Self-efficacy Usage

Experience Habit Strength DeficientSelf-Regulation



These expectations are formed by our own direct experience (enactive learning) or medi-ated by vicarious reinforcement observed through others (vicarious learning).

Outcome expectations, defined as judgments of the likely consequences of behavior(Bandura, 1997), provide incentives for enacting behavior; in contrast, expectations ofaversive outcomes provide disincentives (Bandura, 1986). Expected outcomes are orga-nized around six basic types of incentives for human behavior (Bandura, 1986, p. 232).These include monetary incentives, social incentives (such as obtaining approval from oth-ers), and status incentives. Sensory incentives involve exposure to pleasing or novel sensa-tions. Preferences for enjoyable activities are the basis for activity incentives. There are alsointernal, self-reactive incentives resulting from comparisons of personal actions with stan-dards for behavior. According to LaRose and Eastin (2004), these incentives are theoreti-cally constructed rather than statistically derived from exploratory factor analysis.

Self-Efficacy, Self-Regulation, Experience, and Habit Strength

Other concepts from social cognitive theory that are important for understanding mediatechnology behavior include self-efficacy and self-regulation. Self-efficacy is the belief inone’s capacity to organize and execute a particular course of action (Bandura, 1997). Thosewho perceive themselves to be highly efficacious with reference to a particular task willinvest sufficient levels of effort to achieve successful outcomes, whereas those with lowlevels of self-efficacy will not persist. LaRose and Eastin (2004) posited that self-efficacyis directly and indirectly related to media usage through expected outcomes. Prior experi-ence in turn causally precedes self-efficacy (Eastin & LaRose, 2000), probably through theprocess of enactive mastery (Bandura, 1986). The social cognitive theory construct of self-regulation (Bandura, 1991) describes how individuals monitor their own behavior, judge itin relation to personal and social standards, and apply self-reactive incentives to moderatetheir behavior. Within social cognitive theory, habit is a failure of the self-monitoring sub-function of self-regulation. A concept related to habit is deficient self-regulation, a state inwhich conscious self-control is diminished (LaRose & Eastin, 2004).

Although habit and deficient self-regulation have not been clearly empirically distin-guished in prior research, LaRose, Lin, and Eastin (2003) proposed a possible theoreticaldistinction in which habit represents the failure of self-monitoring and deficient self-regulation represents a failure of the judgmental and self-reactive subfunctions. Accordingto LaRose and Eastin (2004), deficient self-regulation reflects a state of mind distinct fromone in which media consumers are inattentive. This explains how both might have inde-pendent effects on media attendance. Habit strength and deficient self-regulation should berelated by the fact that people with deficient self-control may also be expected to engage inhabitual behavior. Habit strength is expected to influence ongoing behavior. According toLaRose and Eastin, repetition makes us inattentive to the reasoning behind our mediabehavior; our mind no longer devotes attention resources to evaluating it, thereby freeingitself for more important decisions. LaRose and Eastin posited that habit strength should becausally determined by outcome expectations, which precede habit strength in time. Habitstrength should be preceded by self-efficacy, as users are unlikely to be inattentive tobehavior they are still mastering.

78 Social Science Computer Review




Validation of the Model of Media Attendance

To empirically examine the strength of the model of media attendance outside itsAmerican context, Peters et al. (2006) replicated the original study by LaRose and Eastin(2004) on Internet usage in Germany. In contrast to the original study, Peters et al. found anegative direct effect of expected outcomes on Internet habit strength. According to Peterset al., however, this is not in principal conflict with social cognitive theory. Social cogni-tive theory proposes that stronger habitualization leads to diminishing consciousness ofexpected outcomes. This was also reflected in the mean scores for measures of habitstrength and experience. The mean score for habit strength in the original 2004 study was4.19 (SD = 1.49) on a seven-point Likert scale and that for experience was 5.74 years (SD =3.04). The mean score for habit strength in the 2006 validation study was 5.67 (SD = 1.38)on a seven-point Likert scale and that for experience was 7.22 years (SD = 2.75). Thismight indicate that the stage of habitualization influences the relationship betweenexpected outcomes and habit strength. Habitualization is an individual process and anincrease of habitualization during its early stages should lead to increased consciousness ofexpected outcomes. After stronger habitualization occurs, further increases of habitualiza-tion should lead to decreased consciousness of expected outcomes. According to Peters et al.,therefore, a more precise prediction of the correlation between expected outcomes andhabit strength depends on the user’s stage of individual habitualization. Peters et al. alsonoted the importance of the direct integration of habit strength in the model of media atten-dance. A separate test of the causal relationship between the expected outcomes andInternet usage showed that the direct causal connection between expected outcomes andInternet usage was overestimated without the integration of habit strength.

In the Netherlands, Peters et al. (2006) tested the merits of the new model of media attendancein a context of media use other than the Internet. Peters et al. adapted the model of media atten-dance to the context of mobile communication technology (i.e., the use of General Packet RadioServices, GPRS). With the use of GPRS, many extra mobile services become available on amobile phone; these include sending and receiving full-color pictures, sending and receivinge-mail, or even Internet facilities. The study observed a remarkably high percentage of explainedvariance in GPRS usage in comparison to previous studies on mobile communication technologyuse (e.g., Dimmick, Kline, & Stafford, 2000; Leung & Wei, 2000). However, none of the incen-tive categories reconstructed as outcome expectations were significant predictors of GPRS usage.The most significant predictor from the model of media attendance was self-efficacy.

According to Hofstetter et al. (2001), self-efficacy involves a combination of the expectedoutcomes of a task and the belief that one can perform the task adequately. According toPeters et al. (2006), this may partly explain why none of the incentive categories were sig-nificant predictors of GPRS usage. GPRS users apparently have low outcome expectanciesof GPRS despite their high levels of self-efficacy regarding their knowledge of GPRS usage.The use of GPRS was almost insignificant. According to the open-ended question in thestudy by Peters et al., respondents did not seem to have a need for GPRS. GPRS could notcompete with preexisting media like the Internet or e-mail available via personal computers.Apparently, the technology and features of GPRS are not a sufficient driver for GPRS;GPRS do not add value to people’s mobile communication needs.

