- Email: [email protected]
The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture
G Model
AJP-502; No. of Pages 9 Asian Journal of Psychiatry xxx (2013) xxx–xxx
Contents lists available at ScienceDirect
Asian Journal of Psychiatry journal homepage: www.elsevier.com/locate/ajp
The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture A.K.M. Rezaul Karim *, Naima Nigar Department of Psychology, University of Dhaka, 1000 Dhaka, Bangladesh
A R T I C L E I N F O
A B S T R A C T
Article history: Received 18 September 2013 Received in revised form 16 October 2013 Accepted 28 October 2013 Available online xxx
There is growing importance of the Internet Addiction Test (IAT) in Internet addiction research around the world. Since the development of the IAT (Young, 1996, 1998), a number of validation studies have been done in various cultures. The aim of this study was to translate the instrument into Bangla and validate in Bangladeshi culture, a culture vulnerable to Internet addiction. A total of 177 Internet users (77 females and 100 males) participated in the study. Exploratory factor analysis (EFA) of the data from 172 participants (who provided complete responses) identified a four factor structure of the IAT with 18 items. The four factors namely ‘Neglect of duty’, ‘Online dependence’, ‘Virtual fantasies’, and ‘Privacy and self-defense’ together explained 55.68% of the total variance. Problematic (moderate/excessive) users on the IAT scored, on average, higher on each of the four IAT factors as compared to average or nonproblematic (minimal) users consistently across genders. The IAT and its factors showed good internal consistency (Cronbach’s a = .89 for the IAT, and .60–.84 for the factors), strong convergent and discriminant validity. Thus, the Bangla version IAT appears to be valid and reliable and therefore may be used in further research on Internet addiction in the country. ß 2013 Elsevier B.V. All rights reserved.
Keywords: IAT Internet Addiction Bangla Validation
1. Introduction Addictive use of the Internet has emerged as a new and rapidly developing area of addiction research. In researching this topic, a valid and reliable tool is essential. Young (1996, 1998) designed such a research and diagnostic tool, based on DSM-IV criteria for pathological gambling, called the Internet Addiction Test (IAT). The IAT is a widely used tool for assessing the symptoms of Internet addiction defined as any online-related, compulsive behavior which causes impairment of normal social interactions/intimacy (with friends and relatives), and increases daily stress, feeling of loneliness, anxiety and depression. The IAT results can help decide whether someone needs to see a doctor or other mental health professional to discuss his/her addictive use of the Internet and its negative consequences. Because of its growing importance in clinical practice and addiction research, a number of validation studies have been attempted on the English (Widyanto and McMurran, 2004; Widyanto et al., 2011; Jelenchick et al., 2012), Italian (Ferraro et al., 2007), Chinese (Chang and Law, 2008), French (Khazaal et al., 2008), Finnish (Korkeila et al., 2010), German (Barke et al., 2012) and Arabic (Hawi, 2013) versions of the test. These studies have shown moderate-to high-internal consistency of the
* Corresponding author. Fax: +880 2 8615583. E-mail addresses: [email protected], [email protected] (A.K.M. Rezaul Karim).
