Journal of Retailing and Consumer Services 19 (2012) 67–77

Contents lists available at SciVerse ScienceDirect

Journal of Retailing and Consumer Services

0969-69

doi:10.1

n Corr

E-m

b.lang@1 Te

journal homepage: www.elsevier.com/locate/jretconser

Modeling the effect of self-efficacy on game usage and purchase behavior

Robert Davis a,n, Bodo Lang b,1

a Faculty of Creative Industries and Business, Unitec Institute of Technology, Department of Management and Marketing, Private Bag 92025, Auckland, New Zealandb Marketing Department, The University of Auckland Business School, Private Bag 92019, Auckland 1142, New Zealand

a r t i c l e i n f o

Available online 16 December 2011

Keywords:

Self-efficacy

Usage

Purchase

Computer games

Confirmatory factors analysis

Structural equation modeling

89/$ - see front matter & 2011 Elsevier Ltd. A

016/j.jretconser.2011.09.002

esponding author. Tel.: þ649 815 4321.

ail addresses: rdavis@unitec.ac.nz (R. Davis),

auckland.ac.nz (B. Lang).

l.: þ64 9 923 7162.

a b s t r a c t

This research models the relationship between self-efficacy, game purchase and usage. Four-hundred

and ninety three consumers responded to a questionnaire. We deployed confirmatory factors analysis

(CFA) and structural equation modeling (SEM) across 4 game types; original model (all games) and

alternative models, Sports/Simulation/Driving, Role Playing Game/Massively Multiplayer Online Role-

Playing Game/Strategy and Action/Adventure/Fighting. The impact of self-efficacy on usage and

purchase was modeled both individually and simultaneously. For individual effects; models had

adequate fit with Sports/Simulation/Driving showing an impact between self-efficacy on game usage

and purchase. Our results showed no simultaneous relationship. We conclude that self-efficacy does

impact usage or purchase but game type affects this relationship. Research implications are discussed.

& 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Recent advances in games on PC, MAC, Console, Mobile, iPhoneor iPad have increased the consumers purchase and use of theseentertainment related products and services (Prugsamatz et al.,2010). According to the Entertainment Software Association inthe U.S.: (1) computer and video game software sales generated$10.5 billion in 2009, (2) sixty-seven percent of American house-holds play computer or video games, (3) the average game playeris 34 years old and has been playing games for 12 years. Overall,sales of game hardware, software and accessories have eclipsedthose of the US box-office, cementing gaming as a dominant forceof technological consumption (Khan, 2002; Guth, 2003). Europehas also become a significant industry and market. The UK is thethird largest market globally with total sales in 2004 of entertain-ment and leisure software of £1.34b (Boyle and Hibberd, 2005).The interactive entertainment industry in the UK is set to grow by7.5% between 2009 and 2012 (UKIE, 2011).

There are many factors that have fueled this change in theconsumers consumption behavior but it is argued in this researchthat the growth in the importance of games in a consumers enter-tainment experience has been largely attributable to increasedtechnology related self-efficacy (Allan, 2010). Usage and purchasehas grown because the consumer perceives that they have thecapability to be interactive with a game and therefore, other stimuliwithin the game (e.g., Advergames).

ll rights reserved.

As a consequence, marketing practioners and researchers havebecome more interested in the potential of this medium for market-ing. A key focus of this interest is related to three questions, that is,self-efficacy and the fit between the consumer and: (1) the game,(2) the marketing stimulus (e.g., advertisement) and (3) the co-creation of experience with the game and stimulus. All thesequestions place emphasis on the consumers belief in their capabilityto not only play the game as well as interact with the marketingstimulus to accomplish specific objectives but also to be an activeplayer in the co-creation of experience (Bandura, 1982).

A review of the existing research shows that much of the workto date has focused on the effect of advertising within a game onthe consumer (Molesworth, 2006). For example, Prugsamatz et al.(2010) apply the theory of planned behavior by gamer type,showing the effects on purchase intentions. Also, Cauberghe andDe Pelsmacker (2010) replicate the effect of in game advertisingon brand recall and attitude. They also take in to account themediating effects of product involvement, although, we acknowl-edge that the games medium is predominately service oriented.This work is consistent with Nicovich (2005) who have measuredthe relationship between consumer involvement on the advertis-ing effect. Like Cauberghe and De Pelsmacker (2010), many othershave examined the advertising communication effect of productor brand placement in computer games on the consumer (e.g.,Schneider and Cornwell, 2005; Mackay et al., 2009; Chaney et al.,2004; Nelson et al., 2004; Winkler and Buckner, 2006; Yang et al.,2006; Mau, Silberer and Constien, 2008). While this research hasreplicated traditional models, they have ignored two importantfactors. First, the mediating effect of the service experience and,second the difference between a product vs. an entertainmentorientation.

R. Davis, B. Lang / Journal of Retailing and Consumer Services 19 (2012) 67–7768

Recently, other researchers in an attempt to extend our under-standing of the consumer response to marketing in the gameenvironment have started to explore avatar-based advertising (Jinand Bolebruch, 2009). In this study, the overall effects of avatar-based interactive advertising on product involvement and attitudewere tested. It was found that consumers ‘‘perceive human-likespokes-avatars as more attractive, and players who interact with ahuman-like spokes-avatar perceive the iPhone advertisement asmore informative than those who interact with a non-humanspokes-avatar (Jin and Bolebruch, 2009, p57).’’

Despite these developments, most of the existing research hasfocused on the consumer- advertisement response. Many withthe exception of Prugsamatz et al. (2010) have not compareddifferent game genres. Little attention has been given to under-standing the fit between consumer, game and marketing stimulusfrom a self-efficacy perspective and the effect of this on use andpurchase. This is concerning because if a consumer does not havethe belief in their capability to be interactive with the game and/or marketing stimulus concurrently, it is less likely that they willvalue the experience. Self-efficacy plays a key mediating role inthe interactivity between consumer, game and marketing stimu-lus. If consumers do not have a high level of self-efficacy then thismay reduce use and purchase. Also, as some researchers haveargued, there may be negative impacts on the gamers self-andother aspects of cognition (Boyle and Hibberd, 2005; Andersonand Bushman, 2002; Dill and Dill, 1998).

While these perspectives are valuable for our understanding, afundamental research question has not been addressed, such asthose concerning the consumers’ self-efficacy and its relationshipto game purchase and game usage (Kaltcheva et al., 2011). Wemodel these relationships across 4 game types, grouped accordingto the conceptualization of Myers (1990), namely: (1) all gamesrepresenting our original model and then the alternative compet-ing models, (2) Sports/Simulation/Driving, which places emphasison hand/eye co-ordination/reflexes in real world environments,(3) Role Playing Game (RPG)/Massively Multiplayer Online Role-Playing Game (MMORPG)/Strategy, which places emphasis oncharacters that gain experience and power through encounters and(4) Action/Adventure/Fighting, which places emphasis on simula-tions of futuristic and historical warfare and/or violent activity.

This approach is consistent with Apperly (2006, p. 20) andothers (Prugsamatz et al., 2010) who argue that ‘‘strategy androle-playing genres have their roots in pre-computer forms of play,whereas the simulation genre can be compared to non-entertain-ment computer simulations. The action genre is implicitly connectedto cinema through its deployment of the terminology of thatmedium to mark key generic distinctions.’’