According to LaRose and Eastin (2004), the active selection of media that best meetpersonal needs is not the sole mechanism that explains media attendance. Self-efficacy




beliefs about one’s ability to utilize alternative media channels also contribute to mediaselection. Active selection dominates when new media alternatives appear or personal rou-tines are disrupted. Once habits are established, however, users no longer think throughwhether one alternative or another is a better way of obtaining a particular outcome(LaRose & Eastin, 2004). According to Peters et al. (2006), this may also explain theapparent contradiction that the odds for not using GPRS increased when people had moremobile phone experience. Because of the insignificance of GPRS as an alternative mediachannel, there is no need for GPRS users to adjust their normal mobile communicationbehavior based on the experience that their existing mobile phone use perfectly fulfillstheir mobile communication needs.

The findings of both the validation of the original model of media attendance and theexamination of the model of media attendance in another context of use (Peters et al., 2006)partly support the findings of the original study by LaRose and Eastin (2004). Although themodel is to some extent applicable to contexts of media use other than Internet (Peters et al.,2006) and offers some promising steps forward in measuring media usage, the relative impor-tance of expected outcome in explaining media usage was not fully supported. Literaturedescribing people’s motivations for using new media technology indicates that people are ini-tially influenced more strongly by perceptions about expected use. Over time, however, theirinitial expectations become latent because of the quick habituation of new media technology(Peters & Ben Allouch, 2005). When people are surveyed for their expectations on existingusage, therefore, they may no longer be aware of the relative importance of expected out-comes or no longer have expectations because the outcomes are already known from personaluse (i.e., habit strength). According to LaRose and Eastin, outcome expectations reflect cur-rent beliefs about the outcomes of prospective future behavior; habit strength represents pat-terns of behavior established by past thinking about outcome expectations that is no longerrepeated in the present. One might therefore expect that habit strength influences ongoingbehavior, independent of current active thinking about expected outcomes. After initial stronghabitualization, this means that further increases of habitualization would lead to a decreasedconsciousness of expected outcomes. This idea is confirmed by the findings of the study byPeters and Ben Allouch (2005). Within the model of media attendance, this may indicate thatmobile communication technology use is more likely to be explained by habit strength.Conversely, the intention to adopt new mobile communication technology is more likely tobe predicted by outcome expectations. Two hypotheses are proposed:

Hypothesis 1: Habit strength is a stronger predictor in explaining mobile phone use than out-come expectations.

Hypothesis 2: Outcome expectations are a stronger predictor in predicting the intention toadopt mobile video telephony than habit strength.

An Examination of the Model of media Attendance to Explain Mobile Phone Use and to

Predict the Intention to Adopt Mobile Video Telephony

According to Popper (1989), a single test of a proposed model is not sufficient to statethe degree of corroboration; successful tests in other contexts are necessary to raise the




degree of corroboration. Otherwise, a situation that resembles a Sisyphean strategy (Opp,2002), in which the number of isolated models that belong together will increase but allownothing to be said about their degree of corroboration, will occur. A single test of a newlyintroduced model is more likely to be successful because of the proximity of the postulatedhypothesis of the proposed model to the empirical base. According to Peters et al. (2006),therefore, more stringent follow-up tests are needed.

According to a survey commissioned by the European Commission (2006), theNetherlands (91%), Finland (93%), and Sweden (93%) are the countries in Europe with thehighest penetration rate of mobile phones. The average rate of mobile phone penetration inEurope is 80%. The high penetration rate of mobile phones in the Netherlands allows us toexamine the roles of outcome expectations and habit strength in existing mobile communi-cation technology behavior that is almost fully habitualized in everyday life. Recent devel-opments in the mobile communication industry make it possible to add many advancedattributes to mobile communication technology devices. These include video telephony,which allows people to not only talk to but also see each other using the phone’s screen.The use of this newly added feature of mobile video communication technology requires asmall camera built into the face of the phone and wireless broadband access. Equipped withmobile phones using the same technology, it is possible to look at and talk to another per-son. Mobile phone operators in the Netherlands started to promote this new service in early2006. The intention of mobile phone users to begin using this new mobile video telephonytechnology to communicate with other people is a perfect opportunity to examine outcomeexpectations and habit strength within the model of media attendance.

Method

Sample and Procedures

Subscribers of a national panel (N = 867) representing the Dutch population adminis-trated by a profit research and consultancy company were invited via email to voluntaryparticipate in the online survey from May 15, 2006, to May 26, 2006. The 644 mobilephone users (74.28% response rate) who responded to the invitation and completely filledin the questionnaire were divided into two equal subsamples using a stratified random sam-pling method that employed demographics, mobile phone use, and mobile phone experi-ence as strata. Pearson’s chi-square and independent sample t tests were used to test fordifferences between the two subsamples. No significant differences between the two groupsof respondents were found with regard to demographics, mobile phone experience, ormobile phone use (see Table 1). As the two groups of respondents do not differ with regardto the strata, differences observed between the two models may therefore not be attributedto differences between the two groups of respondents. Respondents of Group 1 (n = 334)were surveyed for mobile phone use, and respondents of Group 2 (n = 310) were surveyedfor the intention to adopt mobile video telephony. At the beginning of the mobile videophone survey, a detailed picture of a mobile video phone device with a description of itsfunctions was used to introduce the technology of mobile video telephony.




Measures

The original items by LaRose and Eastin (2004) were rephrased in the context of mobilecommunication technology use. Substituted items were collected from prior mobile com-munication technology studies (i.e., Peters & Ben Allouch, 2005; Peters et al., 2006) andclassified in accordance with the conceptual definitions found in Bandura (1986).

