- Email: [email protected]
Multivariate effects of gender, ownership, and the frequency of use on computer anxiety among high school students
Computers in Human Behavior 24 (2008) 2639–2648
Contents lists available at ScienceDirect
Computers in Human Behavior journal homepage: www.elsevier.com/locate/comphumbeh
Multivariate effects of gender, ownership, and the frequency of use on computer anxiety among high school students Mustafa Balog˘lu *, Vildan Çevik 1 Gaziosmanpasa University, College of Education, Department of Educational Sciences, Tokat, Turkey
a r t i c l e
i n f o
Article history: Available online 28 April 2008
Keywords: Computer anxiety High school students Ownership Frequency of use
a b s t r a c t Studies that address the problems associated with computer anxiety are abundant; however, fewer studies took into account multivariate nature of the construct. Moreover, studies focusing on high school students are even more limited. Thus, the present study investigated the multivariate effects of gender, ownership, and the frequency of computer use on computer anxiety levels, after controlling for the possible effects of trait anxiety among 715 Turkish high school students. The Computer Anxiety Scale and the State-Trait Anxiety Inventory were used to assess computer anxiety and state and trait anxiety levels, respectively. A 2 X 2 X 3 between-subjects factorial multivariate analysis of covariance was used on three dependent variables that are the three dimensions of computer anxiety: Affective Anxiety, Damaging Anxiety, and Learning Anxiety. Independent variables are gender, ownership (i.e., yes or no), and the frequency of computer use (i.e., everyday, several times a week, or once a week or less). Results showed a significant covariate effect of trait anxiety, significant main effects for gender, ownership, and the frequency of computer use on the dependent variables. No three-way or two-way interaction was detected. After the descriptive and comparative analyses, specific suggestions were provided based on the results. Ó 2008 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: +90 356 2521616/3415; fax: +90 356 2521546. E-mail addresses: [email protected], [email protected] (M. Balog˘lu), [email protected] (V. Çevik). 1 Tel.: +90 356 2521616/3415; fax: +90 356 2521546. 0747-5632/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.chb.2008.03.003
2640
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
1. Introduction With the more extensive use in all levels of education (Durndell & Haag, 2002; Imhof, Vollmeyer, & Beierlein, 2007), issues related to computers are more focus of interest among researchers. It is predicted that the higher number of people use computers in their daily lives, the more will face difficulties with them (Beckers & Schmidt, 2001). In this context, the construct of computer anxiety has been studied since the beginnings of the 80’s, mostly focusing on the non-cognitive factors such as attitudes, previous experiences with computers, trait anxiety, or personality-related factors. Even though the construct of computer anxiety has been studied for an extended period of time, there is still no consensus in the literature regarding its definition (Beckers, Wicherts, & Schmidt, 2007). However, according to a commonly used definition, computer anxiety is the fear and apprehension felt by an individual when considering the utilization of computer technology or when actually using computers (Maurer, 1983). A review of the literature showed that most definitions of computer anxiety include a fear component (Chua, Chen, & Wong, 1999). Computerphobia (Rosen, Sears, & Weil, 1987), computer apprehension (Anderson, 1996), computer resistance (Bohlin & Hunt, 1995), or technophobia (Brosnan, 1999) are some of the other terms used interchangeably with computer anxiety. Regardless of the term, there is evidence that a large number of students hold negative attitudes towards and experience high levels of computer anxiety (e.g., North & Noyes, 2002). Bozionelos (2001) found that as high as 40–50% of students experience computer anxiety. More specifically, there is evidence that high school students may experience higher levels of computer anxiety than college students (Loyd & Gressard, 1984). Recently, computer anxiety has been conceptualized as a multi-dimensional construct, including psychological, operational, and sociological components (e.g., Beckers & Schmidt, 2001; Beckers et al., 2007). Owning a personal computer (PC) at home (ownership) and the frequency of computer use were found to be two of the operational components of computer anxiety (e.g., Brosnan, 1999; Keser, 2001; Rosen & Weil, 1995; Üstündag˘, 2001). Arıkan (2002) found that those individuals who owned a PC at home showed less anxiety related to computers. Similarly, Chua et al. (1999) and Chou (2003) found that computer usage was negatively related to computer anxiety. In a relatively earlier study, Selwyn (1997) found that, out of 530 students, 68.5% had a PC at home; but there was no relationship between ownership and frequency of computer use. With a group of Turkish college students, Namlu and Ceyhan (2002) found that 19.8% of the students had a PC at home and those who had a PC showed significantly lower levels of computer anxiety. In addition, they found that as the frequency of computer use increased, computer anxiety levels decreased. This meant that students who used computers more often showed lower levels of computer anxiety compared to those who used computers less frequently. Gender may be the single most widely investigated sociological variable in computer anxiety research (Mcilroy, Bunting, Tierney, & Gordon, 2001). Generally speaking, men have more positive attitudes toward computers than women (Bovee, Voogt, Meelissen, 2007; Dupin-Bryant, 2002). In addition, women show higher levels of computer anxiety than men (e.g., Abdelhamid, 2002; Durndell & Haag, 2002; Mcilroy et al., 2001; Todman, 2000). However, other studies failed to find significant differences between men and women on computer anxiety levels (e.g., Colley, Gale, & Harris, 1994; Rosen & Weil, 1995; Scott & Rockwell, 1997; Tekinarslan, 2008). On the contrary, a few earlier studies found that men showed higher levels of computer anxiety than women (e.g., Brosnan & Lee, 1998; Lever, Sherrod, & Bransford, 1989; Siann, Macleod, Glissov, & Durndell, 1990). Rosen and Maguire (1990) concluded that although some studies have found gender differences, such differences were minimal. Emphasizing the interaction between gender and computer usage, Scott and Rockwell (1997) argue that gender differences on computer anxiety are a function of different levels of computer experience (i.e., the frequency of computer use). Current literature is inconsistent regarding the factors of computer anxiety. In addition, fewer studies have investigated computer anxiety levels among high school students. Chua et al. (1999) reported 15 studies that investigated college students’ computer anxiety levels, whereas, only five studies focused on high school students. A recent search of the literature between 2000 and 2008 on PsycINFO with ‘‘computer anxiety” as a descriptor revealed 47 studies, 35 peer-reviewed articles and 12 dissertations. Of the studies, 44 focused on college students or adult populations whereas, only two studies
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
2641
focused on high school students. Thus, scarcity of the studies on high school populations continues. More limited is multivariate investigation of the construct, even though there seems to be a consensus on the multidimensionality of computer anxiety. Therefore, the purpose of the present study was to investigate the multivariate effects of gender, ownership, and the frequency of computer use on computer anxiety levels, after controlling for the possible effects of trait anxiety among Turkish high school students. 2. Methods 2.1. Participants Seven hundred fifteen Turkish high school students participated in the study. Students’ ages ranged from 14 to 19 years with a mean of 16.18 years (SD = .95). Of the group, 389 (54.4%) were boys and 326 (45.6%) were girls. There were 120 (16.8%) first-graders, 242 (33.8%) sophomores, 293 (41.0%) juniors, and 60 (8.4%) seniors in the group. More than half of the students did not have a PC at home (n = 447, 62.5%). Two hundred thirty seven students (33.2%) reported using computers every day. Half of the students use computers several times in a week (n = 361, 50.6%) and 116 students use computers once a week or less (16.2%). All students in the sample were enrolled in two-hour per-week introduction to computers courses in their high schools. They took courses in computer labs and were learning basic computer knowledge and skills. Courses focused on introductory hardware knowledge and common operating systems. In addition, students were learning basic software applications. They have not had any formal computer course in elementary or middle school prior to enrolling the class in high school. 2.2. Instruments The Computer Anxiety Scale (CAS; Ceyhan & Namlu, 2000) and the State-Trait Anxiety Inventory (STAI; Spielberger, 1983; Öner & LaCompte, 1983) were used to assess computer anxiety and trait anxiety levels, respectively. The CAS is a 28-item, 4-point Likert type self-report instrument that assesses computer anxiety levels with three subscales: Affective Anxiety (13 items), Damaging Anxiety (9 items), and Learning Anxiety (6 items). The Affective Anxiety subscale was developed to measure negative emotions such as fear, worry, and anxiety towards computers and includes statements such as ‘‘I try to avoid computers as much as possible because I do not feel myself close to them.” The Damaging Anxiety subscale is supposed to measure the fear of damaging computers and/or the work being done on computers and includes items such as ‘‘I feel anxious when the computer freezes up.” The Learning Anxiety subscale measures anxiety towards learning computers or computer applications and includes items such as ‘‘I feel anxious that I will not be able to learn how to use computers.” The total scale score is computed by adding responses given to the individual scale items. Higher scores refer to higher anxiety in the respected subscales and the total scale and total scale score ranges from 28 to 112. Psychometric properties of the CAS were studied by Namlu and Ceyhan (2002). Construct validity of the instrument was investigated through a principal component analysis with a varimax rotation, which showed that three components explained 53.0% of the total variance in the CAS items. Of these components, the first had an eigenvalue of 11.03 and accounted for 39.6% of the total variance. The second component (eigenvalue = 2.36) accounted for 8.4% of the total variance. The third component had an eigenvalue of 1.41 and accounted for 5.0% of the total variance. Concurrent validity of the instrument was found by significant correlations between the scores of the CAS and the Computer Attitudes Scale, r = .39, (Deniz, 1994); the State (r = .44)-Trait (r = .43) Anxiety Inventory (Spielberger, 1983); and the Test Anxiety Inventory, r = .32 (Spielberger, 1983). Internal consistency of the total scale was .94: Affective Anxiety = .92, Damaging Anxiety = .89, and Learning Anxiety = .73. In the present study, the total CAS items’ coefficient alpha internal consistency was found to be .92 (Table 1). The STAI, a 40-item, 4-point Likert type instrument, was developed to measure transitory-emotional and relatively stable anxiety reactions (Spielberger, Gorsuch, & Lushene, 1970). The first 20 items in the inventory measure feelings of apprehension, tension, nervousness, and worry about the current situation (A-State) and the last 20 items assess how the respondents feel in general (A-Trait). The later was used in the present study.