Usage and purchase are employed as dependent variables andrelate to the frequency of this behavior. Usage and purchase haveoften been used in this capacity in marketing research. Forexample, Shimp and Kavas (1984) relate the theory of reasonedaction to usage. Usage has also been deployed in an experimentalcontext. Folkes et al. (1993) relate product supply to usage. Desaiand Hoyer (2000) examine the composition of memory-sets todifferent usage. Purchase behavior has also played a key role inmarketing research as a dependent variable (Sriram et al., 2010;Hui et al., 2009; Liu, 2007). For example, Bawa and Shoemaker(1987) develop a model of coupon usage across product classes toexplain the purchasing behavior between coupon-prone and non-coupon-prone households. Also, Sismeiro and Bucklin (2004) usebinary probit models of navigational behavior to predict actualpurchase online.

Our work has implications for current research focusing on thefit between consumer, game and marketing stimulus from a self-efficacy perspective and the effect of this on use and purchase.Through this understanding it provides an important direction for

the advertising of games and for game designers. Through a betterunderstanding of what consumers’ value and whether it drivesusage and purchase, advertising within games may well becomemore effective in terms of reaching communication goals such asbrand recall and awareness. Product and game involvement mayalso increase.

This paper is organized as follows. First, we present the conceptualmodel which begins with a definition of the concept of the gameleading to our hypotheses. The paper follows with the methodologyand results. The paper concludes with the discussion, managerial andresearch implications.

2. Conceptual model

A wide variety of concepts have been applied to conceptualizethe consumers interaction with games such as; narratives andinteractive texts (Juul, 2001, Ryan, 2001), cultural artifacts (Prensky,2001) and technological drivers (Woods, 2003; Bushnell, 1996;Aarseth, 2003). In the context of this research we draw from aconceptual model that defines the game from the consumers’experience (Newman, 2002a, 2004; Manninen, 2003; Aarseth,2003) of the consumption or play of the game (Chen, 2008;Holbrook and Hirschman, 1982). Playing a game involves instanta-neous feedback in visual, auditory and kinesthetic forms. This feed-back helps to create interactivity and shape the consumersexperience in cognition and within the medium, create rich virtualworlds that blur the boundaries between imagination and reality(Jessen, 1999).

The process of consumption is not singular, but rather anexperience that varies with the consumer and their level ofinteraction, both within the game and with other game players.A game has an explicit structure that defines how it is to beplayed (Choi and Kim, 2004), yet it is open to interpretation andexperimentation. It is also a representation of the functional andrecreational desires of the immediate consumer (Newman,2002a). Eber (2001) demonstrates that the choice to interact withthe game may be driven by a hedonic need. This enforces theconcept brought forward by Mortensen (2002) and Fromme(2003) that the attraction of the game depends on the subjectiveinterpretation and desire of the consumers and by their self-concept (Walther, 2003; Gottschalk, 1995).

We propose that when a consumer plays a game they experienceinteractivity. The effect of this feedback is to transform theirperceptions of self-efficacy; the belief they hold in their capabilityto accomplish a task, which, in this respect refers to their ability toplay the game (Agarwal et al., 2000; Bandura, 1982). In essence itchanges their fundamental belief that they are capable throughgame play to achieve the desired goals and outcomes.

This argument is supported by Allan (2010), Bandura (1977,1982) and Smith (2002a,b) who defines four sources of self-efficacy:mastery experiences (performance accomplishments), vicariouslearning and experience, social persuasion and affective states(emotional arousal). Allan (2010, p. 36) posits; ‘‘video games canproduce both positive and negative emotional arousal in those whoplay them. Watching another person play a video game provides theobserver with vicarious experience to make efficacy comparisons.Verbal persuasion influences video game self-efficacy when a playerreceives feedback from others. Finally, video games are generallyperformance accomplishment tasks. They provide a player with aconstant stream of input. This input supplies the player withongoing mastery experience to build video game self-efficacy.’’

These findings are consistent with Newman’s (2002a,b, 2004)continuum of engagement and Vorderer (2003) and Eber (2001),who define a game as a ‘form of mastery’ (i.e. the acquisition andperfection of a skill). Consequently, self-efficacy has primarily been

Table 1Sample characteristics (n¼493).

Variable Categories Percent of sample

Gender Male 82.2

Female 17.8

Age r10 0.4

11–15 4.3

16–20 40.2

21–25 37.1

Z26 18.1

Ethnicity NZ Pakeha 29.4

Maori 7.5

Pacific Islander 6.5

Asian 38.5

European 9.9

Others 8.1

Marital status Single 77.3

Widowed 0.2

Living with partner 13.8

Married 7.3

Divorced/separated 1.4

Education Non-degree 66.1

Degree 33.9

Employment Student 47.7

Full time 25.4

Self-employed 4.9

Unemployed 4.3

Homemaker 0.4

Part-time 6.7

Student/part-time 10.8

R. Davis, B. Lang / Journal of Retailing and Consumer Services 19 (2012) 67–77 69

operationalized in the form of prior experience to represent bothmastery and vicarious learning experiences (Igbaria and Iivari, 1995)and is considered to be dynamic in nature since the consumer isexpected to become more capable in performing a task as theirexposure to the task increases.

We argue that through the games consumption and experi-ence of interactivity; consumers will have positive self-efficacy.Thus, the belief in their capability to be interactive with the gamewill drive the value of the experience, positively impacting usage(Allan, 2010) and purchase. Therefore, it is hypothesized that:

Hu1–4. Self-efficacy has an individual effect on game purchase mea-

sured across four game model types; (1) original model, (2) Sports,

Simulation and Driving; (3) RPG, MMORPG and Strategy and (4) Action,

Adventure and Fighting.

Hp1–4. Self-efficacy has an individual effect on game usage measured

across four game model types; (1) original model, (2) Sports, Simulation

and Driving; (3) RPG, MMORPG and Strategy and (4) Action, Adventure

and Fighting.

H5–8. Self-efficacy has a simultaneous effect on game usage and

purchase measured across four game model types; (1) original model,

(2) Sports, Simulation and Driving; (3) RPG, MMORPG and Strategy

and (4) Action, Adventure and Fighting.

As we have noted in our hypotheses; these hypotheses areextended over the 4 game types so the analysis of the pathcoefficients and SEM model fit will proceed to test 8 hypothesizedrelationships between self-efficacy and; (1) game purchase and(2) game usage. Therefore, 8 models are also compared.

Annual Income o10,000 47.5

10,000–20,000 16.6

20,001–30,000 7.5

30,001–40,000 11.4

40,001–50,000 9.5

50,001–60,000 3.2

60,001–80,000 2.4

Z80,000 1.8

3. Method

Data was gathered through face-to-face interviews with 493consumers in Auckland, New Zealand. All consumers who walkedpast the interviewers were considered to be potential respon-dents. The interviewers were rotated around four locations inAuckland; east, west, south and north. Every potential respondentwas asked to participate so they had an equal chance to completethe survey. Those that agreed to participate were asked to respondto a structured questionnaire. Respondents were screened with twoquestions: (1) In the last week, did you play games on yourcomputer (PC or MAC), or on a games console (perhaps throughthe Internet), such as an Xbox, Playstation or Wii that you pur-chased?’’ If the answer was ‘‘Yes’’, they were asked (2) What gamedid you play most often in the last week?