Undergraduate students (N = 62) from both the departments of communication studiesand psychology at the University of Twente in the Netherlands participated in a pretest ofthe model of media attendance for research experience points. The rephrased items werepretested on legibility and internal consistency (Nunally, 1978). Furthermore, those itemswith highly correlated error variances or poor loading onto unique factors were removed.This procedure resulted in a reduction of the number of observed indicators of the latentconstructs. As a result of the pretest, additional items to explain mobile phone use weredeveloped to permit stronger operationalization of the novel (α = .50) and status (α = .51)expected outcome measures. The internal consistencies of both the other measures(Nunally, 1978) for explaining mobile phone use and the measures for predicting the inten-tion to adopt mobile video telephony were above aspiration level (α > .70).

Mobile phone behavior and experience. To measure mobile phone use, respondents wereasked to estimate the number of times they used a mobile phone to make a phone call on anaverage weekday. Respondents were also asked to estimate the number of times they used amobile phone to send an SMS message on an average weekday. Mobile phone experience wasmeasured based on the number of years the respondents had used a mobile phone.

Table 1Summary of Demographics, Mobile Phone Experience, and Mobile Phone Use

Group 2 (n = 310)Group 1 (n = 334) Mobile Video

Mobile Phone Use (%) Phone Adoption (%)

Gendera: Male 43 44Female 57 56

Ageb < 20 4 420-40 45 3840-60 40 4760> 11 11

Educationc: High school or less 31 33Vocational education 27 26Bachelor’s degree 31 31Master’s degree 11 10

Mobile phone experience (years)d M = 6.90; SD = 2.84 M = 6.60; SD = 2.89Mobile phone use (times a day)e M = 3.29; SD = 5.64 M = 2.88; SD = 3.97

Note:a. χ2 (1, N = 644) = .01; p > .05.b. χ2 (3, N = 644) = 6.41; p > .05c. χ2 (3, N = 644) = .59; p > .05.d. t (644) = 1.36; p > .05.e. t (644) = 1.07; p > .05.




Expected outcomes. In the context of mobile phone use, expected outcomes (i.e., “usinga mobile phone, how likely are you to ___”) were measured on a Likert-type scale thatranged from 1 (very unlikely) to 7 (very likely). The expected outcomes measures includemonetary incentives, social incentives, novel incentives, status incentives, activity incen-tives, and self-reactive incentives. Although the operationalization of monetary incentivesare in terms of benefit and profit (e.g., saving time, doing a better job) rather than money,the labels for the incentives used in this study are the same as those originally defined byBandura (1986) for the sake of distinctness.

Self-efficacy and habit strength. Self-efficacy (e.g., “I can handle my mobile phone with-out help from others”) and habit strength (e.g., “the use of a mobile phone is part of mydaily routine”) were measured on a Likert-type scale that ranged from 1 (fully disagree) to7 (fully agree).

Mobile video phone measures. In the context of the intention to adopt mobile video tele-phony, expected outcomes, self-efficacy, and mobile phone experience were measured in asimilar manner as in the context of mobile phone use. As mobile video telephony is a newtechnology, it is not likely that this new technology has already been habitualized. Therefore,habit strength is operationalized in terms of forethought. For example, the habit strengthitem “the use of a mobile phone is part of my daily routine” is modified into the prospectivehabit strength item “the use of a mobile video phone would be part of my daily routine.”Deficient self-regulation (e.g., “I have tried unsuccessfully to cut down the amount of timeI spend using my mobile phone”) was not included in the instrument to measure intention toadopt mobile video telephony; in contrast to prospective habit strength, a valid judgmentabout deficient self-regulation would imply that respondents should have had experiencewith mobile video telephony. To equally compare both samples, deficient self-regulation wasnot included in the instrument to measure mobile phone use.

Behavioral intention to use a mobile video phone. Behavioral intention (e.g., “I intendto use a mobile video phone within the next 6 months”) was measured with three items ona Likert-type scale that ranged from 1 (fully disagree) to 7 (fully agree).

Data Analysis

Structural equation analysis using Amos 6.0 (Arbuckle, 2005) with maximum likelihoodestimation was used to test the hypothesized model of media attendance both in explainingmobile phone use and to predict the intention to adopt mobile video telephony. In this study,as suggested by Holbert and Stephenson (2002), the following model fit indices will beused: The chi-square estimate with degrees of freedom given that it is still the most com-monly used means by which to make comparisons across models (Hoyle & Panter, 1995).The ratio between chi-square and degrees of freedom should not exceed five for modelswith a good fit (Bentler, 1989). Additionally, the standardized root mean squared residual(SRMR) as a second absolute fit statistic (Hu & Bentler, 1999) in combination with theTucker–Lewis index (TLI) as incremental index and the root mean squared error of approxi-mation (RMSEA; Browne & Cudeck, 1993) are reported. Hu and Bentler (1999) recommend




using a cutoff value close to .95 for TLI in combination with a cutoff value close to .09 forSRMR to evaluate model fit and the RMSEA close to .06 or less.

Fit indexes are relative to progress in the field (Garson, 2006). Although there are rulesof thumb for acceptance of model fit (e.g., that TLI should be at least .95), Bollen (1989)observed that these cut-offs are arbitrary. A more salient criterion may be simply to com-pare the fit of one’s model to the fit of other prior models of the same phenomenon. Forexample, a TLI of .85 may represent progress in a field where the best prior model had afit of .70. In this study the primary goal was to compare two models. Therefore the cutoffvalues of the fit indices will be used more as reference to compare the two models than asabsolute measure of model fit.

The Fornell and Larcker (1981) discriminant validity criterion was used to test discrim-inant validity. The Fornell and Larcker criterion is satisfied when a construct is moreclosely related to its own indicators than to other constructs.

Results

Explaining Mobile Phone Use

Prior to the analyses, data were checked for normality. Because of skewness to the upperend of the distribution of the measures mobile phone usage and SMS usage, a square-roottransformation was performed to correct skew. Because of skewness to the lower end of thedistribution of the measure self-efficacy, an inverse (reciprocal) transformation was per-formed to correct skew (Garson, 2006).