2642
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
Table 1 Means, standard deviations, and correlation coefficients of the computer anxiety and A-trait scales Correlation coefficients*
AA Affective Anxiety (AA) Damaging Anxiety (DA) Learning Anxiety (LA) Total CIS Scores (CAS) A-Trait (TA) Alpha Coefficients
.69 .63 .19 .32 .87
DA
.57 .89 .31 .89
LA
.79 .38 .73
CAS
.37 .92
Groups Total group (n = 715)
Boys (n = 389)
Girls (n = 326)
TA
M
SD
M
SD
M
SD
– .91
17.94 14.08 10.70 42.57 43.64 .92
05.43 05.07 03.12 11.93 7.63 .90
17.22 13.11 10.39 40.72 43.49 .93
04.63 04.17 03.03 10.08 7.09
18.48 15.23 11.06 44.77 43.82
06.20 05.76 03.20 13.51 8.25
All correlation coefficients were significant at p < .001.
The validity and the reliability of the STAI have been extensively studied in the literature. Spielberger et al. (1970) found that A-State anxiety scores were significantly higher in the exam condition than in the normal condition. Coefficient alpha reliability of the A-State was found .91 for men and .93 for women; and A-Trait was found .90 for men and .91 for women (Spielberger, 1983). The test-retest reliability studies proved that trait anxiety was relatively stable over time. Spielberger et al. (1970) also found satisfactory test-retest reliability evidence (.65–.86) for intervals up to 3 months. Spielberger (1980) reported that A-Trait scores of 197 university students had .84 (men) and .76 (women) onehour test-retest reliability. A 20-day test–retest reliability was found to be .86 and .76 for men and women, respectively. The STAI was translated and adapted to Turkish (LaCompte & Öner, 1975) and the Turkish form’s validity and reliability properties were found to be adequate (LaCompte & Öner, 1975; Öner & LaCompte, 1983). In the present study, coefficient alpha internal consistency of the ATrait was found to be .91 (Table 1). 2.3. Procedure After permission to use the CAS and the STAI was obtained, a research package including the CAS and A-Trait items, and a set of demographic questions was assembled. Students were contacted during their classes and informed about the study. Participants signed consent forms. All administrations were completed during one class hour, which took approximately 30 min. Students were given extra course credit for participation in the study. The Statistical Procedures for Social Sciences 10.0 (SPSS Inc, 2000) was used to code and analyze the data. Data were screened for the assumptions of parametric statistics. Normality, homogeneity of variances, and linearity assumptions for each cell were tested at univariate, bivariate, and multivariate levels. In order to test the three dependent variables (i.e., Affective Anxiety, Damaging Anxiety, and Learning Anxiety) simultaneously, a between-subjects factorial multivariate analysis of covariance (MANCOVA) was used with a non-orthogonal design. The effects of trait anxiety were controlled by using the A-Trait scores as the covariate. Independent variables were gender (i.e., boys and girls), ownership (i.e., yes and no), and the frequency of computer use (i.e., everyday, several times in a week, and once a week or less). Thus, a 2 X 2 X 3 between-subjects factorial MANCOVA was performed. When omnibus multivariate significance was found, analysis proceeded with Roy–Bargman stepdown Ftests (Tabachnick & Fidell, 2007) and discriminant analysis to assess the importance of the dependent variables and/or the covariate. Adjusted marginal means [M(adj)] were computed and used in the interpretation of significant main effect(s) associated with the dependent variables. 3. Results Preliminary results showed that more boys than girls had a PC at home (n = 158 vs. n = 110); however, this difference was not statistically significant ðv2ð1Þ ¼ 3:46; p < :07Þ. There were significant associations between ownership and the frequency of computer use ðv2ð2Þ ¼ 185:78; p < :001Þ and the
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
2643
frequency of computer use and gender ðv2ð2Þ ¼ 11:06; p < :01Þ. Boys who owned a PC at home used computers most frequently. Means, standard deviations, and Pearson product-moment correlation coefficients for the CAS and A-Trait scores were computed (Table 1). In addition, as suggested by Henson (2000) and Thompson and Vacha-Haase (2000), internal consistency reliability coefficients (alpha) for the scales and subscales used in the study are reported (Table 1). Girls scored higher in the total and subscale of the CAS, and Trait Anxiety scores. Trait Anxiety comparisons showed that boys and girls did not differ significantly (t = .57, p < .57). When the means were divided by the number of items in each subscale, the highest anxiety was on Learning Anxiety and the lowest was on Affective Anxiety both for boys and girls. In addition, all correlation coefficients were statistically significant among the subscales of the computer anxiety measure (Table 1). A 2 X 2 X 3 between-subjects factorial MANCOVA was performed on three dependent variables that are the three dimensions of computer anxiety. Means and standard deviations on Trait Anxiety, and the subscales of the CAS for men and women, ownership and the frequency of use were reported (Table 2). Boys who owned a PC at home and used it everyday scored lowest on the Affective Anxiety, Damaging Anxiety and Learning Anxiety. Before the multivariate analyses, assumptions of parametric statistics (i.e., normality, homogeneity of variances, linearity, and multicollinearity) were tested visually, numerically, and statistically. A non-orthogonal design (i.e., unequal cell sizes) was used: boys (n = 389), girls (n = 326), own a PC (n = 268), do not own a PC (n = 447) with use computers everyday (n = 237), several times in a week (n = 361), and once a week or less (n = 116). SPSS MANOVA (SPSS Inc, 2000) was used to adjust the non-orthogonality problem before the analyses. However, because of the widely discrepant sample sizes in each cell and the homogeneity of dispersion (Box’s M = 272.05, F = 3.88, p < .001), the homogeneity of variance-covariance assumption was violated. A further investigation of the variances for each cell revealed that cells with larger sample sizes produced larger variances. In such situations, the null hypothesis can still be rejected with confidence (Tabachnick & Fidell, 2007). In addition, an overall test of the homogeneity of regression and tests for the homogeneity of regression for MANCOVA stepdown analyses showed that the homogeneity of regression assumption was met for all the dependent variables (p > .05). For multivariate differences, Wilks k was used to test the null hypothesis. Trait Anxiety appeared to have a significant covariate effect on the combined dependent variables [Wilks’ k = .87; F(3,699) = 35.23; p < .001]. There was a moderate association between the covariate and the combined dependent variables (g2 = .13). The results of dependent variable-covariate stepdown analyses indicated that Trait Table 2 Means and standard deviations on the A-trait and CAS subscale scores for gender, ownership, and the frequency of use Gender
Ownership
Frequency of use Everyday
Affective Anxiety
Boys Girls
Damaging Anxiety
Boys Girls
Learning Anxiety
Boys Girls
Trait Anxiety
Boys Girls
Several times a week
Once a week or less
M
SD
M
SD
M
SD
Yes No Yes No
15.45 16.58 15.35 16.71
3.56 4.18 2.78 4.37
16.06 17.99 17.12 18.38
3.65 4.85 4.17 5.08
16.75 20.90 20.17 23.63
3.01 5.32 2.14 9.11
Yes No Yes No
11.24 13.32 12.38 14.11
2.73 3.80 3.25 3.88
11.58 13.82 14.45 15.37
3.45 3.98 4.21 5.66
11.50 17.21 14.17 19.13
2.78 5.68 5.46 7.20
Yes No Yes No
8.86 10.42 9.21 10.57
2.59 2.72 2.34 3.21
9.56 11.24 10.42 11.35
2.82 2.98 2.42 3.08
8.63 12.33 13.17 12.95
2.00 2.82 1.72 3.50
Yes No Yes No
41.47 44.47 41.75 43.29
7.85 5.99 8.30 8.03
42.28 44.45 43.03 44.40
7.07 6.21 7.56 8.55
42.13 45.74 44.17 45.57
9.41 7.67 3.66 8.03
2644
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
Anxiety was related with Affective Anxiety, Damaging Anxiety, and Learning Anxiety (Table 3). When each dependent variable with covariate as independent variables was entered into separate regressions, results showed that the covariate was significantly related to Affective Anxiety (b = .26, t = 7.71, p < .001), Damaging Anxiety (b = .25, t = 7.51, p < .001), and Learning Anxiety (b = .31, t = 9.48, p < .001). Three-way and 2-way interactions were investigated. Analyses did not detect a significant 3-way gender by ownership by the frequency of computer use interaction effect, after adjusting for the effects of Trait Anxiety [Wilks’ k = .99; F(6,1398) = 1.03; p < .40]. Similarly, there was not any significant 2-way interaction of ownership by the frequency of computer use [Wilks’ k = .99; F(6,1398) = 1.44; p < .20], gender by the frequency of computer use [Wilks’ k = .99; F(6,1398) = 1.63; p < .13], or gender by ownership [Wilks’ k = .99; F(6,699) = 2.29; p < .08]. Because 3-way and 2-way interactions were not statistically significant, stepdown analyses for the interaction effects were not computed. After adjusting for the effects of Trait Anxiety, a significant main effect for gender was found [Wilks’ k = .97; F(6,699) = 5.57; p < .001], which indicated significant differences between boys and girls on the dependent variables. However, association between gender and the combined dependent variables was small (g2 = .02). Because omnibus MANCOVA showed a significant main effect, the nature of the relationships between this main effect and the dependent variables was investigated. Univariate F investigations were not appropriate because, conceptually and statistically, the dependent variables were related (Table 1). However, because all the intercorrelations among dependent variables were in excess of .30 (Tabachnick & Fidell, 2007), Roy–Bargman stepdown analyses (Tabachnick & Fidell, 2007) were performed. Results showed that boys and girls significantly differed on Affective Anxiety, Damaging Anxiety, and Learning Anxiety, after adjusting for Trait Anxiety (Table 3). Regression analyses showed that gender was significantly related to Affective Anxiety (b = .64, t = 2.21, p < .03), Damaging Anxiety (b = .91, t = 3.43, p < .001), and Learning Anxiety (b = .55, t = 3.31, p < .001).