Question 1 established that the respondent was a regular gameplayer of games they had actually purchased and, Question2 checked that the game was not a game preloaded on a computersuch as Solitaire. Four-hundred and ninety three respondentsprovided usable data. Eighty-two percent of the respondents weremale and 18% were female (Table 1). The majority of the respon-dents (77%) were 25 years and under. About 66% of the respondentshad not received a degree and 77% were single. Thirty-nine percentof respondents were Asians and 48% of the respondents werestudents. Forty-eight percent of the respondents had an annualincome of less than $10,000. The samples demographics are gen-erally consistent with the recent research by INZ (2010) on theNew Zealand gaming consumer (N¼1958).

The questionnaire (see Table 2) was designed to measuremulti-item constructs. Throughout the whole questionnaire aseven point scale was used to measure the constructs of interest(1¼ ‘‘strongly disagree’’, 7¼ ‘‘strongly agree’’). To operationalizeself-efficacy we use Smith (2002a,b) with an adapted form ofTorkzadeh and Koufteros (1994) computer self-efficacy (CSE)scale based upon Bandura’s (1977, 1982) guidelines on self-

efficacy and social cognitive theory. Purchase behavior is basedon an adaption of the scale of Bristol and Mangleburg (2005).Usage behavior is based on the Technology Acceptance Model(Venkatesh et al., 2003). Game categories for usage and purchaseare derived from Myers (1990) and the retail categories commonlyused in consumer purchases (http://store.steampowered.com/).

4. Analysis

The analysis tested the proposed conceptual model withconfirmatory factor analysis (CFA) and structural equation mod-eling (SEM). The commonly used approach was employed as wewanted to use an analysis method that not only supported modelrefinement but could rigorously assess model fit across four gamingtypes. It also helped us measure the individual and simultaneouseffects in the relationship between self-efficacy, usage and purchase.

5. Confirmatory factor analysis

This study adopted a two-stage process (Kline, 1998). The firststage of the process was to construct separate measurement modelsfor each latent variable. The structural model is constructed as thesecond stage of the process. Initial data screening was done formissing values, outliers and the normality of the dataset was tested.We examined all scale items and reverse-coded when applicable toreflect the hypothesized directions.

Table 2Questionnaire items.

Screen question: in the last week, did you play

games on your computer (PC or MAC), or on a

games console (perhaps through the Internet),

such as an Xbox, Playstation or Wii that you

purchased? (check the game was purchased)

PC/Mac Xbox PS Internet

SCREEN question: if yes—what game did you

play most often in the last week? [(check the

game is not a game preloaded on a computer

such as solitaire, etc.)

Name of game

This questionnaire is about games you can play

on your computer (PC or MAC) or on a games

console, such as an Xbox, Playstation or Wii.

We will call these console games, simply

‘‘games’’ in this questionnaire

Very rarely Very Often CODE

Purchase behavior: thinking about the types of

games you buy please answer the following

questions by providing a number between

1 and 7 where 1 means ‘very rarely’ and

7 means ‘very often’.

1. How often do you buy games? 1 2 3 4 5 6 7 PB1

2. How often do you buy the following game

types?

Very Rarely Very Often

Action 1 2 3 4 5 6 7 PB2

Adventure 1 2 3 4 5 6 7 PB3

Driving 1 2 3 4 5 6 7 PB4

Fighting 1 2 3 4 5 6 7 PB5

Children 1 2 3 4 5 6 7 PB6

Educational 1 2 3 4 5 6 7 PB7

Massively Multiplayer Online Role Playing

Game (MMORPG)

1 2 3 4 5 6 7 PB8

Role Playing Game (RPG) 1 2 3 4 5 6 7 PB9

Simulation 1 2 3 4 5 6 7 PB10

Strategy 1 2 3 4 5 6 7 PB11

Sports 1 2 3 4 5 6 7 PB12

3. How many games do you own in total? PB13

1 game 2 games 3–5 games

6–10 games 11–15 games 16–20 games

21–30 games 31–40 games More than 40

Play usage behavior: thinking about the types of

games you play please answer the following

questions by providing a number between

1 and 7 where 1 means ‘very rarely’ and

7 means ‘very often’.

4. How often do you play games on each of

the following platforms?

Very Rarely Very Often

PC/MAC 1 2 3 4 5 6 7 PU1

Xbox 1 2 3 4 5 6 7 PU2

Playstation 1 2 3 4 5 6 7 PU3

Connected to the Internet 1 2 3 4 5 6 7 PU4

Wii 1 2 3 4 5 6 7 PU5

5. In a typical week, how many hours do you

play games?

PU6

Less than 2 h 3–5 h 6–10 h

R.

Da

vis,

B.

Lan

g/

Jou

rna

lo

fR

etailin

ga

nd

Co

nsu

mer

Services

19

(20

12

)6

7–

77

70

11–20 h 21–30 h More than 30 h

6. How long have you been playing games? PU7

Less than 6 months 7–11 months 1–2 years

3–5 years 6–10 years 11–15 years

16–20 years More than 20 years

7. How often do you play the following game

types?

Very Rarely Very Often

Action 1 2 3 4 5 6 7 PU8

Adventure 1 2 3 4 5 6 7 PU9

Driving 1 2 3 4 5 6 7 PU10

Fighting 1 2 3 4 5 6 7 PU11

Children 1 2 3 4 5 6 7 PU12

Educational 1 2 3 4 5 6 7 PU13

Massively Multiplayer Online Role Playing

Game (MMORPG)

1 2 3 4 5 6 7 PU14

Role Playing Game (RPG) 1 2 3 4 5 6 7 PU15

Simulation 1 2 3 4 5 6 7 PU16

Strategy 1 2 3 4 5 6 7 PU17

Sports 1 2 3 4 5 6 7 PU18

8. If not clear from Q7, ask and circle: which

one of these game types do you play most?

PU19

9. Which one game do you play most from

that group (Q8)? Write down the name

PU20

Skill level 10. When thinking about insert name of game from Q9 how would you rate your skill level? Beginner 1 2 3 4 5 6 7 Expert SK1

Thinking about game from Q9, please answer the following questions by providing a number between 1 and 7 where 1 means ‘strongly disagree’ and7 means ‘strongly agree’.