Measurement model. Using a first-order confirmatory factor analysis, the measurementmodel estimated the extent to which the observed items loaded onto their respective latentvariables. Because experience was measured with a single observed item, it was notincluded in the measurement model. All latent constructs but no observed error varianceswere allowed to covary with one another. The initial measurement model generated poorfit, χ2 (876) = 2546.21; χ2 /df = 2.91; SRMR = .113; TLI = .814; RMSEA = .076 (90% con-fidence interval, CI = .72, .79).

Items with highly correlated error variances identified by post hoc modification indices anditems that loaded poorly onto its unique factor were removed. This procedure resulted in thereduction of the number of observed indicators of the latent constructs to better fit the mea-surement model. The internal consistency of the measures (Nunally, 1978) to explain mobilephone use was above aspiration level (α > .70), except for novel outcomes (α = .53) and sta-tus outcomes (α = .61). Both novel outcomes and status outcomes were excluded from fur-ther analysis. The modified measurement model generated a good fit, χ2 (149) = 334.99;χ2 /df = 2.25; SRMR = .053; TLI = .945; RMSEA = .061 (CI = .052, .070). The correlationmatrix of the observed variables, mobile phone usage, and SMS usage is shown in Table 2.

Structural model. The results obtained from testing the validity of a causal structure ofthe hypothesized model showed that the initial model did not fit the data, χ2 (178) = 508.75;χ2 /df = 2.89; SRMR = .094; TLI = .905; RMSEA = .075 (CI = .068, .083). Post hoc mod-ification indices suggested an improved fit by correlating the error terms of habit strengthwith monetary outcomes (r = .63) and self-reactive outcomes with activity outcomes (r = .71).



85

Tabl

e 2

Cor

rela

tion

Mat

rix

of t

he O

bser

ved

Var

iabl

es,M

obile

Pho

ne U

sage

,and

SM

S U

sage

12

34

56

78

910

1112

1314

1516

1718

1920

1. S

elf-

reac

tive

1.–

.76*

.73*

.39*

.42*

.52*

.09

.15*

.12*

.61*

.60*

.71*

.29*

.21*

.35*

.04

.05

.06

.08

.11

2. S

elf-

reac

tive

2–

.74*

.38*

.41*

.48*

.08*

.13*

.08

.59*

.60*

.70*

.25*

.18*

.36*

.05

.05

.05

.06

.08

3. S

elf-

reac

tive

3–

.35*

.39*

.48*

.13*

.17*

.16*

.55*

.56*

.67*

.29*

.25*

.40*

.09

.11

.08

.16*

.23*

4. S

ocia

l 1–

.73*

.62*

.19*

.34*

.19*

.54*

.52*

.53*

.36*

.16*

.34*

.10

.08

.09

.03

.13*

5. S

ocia

l 2–

.63*

.18*

.36*

.22*

.54*

.60*

.61*

.40*

.20*

.37*

.09

.07

.10

.09

.14*

6. S

ocia

l 3–

.17*

.26*

.15*

.58*

.55*

.60*

.29*

.12*

.32*

.03

.01

.01

.04

.16*

7. M

onet

ary

1–

.49*

.60*

.24*

.21*

.18*

.45*

.45*

.49*

.06

.01

.01

.17*

.08

8. M

onet

ary

2–

.70*

.30*

.28*

.26*

.45*

.33*

.47*

.10

.13*

.10

.15*

.07

9. M

onet

ary

3–

.19*

.21*

.17*

.46*

.42*

.49*

.03

.06

.06

.15*

.06

10. A

ctiv

ity 1

–.7

1*.7

4*.3

5*.2

8*.3

9*.0

1.0

5.0

4.0

6.0

9

11. A

ctiv

ity 2

–.7

8*.3

7*.3

0*.4

6*.1

1*.1

0.0

9.0

8.1

7*

12. A

ctiv

ity 3

–.4

0*.3

1*.4

2*.1

0.0

8.0

7.0

5.2

1*

13. H

abit

stre

ngth

1–

.61*

.67*

.12*

.13*

.09

.18*

.10

14. H

abit

stre

ngth

2–

.63*

.17*

.26*

.20*

.17*

.09

15. H

abit

stre

ngth

3–

.18*

.16*

.15*

.34*

.15*

16. S

elf-

effi

cacy

1–

.64*

.65*

.03

.05

17. S

elf-

effi

cacy

2–

.84*

.05

.06

18. S

elf-

effi

cacy

3–

.04

.06

19. M

obile

pho

ne u

se−

.14*

20. S

MS

use

–

Not

e:*p

< .0

5.

α




The respecified model generated a good fit, χ2 (176) = 405.99; χ2 /df = 2.31; SRMR = .061;TLI = .937; RMSEA = .063 (CI = .055, .071). Table 3 summarizes the original (uncor-rected) mean and standard deviation, Cronbach’s α, the factor loading (β), and the squaredmultiple correlation (R2) of the observed indicators to explain mobile phone use. The pathmodel with standardized path coefficients is featured in Figure 2.

Table 3Descriptive Statistics, Factor Loadings, Squared Multiple Correlations, and

Cronbach’s α for the Observed Indicators to Explain Mobile Phone Use

M SD β R2

Usage .67Mobile phone (typical weekday) 3.29 5.64 .52 .27SMS (typical weekday) 2.26 11.13 .52 .28

Social outcomes (α = .85) .70To keep my family and friends up-to-date 3.96 2.08 .82 .67To keep up contact with my family and friends 3.95 2.07 .86 .74To strengthen my relations with family and friends 3.34 1.99 .76 .58

Activity outcomes (α = .90) .89Because I like to be called 2.97 2.00 .81 .66To have a nice conversation 2.93 2.01 .85 .73Because it’s a pleasant activity 2.46 1.75 .92 .85

Monetary outcomes (α = .81) .11To save time because I am accessible everywhere 4.74 2.11 .69 .48To be more quickly accessible 5.31 1.89 .78 .61To be always accessible 5.74 1.72 .88 .77

Self-reactive outcomes (α = .89) .53To relax 1.87 1.37 .88 .77To pass the time 1.82 1.38 .87 .76When I don’t have anything to do 2.20 1.64 .84 .70