Table 3 Test of the covariate, main effects, and interaction effects* Effect
Dependent variables
Roy–Bargman stepdown Fa
df
Sig. of F
Covariate (Trait Anxiety)
Affective Anxiety Damaging Anxiety Learning Anxiety
59.37 12.60 5.11
2/701 2/700 2/699
.001* .001* .001*
GXOXF
Affective Anxiety Damaging Anxiety Learning Anxiety
.16 .10 2.85
2/701 2/700 2/699
.86 .91 .06
OXF
Affective Anxiety Damaging Anxiety Learning Anxiety
1.34 2.48 .52
2/701 2/700 2/699
.27 .09 .59
GXF
Affective Anxiety Damaging Anxiety Learning Anxiety
1.63 1.02 2.25
2/701 2/700 2/699
.20 .36 .11
GXO
Affective Anxiety Damaging Anxiety Learning Anxiety
.01 .37 2.51
2/701 2/700 2/699
.92 .55 .09
Gender (G)
Affective Anxiety Damaging Anxiety Learning Anxiety
4.84 7.20 4.57
2/701 2/700 2/699
.03* .01* .03*
Ownership (O)
Affective Anxiety Damaging Anxiety Learning Anxiety
10.68 14.43 03.22
2/701 2/700 2/699
.001* .001* .07
The frequency of computer use (F)
Affective Anxiety Damaging Anxiety Learning Anxiety
14.00 00.27 01.87
2/701 2/700 2/699
.001* .77 .16
a *
Because dependent variables were correlated univariate Fs were not evaluated. Significant group differences.
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
2645
The effect of gender on the dependent variables was also investigated through discriminant analysis. Structure coefficients showed that Affective Anxiety (.72), Damaging Anxiety (.05), and Learning Anxiety (.38) correlated with the discriminant function. Bonferroni adjusted pairwise comparisons (Hays, 1994) showed that girls (M(adj) = 18.54, SE = .41) scored significantly higher than boys (M(adj) = 17.33, SE = .37) on Affective Anxiety (p < .03). Similarly, on Damaging Anxiety, girls (M(adj) = 14.92, SE = .38) scored significantly higher than boys (M(adj) = 13.15, SE = .34, p < .001). Girls (M(adj) = 11.27, SE = .24) scored significantly higher than boys on Learning Anxiety as well (M(adj) = 10.20, SE = .21, p < .001). A significant main effect for ownership was found, after adjusting for the effects of Trait Anxiety [Wilks’ k = .96; F(6,699) = 9.54; p < .001, g2 = .04]. Stepdown analyses showed that students who owned a PC differed significantly from those who did not on Affective Anxiety and Damaging Anxiety (Table 3). Regression analyses showed that this variable was significantly related to Affective Anxiety (b = 1.11, t = 3.85, p < .001) and Damaging Anxiety (b = 1.47, t = 5.53, p < .001). Structure coefficients showed that Affective Anxiety (.07), Damaging Anxiety (.78), and Learning Anxiety (.42) correlated with the discriminant function. After a Bonferroni adjustment for multiple comparisons, pairwise comparisons showed that students who do not have a PC (M(adj) = 18.84, SE = .27 and M(adj) = 15.32, SE = .25) scored significantly higher than those who had (M(adj) = 17.03, SE = .48 and M(adj) = 12.75, SE = .45) on Affective Anxiety (p < .001) and Damaging Anxiety (p < .001), respectively. Finally, the frequency of computer use showed a significant main effect on the dependent variables, after adjusting for the effects of Trait Anxiety [Wilks’ k = .96; F(6,1398) = 5.33; p < .001, g2 = .02]. However, stepdown analyses showed a significant effect only on Affective Anxiety (Table 3). Regression analyses showed that this main effect was only related to Affective Anxiety (b = 1.89, t = 5.33, p < .001), but not to Damaging Anxiety (b = .22, t = .68, p < .83) or Learning Anxiety (b = .08, t = .421, p < .67). Bonferroni adjusted pairwise comparisons showed that students who used computers everyday showed the lowest affective anxiety (M(adj) = 16.18, SE = .35) followed by those who used computers several times in a week (M(adj) = 17.40, SE = .30). Students who used computers once a week or less showed the highest levels of affective anxiety (M(adj) = 20.21, SE = .69). 4. Discussion Studies that address the problems associated with and correlates of computer anxiety are abundant; however, a review of the literature on the topic shows inconsistent results with regard to gender effects. Moreover, fewer studies have focused on high school students, even though they may experience high levels of computer anxiety (Loyd & Gressard, 1984). Lastly, the multivariate nature of the construct of computer anxiety has not been adequately integrated into current literature. Therefore, there is still need for clarification in the area of computer anxiety and the present study was an attempt in this direction. Literature suggests that computer anxiety is a multidimensional construct (e.g., Beckers & Schmidt, 2001; Beckers et al., 2007; Deane, Henderson, Barrelle, Saliba, & Mahar, 1995). Thus, its multivariate nature should be taken into account when designing research studies on computer anxiety. Nonetheless, the review of the literature showed that most computer anxiety research disregarded the issue and usually investigations have been done either on a single computer anxiety scale score or without appropriate adjustments for multiple comparisons. In the present study, a linear combination of the three dimensions of computer anxiety (i.e., Affective Anxiety, Damaging Anxiety, and Learning Anxiety) and their interactions were investigated in multivariate context. We studied three of the commonly investigated variables (i.e., gender, ownership, and the frequency of computer use) in computer anxiety research. We looked at the multivariate effects of gender, ownership, and the frequency of computer use on computer anxiety levels, after controlling for the effects of trait anxiety. Results showed that 37.5% of the high school students had a PC at home and that 83.8% used computers everyday or several times in a week. Therefore, we conclude that the sample of the study had accessibility and familiarity with computers. When these findings are compared with pervious studies conducted in Turkey, interesting results emerged. For example, Namlu and Ceyhan (2002) reported that 19.8% Turkish college students had a PC at home and 52.7% used computers everyday or several times in a week. In a more recent study, Tekinarslan (2008) found increased ownership (47.95%)
2646
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
among college students. Both studies concur with the present study that students have accessibility and familiarity with computers. In addition, the present study is in agreement with those of Panagiotakopoulos and Koustourakis (2001), Namlu and Ceyhan (2002) and Tekinarslan (2008) that computer anxiety levels differ according to ownership and frequency of use. Even though they did not differ statistically, more boys than girls had a PC at home (n = 158 vs. n = 110, respectively). These results are similar to those found by Selwyn (1997) and Karavidas, Lim, and Katsikas (2005). Because we did not inquire about the sibling status in the family, it is not clear from these results whether parents prefer to buy a PC if the child is a boy. This point should be investigated in future studies. Contrary to Selwyn, we found significant association between ownership and the frequency of computer use. However, multivariate analysis showed no significant interaction effect between the two variables on computer anxiety levels. Girls and boys did not differ on stable anxiety reactions (i.e., trait anxiety). It can be argued that individuals who are anxious in general might show higher levels of computer anxiety as well. Most recently, Beckers et al. (2007) found that computer anxiety was more strongly related to trait anxiety. In order to control such an effect, if any, A-Trait scores are statistically adjusted for by using them as covariate. It was found that trait anxiety had a significant covariate effect on all three dependent variables. Future researchers should take the information into account when studying computer anxiety among different populations. We suggest using trait anxiety scores as covariate before group comparisons on computer anxiety levels. Gender is one of the most studied variables in computer anxiety research; but, gender-related research has not reached a consensus. A review of the literature on the topic shows inconsistent results with regard to gender effects. There may be several possible explanations as to why the literature shows inconsistent effects of gender. First, gender effects on computer anxiety might be sampledependent. For example, high school girls and boys might experience computer anxiety differently from college men and women. This is one of the reasons why we focused on high school students and their computer anxiety experiences. Second, gender might interact with other variable(s) in affecting computer anxiety levels. For example, boys who own a PC might face different aspects of computer anxiety than girls who do not own a PC. This is another reason why the present study attempted to integrate several aspects of computer anxiety in relation to gender. Third, time might be a significant variable in relation to computer anxiety and gender investigations. For example, nowadays men (or women) might be experiencing anxiety differently than men (or women) who lived some time ago. We attempted to study contemporary gender differences, especially in an era when computers are more commonly integrated into our daily lives. Gender effects found in the present study support studies that found girls have higher levels of computer anxiety compared to boys (i.e., Karavidas et al., 2005; Mcilroy et al., 2001; Namlu & Ceyhan, 2002; Palaigeorgiou, Siozos, Konstantakis, & Tsoukalas, 2005; Rosen & Weil, 1995; Todman, 2000). In all three dimensions (i.e., Affective Anxiety, Damaging Anxiety, and Learning Anxiety) girls outscored boys regardless of their trait anxiety levels. However, no interaction between gender and ownership or the frequency of computer use on computer anxiety levels was found. We could not find any study that investigated such interactions in the literature to compare these results. Future studies should focus on complex interactions similar to the ones studied in the present investigation. It can be argued that factors such as owning a PC at home or the frequency of computer use might confound computer anxiety (e.g., Brosnan 1999; Chua et al., 1999; Keser, 2001; Rosen & Weil, 1995; Scott & Rockwell, 1997; Tuluk, 1999). For example, Teo (2006) found that students who owned a PC at home showed lower computer anxiety levels compared to those who did not. The present results support this argument in general. Both ownership and the frequency of computer use had significant effects on different aspects of computer anxiety. However, complex relationships were detected. For example, ownership had a significant effect on affective anxiety and damaging anxiety, but not on learning anxiety. Therefore, we conclude that students who do not have a PC at home fear most damaging computers; but they do not differ when it comes to learning computers or computer-relevant information. Similarly, the frequency of computer use did not have any significant effect on damaging anxiety or learning anxiety. On the contrary, it did have a significant effect on affective anxiety. In sum, these results make it clear that the effect of any single variable on computer anxiety should be studied in relation with other relevant variables. Computer anxiety seems to be affected by multiple variables