IAT, with concurrent validity in British culture (Widyanto and McMurran, 2004), convergent and discriminant validity in Chinese (Chang and Law, 2008) and Finish (Korkeila et al., 2010) cultures, and construct validity in American culture (Jelenchick et al., 2012). All these versions have further demonstrated factorial validity. However, there is a wide variation in its factor structure. As reported, it has both the uni-factorial and multi-factorial structures. Specifically, the French and the Arabic versions have shown to comprise only one factor. The study of Korkeila et al. (2010) supported for both a single-factor model (dependent use) and a two-factor model (salient use and loss of control) of the Finnish version. On the other hand, the English version has a 6factor structure (factors: salience, excessive use, neglect of work, anticipation, lack of control, and neglect of social life; Widyanto and McMurran, 2004) and a 3-factor structure (factors: emotional/ psychological conflict, time management issues and mood modification; Widyanto et al., 2011) in British culture as well as a 2-factor structure (factors: dependent use and excessive use; Jelenchick et al., 2012) in American culture. The ‘dependent use’ and ‘excessive use’ of the 2-factor structure of the English version resembles the ‘emotional and cognitive preoccupation’ and ‘loss of control and interference with daily life’ of the 2-factor structure of the German version respectively. However, other versions of the IAT are widely different from each other even when they comprise the same number of factors. Thus factor structure of the IAT varies from study to study, from culture to culture and even within the same culture, and therefore it requires a validation study to be
1876-2018/$ – see front matter ß 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ajp.2013.10.011
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011
G Model
AJP-502; No. of Pages 9 2
A.K.M. R. Karim, N. Nigar / Asian Journal of Psychiatry xxx (2013) xxx–xxx
usable in a new culture. Examining factor structure or dimensionality of an instrument in a new culture is important for accurate specifications of theories (Smith and McCarthy, 1995) and theorydriven research. The dimensions can provide a greater detail of the nature of culture-based addiction than just using Internet addiction as an overall concept. For example, addicts may suffer different levels of negative impacts in different aspects of Internet addiction. Knowing the levels can help us to focus remedy and treatment service on the areas that require special attention. The Internet has broken down geographic and cultural boundaries, expanding from the highly developed to the less- or under-developed countries. However, since its development with the work of Young (1996, 1998) research on Internet addiction has been confined to a few developed countries only. Thus it has been partially and poorly understood. In order to advance our understanding of the phenomenon, research in other countries, especially in the less- and under-developed countries should be initiated. Bangladesh is such a country where the first connectivity of Internet became a reality in 1996. Though it was somehow late, over the last few years the growth has been rapid. The number of Internet users in the country grew from 100,000 in 2000 to 450,000 in 2007, to 556,000 in 2009, to 617,300 in 2010, to 5,501,609 in 2011, and to 8,054,190 in 2012 (Internet World Stats, 2013). According to latest reports of Bangladesh Telecommunication Regulatory Commission (BTRC, 2013), there are about 34,711,101 mobile Internet users, 1,222,620 ISP (Internet Service Provider) plus PSTN (Public circuit-Switched Telephone Networks) users and 315,297 WiMAX (Worldwide Interoperability for Microwave Access) users in Bangladesh. The Internet provides a constant, ever-changing source of information and entertainment, and can be accessed from most smartphones, tablets, laptops and desktop computers. Email, blogs, social networks, instant messaging, and message boards allow for both public and anonymous communication about any topics. While time spent online can be hugely productive, compulsive Internet use (e.g., compulsively checking smartphones, or other mobile devices) can interfere with daily life, work, and relationships. However, due to a lack of scientific enquiry we still do not know so much about this negative side, i.e., the presence and severity of addictive use of Internet and its impact on personal, academic, professional and social life in Bangladesh. We have therefore developed active interest to create a field of addiction research in the country, aiming to focus on pathological use of Internet among the people. As a first step toward this end, we validated the IAT within the socio-cultural context of Bangladesh. We used an exploratory factor analysis (EFA) to examine the psychometric properties and establish our culture-based factor structure of the IAT for university students. University students have easy access to Internet through their own computers (desktops, laptops, and tablets), and/or via mobile phones. We chose university students because we understand from the broadcast of different media (e.g., newspapers, Internet blogs, and T.V.) that they are vulnerable to a variety of social networks such as Facebook, hi5, and Twitter. They are vulnerable because of their age characterized by psychological and developmental dynamics, ready access to the Internet, and an expectation of computer/Internet use. However, future research using this tool will advance our understanding of the motives underlying Internet addiction among university students and help design effective recovery programs for them. 2. Method 2.1. Participants A total of 177 students (77 females and 100 males) studying at the University of Dhaka voluntarily participated in this study. All