Novel outcomes (excluded from analysis)To get immediate knowledge of the latest news 2.05 1.53To take pictures 2.40 1.84To send text messages 4.14 2.23

Status outcomes (excluded from analysis)Fits my lifestyle 2.73 1.88Because it is a modern way to communicate 3.34 2.02 Get up to date with new technology 2.34 1.63

Self-efficacy (α = .92) .02I can handle my mobile phone without help from others 6.57 1.02 .78 .63It is no problem for me to operate my mobile phone 6.56 .95 .97 .94I have the knowledge and skills to operate my mobile phone 6.56 .94 .93 .86

Habit strength (α = .84) .35The use of a mobile phone is part of my daily routine 4.36 2.15 .88 .78I always carry my mobile phone with me 5.66 1.74 .71 .51I would miss a mobile phone if it were not available 5.08 1.94 .77 .60

Note: The R2 of a latent dependent predictor is the percentage of the variance in the latent dependent variableaccounted for by the latent independent variable. The R2 of an observed indicator is the estimated percentagevariance explained in that variable. In other words, the error variance of a variable is approximately 1 minus thepercentage of the variance of the variable itself. M, mean; SD, standard deviation; β, factor loading; R2, squaredmultiple correlation.




Hypothesis 1 predicted habit strength as a stronger predictor in explaining mobile phoneuse than outcome expectations. The standardized path coefficients in Figure 2 show a sig-nificant direct effect of habit strength on mobile phone usage and a nonsignificant directeffect of expected outcomes. Also, Figure 2 shows significant direct effects of outcomeexpectations and self-efficacy on habit strength and a significant direct effect of experienceon self-efficacy. The indirect effect of expected outcomes on mobile phone usage (β = .38)is mediated by the direct effect of outcome expectations on habit strength. The direct effectof habit strength on mobile phone usage surpasses the indirect effect of expected outcomeson mobile phone usage. The first hypothesis is supported.

The indirect effect of self-efficacy on mobile phone usage (β = .09) is mediated by thedirect effect of self-efficacy on habit strength. The indirect effect of experience on mobilephone usage (β = .01) is mediated via the consecutive effect of self-efficacy and habitstrength on mobile phone use. Squared multiple correlations provide information about thevariance accounted for by the complete set of variables and showed that mobile phone usewas accounted for 67% (see Table 3).

Predicting the Intention to Adopt Mobile Video Telephony

Prior to the analyses, data were checked for normality. Because of skewness to the lowerend of the distribution of the mobile video phone measures (expect for self-efficacy), an

Figure 2Standardized Path Coefficients of the Model ofMedia Attendance to Explain Mobile Phone Use

ExpectedOutcomes

Self-efficacy Usage

Experience Habit Strength

.68***

.56***

.13**

.09

.13***

.05

.11 .19

Note: The observed indicators of the latent construct are not shown (see Table 3). Dotted lines are nonsignificantpaths (nonsignificant factor loadings in italic).**p < .01. ***p < .001.




inverse (reciprocal) transformation was performed to correct skew (Garson, 2006). Toequally compare both samples, novel outcomes and status outcomes were excluded fromfurther analysis in both samples.

Measurement model. Using a first-order confirmatory factor analysis, the measurementmodel estimated the extent to which the observed items loaded onto their respective latentvariables. Because experience was measured with a single observed item, it was notincluded in the measurement model. All latent constructs but no observed error varianceswere allowed to covary with one another. The initial measurement model generated poorfit, χ2 (866) = 3509.93; χ2 /df = 4.05; SRMR = .069; TLI = .813; RMSEA = .099 (CI =.096, .103). Items with highly correlated error variances identified by post hoc modifica-tion indices and items that loaded poorly onto its unique factor were removed. This pro-cedure resulted in a reduction in the number of observed indicators of the latent constructsto better fit the measurement model. The internal consistency of the measures (Nunally,1978) to predict the intention to adopt mobile video telephony was above aspiration level(α > .70). The correlation matrix of the observed variables and the intention to adoptmobile video telephony is shown in Table 4. The modified measurement model generateda good fit, χ2 (168) = 397.84; χ2 /df = 2.37; SRMR = .032; TLI = .961; RMSEA = .067(CI = .058, .075).

Structural model. The results obtained from testing the validity of a causal structure ofthe hypothesized model showed a good fit, χ2 (198) = 443.04; χ2 /df = 2.24; SRMR = .035;TLI = .962; RMSEA = .063 (CI = .055, .071).

Table 5 summarizes the original (uncorrected) mean and standard deviation, Cronbach’sα, the factor loading (β), and the squared multiple correlation (R2) of the observed indica-tors to predict the intention to adopt mobile video telephony. The path model with stan-dardized path coefficients is featured in Figure 3.

Hypothesis 2 predicted outcome expectations as a stronger predictor in predicting theintention to adopt mobile video telephony than habit strength. The standardized path coef-ficients in Figure 3 show significant direct effects of expected outcomes and prospectivehabit strength on the intention to adopt mobile video telephony. Figure 3 also shows sig-nificant direct effects of expected outcomes on prospective habit strength, mobile phoneexperience on self-efficacy, and self-efficacy on expected outcomes. Although the directeffect of prospective habit strength on the intention to adopt mobile video telephony isstronger than the effect of expected outcomes, the total effect of expected outcomes (β =.56) on the intention to adopt mobile video telephony surpasses the direct effect of prospec-tive habit strength. The second hypothesis is supported.

Further total effects on the intention to adopt mobile video telephony were self-efficacy(β = .10) and mobile phone experience (β = .02). Furthermore, there were significant totaleffects of mobile phone experience (β = .03) and self-efficacy (β = .12) on prospective habitstrength, and a significant total effect of mobile phone experience (β = .04) on expectedoutcomes. Squared multiple correlations provide information about the variance accountedfor by the complete set of variables and showed that the intention to adopt mobile videotelephony was accounted for 40% (see Table 5).