Strongly disagree Strongly agree CODE

Self-efficacy 11. I expect to become proficient in playing this game. 1 2 3 4 5 6 7 SE1

12. I feel comfortable playing this game. 1 2 3 4 5 6 7 SE2

13. I am skilled at playing this game. 1 2 3 4 5 6 7 SE3

14. I know how to do what I want to do with this game. 1 2 3 4 5 6 7 SE4

15. I know more about the game than most other people who play this game. 1 2 3 4 5 6 7 SE5

16. I can play this game if

I can call someone for help if I get stuck. 1 2 3 4 5 6 7 SE6

I have the manual for reference. 1 2 3 4 5 6 7 SE7

I have a lot of time to practice. 1 2 3 4 5 6 7 SE8

I have the built-in help assistance. 1 2 3 4 5 6 7 SE9

I have never played a similar game like it before. 1 2 3 4 5 6 7 SE10

I have never played it before. 1 2 3 4 5 6 7 SE11

I have not seen anyone play it before. 1 2 3 4 5 6 7 SE12

I have played a similar game like this one before. 1 2 3 4 5 6 7 SE13

I have seen someone else play it before I play. 1 2 3 4 5 6 7 SE14

Someone else has helped me to get started. 1 2 3 4 5 6 7 SE15

Someone showed me how to play it first. 1 2 3 4 5 6 7 SE16

There was no one to help me to show me what to do. 1 2 3 4 5 6 7 SE17

R.

Da

vis,

B.

Lan

g/

Jou

rna

lo

fR

etailin

ga

nd

Co

nsu

mer

Services

19

(20

12

)6

7–

77

71

R. Davis, B. Lang / Journal of Retailing and Consumer Services 19 (2012) 67–7772

Subsequently, the data was subjected to multivariate normal-ity testing. Results show that the Mardia coefficient was greaterthan 15, very much higher than the 3.0 cutoff advised by Wothke(1993). Thereby, the Bollen–Stine bootstrap method was used(Bollen and Stine, 1992). Cunningham (2008) stresses that if theBollen–Stine (B–S) p value is less than 0.05, the model should berejected.

Convergent and discriminant validity of the constructs weretested using the confirmatory factor analysis (CFA) that combinedall constructs concurrently. Maximum likelihood estimation (MLE)was used to fit the models. MLE is a procedure that improvesparameter estimates in a way that minimizes the differencesbetween the observed and estimated covariance matrices (Pampel,2000). Construct refinement was enabled by an analysis of covar-iance residuals and modification indices and exclusion of items untilthe goodness-of-fit was achieved. Following Baumgartner andHomburg (1996), conventional measures were used to assess themodel fit: goodness-of-fit indices, chi-squared (X2), the comparativefit index (CFI) and normalized fit index (NFI). For CFI and NFI valuesclose to 1 are indicative of good model fit (Bentler, 1990). The rootmean square error of approximation (RMSEA) was calculated for theoverall model and according to Bentler (1990), values below 0.05indicate close fit and values up to 0.08 are reasonable. Finally, thestandardized root mean squared residual (SRMR) as described by Huand Bentler (1995) computes how much the model explains thecorrelations to within an average error. Bentler (1990) argues that amodel is regarded as having an acceptable fit if the SRMR is less than0.10, while a SRMR of 0 indicates a perfect fit (Browne and Cudeck,1993).

The final measurement models show a reasonably good fit andmost of the fit indices are above or close to the required minimumthreshold level. The ratio of minimum discrepancy to degree offreedom (chi-square/DF ratio) should be less than 5 or preferablyless than 2 (Bentler, 1990). The GFI index is above the threshold of0.90 (Hair et al., 2009), and CFI is close to 1 (Bentler, 1990) forevery construct.

Composite reliability is an indicator of the shared varianceamong the set of observed variables used as indicators of a latentconstruct (Bacon et al., 1995; Kandemir et al., 2006). The threeitems included in self-efficacy are: (1) respondents have a manualfor reference; (2) respondents have the built-in help assistanceand (3) respondents have never played this game before. The con-struct reliability for these self-efficacy items was 0.83, above therecommended value of 0.70 or higher. In addition, the coefficientalpha value was 0.83, above the threshold value of 0.70 thatNunnally (1978) recommends. The average variance extracted(AVE) value was 0.63. It reflects the average communality for eachlatent factor and is used to establish convergent validity. The AVEvalue is above the threshold value of 0.50 (Chin, 1998; Hock andRingle, 2006; Fornell and Larcker, 1981).

Table 3SEM model fit (step 1): individual effect.

Dependent variable Game group X2 (DF) X2/DF ra

Game, usage Original 401.77 (53) 7.58

Game, purchase Original 440.44 (53) 8.31

Game, usage Sports, Simulation, Driving 194.67 (8) 1.83

Game, purchase Sports, Simulation, Driving 10.90 (8) 1.36

Game, usage RPG, MMORPG, Strategy 43.51 (8) 5.44

Game, purchase RPG, MMORPG, Strategy 21.85 (8) 2.73

Game, usage Action, Adventure, Fighting 23.12 (8) 2.89

Game, purchase Action, Adventure, Fighting 27.97 (8) 3.50

X2—chi-square; CFI—comparative fit index; TLI—Tucker Lewis index; GFI—goodness-

dized root-mean-squared residual; B—S p—Bollen–Stine bootstrap p; DF—degrees of f

6. Structural equation modeling

The structural equation modeling process had two competingsteps. The first step assessed the conceptual model measuring theindividual effects of self-efficacy on purchase and usage sepa-rately. The second step measured the simultaneous effect of self-efficacy on purchase and usage together.

6.1. Individual effects

In the first step SEM, the same conventional measures wereused to assess the model fit as in the CFA, that is, the goodness-of-fit indices (GFI), the chi-squared (X2)/degrees of freedom (DF)ratio, the comparative fit index (CFI), the normalized fit index(NFI), the root mean squared error of approximation (RMSEA), thestandardized RMR (SRMR) and the Bollen–Stine (B–S) p value.

The SEM focused on the analysis of the hypotheses of the fourcompeting forms of this model; (1) the original model includes allthe game types while the alternative models focus on each gamegenre, namely (2) Sports, Simulation and Driving; (3) RPG, MMORPGand Strategy and (4) Action, Adventure and Fighting. The results ofthe SEM analysis for both models are displayed in Tables 3 and 4.The final model met suggestions from the literature regarding theminimum number of items attached to a construct (Hair et al.,2009).

For the original model: the game usage results indicate inade-quate model fit (GFI¼0.88, CFI¼0.75, TLI¼0.69, RMSEA¼0.12,SRMR¼0.09, X2/DF¼7.58 and B–S p¼0.00). Similarly, the self-efficacy results for game purchase were inadequate (GFI¼0.86,CFI¼0.81, TLI¼0.76, RMSEA¼0.12, SRMR¼0.08, X2/DF¼8.31 andB–S p¼0.00). With poor fit indices results and unacceptable B–S p

values, the models should be rejected. The standardized factorloadings for self-efficacy (game usage) ranged from 0.69 to 0.87and were highly significant (po0.001). The standard factor loadingfor self-efficacy (game purchase) were similar and highly significant(po0.001). The average variance extracted (AVE) value was 0.63.