the participants had connections to Internet through their personal computers and/or via mobile phones. However, because of incomplete responses 5 participants (2 females and 3 males) were dropped. The age of the remaining 172 participants ranged from 18 to 25 years with a mean of 22.33 and standard deviation (SD) of 2.01. Of the participants, 43.6% were females and 56.4% were males. The female participants’ age ranged from 18 to 25 years with a mean of 22.11 and SD of 1.89 and the male participants’ age ranged from 18 to 25 years with a mean of 22.49 and SD of 2.09. 83.1% of the participants were single, 7.6% were married, 6.4% were in a relationship, 1.2% were divorced, and 1.7% were separated. 2.2. Measures 2.2.1. Internet Addiction Test Developed by Young (1996), Internet Addiction Test (IAT) is the first validated and reliable measure of addictive use of the Internet. It is a 20-item questionnaire that measures psychological dependence, compulsive use, and withdrawal, as well as related problems of school, sleep, family, and time management. The IAT is a 5-point Likert type scale ranging from 1 (rarely) to 5 (always). The minimum obtainable score on the IAT is 20 and the maximum is 100. The higher the score the greater is the level of Internet addiction. Based on the severity of addictive behavior individuals on the IAT can be classified as minimal users (20–39), moderate users (40–69) or excessive users (70–100) (Widyanto and McMurran, 2004; Widyanto et al., 2011). Minimal users are the average online users who have complete control over their Internet usage. Moderate users are those experiencing occasional or frequent problems due to Internet usage. Excessive users are those having significant problems caused by Internet usage. As stated earlier in this article, the English version IAT has exhibited good psychometric properties in various studies (Widyanto and McMurran, 2004; Jelenchick et al., 2012). 2.2.2. Miller Social Intimacy Scale The Bangla version of the Miller Social Intimacy Scale (MSIS) was used to examine the discriminant validity of the Bangla version IAT. The original MSIS is a 17-item tool designed to assess intimacy in a variety of relationships such as friendships, family and spouse (Miller and Lefcourt, 1982). The Bangla version instrument comprises 14 items; 5 items to measure the frequency of intimate contacts and 9 items to assess the intensity of intimate relations (Nigar, 2012). Each of the items is rated on a 10-point scale, ranging from 1 (very rarely) to 10 (almost always). The higher the score on MSIS the higher is the intimacy. Internal consistency (Cronbach’s a) of the original MSIS ranges from .86 to .91, the test–retest reliability coefficients being .84 for one-month interval and .96 for a twomonth interval (Miller and Lefcourt, 1982). The psychometric properties of the Bangla version MSIS also support that it is a reliable and valid measure of social intimacy (Nigar, 2012). 2.3. Procedures 2.3.1. Translating the IAT into Bangla The IAT items were first translated into Bangla, called the first draft. It was then given to four judges, including one expert in Bangla, one expert in English and two experts in Psychology/ Psychometrics. Their native language was Bangla, but being professors/lecturers of a university or college they had also very good command in English. Their task was to judge the accuracy of translation and relevance/suitability of each item for measuring Internet addiction in the socio-cultural context of Bangladesh. Each expert independently rated the translation using a 2-point scale (0 = Not correct, 1 = Correct) and the relevance of each item using
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011
G Model
AJP-502; No. of Pages 9 A.K.M. R. Karim, N. Nigar / Asian Journal of Psychiatry xxx (2013) xxx–xxx
3
responses and were therefore excluded from further processing. Data for the remaining 172 participants were fed into computer for factor analysis on IBM SPSS Statistics 20. According to standard textbook authors and researchers, the minimum sample size for factor analysis varies from 100 (e.g., Kline, 1979; Gorsuch, 1983) to 250 (e.g., Cattell, 1978), and there is practice of applying factor analysis even to the data for less than 100 participants (e.g., Widyanto and McMurran, 2004). There is another set of recommendations varying from a minimum SV (subjects-tovariables) ratio of 2:1 (e.g., Guilford, 1956; Kline, 1979) to 10:1 (e.g., Everitt, 1975; Kunce et al., 1975; Nunnally, 1978; Marascuilo and Levin, 1983). The number of participants in this study was more than 8 times the number of IAT items/variables (20). Thus the sample size required for factor analysis was satisfied. However, before carrying out factor analysis we examined the response distributions of all IAT items and estimated their internal consistency by investigating inter-item correlations and itemtotal correlations. Then we analyzed the data in EFA, a method widely used to uncover the underlying structure of a relatively large set of variables (e.g., Gratz and Roemer, 2004; Hawi, 2013; Korkeila et al., 2010; Muris, 2001; Widyanto et al., 2011; Widyanto and McMurran, 2004). We also examined the discriminant validity by correlating the IAT and IAT subtests with the MSIS and the reliability by estimating internal consistency (Cronbach’s a) of the full test and subtests as well.