89

Tabl

e 4

Cor

rela

tion

Mat

rix

of t

he O

bser

ved

Var

iabl

es a

nd M

obile

Vid

eo P

hone

Int

enti

on

12

34

56

78

910

1112

1314

1516

1718

1920

21

1. S

elf-

reac

tive

1–

.79*

.74*

.50*

.47*

.49*

.47*

.52*

.49*

.53*

.64*

.65*

.31*

.31*

.44*

.18*

.14*

.15*

.30*

.30*

.30*

2. S

elf-

reac

tive

2–

.85*

.62*

.52*

.57*

.53*

.55*

.52*

.52*

.63*

.60*

.38*

.43*

.52*

.13*

.11

.13*

.34*

.34*

.32*

3. S

elf-

reac

tive

3–

.65*

.55*

.56*

.55*

.55*

.51*

.52*

.65*

.64*

.42*

.47*

.51*

.13*

.10

.14*

.31*

.32*

.30*

4. S

ocia

l 1–

.76*

.66*

.55*

.57*

.56*

.63*

.62*

.64*

.45*

.44*

.53*

.15*

.15*

.17*

.32*

.34*

.34*

5. S

ocia

l 2–

.69*

.52*

.61*

.55*

.64*

.56*

.63*

.49*

.43*

.53*

.16*

.15*

.17*

.33*

.33*

.30*

6. S

ocia

l 3–

.52*

.55*

.57*

.66*

.56*

.61*

.40*

.40*

.48*

.16*

.16*

.17*

.31*

.29*

.27*

7. M

onet

ary

1–

.78*

.70*

.66*

.68*

.65*

.48*

.41*

.49*

.17*

.17*

.20*

.33*

.32*

.29*

8. M

onet

ary

2–

.72*

.69*

.65*

.68*

.53*

.49*

.52*

.19*

.16*

.20*

.42*

.42*

.39*

9. M

onet

ary

3–

.68*

.57*

.71*

.47*

.45*

.47*

.13*

.12*

.15*

.39*

.39*

.37*

10. A

ctiv

ity 1

–.7

3*.7

4*.4

3*.4

2*.4

9*.2

3*.2

0*.2

2*.4

4*.4

2*.4

1*

11. A

ctiv

ity 2

–.8

2*.4

0*.3

9*.5

1*.1

6*.1

2*.1

4*.4

1*.4

0*.4

0*

12. A

ctiv

ity 3

–.4

9*.4

5*.5

4*.1

7*.1

6*.1

7*.4

4*.4

3*.4

2*

13. H

abit

stre

ngth

1–

.72*

.73*

.10

.04

.07

.48*

.48*

.46*

14. H

abit

stre

ngth

2–

.83*

.04

.01

.05

.48*

.49*

.46*

15. H

abit

stre

ngth

3–

.08

.07

.09

.45*

.48*

.45*

16. S

elf-

effi

cacy

1–

.79*

.76*

.12*

.11

.12*

17. S

elf-

effi

cacy

2–

.92*

.08

.08

.09

18. S

elf-

effi

cacy

3–

.12*

.11

.11

19. I

nten

tion

1–

.94*

.91*

20. I

nten

tion

2–

.96*

21. I

nten

tion

3–

Not

e:*p

< .0

5.




Table 5Descriptive Statistics, Factor Loadings, Squared Multiple Correlations,

and Cronbach’s α for the Observed Indicators to Predict theIntention to Adopt Mobile Video Telephony

M SD β R2

Intention (α = .98) .40I plan to use mobile video phone within the next 6 months 1.47 1.08 .98 .96I intend to use mobile video phone within the next 6 months 1.43 1.02 .98 .96I will use mobile video phone within the next 6 months 1.39 .96 .95 .91

Social outcomes (α = .89) .86To keep my family and friends up-to-date 2.09 1.57 .83 .70To keep up visual contact with family and friends 2.01 1.63 .87 .76To strengthen my relations with family and friends 1.75 1.41 .87 .75

Activity outcomes (α = .91) .96Because of the possibility to call with video 2.49 1.89 .85 .73To have a nice conversation 2.22 1.78 .90 .81Because it’s a pleasant activity 2.07 1.62 .87 .77

Monetary outcomes (α = .91) .91To communicate in a more understandable manner 2.33 1.78 .86 .74To not just have to communicate with voice only 2.10 1.68 .91 .82To better communicate 2.10 1.63 .86 .75

Self-reactive outcomes (α = .95) .72To relax 1.64 1.19 .94 .89To pass the time 1.58 1.13 .95 .90When I don’t have anything to do 1.72 1.35 .92 .84

Novel outcomes (excluded from analysis)To capture video clips 2.31 1.82To take pictures 2.90 2.06To send video clips 2.16 1.69

Status outcomes (excluded from analysis)Fits my lifestyle 1.66 1.29Because it is a modern way to communicate 2.20 1.78Get up-to-date with new technology 2.33 1.79

Self-efficacy (α = .95) .06I would handle mobile video phone without the help from others 4.70 2.25 .81 .65It would be no problem for me to operate mobile video phone 5.17 2.02 .98 .96I have the knowledge and skills to operate mobile video phone 5.21 2.04 .94 .89

Prospective habit strength (α = .91) .51The use of mobile video phone would be part of my daily routine 1.58 .96 .80 .64I would always make phone calls with mobile video phone 1.48 1.02 .92 .84I would miss mobile video phone if it were not available 1.47 1.08 .92 .85

Note: The R2 of a latent dependent predictor is the percentage of the variance in the latent dependent variableaccounted for by the latent independent variable. The R2 of an observed indicator is the estimated percentagevariance explained in that variable. In other words, the error variance of a variable is approximately 1 minus thepercentage of the variance of the variable itself. M, mean; SD, standard deviation; β, factor loading; R2, squaredmultiple correlation.




Discussion

This study examined the triadic relationship between expected outcomes, habit strength,and mobile communication technology use and adoption within the model of media atten-dance (LaRose & Eastin, 2004). Within the model of media attendance, the findings of thisstudy support the assumptions that (a) mobile phone use is more likely to be explained byhabit strength and (b) the intention to adopt mobile video telephony is more likely to bepredicted by outcome expectations.