For the Sports, Simulation and Driving Model: The game usageresults suggest adequate model fit (GFI¼0.99, CFI¼0.99, TLI¼0.98,RMSEA¼0.04, SRMR¼0.03, X2/DF¼1.83 and B–S p¼0.45). Similarlythe self-efficacy results for game purchase were adequate (GFI¼0.99, CFI¼0.99, TLI¼0.99, RMSEA¼0.03, SRMR¼0.02, X2/DF¼1.36and B–S p¼0.79). The standardized factor loadings for self-efficacy(game usage) ranged from 0.69 to 0.87 and were highly significant(po0.001). The standard factor loading for self-efficacy (gamepurchase) were similar and highly significant (po0.001). Theaverage variance extracted (AVE) value was 0.63. With these results,both models (game usage and game purchase) in the Sports,Simulation and Driving genre are accepted. The results for Sports,Simulation and Driving game classification reveal that a significantpositive relationship for the path between self-efficacy and game

tio p CFI TLI GFI RMSEA SRMR B-S p

0.00 0.75 0.69 0.88 0.12 0.09 0.00

0.00 0.81 0.76 0.86 0.12 0.08 0.00

0.07 0.99 0.98 0.99 0.04 0.03 0.45

0.21 0.99 0.99 0.99 0.03 0.02 0.79

0.00 0.95 0.91 0.97 0.09 0.06 0.00

0.01 0.98 0.97 0.99 0.06 0.04 0.09

0.00 0.98 0.97 0.99 0.06 0.04 0.06

0.00 0.98 0.96 0.98 0.07 0.04 0.02

of-fit-index; RMSEA—root-mean-square error of approximation; SRMR—standar-

reedom.

Ta

ble

4S

EM

pa

thco

effi

cie

nts

(ste

p1

):in

div

idu

al

eff

ect

.

Ga

me

gro

up

Ind

ica

tor

Dir

ect

ion

Co

nst

ruct

Sta

nd

ard

ize

d

loa

din

g

Un

-sta

nd

ard

ize

d

loa

din

g

S.E

.t-

Va

lue

pH

yp

oth

esi

sC

on

clu

sio

n

Sp

ort

s,S

imu

lati

on

,D

riv

ing

Ga

me

usa

ge

(GU

)’

Se

lf-e

ffica

cyG

U.2

4.1

3.0

43

.55

.00

H1

uS

up

po

rte

d

Ga

me

pu

rch

ase

(GP

)S

elf

-effi

cacy

GP

.20

.15

.04

3.4

3.0

0H

1p

Su

pp

ort

ed

RP

G,

MM

OR

PG

,S

tra

teg

yG

am

eu

sag

e(G

U)

Se

lf-e

ffica

cyG

U.2

4.2

5.0

64

.58

.00

H2

uM

od

el

reje

cte

d,

B–

Spo

0.5

Ga

me

pu

rch

ase

(GP

)S

elf

-effi

cacy

GP

.15

.14

.05

2.7

0.0

1H

2p

Mo

de

lre

ject

ed

,B

–S

po

0.5

Act

ion

,A

dv

en

ture

,Fi

gh

tin

gG

am

eu

sag

e(G

U)

Se

lf-e

ffica

cyG

U�

.01

�.0

1.0

4�

.13

.90

H3

uM

od

el

reje

cte

d,

B–

Spo

0.5

Ga

me

pu

rch

ase

(GP

)S

elf

-effi

cacy

GP

�.0

3�

.02

.04

�.4

9.6

3H

3p

Mo

de

lre

ject

ed

,B

–S

po

0.5

Ori

gin

al

mo

de

lG

am

eu

sag

e(G

U)

Se

lf-e

ffica

cyG

U.1

4.0

8.0

32

.44

.02

H4

uM

od

el

reje

cte

d,

B–

Spo

0.5

Ga

me

pu

rch

ase

(GP

)S

elf

-effi

cacy

GP

.11

.07

.03

2.0

0.0

5H

4p

Mo

de

lre

ject

ed

,B

–S

po

0.5

SE

—st

an

da

rde

rro

r.

R. Davis, B. Lang / Journal of Retailing and Consumer Services 19 (2012) 67–77 73

usage. Likewise a significant positive relationship exists for the pathbetween self-efficacy and game purchase.

For the RPG, MMORPG and Strategy Model: The game usage resultsin an adequate model fit (GFI¼0.97, CFI¼0.95, TLI¼0.91,RMSEA¼0.09, SRMR¼0.06, X2/DF¼5.44 and B–S p¼0.00). Similarlythe self-efficacy results for game purchase were adequate (GFI¼0.99,CFI¼0.98, TLI¼0.97, RMSEA¼0.06, SRMR¼0.04, X2/DF¼2.73 andB–S p¼0.09). The standardized factor loadings for self-efficacy (gameusage) ranged from 0.69 to 0.87 and were highly significant(po0.001). The standard factor loading for self-efficacy (gamepurchase) were similar and highly significant (po0.001). The averagevariance extracted (AVE) value was 0.63. Considering the Bollen–Stine (B–S) p values of these models, the game usage and purchasemodels are rejected. We note that there is a significant relationshipbetween self-efficacy and game purchase.

For the Action, Adventure and Fighting Model: The game usageresults an adequate model fit (GFI¼0.99, CFI¼0.98, TLI¼0.97,RMSEA¼0.06, SRMR¼0.04, X2/DF¼2.89 and B–S p¼0.06). Similarlythe self-efficacy results for game purchase were adequate (GFI¼0.98, CFI¼0.98, TLI¼0.96, RMSEA¼0.07, SRMR¼0.04, X2/DF¼3.50and B–S p¼0.02). The standardized factor loadings for self-efficacy(game usage) ranged from 0.69 to 0.87 and were highly significant(po0.001). The standard factor loading for self-efficacy (gamepurchase) were similar and highly significant (po0.001). Theaverage variance extracted (AVE) value was 0.63. Considering theBollen Stine (B–S) p values of both models, they are rejected.

6.2. Simultaneous effect

We have also investigated the impact of self-efficacy on gameusage and purchase behavior simultaneously across the game typesand the original model. Given the Bollen–Stine (B–S) p values areless than 0.5 all models should be rejected (see Tables 5 and 6).

7. Discussion

We investigated the impact of self-efficacy on game usage andpurchase behavior, both individually and simultaneously. It wasconcluded in the individual effects analysis that self-efficacyimpacts game usage and purchase for only the Sports/Simula-tion/Driving genre. Our results showed no simultaneous relation-ship across all games types. Overall, we conclude that consumersself-efficacy does impact usage and/or purchase behavior butgame type has a significant impact on this relationship. The gametypes that showed no relationship between self-efficacy andusage or purchase were:

1.

All game genres combined. 2. Action/Adventure/Fighting. 3. Role Playing Game/Massively Multiplayer Online Role-Playing

Game/Strategy.

The positive relationship between self-efficacy and consumervalue evaluations and usage intentions is supported by VanBeuningen et al. (2009) and Dash and Saji (2007). More recently,Allan (2010, p. 36) concurred with our findings, stating that ‘‘self-efficacy may not be the only determinant of one’s motivation toplay a video game, but it appears to be an important one.’’ It wasalso argued that; (1) males had higher video game self-efficacyand (2) usage frequency was related to video game self-efficacy.In our study, Eighty-two percent of the respondents were male sowe suggest a similar effect to Allan’s (2010) gender correlations.Given the game type, that is, Sports/Simulation/Driving showed asignificant model fit, we further contend subjectively that ourresults may be influenced by gender. Also, it is not surprising that

Table 6SEM path coefficients (step 2): simultaneous effect.