another 2-point scale (0 = Not relevant, 1 = Relevant). Following their evaluation, accuracy of the translation was examined by calculating for each item the Accuracy Index (AI = Number of rating at 1/Number of experts). The item yielding an AI of 1 (AI = 4/4) was considered to be correctly and reliably translated. All the four experts rated 16 items’ translation at 1, the AI for each of them becoming 1. The remaining 4 items’ yielded an AI of less than 1. The experts suggested some corrections to the clarity, wording and organization of these items. By reviewing those items in the light of their comments and suggestions the accuracy of translation was ensured. The relevance/suitability of the items in Bangladeshi culture was examined by calculating for each item the Relevance Index (RI = Number of rating at 1/Number of experts). The item yielding an RI of 1 or .75 (RI = 4/4 or RI = 3/4) was considered to be relevant or suitable. All the four experts rated the relevance of each item at 1, the RI for them becoming 1. Thus the second draft of the Bangla version IAT was finalized to administer on the selected participants. 2.3.2. Data acquisition Each participant was briefed about the general study purpose and requested to volunteer in the study. Participants were assured that their responses would be kept confidential and used only for research purposes. After getting their verbal consent the paperbased survey (2nd draft) was administered in person. The survey components included an informed consent statement, sociodemographic section, the IAT and the MSIS. Participants were asked to sign on the consent paper, record the socio-demographic information (age, gender, education, marital status and SES), and read carefully the standard instructions of how to respond before going through the items or questions of the test/scale. Thus the surveys were administered and data were collected over a 6-week period from all the participants.
3. Results 3.1. Factor structure 3.1.1. Item analysis Response distributions of the IAT items indicated that none of these variables/items were excessively skewed or kurtotic (Kendall and Stuart, 1958). The correlation matrix (R-matrix, Table 1) contained no negative values and out of 190 inter-item correlation coefficients 179 were significant, the average inter-item coefficient being .29. The item-total correlations were significant and ranged from .33 to .68 with a mean of .57. So, no item was excluded on the basis of the item response distributions or correlation matrix.