The empirical results also show a strong effect of expected outcomes on prospectivehabit strength with regard to predicting the adoption of mobile video telephony. This resultsupports the notion that habit strength is causally determined by outcome expectations,which precede habit strength in time when media use is not fully habitualized (LaRose &Eastin, 2004).

On the basis of the mediating effect of habit strength on the influence of expected out-comes on mobile phone usage, one might conclude that the relationship between expectedoutcomes and habit strength depends on the stage of individual habitualization. This conclu-sion makes sense. In the Netherlands, the habitualization process for the mobile phone isalmost complete. Once mobile communication technology behavior is more strongly deter-mined by habit strength, the effect of outcome expectations may no longer greatly influencepeople’s mobile communication technology behavior. This occurs either because people nolonger consciously evaluate expected outcomes as the adoption decision is already made orbecause people no longer have expectations as the outcomes are already known.

Figure 3Standardized Path Coefficients of the Model of Media Attendance

to Predict the Intention to Adopt Mobile Video Telephony

ExpectedOutcomes

Self-efficacy Intentionto Adopt

Experience ProspectiveHabit Strength

.43***

.25**

.73***

.17**

.25***

.01

.01

.08

Note: The observed indicators of the latent construct are not shown (see Table 5). Dotted lines are nonsignifi-cant paths (nonsignificant factor loadings in italic).**p < .01. ***p < .001.




The results of testing the model of media attendance support the findings described byPeters and Ben Allouch (2005). These findings noted that people are initially influencedmore strongly by perceptions about the expected use; over time, however, initial expecta-tions become latent because of the rapid habituation of new media technology.

Initial expectations are apparently reflections of a relatively short moment in time, and theyare subjected to changes over time. Once outcome expectations become latent, it becomes dif-ficult to explain media behavior solely with expected outcomes. Aarts, Verplanken, and VanKnippenberg (1998) posited that when behavior is performed repeatedly and becomeshabitual, this habitual behavior is guided by automated processes rather than elaborate deci-sion processes. Therefore, the relative importance of expected outcomes as described byLaRose and Eastin (2004, p. 371) to predict media consumption is only supported whenhabit strength is not already very pronounced.

This conclusion is also reflected in the path from mobile phone experience via self-efficacyto habit strength with regard to explaining mobile phone use. For predicting the intention toadopt mobile video telephony, however, the path from mobile phone experience via self-effi-cacy goes to expected outcomes. To adopt mobile video telephony, users still need to learnhow to successfully obtain the expected outcomes; in this context, therefore, self-efficacyincreases with experience. Once users achieve satisfactory means for attaining those out-comes, LaRose and Eastin (2004) suggest that they should become increasingly inattentive tospecific behaviors that support them. In cases where mobile communication technology useis almost habitualized, self-efficacy no longer influences expected outcomes. This occursbecause habitualized users no longer have to learn how to obtain successful outcomes.

The findings of this study may also offer an alternative explanation of the innovationcluster concept used within the diffusion of innovations paradigm. According to the dif-fusion of innovations paradigm, the decision to adopt an innovation is predicted primarilyby the perceived attributes of an innovation and only to a lesser extent by the personalityof the potential innovator (Rogers, 2003). Recent empirical evidence, however, suggeststhat the adoption of technological innovations is better predicted by the ownership ofrelated innovations (Vishwanath & Chen, 2006). This suggests that innovations are viewednot singularly, but rather as interrelated bundles of new ideas (Rogers, 2003). Innovationsthat form a cluster tend to be compatible with one another and possess similar attributes(Rogers, 2003); in other words, they presumably satisfy the same underlying need(LaRose & Atkin, 1992). According to Vishwanath and Chen (2006), this suggests anassociational process in which individuals consciously or subconsciously relate technolo-gies to each other. The adoption of any one technology from a cluster spurs the adoptionof related technologies from within the same cluster. From a social cognitive view, thisadoption behavior can be explained by the mechanisms in the model of media attendance.Adoption of new media technology is determined by the expected outcomes that followfrom media consumption. Outcome expectations reflect current beliefs about the out-comes of prospective future behavior, but they are predicated on comparisons betweenincentives expected and incentives attained in the past. This study indicates that existingmedia technology use is more likely to be explained by habit strength, which suggests thatadopting related technologies from within the same cluster is also more likely to beexplained by habit strength.




Limitations of the Study

Clearly, this study has limitations. First, the measurement of the novel and status out-comes as latent indicators to explain mobile phone use was limited in terms of reliability.Although the internal consistencies of both the novel and status outcomes improved rela-tive to those in the pretest, the measures were still below the aspiration level and wereexcluded from further analysis. Extended item batteries should be developed more specif-ically to match the media technology in question to permit a stronger operationalization ofthe latent constructs. Second, to improve the fit of the model to explain existing mediabehavior, post hoc modification indices suggested the correlation of the error terms of habitstrength with monetary outcomes and the self-reactive outcomes with activity outcomes.The relationship between self-reactive outcomes and activity outcomes has already beendescribed by LaRose and Eastin (2004) within the context of Internet usage, where self-reactive outcomes of Internet usage were positively related to Internet activity outcomes.

Conclusions

Understanding media technology behavior is important for allowing the media technol-ogy industry to react accurately to the changing behavior of its customers. Understandingcustomers’ needs and desires is vital for the generation of products and services that con-sumers will actually use. For both academia and the media technology industry, under-standing the behavior of the media technology consumer is important because it offersinsight into the processes of technological innovation and diffusion as well as the use ofmedia technology. The findings of this study suggest that people’s behavior for adoptingand using mobile communication technology is determined primarily by a particularmobile communication technology’s stage of development and diffusion. Depending onwhether the mobile communication technology is an already well-accepted technology ora new innovative technology, different factors (e.g., habit strength or outcome expectations)will influence the public in terms of both mobile communication technology use and theintention to adopt new mobile communication technology. The mobile communicationindustry should account for these factors, for example, by adjusting promotions towardtheir customers when launching new products and updates. The results of this study showthat the model of media attendance is capable of visualizing the dynamics between the dif-ferent factors that are important for both explaining mobile phone use and predicting theintention to adopt mobile video telephony. The model of media attendance offers somepromising steps forward in measuring media technology usage and contributes to existingresearch on media use behavior. According to LaRose and Eastin (2004), it was possible topredict media consumption to an unprecedented degree via the model of media attendance.The results of the present study suggest that this notion has to be refined, in particular whenoutcome expectations become latent. The model of media attendance is grounded in socialcognitive theory and therefore offers a fruitful alternative for measuring media selectivityand usage. This is true both from the perspective of explaining and predicting media usageand from the perspective of validating and extending theory. More stringent follow-up testsof the model of media attendance, which extend it to other media technologies and differ-ent contexts of media use, are needed to further determine the degree of corroboration.