Game group Indicator Direction Construct Standardized

loading

Un-standardized

loading

S.E. t-Value p Hypothesis Conclusion

Sports Simulation Driving Game usage (GU) ’ Self-Efficacy GU 0.24 0.25 0.07 3.70 0.00 H5u Model rejected, B–S po0.5

Game purchase (GP) 0.20 $0.20 0.06 3.33 0.00 H5p Model rejected, B–S po0.5

RPG MMORPG Strategy Game usage (GU) Self-Efficacy GP 0.24 0.24 0.05 4.79 0.00 H6u Model rejected, B–S po0.5

Game Purchase (GP) 0.16 0.16 0.05 3.31 0.00 H6p Model rejected, B–S po0.5

Action Adventure Fighting Game usage (GU) Self-Efficacy GP 0.003 0.003 0.06 0.05 0.96 H7u Model rejected, B–S po0.5

Game purchase (GP) �0.01 �0.01 0.05 �0.11 0.92 H7p Model rejected, B–S po0.5

Original model Game usage (GU) Self-Efficacy GP 0.15 0.16 0.06 2.69 0.01 H8u Model rejected, B–S po0.5

Game purchase (GP) 0.11 0.11 0.05 2.01 0.04 H8p Model rejected, B–S po0.5

SE—standard error; the above models are rejected with, B–S po0.5.

Table 5SEM model fit (step 2): simultaneous effect.

Dependent variables Game group X2 (DF) X2/DF ratio p CFI TLI GFI RMSEA SRMR B–S p

Game, usage Sports, Simulation, Driving 509.63 (24) 21.24 0.00 0.74 0.62 0.85 0.20 0.08 0.00

Game, purchase

Game, usage RPG, MMORPG, Strategy 65.89 (24) 2.75 0.00 0.97 0.96 0.97 0.06 0.05 0.01

Game, purchase

Game, usage Action, Adventure, Fighting 584.40 (24) 24.35 0.00 0.74 0.61 0.82 0.22 0.09 0.00

Game, Purchase

Game, usage Original 2961.16 (186) 15.92 0.00 0.47 0.40 0.63 0.17 0.13 0.00

Game, purchase

X2—chi-square; CFI—Comparative fit index; TLI—Tucker Lewis index; GFI-goodness-of-fit-index; RMSEA—root-mean-square error of approximation; SRMR—standardized root-mean-squared residual; B—S p—Bollen–Stine

bootstrap p; DF—degrees of freedom; the above models are rejected with, B–S po0.5.

R.

Da

vis,

B.

Lan

g/

Jou

rna

lo

fR

etailin

ga

nd

Co

nsu

mer

Services

19

(20

12

)6

7–

77

74

R. Davis, B. Lang / Journal of Retailing and Consumer Services 19 (2012) 67–77 75

self-efficacy has an impact because this game type places empha-sis on hand/eye co-ordination/reflexes in real world environments(Myers, 1990). Consumers must have a belief in their capability toaccomplish tasks, play the game and achieve defined objectives(Agarwal et al., 2000; Bandura, 1982). Such games also have ahigh level of interactivity between consumer and game during theconsumption process.

What is interesting to explore is why the consumer’s processof self-efficacy with Action/Adventure/Fighting games, whichplace emphasis on simulations of futuristic and historical warfareand/or violent activity did not affect purchase or usage. It wouldappear that there is no match between the actual self and theideal self when they experience these games. This finding mayalso indicate that players of this genre differ from players of othergenres. For example, gamers in the Action, Adventure and Fight-ing genre may engage in gaming to a greater extent and thus,have a smaller gap between their actual and ideal self in thegame. That is, they are highly proficient already and self-efficacyis not a key driver of purchase. It may also suggest that suchgames do not impact self-concept and their maybe a low level ofinteractivity. This finding may conflict with the view that, forexample, violent computer games create violent consumers. Thisview maybe tempered by other findings. For example Allan (2010,p. 4) and others (Carnagey et al., 2007; Anderson et al., 2003)argues that ‘‘violent video games have been shown to increaseaggression and physiological arousal of those who play themy

attributed to the desensitization effect.’’A similar non-significant result was found for Role Playing

Game/Massively Multiplayer Online Role-Playing Game/Strategygames, where self-efficacy was not related to usage or purchase.As with Action/Adventure/Fighting games, this may be related tothe effect of multiple self’s. It is proposed that with the consumerthere may be some confusion about which character is supposedto have game self-efficacy. Is it the consumer or the game player(character within the game)? These types of games do not havewell defined goals. A lot more emphasis is placed on explorationand experimentation. It may be more difficult for a consumer toassess self-efficacy with this level of ambiguity.

One of the key managerial findings of the study relates tomarketing stimuli within a game. Our findings suggest that market-ers and gamer developers must first consider the mediating effect ofself-efficacy on the effectiveness of their advertisement or product/service placement within the gaming environment. Simply put, if theconsumer does not perceive they have the capabilities to play thegame, their purchase and usage behavior will be affected. Practionersalso should consider the impact of game type. While our findings areonly related to self-efficacy, we suggest that different game typeswill reveal different results when measuring the consumers’ cogni-tive response to game consumption and experience.

8. Limitations

Future research may wish to ascertain the applicability of theresults to other geographical areas. Also, it could be argued thatgrouping the games together in terms of genre types is a limitationof the data analysis. We believe that grouping the games isappropriate as they exhibit similar characteristics and thus repre-sent similar acts of consumption. Our study also differentiated gametypes but did not examine the differences of online vs. offlinegaming. Would we expect a difference in the results? Furtherstudies may uncover differences but we are yet to uncover anyconvincing evidence. We note that the sample is biased towardsmales. We could have controlled for this during data collection, butthis would have manipulated the randomly generated sample.It could be argued that having a male biased sample may be more

representative of the market population for computer games. USmarket statistics from the Entertainment Software Associationshowed that in 2008 sixty percent of all game players are men.We acknowledge that a balance will evolve between the numbers ofmale and female gamers over time as more games are developedwith a specific gender orientation. Future research should also takeaccount of this change.

9. Future research

Future studies should now introduce specific marketing stimuliwithin different types of games and measure the mediating effect ofself-efficacy on involvement, brand recall and awareness. There isalso the need to clarify the relationship between self-efficacy andmultiple self-concepts. Given that the act of playing a game is alearning experience that is often concerned with the mastery of askill, Prensky’s (2001) research on consumer learning styles may beintegrated to classify gamers using alternative criteria. The focuscould be on defining the consumer’s personality and learning styleto support the self-concept as key antecedents of game selection andgaming behavior. Another extension to the research model would beto focus on the three dimensions of the game (game-play, game-structure and game-world). Such research would require thesedimensions to be expanded further to identify the key elementsthat constitute each of these dimensions. For example, game-worldcould be expanded into elements such as the use of 3D graphics,based on real-life/fantasy, exploratory world/restrictive world andgame-play could be expanded using elements of interactivity,competition and teamwork. Given this conceptual model is newwithin this research context it may be argued that there is a lack ofqualitative data to support its development and use. This wouldconsist of a phenomenological design utilizing grounded theory asthe primary research methodology of both new and experiencedplayers. Future research should extend the model into othersamples, different from the New Zealand context.

Acknowledgments

Manukau Institute of Technology, Chi Main Ong and JosephinoSan Diego.