2.4. Data analyses Each participant’s responses to the test items were scored according to the scoring principles of the IAT and the MSIS. Five participants left a few IAT items and/or MSIS items with missing
Table 1 Correlation matrix (R-matrix) for IAT. Item
01
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20
1 .60** 1 .32** .32** 1 .26** .15* .31** 1 .40** .45** .25** .19** 1 .48** .62** .33** .19** .45** 1 .27** .11 .15* .06 .15* .07 1 .35** .51** .31** .20** .42** .58** .17* 1 .22** .17* .25** .20** .30** .24** .13 .25** 1 .21** .26** .40** .28** .19** .36** .05 .35** .25** 1 .35** .39** .34** .33** .26** .40** .06 .44** .19** .35** 1 .33** .26** .32** .30** .27** .32** .15* .27** .19** .24** .43** 1 .29** .29** .16* .17* .18** .24** .10 .25** .33** .24** .19** .30** 1 .48** .46** .22** .32** .36** .47** .18** .38** .31** .26** .33** .38** .51** 1 .36** .45** .24** .20** .24** .32** .17* .46** .17* .40** .29** .42** .27** .33** 1 .39** .38** .23** .24** .38** .39** .15* .38** .28** .30** .41** .42** .36** .41** .45** 1 .22** .33** .11 .15* .30** .32** .15* .44** .39** .17* .20** .22** .27** .30** .29** .43** 1 .26** .30** .19** .17* .17* .36** .08 .40** .35** .20** .27** .27** .31** .26** .24** .21** .38** 1 .21** .26** .28** .09 .34** .29** .14* .29** .17* .26** .28** .34** .27** .23** .25** .33** .30** .21** 1 .28** .39** .30** .15* .22** .35** .08 .44** .14* .29** .29** .33** .34** .22** .47** .43** .42** .51** .48** 1
IAT total
.65**
02
.67**
03
.50**
04
.43**
05
.58**
06
.67**
07
.33**
08
.68**
09
.49**
10
11
.52**
.59**
12
.60**
13
.55**
14
.67**
15
.62**
16
.67**
17
.56**
18
.53**
19
.52**
20
IAT Total
.61** 1
Note. N = 172; average inter-item correlation = .29; average item-total correlation = .57. * p < .05 (one-tailed). ** p < .01 (one-tailed). Determinant = .001.
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011
G Model
AJP-502; No. of Pages 9 4
A.K.M. R. Karim, N. Nigar / Asian Journal of Psychiatry xxx (2013) xxx–xxx
3.1.2. Factor analysis First, in order to examine whether data were suitable for factor analysis, measures of sampling adequacy were carried out on the 20-item IAT. Inspection of the R-matrix revealed a substantial number (46.84%) of coefficients .30 and above. The determinant of the R-matrix was .001 (>.00001, Field, 2005), indicating that there was no multicolinearity (very highly correlated variables) or singularity (perfectly correlated variables) problem. The Kaiser– Meyer–Olkin (KMO) measure indicated a value of .87 which exceeded the recommended value of .60 (Kaiser, 1970) and Bartlett’s test of sphericity indicated a x2 value of 1197.83 (p < .001). All this together supports the factorability of the Rmatrix. Data for the full set of IAT items were therefore subjected to EFA. Method of principal component (PC) with varimax rotation was used. The initial analysis with eigenvalue > 1.00 (the Kaiser– Guttman criterion) extracted 5 factors, accounting for 57.91% of the total variance (data not shown). However, Floyd and Widaman (1995) suggested that the scree test (Cattell, 1966) is a more accurate method for retaining factors than the Kaiser–Guttman criterion. Inspection of the scree plot revealed a clear break after the 4th component (Fig. 1a), leading us to retain 4 components. Therefore, we analyzed the data in another EFA, limiting the number of factors to 4 with all factor loadings < .40 suppressed. The four factors together accounted for 52.56% of the total variance (data not shown). Results further revealed that two items (item 07and item 16) loaded at < .40. These low factor loadings indicated that perhaps the items do not represent a symptom of problematic Internet use in Bangladeshi culture. After discarding these two items, data were subjected to a final EFA with all factor loadings < .40 suppressed. Though the EFA was run this time without specifying the number of factors a four-factor structure of the IAT with 18 items was identified, consistently with the scree plot (Fig. 1b). These four factors, which were rotated to position of maximum orthogonality in eight iterations, explained together 55.68% of the total variance (Table 2). Factor 1 accounts for 17.44% of the variance, Factor 2 accounts for 13.73% of the variance, Factor 3 accounts for 13.58% of the variance, and Factor 4 accounts for