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
2647
and univariate investigations of the variables will not convey the most accurate picture. Appropriate interactions should be taken into consideration when studying computer anxiety. Researchers predict that computing and computers will be more involved in students’ lives (Mcilroy et al., 2001). Therefore, information provided in the present study should particularly be of great value for students and computer instructors alike. Computer instructors should be aware that students who have limited access to or lesser use of computers might experience more difficulties. These difficulties might be present in the form of affective anxiety, fear of damaging computers, or learning computers. One shortcoming of the present study is that it was limited with its sample characteristics. The sample of the study included high school students who were enrolled in classes in northern Turkey. In addition, the study is limited with the quantitative data on the frequency of computer use. Qualitative data such as for what purposes students use computers (e.g., programming, games-playing, surfing the Internet, or communication) were not sought. For example, some students might use computers everyday for game playing, whereas, others use once a week for complex programming. This type of information might be useful in studying computer anxiety because the literature indicates that task-relevant use of computers and anxiety might interact (Humphreys & Revelle, 1984). Finally, the present study was not an experimental design; therefore, the variables significantly related to computer anxiety are not the ‘‘causes” of computer anxiety. We cannot infer, for example, that having a PC at home ‘‘causes” less affective or damaging anxiety. We can simply infer that there is a significant difference on affective and damaging anxiety and students who own a PC at home show lesser levels of anxiety. Causality cannot be determined as a result of statistical analyses. It can only be determined as a result of design or theory (Trochim & Donnelly, 2006). Therefore, experimental studies are needed to identify the causes of computer anxiety, if that is the purpose. Acknowledgements The authors express their gratitude to two anonymous reviewers, whose comments and suggestions led to a more comprehensive paper. This study is supported by the Scientific and Technical Research Council of Turkey. References Abdelhamid, I. S. (2002). Attitudes towards computers: A study of gender differences and other variables. Journal of the Social Sciences, 30, 285–316. Anderson, A. A. (1996). Predictors of computer anxiety and performance in information systems. Computers in Human Behavior, 12, 61–77. Arıkan, D. (2002). Sınıf ög˘retmeni adaylarının bilgisayara yönelik tutumları, bilgisayar kaygı düzeyleri ve bilgisayar dersine ilisßkin deg˘erlendirmeleri [Attitudes toward computers, computer anxiety, views of a computer course]. Unpublished master’s _ thesis, Dokuz Eylül University, Izmir, Turkey. Beckers, J. J., & Schmidt, H. G. (2001). The structure of computer anxiety: A six-factor model. Computers in Human Behavior, 17, 35–49. Beckers, J. J., Wicherts, J. M., & Schmidt, H. G. (2007). Computer anxiety: Trait or state. Computers in Human Behavior, 23, 2851–2862. Bohlin, R. M., & Hunt, N. P. (1995). Course structure effects on students’ computer anxiety. Journal of Educational Computing Research, 13(3), 263–270. Bovee, C., Voogt, J., & Meelissen, M. (2007). Computer attitudes of primary and secondary students in South Africa. Computers in Human Behavior, 23, 1762–1776. Bozionelos, N. (2001). Computer anxiety: Relationships with computer experience and prevalence. Computers in Human Behavior, 17, 213–224. Brosnan, M. J. (1999). Modeling technophobia: A case for word processing. Computers in Human Behavior, 15, 105–121. Brosnan, M. J., & Lee, W. (1998). A cross-cultural comparison of gender differences in computer attitudes and anxieties: The United Kingdom and Hong Kong. Computers in Human Behavior, 14, 559–577. Ceyhan, E., & Namlu, A. G. (2000). Bilgisayar Kaygısı Ölcßeg˘i: Gecßerlik ve güvenirlik cßalısßması [The computer anxiety scale: Validity and reliability]. Anadolu Üniversitesi Eg˘itim Fakültesi Dergisi, 10(2), 77–93. Chou, C. (2003). Incidences and correlates of Internet anxiety among high school teachers in Taiwan. Computers in Human Behavior, 19, 731–749. Chua, S. L., Chen, D. T., & Wong, F. L. (1999). Computer anxiety and its correlates: A meta analysis. Computers in Human Behavior, 15, 609–623. Colley, A. M., Gale, M. T., & Harris, T. A. (1994). Effects of gender role identity and experience on computer attitude components. Journal of Educational Computing Research, 10(2), 129–137.
2648
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
Deane, F., Henderson, R., Barrelle, K., Saliba, A., & Mahar, D. (1995). Construct validity of computer anxiety measured by the computer attitudes scale. In Y. Anzai, K. Ogawa, & H. Mori (Eds.), Advances in human factors/ergonomics 20A symbiosis of human and artifact (pp. 581–586). Amsterdam: Elsevier Science. Deniz, L. (1994). Bilgisayar Tutum Ölcßeg˘i’nin gecßerlik, güvenirlik ve norm cßalısßması ve örnek bir uygulama [Development of the attitudes toward computers: A preliminary study]. Unpublished doctoral dissertation, Marmara University, Istanbul, Turkey. Dupin-Bryant, P. (2002). Reducing computer anxiety in adults learning to use microcomputers. Journal of Extension, 40(5). Available from Retrieved 17.10.06. Durndell, A., & Haag, Z. (2002). Computer self-efficacy, computer anxiety, attitudes towards the Internet and reported experience with the Internet, by gender, in an East European sample. Computers in Human Behavior, 18, 521–535. Hays, W. L. (1994). Statistics (5th ed.). Austin, TX: Harcourt Brace College Publishers. Henson, R.K. (2000). A primer on coefficient alpha. In: Paper presented at the annual meeting of the American Educational Research Association, San Francisco, CA. (ERIC Document Reproduction Service No. ED314483). Humphreys, M. S., & Revelle, W. (1984). Personality, motivation, and performance: A theory of the relationship between individual differences and information processing. Psychological Review, 91, 153–184. Imhof, M., Vollmeyer, R., & Beierlein, C. (2007). Computer use and the gender gap: The issue of access, use, motivation, and performance. Computers in Human Behavior, 23, 2823–2837. Karavidas, M., Lim, N. K., & Katsikas, S. L. (2005). The effects of computers on older adults. Computers in Human Behavior, 21, 697–711. Keser, H. (2001). Yönetici adaylarının teknolojiye yönelik tutumları [Attitudes towards technology among candidate administrators]. Ankara: MEB Publishing. LaCompte, W.A., & Öner, N. (1975). Durumluk Kaygı-Sürekli Kaygı Envanterinin Türkcße’ye adaptasyonu ve standardizasyonu ile ilgili cßalısßma [A study of the state-trait anxiety inventory: Adaptation to Turkish]. XI. In: The annual meeting of the Nervepsychiatry Congress. Lever, S., Sherrod, K. B., & Bransford, J. (1989). The effects of logo instruction on elementary students’ attitudes toward computers and school. Computers in Schools, 6, 45–58. Loyd, B.H., & Gressard, C. (1984). The effects of sex, age, and computer experience on computer attitudes. In: Paper presented at the Eastern Educational Research Association annual meeting. Maurer, M. M. (1983). Development and validation of a measure of computer anxiety. Unpublished master’s thesis, Iowa State University, Ames, IA. Mcilroy, D., Bunting, B., Tierney, K., & Gordon, M. (2001). The relation of gender and background experience to self-reported computing anxieties and cognition. Computers in Human Behavior, 17, 21–33. Namlu, A., & Ceyhan, E. (2002). Bilgisayar kaygısı: Üniversite ög˘rencileri üzerinde bir cßalısßma [Computer anxiety: A study on university students]. Eskisehir: Anadolu University Publishing. North, A. S., & Noyes, J. M. (2002). Gender influences on children’s computer attitudes and cognitions. Computers in Human Behavior, 18, 135–150. Öner, N., & LaCompte, A. (1983). Durumluk-Sürekli Kaygı Envanteri El Kitabı [The state-trait anxiety inventory manual]. Istanbul: Bogazici University Press. Palaigeorgiou, G. E., Siozos, P. D., Konstantakis, N. I., & Tsoukalas, I. A. (2005). A computer attitudes scale for computer science freshmen and its educational implications. Journal of Computer Assisted Learning, 21, 330–342. Panagiotakopoulos, C., & Koustourakis, G. (2001). Dealing with 1st year university students’ computer anxiety. Themes in Education, 2(1), 35–46. Rosen, L. D., & Maguire, P. (1990). Myth and realities of computerphobia: A meta-analysis. Anxiety Research, 3, 175–191. Rosen, L. D., Sears, D. C., & Weil, M. M. (1987). Behavior research methods, instruments, and computers. Computerphobia, 19(2), 167–179. Rosen, L. D., & Weil, M. M. (1995). Computer anxiety: A cross-cultural comparison of university students in ten countries. Computers in Human Behavior, 11(1), 9–31. Scott, C. R., & Rockwell, S. C. (1997). The effect of communication, writing and technology apprehension on the likelihood to use new communication technologies. Communication Education, 46(1), 44–62. Selwyn, N. (1997). The effect of using computer at home on students’ school use of it. Research in Education, 58, 79–81. Siann, G., Macleod, J., Glissov, P., & Durndell, A. (1990). The effect of computer use on gender differences in attitudes to computers. Computers in Education, 14, 183–191. Spielberger, C. D. (1980). Test anxiety inventory. CA: Mind Garden Inc. Spielberger, C. D. (1983). State-trait anxiety inventory. CA: Mind Garden Inc. Spielberger, C. D., Gorsuch, R. L., & Lushene, R. E. (1970). Manual for the state-trait anxiety inventory. Palo Alto, California: Consulting Psychologists Press. SPSS Inc. (2000). SPSS for Windows User’s Guide (Version 10.0) [Computer software]. Chicago, IL: SPSS Inc. Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Needham Heights, MA: Springer. Tekinarslan, E. (2008). Computer anxiety: A cross-cultural comparative study of Dutch and Turkish university students. Computers in Human Behavior, 24, 1572–1584. Teo, T. (2006). Attitudes toward computers: A study of post-secondary students in Singapore. Interactive Learning Environments, 14, 17–24. Thompson, B., & Vacha-Haase, T. (2000). Psychometrics is datametrics: The test is not reliable. Educational and Psychological Measurement, 60, 174–195. Todman, J. (2000). Gender differences in computer anxiety among university entrants since 1992. Computers & Education, 34, 27–35. Trochim, W., & Donnelly, J. P. (2006). The research methods knowledge base (3rd ed.). Atomicdog Publishing. Tuluk, G. (1999). Bilgisayar destekli matematik ög˘retiminin ög˘renci üzerindeki etkisi [The effects of computer-based mathematics teaching on students]. In: Proceedings of the 3rd National physical Sciences Meeting. Üstündag˘, N. (2001). Müfredat labaratuar okullarında görev yapan yönetici ve ög˘retmenlerin bilgisayar tutumları ile kaygı düzeyleri arasındaki ilisßkinin incelenmesi [Investigating the relationship between computer attitudes and anxiety levels of principals _ and teachers]. Unpublished master’s thesis, Marmara University, Istanbul.
Contents lists available at ScienceDirect
Computers in Human Behavior journal homepage: www.elsevier.com/locate/comphumbeh
Multivariate effects of gender, ownership, and the frequency of use on computer anxiety among high school students Mustafa Balog˘lu *, Vildan Çevik 1 Gaziosmanpasa University, College of Education, Department of Educational Sciences, Tokat, Turkey
a r t i c l e
i n f o
Article history: Available online 28 April 2008
Keywords: Computer anxiety High school students Ownership Frequency of use
a b s t r a c t Studies that address the problems associated with computer anxiety are abundant; however, fewer studies took into account multivariate nature of the construct. Moreover, studies focusing on high school students are even more limited. Thus, the present study investigated the multivariate effects of gender, ownership, and the frequency of computer use on computer anxiety levels, after controlling for the possible effects of trait anxiety among 715 Turkish high school students. The Computer Anxiety Scale and the State-Trait Anxiety Inventory were used to assess computer anxiety and state and trait anxiety levels, respectively. A 2 X 2 X 3 between-subjects factorial multivariate analysis of covariance was used on three dependent variables that are the three dimensions of computer anxiety: Affective Anxiety, Damaging Anxiety, and Learning Anxiety. Independent variables are gender, ownership (i.e., yes or no), and the frequency of computer use (i.e., everyday, several times a week, or once a week or less). Results showed a significant covariate effect of trait anxiety, significant main effects for gender, ownership, and the frequency of computer use on the dependent variables. No three-way or two-way interaction was detected. After the descriptive and comparative analyses, specific suggestions were provided based on the results. Ó 2008 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: +90 356 2521616/3415; fax: +90 356 2521546. E-mail addresses: [email protected], [email protected] (M. Balog˘lu), [email protected] (V. Çevik). 1 Tel.: +90 356 2521616/3415; fax: +90 356 2521546. 0747-5632/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.chb.2008.03.003
2640
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
1. Introduction With the more extensive use in all levels of education (Durndell & Haag, 2002; Imhof, Vollmeyer, & Beierlein, 2007), issues related to computers are more focus of interest among researchers. It is predicted that the higher number of people use computers in their daily lives, the more will face difficulties with them (Beckers & Schmidt, 2001). In this context, the construct of computer anxiety has been studied since the beginnings of the 80’s, mostly focusing on the non-cognitive factors such as attitudes, previous experiences with computers, trait anxiety, or personality-related factors. Even though the construct of computer anxiety has been studied for an extended period of time, there is still no consensus in the literature regarding its definition (Beckers, Wicherts, & Schmidt, 2007). However, according to a commonly used definition, computer anxiety is the fear and apprehension felt by an individual when considering the utilization of computer technology or when actually using computers (Maurer, 1983). A review of the literature showed that most definitions of computer anxiety include a fear component (Chua, Chen, & Wong, 1999). Computerphobia (Rosen, Sears, & Weil, 1987), computer apprehension (Anderson, 1996), computer resistance (Bohlin & Hunt, 1995), or technophobia (Brosnan, 1999) are some of the other terms used interchangeably with computer anxiety. Regardless of the term, there is evidence that a large number of students hold negative attitudes towards and experience high levels of computer anxiety (e.g., North & Noyes, 2002). Bozionelos (2001) found that as high as 40–50% of students experience computer anxiety. More specifically, there is evidence that high school students may experience higher levels of computer anxiety than college students (Loyd & Gressard, 1984). Recently, computer anxiety has been conceptualized as a multi-dimensional construct, including psychological, operational, and sociological components (e.g., Beckers & Schmidt, 2001; Beckers et al., 2007). Owning a personal computer (PC) at home (ownership) and the frequency of computer use were found to be two of the operational components of computer anxiety (e.g., Brosnan, 1999; Keser, 2001; Rosen & Weil, 1995; Üstündag˘, 2001). Arıkan (2002) found that those individuals who owned a PC at home showed less anxiety related to computers. Similarly, Chua et al. (1999) and Chou (2003) found that computer usage was negatively related to computer anxiety. In a relatively earlier study, Selwyn (1997) found that, out of 530 students, 68.5% had a PC at home; but there was no relationship between ownership and frequency of computer use. With a group of Turkish college students, Namlu and Ceyhan (2002) found that 19.8% of the students had a PC at home and those who had a PC showed significantly lower levels of computer anxiety. In addition, they found that as the frequency of computer use increased, computer anxiety levels decreased. This meant that students who used computers more often showed lower levels of computer anxiety compared to those who used computers less frequently. Gender may be the single most widely investigated sociological variable in computer anxiety research (Mcilroy, Bunting, Tierney, & Gordon, 2001). Generally speaking, men have more positive attitudes toward computers than women (Bovee, Voogt, Meelissen, 2007; Dupin-Bryant, 2002). In addition, women show higher levels of computer anxiety than men (e.g., Abdelhamid, 2002; Durndell & Haag, 2002; Mcilroy et al., 2001; Todman, 2000). However, other studies failed to find significant differences between men and women on computer anxiety levels (e.g., Colley, Gale, & Harris, 1994; Rosen & Weil, 1995; Scott & Rockwell, 1997; Tekinarslan, 2008). On the contrary, a few earlier studies found that men showed higher levels of computer anxiety than women (e.g., Brosnan & Lee, 1998; Lever, Sherrod, & Bransford, 1989; Siann, Macleod, Glissov, & Durndell, 1990). Rosen and Maguire (1990) concluded that although some studies have found gender differences, such differences were minimal. Emphasizing the interaction between gender and computer usage, Scott and Rockwell (1997) argue that gender differences on computer anxiety are a function of different levels of computer experience (i.e., the frequency of computer use). Current literature is inconsistent regarding the factors of computer anxiety. In addition, fewer studies have investigated computer anxiety levels among high school students. Chua et al. (1999) reported 15 studies that investigated college students’ computer anxiety levels, whereas, only five studies focused on high school students. A recent search of the literature between 2000 and 2008 on PsycINFO with ‘‘computer anxiety” as a descriptor revealed 47 studies, 35 peer-reviewed articles and 12 dissertations. Of the studies, 44 focused on college students or adult populations whereas, only two studies
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
2641