References

Aarseth, E., 2003. Playing Research: Methodological Approaches to Game Analysis.In: Proceedings of Conference of Digital Arts and Cultures (DAC). Melbourne,VIC.

Agarwal, R., Sambamurthy, V., Stair, R.M., 2000. Research report: the evolvingrelationship between general and specific computer self-efficacy—an empiri-cal assessment. Information Systems Research 11 (4), 418–430.

Allan, J.D., 2010. An Introduction to Video Game Self-Efficacy. Masters Thesis.Faculty of California State University, Chico.

Anderson, C., Berkowitz, L., Donnerstein, E., Huesmann, L., Johnson, J., Linz, D.,Wartella, E., 2003. The influence of media violence on youth. PsychologicalScience in the Public Interest 4 (3), 81–110.

Anderson, C.A., Bushman, B.J., 2002. The effects of media violence on society.Science 295, 2377–2379.

Apperley, T.H., 2006. Genre and game studies: toward a critical approach to videogame genres. Simulation and Gaming 37 (1), 6–23.

Bacon, D.R., Sauer, P.L., Young, M., 1995. Composite reliability in structuralequation modeling. Educational and Psychological Measurement 55, 394–406.

Bandura, A., 1977. Self-efficacy: toward a unifying theory of behavioural change.Psychological Review 84, 191–215.

Bandura, A., 1982. Self-efficacy mechanism in human agency. American Psychol-ogist 37 (2), 122–147.

Baumgartner, H., Homburg, C., 1996. Applications of structural equation modelingin marketing and consumer research: a review. International Journal ofResearch in Marketing 13, 139–161.

Bawa, K., Shoemaker, R.W., 1987. The coupon-prone consumer: some findingsbased on purchase behavior across product classes. Journal of Marketing 51(4), 99–110.

R. Davis, B. Lang / Journal of Retailing and Consumer Services 19 (2012) 67–7776

Bentler, P.M., 1990. Comparative fit indexes in structural models. PsychologicalBulletin 107, 238–246.

Bollen, Kevin A., Stine, Robert A., 1992. Bootstrapping goodness-of-fit measuresin structural equation models. Sociological Methods and Research 21,205–229.

Boyle, Raymond, Hibberd, Mathew, 2005. Review of research on the impact ofviolent computer games on young people. Her Majesty’s Department of Tradeand Industry/Department of Culture, Media and Sport, London.

Browne, M.W., Cudeck, R., 1993. Alternative ways of assessing model fit. In: Bollen,K.A., Long, J.S. (Eds.), Testing structural Equation Models, SAGE, Newbury Park,CA, pp. 136–162.

Bristol, Terry, Mangleburg, Tamara F., 2005. Not telling the whole story: teendeception in purchasing. Journal of the Academy of Marketing Science 33 (1),79–95.

Bushnell, N., 1996. Relationships between fun and the computer business.Communications of the ACM 39 (8), 31–37.

Carnagey, N., Anderson, C., Bushman, B., 2007. The effect of video game violence onphysiological desensitization to real-life violence. Journal of ExperimentalSocial Psychology 43 (3), 489–496.

Cauberghe, V., De Pelsmacker, P., 2010. Advergames: the impact of brandprominence and game repetition on brand responses. Journal of Advertising31 (1), 5–18.

Chaney, I.M., Lin, K., Chaney, J., 2004. The effect of billboards within the gamingenvironment. The Journal of Interactive Advertising 5 (1), 37–45.

Chen, M.K., 2008. Rationalization and Cognitive Dissonance: Do Choices Affect orReflect Preferences? Cowles Foundation Discussion Paper No. 1669. YaleUniversity, New Haven, CT.

Chin, W.W., 1998. The partial least squares approach for structural equationmodeling. In: Macoulides, G.A. (Ed.), Modern Methods for Business Research,Lawrence Erlbaum Associates, Mahwah, NJ, pp. 295–336.

Choi, D., Kim, J., 2004. Why people continue to play online games: in search ofcritical design factors to increase customer loyalty to online contents.Cyberpsychology and Behaviour 7 (1), 11–24.

Cunningham, E., 2008. A Practical Guide to Structural Equation Modeling UsingAmosStatsline, Melbourne, VIC.

Dash, S., Saji, K.B., 2007. The role of consumer self-efficacy and website social-presence in customers’ adoption of B2C online shopping: an empirical study inthe Indian context. Journal of International Consumer Marketing 20 (2), 33–48.

Desai, K.K., Hoyer, W.D., 2000. Descriptive characteristics of memory-basedconsideration sets: influence of usage occasion frequency and usage locationfamiliarity. Journal of Consumer Research 27 (3), 309–323.

Dill, Karen E., Dill, Jody C., 1998. Video game violence: a review of the empiricalliterature. Aggression and Violent Behavior: A Review Journal 3, 407–428.

Eber, D.E., 2001. Towards computer game aesthetics—editorial. Digital Creativity12 (3), 129–132.

Folkes, V.S., Martin, I.M., Gupta, K., 1993. When to say when: effects of supply onusage. Journal of Consumer Research 20 (3), 467–477.

Fornell, C., Larcker, D., 1981. Structural equation models with unobservable variablesand measurement error. Journal of Marketing Research 18 (1), 39–50.

Fromme, J., 2003. Computer games as a part of children’s culture. InternationalJournal of Computer Game Research 3 (1) (available at: accessed May 2010)/http://www.gamestudies.org/0301/Fromme/S.

Gottschalk, S., 1995. Videology: video-games as postmodern sites/sights ofideological reproduction. Symbolic Interaction 18 (1), 1–18.

Guth, R.A., 2003. Choosing sides: game giant links with Sony, snubbingMicrosoft—electronic arts show its clout and wariness of allowing Xboxto dominate market—John Madden on play-by-play. Wall Street Journal(New York, 12 May).

Hair, Joseph F., Black, William, Babin, Barry, Anderson, Rolph E., 2009. MultivariateData Analysis, seventh edition. Prentice Hall, Upper Saddle River, NJ.

Hock, Michael, Ringle, Christian M., 2006. Strategic Networks in the SoftwareIndustry: An Empirical Analysis of the Value Continuum. Paper Presented atthe IFSAM Viiith World Congress. 2006, Berlin. Available at: /http://www.Ibl-Unihh.De/IFSAM06.PdfS (accessed 11.06.10.).

Holbrook, M.B., Hirschman, E.C., 1982. The experiential aspects of consumption:consumer fantasies, feelings, and fun. Journal of Consumer Research 9 (2),132–140.

Hu, L.-T., Bentler, P.M., 1995. Evaluating model fit. In: Hoyle, R.H. (Ed.), StructuralEquation Modeling: Concepts, Issues and Applications, SAGE, Thousand Oaks,CA, pp. 76–99.

Hui, S.K., Bradlow, E.T., Fader, P.S., 2009. Testing behavioral hypotheses using anintegrated model of grocery store shopping path and purchase behavior.Journal of Consumer Research 36 (3), 478–493.

INZ, 2010. National Research Prepared by Bond University for the InteractiveGames and Entertainment Association.

Igbaria, M., Iivari, J., 1995. The effects of self-efficacy on computer usage. Omega.International Journal of Management Science 23 (6), 587–605.