10.93% of the variance. Before labeling the factors we identified three pairs of cross-loadings between the factors. Specifically, item 08 was cross loaded on Factor 1 and Factor 2 with the loadings of .54 and .49 respectively; item 14 was cross-loaded on Factor 1 and Factor 4 with the loadings of .55 and .52 respectively; and item 17 was cross-loaded on Factor 2 and Factor 4 with the loadings of .52 and .49 respectively. We grouped both item 08 and item 14 under Factor 1, the factor of their greater loadings and best conceptual fit; and item 18 under Factor 4, the factor of its smaller loading but best conceptual fit. Thus Factor 1 comprises items 1, 2, 5, 6, 8 and 14 which we termed as ‘Neglect of duty’, Factor 2 comprises items 15, 17, 19 and 20 which we termed as ‘Online dependence’, Factor 3 comprises items 3, 4, 10, 11 and 12 which we termed as ‘Virtual fantasies’, and Factor 4 comprises items 9, 13 and 18 which we termed as ‘Privacy and self-defense’. 3.2. IAT scores and IAT factor scores We proposed for the Bangla version 18-item IAT a cutoff scores system equivalent to that of the English version 20-item IAT (Table 3). Using this cutoff scores system, the majority of participants i.e., 63.95% were categorized as ‘minimal users’ (IAT score < 36), 34.30% were categorized as moderate users (62 IAT score 36) and only 1.74% as excessive users (IAT score > 62). Of the female participants, 77.33% were categorized as ‘minimal users’, 21.33% as moderate users and only 1.33% as excessive users. Of the male participants, 53.61% were categorized as ‘minimal users’, 44.33% as moderate users and only 2.06% as excessive users. Thus the percentage of males is about double the percentage of females falling in the moderate/higher level of Internet addiction. The mean IAT score calculated over all participants was 33.75 11.14. Table 4 provides detailed descriptive statistics for the full IAT and four IAT factors. Regardless of participant’s sex, the mean IAT score was 26.93 4.89 for minimal users, 44.80 7.25 for moderate users and 66.67 1.53 for excessive users (Table 4). The mean IAT score was 25.76 4.88 for female minimal users, 46.00 7.93 for female moderate users and 65.00 for female
Fig. 1. The scree plots generated in EFA: (a) for 20 items, and (b) for 18 items.
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011
G Model
AJP-502; No. of Pages 9 A.K.M. R. Karim, N. Nigar / Asian Journal of Psychiatry xxx (2013) xxx–xxx
5
Table 2 Rotated factor matrix for a reduced set of IAT items (item 07 and item 16 discarded). IAT items in brief
Factor loadings F1
Item Item Item Item Item Item Item Item Item Item Item Item Item Item Item Item Item Item
01 02 05 06 08 14 15 17 19 20 03 04 10 11 12 09 13 18
Stays online longer than intended Neglects household chores to spend more time online Others in life complain about the time spent online Grade or school work suffers due to spending time online Performance/productivity suffers because of the Internet Loses sleep due to late night logins Fantasizes about being online when actually offline Tries to cut down time to spend online, but fails Chooses time to spend more online than with others Feels depressed offline and it goes away once back online Prefers excitement of the Internet to close one’s intimacy Forms new relationships with fellow online users Blocks out disturbing thoughts with those of the Internet Finds anticipating when to go online again Fears of empty and joyless life without the Internet Becomes defensive when asked about online activities Snaps, yells or acts annoyed if bothered while online Tries to hide the amount of time spent online
F2 .72 .80 .64 .72 .54 .55
F4
(.49) (.52) .47 .52 .60 .81
(.49)
.66 .66 .62 .58 .56 .74 .65 .47
(.53) 3.14 17.44% .84
Eigenvalue Variance explained Cronbach’s (standardized) a
F3
2.47 13.73% .70
2.45 13.58% .71
1.97 10.93% .60
Note. N = 172. Factor loadings < .40 were suppressed. Items corresponding to the parenthesized loadings did not conceptually fit with the corresponding factors. Extraction method: principal component analysis. Rotation method: varimax with Kaiser normalization. Rotation converged in 8 iterations.