References

Aarts, H., Verplanken, B., & Van Knippenberg, A. (1998). Predicting behavior from actions in the past:Repeated decision making or a matter of habit? Journal of Applied Social Psychology, 28(15), 1355-1374.

Arbuckle, J. L. (2005). Amos 6.0 user’s guide. Chicago: SPSS.Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ:

Prentice Hall.Bandura, A. (1991). Social cognitive theory of self-regulation. Organizational Behavior and Human Decision

Processes, 50, 248-287.Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman.Bandura, A. (2002). Social cognitive theory of mass communication. In J. Bryant and D. Zillman (Eds.), Media

effects: Advances in theory and research (2nd ed., pp. 121-155). Mahwah, NJ: Erlbaum.Bentler, M. (1989). EQS structural equations program manual. Los Angeles: BMDP Statistical Software.Bollen, K. A. (1989). Structural equations with latent variables. New York: Wiley.Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K. A. Bollen & J. S. Long

(Eds.), Testing structural equation models (pp.136-162). Thousand Oaks, CA: Sage.Dimmick, J., Kline, S., & Stafford, L. (2000). The gratification niches of personal e-mail and the telephone:

Competition, displacement, and complementarity. Communication Research, 27(2), 227-248.Eastin, M. S., & LaRose, R. (2000). Internet self-efficacy and the psychology of the digital divide. Journal of

Computer Mediated Communication, 6(1). Retrieved October 9, 2006, from http://jcmc.indiana.edu/vol6/issue1/eastin.html

European Commission. (2006). Special Eurobarometer: E-communications household survey. Retrieved October 9,2006, from http://ec.europa.eu/information_society/policy/ecomm/doc/info_centre/studies_ext_consult/ecomm_household_study/eb_jul06_main_report_en.pdf

Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables andmeasurement error. Journal of Marketing Research, 28, 39-50.

Garson, G. D. (2006). Statnotes: Topics in multivariate analysis. Retrieved October 9, 2006, fromhttp://www2.chass.ncsu.edu/garson/pa765/structur.htm

Hofstetter, C. R., Zuniga, S., & Dozier, D. M. (2001). Media self-efficacy: Validation of a new concept. MassCommunication and Society, 4(1), 61-78.

Holbert, R. L., & Stephenson, M. T. (2002), Structural equation modeling in the communication sciences, 1995-2000. Human Communication Research, 28(4), 531-551.

Hoyle, R. H., & Panter, A. T. (1995). Writing about structural equation models. In R. H. Hoyle (Ed.), Structuralequation modeling: Comments, issues, and applications (pp. 158-176). Thousand Oaks, CA: Sage.

Hu, L.-T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventionalcriteria versus new alternatives. Structural Equation Modeling, 6(1), 1-55.

LaRose, R., & Atkin, D. (1992). Audiotext and the re-invention of the telephone as a mass medium. JournalismQuarterly, 69, 413-421.

LaRose, R., & Eastin, M. S. (2004). A social cognitive theory of Internet uses and gratifications: Toward a newmodel of media attendance. Journal of Broadcasting and Electronic Media, 48(3), 358- 377.

LaRose, R., Lin, C. A., & Eastin, M. S. (2003). Unregulated Internet usage: Addiction, habit, or deficient self-regulation? Media Psychology, 5, 224-253.

LaRose, R., Mastro, D. A., & Eastin, M. S. (2001). Understanding Internet usage: A social cognitive approachto uses and gratifications. Social Science Computer Review, 19, 395-413.

Leung, L., & Wei, R. (2000). More than just talk on the move: Uses and gratifications of the cellular phone.Journalism and Mass Communication Quarterly, 77(2), 308-320.

Nunally, J. C. (1978). Psychometric theory (2nd ed.). New York: McGraw-Hill.Opp, K. D. (2002). Methodologie der Sozialwissenschaften. Einführung in die Probleme ihrer Theoriebildung

und praktischen Anwendung [The methodology of social sciences. Introduction into the problems of theory-building and their practical application] (5th rev. ed.). Wiesbaden, Germany: Westdeutscher Verlag.

Peters, O., & Ben Allouch, S. (2005). Always connected: A longitudinal field study of mobile communication.Telematics and Informatics, 22(3), 239-256.



Peters, O., Rickes, M., Jöckel, S., Von Criegern, C., & Van Deursen, A. (2006). Explaining and analyzing audi-ences: A social cognitive approach to selectivity and media use. Communications, 31(3), 279-308.

Popper, K. R. (1989). Logik der Forschung [The logic of scientific discovery] (9th rev. ed.). Tubingen,Germany: Mohr.

Rogers, E. M. (2003). Diffusion of innovations (5th ed.). New York: Free Press.Vishwanath, A., & Chen, H. (2006). Technology clusters: Using multidimensional scaling to evaluate and struc-

ture technology clusters. Journal of the American Society for Information Science and Technology, 57(11),1451-1460.

Oscar Peters, PhD, is an assistant professor at the University of Twente, Department of Media, Communication, andOrganization (MCO) at the Faculty of Behavioral Sciences. He is also the managing director of the Institute forBehavioral Research (IBR) at the University of Twente in the Netherlands. He may be contacted [email protected]




Documents

A Social Cognitive Perspective on Mobile Communication Technology Use and Adoption