focused on high school students. Thus, scarcity of the studies on high school populations continues. More limited is multivariate investigation of the construct, even though there seems to be a consensus on the multidimensionality of computer anxiety. Therefore, the purpose of the present study was to investigate the multivariate effects of gender, ownership, and the frequency of computer use on computer anxiety levels, after controlling for the possible effects of trait anxiety among Turkish high school students. 2. Methods 2.1. Participants Seven hundred fifteen Turkish high school students participated in the study. Students’ ages ranged from 14 to 19 years with a mean of 16.18 years (SD = .95). Of the group, 389 (54.4%) were boys and 326 (45.6%) were girls. There were 120 (16.8%) first-graders, 242 (33.8%) sophomores, 293 (41.0%) juniors, and 60 (8.4%) seniors in the group. More than half of the students did not have a PC at home (n = 447, 62.5%). Two hundred thirty seven students (33.2%) reported using computers every day. Half of the students use computers several times in a week (n = 361, 50.6%) and 116 students use computers once a week or less (16.2%). All students in the sample were enrolled in two-hour per-week introduction to computers courses in their high schools. They took courses in computer labs and were learning basic computer knowledge and skills. Courses focused on introductory hardware knowledge and common operating systems. In addition, students were learning basic software applications. They have not had any formal computer course in elementary or middle school prior to enrolling the class in high school. 2.2. Instruments The Computer Anxiety Scale (CAS; Ceyhan & Namlu, 2000) and the State-Trait Anxiety Inventory (STAI; Spielberger, 1983; Öner & LaCompte, 1983) were used to assess computer anxiety and trait anxiety levels, respectively. The CAS is a 28-item, 4-point Likert type self-report instrument that assesses computer anxiety levels with three subscales: Affective Anxiety (13 items), Damaging Anxiety (9 items), and Learning Anxiety (6 items). The Affective Anxiety subscale was developed to measure negative emotions such as fear, worry, and anxiety towards computers and includes statements such as ‘‘I try to avoid computers as much as possible because I do not feel myself close to them.” The Damaging Anxiety subscale is supposed to measure the fear of damaging computers and/or the work being done on computers and includes items such as ‘‘I feel anxious when the computer freezes up.” The Learning Anxiety subscale measures anxiety towards learning computers or computer applications and includes items such as ‘‘I feel anxious that I will not be able to learn how to use computers.” The total scale score is computed by adding responses given to the individual scale items. Higher scores refer to higher anxiety in the respected subscales and the total scale and total scale score ranges from 28 to 112. Psychometric properties of the CAS were studied by Namlu and Ceyhan (2002). Construct validity of the instrument was investigated through a principal component analysis with a varimax rotation, which showed that three components explained 53.0% of the total variance in the CAS items. Of these components, the first had an eigenvalue of 11.03 and accounted for 39.6% of the total variance. The second component (eigenvalue = 2.36) accounted for 8.4% of the total variance. The third component had an eigenvalue of 1.41 and accounted for 5.0% of the total variance. Concurrent validity of the instrument was found by significant correlations between the scores of the CAS and the Computer Attitudes Scale, r = .39, (Deniz, 1994); the State (r = .44)-Trait (r = .43) Anxiety Inventory (Spielberger, 1983); and the Test Anxiety Inventory, r = .32 (Spielberger, 1983). Internal consistency of the total scale was .94: Affective Anxiety = .92, Damaging Anxiety = .89, and Learning Anxiety = .73. In the present study, the total CAS items’ coefficient alpha internal consistency was found to be .92 (Table 1). The STAI, a 40-item, 4-point Likert type instrument, was developed to measure transitory-emotional and relatively stable anxiety reactions (Spielberger, Gorsuch, & Lushene, 1970). The first 20 items in the inventory measure feelings of apprehension, tension, nervousness, and worry about the current situation (A-State) and the last 20 items assess how the respondents feel in general (A-Trait). The later was used in the present study.
2642
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
Table 1 Means, standard deviations, and correlation coefficients of the computer anxiety and A-trait scales Correlation coefficients*
AA Affective Anxiety (AA) Damaging Anxiety (DA) Learning Anxiety (LA) Total CIS Scores (CAS) A-Trait (TA) Alpha Coefficients
.69 .63 .19 .32 .87
DA
.57 .89 .31 .89
LA
.79 .38 .73
CAS
.37 .92
Groups Total group (n = 715)
Boys (n = 389)
Girls (n = 326)
TA
M
SD
M
SD
M
SD
– .91
17.94 14.08 10.70 42.57 43.64 .92
05.43 05.07 03.12 11.93 7.63 .90
17.22 13.11 10.39 40.72 43.49 .93
04.63 04.17 03.03 10.08 7.09
18.48 15.23 11.06 44.77 43.82
06.20 05.76 03.20 13.51 8.25
All correlation coefficients were significant at p < .001.
The validity and the reliability of the STAI have been extensively studied in the literature. Spielberger et al. (1970) found that A-State anxiety scores were significantly higher in the exam condition than in the normal condition. Coefficient alpha reliability of the A-State was found .91 for men and .93 for women; and A-Trait was found .90 for men and .91 for women (Spielberger, 1983). The test-retest reliability studies proved that trait anxiety was relatively stable over time. Spielberger et al. (1970) also found satisfactory test-retest reliability evidence (.65–.86) for intervals up to 3 months. Spielberger (1980) reported that A-Trait scores of 197 university students had .84 (men) and .76 (women) onehour test-retest reliability. A 20-day test–retest reliability was found to be .86 and .76 for men and women, respectively. The STAI was translated and adapted to Turkish (LaCompte & Öner, 1975) and the Turkish form’s validity and reliability properties were found to be adequate (LaCompte & Öner, 1975; Öner & LaCompte, 1983). In the present study, coefficient alpha internal consistency of the ATrait was found to be .91 (Table 1). 2.3. Procedure After permission to use the CAS and the STAI was obtained, a research package including the CAS and A-Trait items, and a set of demographic questions was assembled. Students were contacted during their classes and informed about the study. Participants signed consent forms. All administrations were completed during one class hour, which took approximately 30 min. Students were given extra course credit for participation in the study. The Statistical Procedures for Social Sciences 10.0 (SPSS Inc, 2000) was used to code and analyze the data. Data were screened for the assumptions of parametric statistics. Normality, homogeneity of variances, and linearity assumptions for each cell were tested at univariate, bivariate, and multivariate levels. In order to test the three dependent variables (i.e., Affective Anxiety, Damaging Anxiety, and Learning Anxiety) simultaneously, a between-subjects factorial multivariate analysis of covariance (MANCOVA) was used with a non-orthogonal design. The effects of trait anxiety were controlled by using the A-Trait scores as the covariate. Independent variables were gender (i.e., boys and girls), ownership (i.e., yes and no), and the frequency of computer use (i.e., everyday, several times in a week, and once a week or less). Thus, a 2 X 2 X 3 between-subjects factorial MANCOVA was performed. When omnibus multivariate significance was found, analysis proceeded with Roy–Bargman stepdown Ftests (Tabachnick & Fidell, 2007) and discriminant analysis to assess the importance of the dependent variables and/or the covariate. Adjusted marginal means [M(adj)] were computed and used in the interpretation of significant main effect(s) associated with the dependent variables. 3. Results Preliminary results showed that more boys than girls had a PC at home (n = 158 vs. n = 110); however, this difference was not statistically significant ðv2ð1Þ ¼ 3:46; p < :07Þ. There were significant associations between ownership and the frequency of computer use ðv2ð2Þ ¼ 185:78; p < :001Þ and the
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
2643
frequency of computer use and gender ðv2ð2Þ ¼ 11:06; p < :01Þ. Boys who owned a PC at home used computers most frequently. Means, standard deviations, and Pearson product-moment correlation coefficients for the CAS and A-Trait scores were computed (Table 1). In addition, as suggested by Henson (2000) and Thompson and Vacha-Haase (2000), internal consistency reliability coefficients (alpha) for the scales and subscales used in the study are reported (Table 1). Girls scored higher in the total and subscale of the CAS, and Trait Anxiety scores. Trait Anxiety comparisons showed that boys and girls did not differ significantly (t = .57, p < .57). When the means were divided by the number of items in each subscale, the highest anxiety was on Learning Anxiety and the lowest was on Affective Anxiety both for boys and girls. In addition, all correlation coefficients were statistically significant among the subscales of the computer anxiety measure (Table 1). A 2 X 2 X 3 between-subjects factorial MANCOVA was performed on three dependent variables that are the three dimensions of computer anxiety. Means and standard deviations on Trait Anxiety, and the subscales of the CAS for men and women, ownership and the frequency of use were reported (Table 2). Boys who owned a PC at home and used it everyday scored lowest on the Affective Anxiety, Damaging Anxiety and Learning Anxiety. Before the multivariate analyses, assumptions of parametric statistics (i.e., normality, homogeneity of variances, linearity, and multicollinearity) were tested visually, numerically, and statistically. A non-orthogonal design (i.e., unequal cell sizes) was used: boys (n = 389), girls (n = 326), own a PC (n = 268), do not own a PC (n = 447) with use computers everyday (n = 237), several times in a week (n = 361), and once a week or less (n = 116). SPSS MANOVA (SPSS Inc, 2000) was used to adjust the non-orthogonality problem before the analyses. However, because of the widely discrepant sample sizes in each cell and the homogeneity of dispersion (Box’s M = 272.05, F = 3.88, p < .001), the homogeneity of variance-covariance assumption was violated. A further investigation of the variances for each cell revealed that cells with larger sample sizes produced larger variances. In such situations, the null hypothesis can still be rejected with confidence (Tabachnick & Fidell, 2007). In addition, an overall test of the homogeneity of regression and tests for the homogeneity of regression for MANCOVA stepdown analyses showed that the homogeneity of regression assumption was met for all the dependent variables (p > .05). For multivariate differences, Wilks k was used to test the null hypothesis. Trait Anxiety appeared to have a significant covariate effect on the combined dependent variables [Wilks’ k = .87; F(3,699) = 35.23; p < .001]. There was a moderate association between the covariate and the combined dependent variables (g2 = .13). The results of dependent variable-covariate stepdown analyses indicated that Trait Table 2 Means and standard deviations on the A-trait and CAS subscale scores for gender, ownership, and the frequency of use Gender