Jessen, Carsten, 1999. Computer Games and Play Culture—An Outline of anInterpretative Framework. Available at: /http://www.carsten-jessen.dk/compgames.htmlS.

Jin, S.A., Bolebruch, J., 2009. Avatar-based advertising in second life: the role ofpresence and attractiveness of virtual spokespersons. The Journal of Inter-active Advertising 10 (1), 51–60.

Juul, J., 2001. Games Telling stories?—a brief note on games and narratives.International Journal of Computer Game Research. Available at /http://www.gamestudies.org/0101/Juul-Gts/SS (last accessed July 2010).

Kaltcheva, Velitchka D., Patino, Anthony D., Chebat, Jean-Charles, 2011. The impactof retail environment extraordinariness on customer self-concept. Journal ofBusiness Research 64 (6), 551–557.

Kandemir, Destan, Attila, Yaprak, Tamer Cavusgil, S., 2006. Alliance orientation:conceptualization, measurement, and impact on market performance. Journalof the Academy of Marketing Science 34 (3), 324–341.

Khan, T.L.Tran, 2002. U.S. game industry posts record sales. Wall Street Journal 7(February).

Kline, R.B., 1998. Principles and Practice of Structural Equation ModelingGuildfordPress, New York.

Liu, Y., 2007. The long-term impact of loyalty programs on consumer purchasebehavior and loyalty. Journal of Marketing 71 (4), 19–35.

Mackay, T., Ewing, M., Newton, F., Windisch, L., 2009. The effect of productplacement in computer games on brand attitude and recall. InternationalJournal of Advertising 28 (3), 423–438.

Manninen, T., 2003. Conceptual, communicative and pragmatic aspects of inter-action forms—rich interaction model for collaborative virtual environments.In: Proceedings of Computer Animation and Social Agents (CASA) Conference.IEEE Computer Society Press, pp. 168–174.

Mau, G., Silberer, G., Constien, C., 2008. Communicating brands playfully: effects ofin-game advertising for familiar and unfamiliar brands. International Journalof Advertising 27 (5), 827–851.

Molesworth, Mike, 2006. Real brands inimaginary worlds: investigating players’experiences of brand placement indigital games. Journal of Consumer Beha-viour 5 (4), 355–366.

Mortensen, T., 2002. Playing with players. Potential methodologies for muds.International Journal of Computer Game Research 2 (1) (available at: accessedJuly 2010)/http://www.gamestudies.org/0102/Mortensen/S.

Myers, D., 1990. Computer game genres. Play and Culture 3, 286–301.Nelson, M.R., Keum, H., Yaros, R.A., 2004. Advertainment or adcreep? Game

players’ attitudes toward advertising and product placements in computergames. Journal of Interactive Advertising 5 (1), 3–21.

Newman, J., 2002a. In search of the game player—the lives of Mario. New Mediaand Society 4 (3), 405–422.

Newman, J., 2002b. The myth of the ergodic Game. International Journal ofComputer Game Research 2 (1) (available at: accessed July 2010)/http://www.gamestudies.org/0102/Newman/S.

Newman, J., 2004. GamesTaylor Francis Group, Routledge.Nicovich, S.G., 2005. The effect of involvement on ad judgment in a video game

environment—the mediating role of presence. Journal of Interactive Advertis-ing 6 (1), 29–39.

Nunnally, J.C., 1978. Psychometric Theory, second ed. McGraw-Hill, New York.Pampel, Fred C., 2000. Logistic Regression: A Primer. Sage Quantitative Applica-

tions in the Social Sciences Series #132Sage Publications, Thousand Oaks, CA.Prensky, M., 2001. Fun, Play and Games: What Makes Games Engaging. Digital

Game-Based LearningMcgraw-Hill, NY.Prugsamatz, S., Lowe, B., Alpert, F., 2010. Modelling consumer entertainment

software choice: an exploratory examination of key attributes, and differencesby gamer segment. Journal of Consumer Behaviour 9 (5), 381–392.

Ryan, M., 2001. Beyond myth and metaphor*—the case of narrative in digitalmedia. International Journal of Computer Game Research 1 (1) (available at:accessed July 2010)/http://www.gamestudies.org/0101/Ryan/S.

Schneider, L.P., Cornwell, T.B., 2005. Cashing in on crashes via brand place-ment in computer games. International Journal of Advertising 24 (3),321–343.

Shimp, T.A., Kavas, A., 1984. The theory of reasoned action applied to couponusage. Journal of Consumer Research 11 (3), 795–809.

Sismeiro, C., Bucklin, R.E., 2004. Modeling purchase behavior at an E-commerceweb site: a task-completion approach. Journal of Marketing Research 41 (3),306–323.

Smith, S.M., 2002a. The role of social cognitive career theory in informationtechnology based academic performance. Information Technology, Learning,and Performance Journal 20 (2), 1–10.

Smith, S.M., 2002b. Using the social cognitive model to explain vocational interestin information technology. Information Technology, Learning, and Perfor-mance Journal 20 (1), 1–9.

Sriram, S., Chintagunta, P.K., Agarwal, M.K., 2010. Investigating consumer purchasebehavior in related technology product categories. Marketing Science 29 (2),291–314.

Torkzadeh, G., Koufteros, X., 1994. Factorial validation of a computer self-efficacyscale and the impact of computer training. Education and PsychologicalMeasurement 54 (3), 813–821.

UKIE, 2011. The Association for UK Interactive Entertainment, hhttp://www.ukie.infoi.

Van Beuningen, J., de Ruyter, K., Wetzelfs, M., Streukens., S., 2009. Customer self-efficacy in technology- based self-service. Journal of Service Research,407–428.

Venkatesh, Viswanath, Morris, Michael G., Davis, Gordon B., Davis, Fred D., 2003.User acceptance of information technology: toward a unified view. MISQuarterly 27 (3), 425–478.

Vorderer, P., 2003. Explaining the enjoyment of playing video games: the role ofcompetition. In: Proceedings of the Second International Conference onEntertainment Computing (May).

Walther, B.K., 2003. Playing and gaming. Reflections and classifications. Interna-tional Journal of Computer Game Research 3 (1) (available at: accessed May2010)/http://www.gamestudies.org/0301/Walther/S.

R. Davis, B. Lang / Journal of Retailing and Consumer Services 19 (2012) 67–77 77

Winkler, T., Buckner, K., 2006. Receptiveness of gamers to embedded brandmessages in advergames: attitudes towards product placement. Journal ofInteractive Advertising 7 (1), 24–32.

Woods, S., 2003. An Investigation into the Cultural Significance of ComputerMediated Games. Candidacy Proposal. Doctor of Philosophy (Media andInformation). Curtin University of Technology, Perth, WA.

Wothke, Werner, 1993. Non positive definite matrices in structural modeling. In:Bollen, K.A., Long, J.S. (Eds.), Testing Structural Equation Models, Sage Pub-lication, Newbury Park, CA.

Yang, M., Roskos-Ewoldsen, D.R., Dinu, L., Arpan, L.M., 2006. The effectiveness of‘‘in-game’’ advertising: comparing college students’ explicit and implicitmemory for brand names. Journal of Advertising 35 (4), 143–152.