Ownership
Frequency of use Everyday
Affective Anxiety
Boys Girls
Damaging Anxiety
Boys Girls
Learning Anxiety
Boys Girls
Trait Anxiety
Boys Girls
Several times a week
Once a week or less
M
SD
M
SD
M
SD
Yes No Yes No
15.45 16.58 15.35 16.71
3.56 4.18 2.78 4.37
16.06 17.99 17.12 18.38
3.65 4.85 4.17 5.08
16.75 20.90 20.17 23.63
3.01 5.32 2.14 9.11
Yes No Yes No
11.24 13.32 12.38 14.11
2.73 3.80 3.25 3.88
11.58 13.82 14.45 15.37
3.45 3.98 4.21 5.66
11.50 17.21 14.17 19.13
2.78 5.68 5.46 7.20
Yes No Yes No
8.86 10.42 9.21 10.57
2.59 2.72 2.34 3.21
9.56 11.24 10.42 11.35
2.82 2.98 2.42 3.08
8.63 12.33 13.17 12.95
2.00 2.82 1.72 3.50
Yes No Yes No
41.47 44.47 41.75 43.29
7.85 5.99 8.30 8.03
42.28 44.45 43.03 44.40
7.07 6.21 7.56 8.55
42.13 45.74 44.17 45.57
9.41 7.67 3.66 8.03
2644
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
Anxiety was related with Affective Anxiety, Damaging Anxiety, and Learning Anxiety (Table 3). When each dependent variable with covariate as independent variables was entered into separate regressions, results showed that the covariate was significantly related to Affective Anxiety (b = .26, t = 7.71, p < .001), Damaging Anxiety (b = .25, t = 7.51, p < .001), and Learning Anxiety (b = .31, t = 9.48, p < .001). Three-way and 2-way interactions were investigated. Analyses did not detect a significant 3-way gender by ownership by the frequency of computer use interaction effect, after adjusting for the effects of Trait Anxiety [Wilks’ k = .99; F(6,1398) = 1.03; p < .40]. Similarly, there was not any significant 2-way interaction of ownership by the frequency of computer use [Wilks’ k = .99; F(6,1398) = 1.44; p < .20], gender by the frequency of computer use [Wilks’ k = .99; F(6,1398) = 1.63; p < .13], or gender by ownership [Wilks’ k = .99; F(6,699) = 2.29; p < .08]. Because 3-way and 2-way interactions were not statistically significant, stepdown analyses for the interaction effects were not computed. After adjusting for the effects of Trait Anxiety, a significant main effect for gender was found [Wilks’ k = .97; F(6,699) = 5.57; p < .001], which indicated significant differences between boys and girls on the dependent variables. However, association between gender and the combined dependent variables was small (g2 = .02). Because omnibus MANCOVA showed a significant main effect, the nature of the relationships between this main effect and the dependent variables was investigated. Univariate F investigations were not appropriate because, conceptually and statistically, the dependent variables were related (Table 1). However, because all the intercorrelations among dependent variables were in excess of .30 (Tabachnick & Fidell, 2007), Roy–Bargman stepdown analyses (Tabachnick & Fidell, 2007) were performed. Results showed that boys and girls significantly differed on Affective Anxiety, Damaging Anxiety, and Learning Anxiety, after adjusting for Trait Anxiety (Table 3). Regression analyses showed that gender was significantly related to Affective Anxiety (b = .64, t = 2.21, p < .03), Damaging Anxiety (b = .91, t = 3.43, p < .001), and Learning Anxiety (b = .55, t = 3.31, p < .001).
Table 3 Test of the covariate, main effects, and interaction effects* Effect
Dependent variables
Roy–Bargman stepdown Fa
df
Sig. of F
Covariate (Trait Anxiety)
Affective Anxiety Damaging Anxiety Learning Anxiety
59.37 12.60 5.11
2/701 2/700 2/699
.001* .001* .001*
GXOXF
Affective Anxiety Damaging Anxiety Learning Anxiety
.16 .10 2.85
2/701 2/700 2/699
.86 .91 .06
OXF
Affective Anxiety Damaging Anxiety Learning Anxiety
1.34 2.48 .52
2/701 2/700 2/699
.27 .09 .59
GXF
Affective Anxiety Damaging Anxiety Learning Anxiety
1.63 1.02 2.25
2/701 2/700 2/699
.20 .36 .11
GXO
Affective Anxiety Damaging Anxiety Learning Anxiety
.01 .37 2.51
2/701 2/700 2/699
.92 .55 .09
Gender (G)
Affective Anxiety Damaging Anxiety Learning Anxiety
4.84 7.20 4.57
2/701 2/700 2/699
.03* .01* .03*
Ownership (O)
Affective Anxiety Damaging Anxiety Learning Anxiety
10.68 14.43 03.22
2/701 2/700 2/699
.001* .001* .07
The frequency of computer use (F)
Affective Anxiety Damaging Anxiety Learning Anxiety
14.00 00.27 01.87
2/701 2/700 2/699
.001* .77 .16
a *
Because dependent variables were correlated univariate Fs were not evaluated. Significant group differences.
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
2645
The effect of gender on the dependent variables was also investigated through discriminant analysis. Structure coefficients showed that Affective Anxiety (.72), Damaging Anxiety (.05), and Learning Anxiety (.38) correlated with the discriminant function. Bonferroni adjusted pairwise comparisons (Hays, 1994) showed that girls (M(adj) = 18.54, SE = .41) scored significantly higher than boys (M(adj) = 17.33, SE = .37) on Affective Anxiety (p < .03). Similarly, on Damaging Anxiety, girls (M(adj) = 14.92, SE = .38) scored significantly higher than boys (M(adj) = 13.15, SE = .34, p < .001). Girls (M(adj) = 11.27, SE = .24) scored significantly higher than boys on Learning Anxiety as well (M(adj) = 10.20, SE = .21, p < .001). A significant main effect for ownership was found, after adjusting for the effects of Trait Anxiety [Wilks’ k = .96; F(6,699) = 9.54; p < .001, g2 = .04]. Stepdown analyses showed that students who owned a PC differed significantly from those who did not on Affective Anxiety and Damaging Anxiety (Table 3). Regression analyses showed that this variable was significantly related to Affective Anxiety (b = 1.11, t = 3.85, p < .001) and Damaging Anxiety (b = 1.47, t = 5.53, p < .001). Structure coefficients showed that Affective Anxiety (.07), Damaging Anxiety (.78), and Learning Anxiety (.42) correlated with the discriminant function. After a Bonferroni adjustment for multiple comparisons, pairwise comparisons showed that students who do not have a PC (M(adj) = 18.84, SE = .27 and M(adj) = 15.32, SE = .25) scored significantly higher than those who had (M(adj) = 17.03, SE = .48 and M(adj) = 12.75, SE = .45) on Affective Anxiety (p < .001) and Damaging Anxiety (p < .001), respectively. Finally, the frequency of computer use showed a significant main effect on the dependent variables, after adjusting for the effects of Trait Anxiety [Wilks’ k = .96; F(6,1398) = 5.33; p < .001, g2 = .02]. However, stepdown analyses showed a significant effect only on Affective Anxiety (Table 3). Regression analyses showed that this main effect was only related to Affective Anxiety (b = 1.89, t = 5.33, p < .001), but not to Damaging Anxiety (b = .22, t = .68, p < .83) or Learning Anxiety (b = .08, t = .421, p < .67). Bonferroni adjusted pairwise comparisons showed that students who used computers everyday showed the lowest affective anxiety (M(adj) = 16.18, SE = .35) followed by those who used computers several times in a week (M(adj) = 17.40, SE = .30). Students who used computers once a week or less showed the highest levels of affective anxiety (M(adj) = 20.21, SE = .69). 4. Discussion Studies that address the problems associated with and correlates of computer anxiety are abundant; however, a review of the literature on the topic shows inconsistent results with regard to gender effects. Moreover, fewer studies have focused on high school students, even though they may experience high levels of computer anxiety (Loyd & Gressard, 1984). Lastly, the multivariate nature of the construct of computer anxiety has not been adequately integrated into current literature. Therefore, there is still need for clarification in the area of computer anxiety and the present study was an attempt in this direction. Literature suggests that computer anxiety is a multidimensional construct (e.g., Beckers & Schmidt, 2001; Beckers et al., 2007; Deane, Henderson, Barrelle, Saliba, & Mahar, 1995). Thus, its multivariate nature should be taken into account when designing research studies on computer anxiety. Nonetheless, the review of the literature showed that most computer anxiety research disregarded the issue and usually investigations have been done either on a single computer anxiety scale score or without appropriate adjustments for multiple comparisons. In the present study, a linear combination of the three dimensions of computer anxiety (i.e., Affective Anxiety, Damaging Anxiety, and Learning Anxiety) and their interactions were investigated in multivariate context. We studied three of the commonly investigated variables (i.e., gender, ownership, and the frequency of computer use) in computer anxiety research. We looked at the multivariate effects of gender, ownership, and the frequency of computer use on computer anxiety levels, after controlling for the effects of trait anxiety. Results showed that 37.5% of the high school students had a PC at home and that 83.8% used computers everyday or several times in a week. Therefore, we conclude that the sample of the study had accessibility and familiarity with computers. When these findings are compared with pervious studies conducted in Turkey, interesting results emerged. For example, Namlu and Ceyhan (2002) reported that 19.8% Turkish college students had a PC at home and 52.7% used computers everyday or several times in a week. In a more recent study, Tekinarslan (2008) found increased ownership (47.95%)
2646
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
among college students. Both studies concur with the present study that students have accessibility and familiarity with computers. In addition, the present study is in agreement with those of Panagiotakopoulos and Koustourakis (2001), Namlu and Ceyhan (2002) and Tekinarslan (2008) that computer anxiety levels differ according to ownership and frequency of use. Even though they did not differ statistically, more boys than girls had a PC at home (n = 158 vs. n = 110, respectively). These results are similar to those found by Selwyn (1997) and Karavidas, Lim, and Katsikas (2005). Because we did not inquire about the sibling status in the family, it is not clear from these results whether parents prefer to buy a PC if the child is a boy. This point should be investigated in future studies. Contrary to Selwyn, we found significant association between ownership and the frequency of computer use. However, multivariate analysis showed no significant interaction effect between the two variables on computer anxiety levels. Girls and boys did not differ on stable anxiety reactions (i.e., trait anxiety). It can be argued that individuals who are anxious in general might show higher levels of computer anxiety as well. Most recently, Beckers et al. (2007) found that computer anxiety was more strongly related to trait anxiety. In order to control such an effect, if any, A-Trait scores are statistically adjusted for by using them as covariate. It was found that trait anxiety had a significant covariate effect on all three dependent variables. Future researchers should take the information into account when studying computer anxiety among different populations. We suggest using trait anxiety scores as covariate before group comparisons on computer anxiety levels. Gender is one of the most studied variables in computer anxiety research; but, gender-related research has not reached a consensus. A review of the literature on the topic shows inconsistent results with regard to gender effects. There may be several possible explanations as to why the literature shows inconsistent effects of gender. First, gender effects on computer anxiety might be sampledependent. For example, high school girls and boys might experience computer anxiety differently from college men and women. This is one of the reasons why we focused on high school students and their computer anxiety experiences. Second, gender might interact with other variable(s) in affecting computer anxiety levels. For example, boys who own a PC might face different aspects of computer anxiety than girls who do not own a PC. This is another reason why the present study attempted to integrate several aspects of computer anxiety in relation to gender. Third, time might be a significant variable in relation to computer anxiety and gender investigations. For example, nowadays men (or women) might be experiencing anxiety differently than men (or women) who lived some time ago. We attempted to study contemporary gender differences, especially in an era when computers are more commonly integrated into our daily lives. Gender effects found in the present study support studies that found girls have higher levels of computer anxiety compared to boys (i.e., Karavidas et al., 2005; Mcilroy et al., 2001; Namlu & Ceyhan, 2002; Palaigeorgiou, Siozos, Konstantakis, & Tsoukalas, 2005; Rosen & Weil, 1995; Todman, 2000). In all three dimensions (i.e., Affective Anxiety, Damaging Anxiety, and Learning Anxiety) girls outscored boys regardless of their trait anxiety levels. However, no interaction between gender and ownership or the frequency of computer use on computer anxiety levels was found. We could not find any study that investigated such interactions in the literature to compare these results. Future studies should focus on complex interactions similar to the ones studied in the present investigation. It can be argued that factors such as owning a PC at home or the frequency of computer use might confound computer anxiety (e.g., Brosnan 1999; Chua et al., 1999; Keser, 2001; Rosen & Weil, 1995; Scott & Rockwell, 1997; Tuluk, 1999). For example, Teo (2006) found that students who owned a PC at home showed lower computer anxiety levels compared to those who did not. The present results support this argument in general. Both ownership and the frequency of computer use had significant effects on different aspects of computer anxiety. However, complex relationships were detected. For example, ownership had a significant effect on affective anxiety and damaging anxiety, but not on learning anxiety. Therefore, we conclude that students who do not have a PC at home fear most damaging computers; but they do not differ when it comes to learning computers or computer-relevant information. Similarly, the frequency of computer use did not have any significant effect on damaging anxiety or learning anxiety. On the contrary, it did have a significant effect on affective anxiety. In sum, these results make it clear that the effect of any single variable on computer anxiety should be studied in relation with other relevant variables. Computer anxiety seems to be affected by multiple variables
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
2647
and univariate investigations of the variables will not convey the most accurate picture. Appropriate interactions should be taken into consideration when studying computer anxiety. Researchers predict that computing and computers will be more involved in students’ lives (Mcilroy et al., 2001). Therefore, information provided in the present study should particularly be of great value for students and computer instructors alike. Computer instructors should be aware that students who have limited access to or lesser use of computers might experience more difficulties. These difficulties might be present in the form of affective anxiety, fear of damaging computers, or learning computers. One shortcoming of the present study is that it was limited with its sample characteristics. The sample of the study included high school students who were enrolled in classes in northern Turkey. In addition, the study is limited with the quantitative data on the frequency of computer use. Qualitative data such as for what purposes students use computers (e.g., programming, games-playing, surfing the Internet, or communication) were not sought. For example, some students might use computers everyday for game playing, whereas, others use once a week for complex programming. This type of information might be useful in studying computer anxiety because the literature indicates that task-relevant use of computers and anxiety might interact (Humphreys & Revelle, 1984). Finally, the present study was not an experimental design; therefore, the variables significantly related to computer anxiety are not the ‘‘causes” of computer anxiety. We cannot infer, for example, that having a PC at home ‘‘causes” less affective or damaging anxiety. We can simply infer that there is a significant difference on affective and damaging anxiety and students who own a PC at home show lesser levels of anxiety. Causality cannot be determined as a result of statistical analyses. It can only be determined as a result of design or theory (Trochim & Donnelly, 2006). Therefore, experimental studies are needed to identify the causes of computer anxiety, if that is the purpose. Acknowledgements The authors express their gratitude to two anonymous reviewers, whose comments and suggestions led to a more comprehensive paper. This study is supported by the Scientific and Technical Research Council of Turkey. References Abdelhamid, I. S. (2002). Attitudes towards computers: A study of gender differences and other variables. Journal of the Social Sciences, 30, 285–316. Anderson, A. A. (1996). Predictors of computer anxiety and performance in information systems. Computers in Human Behavior, 12, 61–77. Arıkan, D. (2002). Sınıf ög˘retmeni adaylarının bilgisayara yönelik tutumları, bilgisayar kaygı düzeyleri ve bilgisayar dersine ilisßkin deg˘erlendirmeleri [Attitudes toward computers, computer anxiety, views of a computer course]. Unpublished master’s _ thesis, Dokuz Eylül University, Izmir, Turkey. Beckers, J. J., & Schmidt, H. G. (2001). The structure of computer anxiety: A six-factor model. Computers in Human Behavior, 17, 35–49. Beckers, J. J., Wicherts, J. M., & Schmidt, H. G. (2007). Computer anxiety: Trait or state. Computers in Human Behavior, 23, 2851–2862. Bohlin, R. M., & Hunt, N. P. (1995). Course structure effects on students’ computer anxiety. Journal of Educational Computing Research, 13(3), 263–270. Bovee, C., Voogt, J., & Meelissen, M. (2007). Computer attitudes of primary and secondary students in South Africa. Computers in Human Behavior, 23, 1762–1776. Bozionelos, N. (2001). Computer anxiety: Relationships with computer experience and prevalence. Computers in Human Behavior, 17, 213–224. Brosnan, M. J. (1999). Modeling technophobia: A case for word processing. Computers in Human Behavior, 15, 105–121. Brosnan, M. J., & Lee, W. (1998). A cross-cultural comparison of gender differences in computer attitudes and anxieties: The United Kingdom and Hong Kong. Computers in Human Behavior, 14, 559–577. Ceyhan, E., & Namlu, A. G. (2000). Bilgisayar Kaygısı Ölcßeg˘i: Gecßerlik ve güvenirlik cßalısßması [The computer anxiety scale: Validity and reliability]. Anadolu Üniversitesi Eg˘itim Fakültesi Dergisi, 10(2), 77–93. Chou, C. (2003). Incidences and correlates of Internet anxiety among high school teachers in Taiwan. Computers in Human Behavior, 19, 731–749. Chua, S. L., Chen, D. T., & Wong, F. L. (1999). Computer anxiety and its correlates: A meta analysis. Computers in Human Behavior, 15, 609–623. Colley, A. M., Gale, M. T., & Harris, T. A. (1994). Effects of gender role identity and experience on computer attitude components. Journal of Educational Computing Research, 10(2), 129–137.
2648
M. Balog˘lu, V. Çevik / Computers in Human Behavior 24 (2008) 2639–2648
Deane, F., Henderson, R., Barrelle, K., Saliba, A., & Mahar, D. (1995). Construct validity of computer anxiety measured by the computer attitudes scale. In Y. Anzai, K. Ogawa, & H. Mori (Eds.), Advances in human factors/ergonomics 20A symbiosis of human and artifact (pp. 581–586). Amsterdam: Elsevier Science. Deniz, L. (1994). Bilgisayar Tutum Ölcßeg˘i’nin gecßerlik, güvenirlik ve norm cßalısßması ve örnek bir uygulama [Development of the attitudes toward computers: A preliminary study]. Unpublished doctoral dissertation, Marmara University, Istanbul, Turkey. Dupin-Bryant, P. (2002). Reducing computer anxiety in adults learning to use microcomputers. Journal of Extension, 40(5). Available from