excessive users. The mean IAT score was 28.23 4.61 for male minimal users, 44.35 7.02 for male moderate users and 67.50 0.71 for male excessive users. Thus males, on average, scored higher in the higher level whereas females scored higher in the moderate level of Internet addiction. Figures in Table 4 further indicate that problematic (both moderate and excessive) users on the IAT scored, on average, higher on each of the four IAT factors consistently across sexes as compared to average or non-problematic (minimal) users. 3.3. Validity 3.3.1. Convergent validity Convergent validity of the Bangla version IAT was examined by estimating inter-factor correlations and the factor-IAT total correlations. The inter-factor correlations (Pearson’s r) were all significant, the coefficient ranging from .43 to .58 (Table 5). The strongest correlation was found between F1 (Neglect of duty) and F2 (Online dependence) and the weakest between F3 (Virtual fantasies) and F4 (Privacy and self-defense). The four IAT factors also significantly correlated with the full IAT, with the coefficients ranging from .71 to .88. The IAT had strongest correlation with F1 Table 3 Cutoff scores for categorizing Internet users on the IAT. User category
20-Item English IATa,b
18-Item Bangla IAT
Minimal users Moderate users Excessive users
20–39 40–69 70–100
18–35 36–62 63–90
Minimal users: Lower score = Number of items 1.0, Upper score = Number of items 2.0–1.0. Moderate users: Lower score = Number of items 2.0, Upper score = Number of items 3.5–1.0. Excessive users: Lower score = Number of items 3.5, Upper score = Number of items 5.0. a Widyanto et al. (2011). b Widyanto and McMurran (2004).
(Neglect of duty) and the weakest with F4 (Privacy and selfdefense). 3.3.2. Discriminant validity Discriminant validity of the Bangla version IAT was examined by correlating the IAT and IAT factors with the Bangla version MSIS and MSIS dimensions. Negative or non-significant correlations were hypothesized between the measures in order for them to have discriminant validity. As hypothesized, the IAT and IAT factors were correlated significantly and negatively with the MSIS (r = .16 to .25) and MSIS: Intensity (r = .21 to .29, Table 6). The IAT and two of the IAT factors (F2: Online dependence, F4: Privacy and selfdefense) significantly and negatively correlated with the MSIS: Frequency (r = .15 to .20), and the other two Factors (F1: Neglect of duty, F3: Virtual fantasies) non-significantly correlated with the MSIS: Frequency (r = .06 to .11). All these correlations strongly support the discriminant validity of the measures. 3.4. Reliability 3.4.1. Internal consistency The interitem correlation matrix (Table 1) contained no negative values, indicating that the items were measuring the same characteristic. The reliability of the Bangla version IAT was further examined by estimating internal consistency. The coefficients of Cronbach’s a were calculated. Cronbach’s a (standardized) for the full IAT was .89. The coefficients for the ‘Neglect of duty’, ‘Online dependence’, ‘Virtual fantasies’, and ‘Privacy and self- defense’ subtests were .84, .70, .71, and .60 respectively (Table 2). 4. Discussion This study was designed to investigate the psychometric properties of a Bangla version of the IAT. Analyzing the data in EFA we identified a four factor model for the IAT comprising 18
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011
G Model
AJP-502; No. of Pages 9
(1.00) (2.31) (1.53) (1.00) (1.53) 23.00 13.33 18.33 12.00 66.67 Note. (–) SD was not possible to determine as n = 1. The minimum obtainable score on the ‘Neglect of duty’ is 6 and the maximum score is 30. The minimum obtainable score on the ‘Online dependence’ is 4 and the maximum score is 20. The minimum obtainable score on the ‘Virtual fantasies’ is 5 and the maximum score is 25. The minimum obtainable score on the ‘Privacy and self-defense’ is 3 and the maximum score is 15. The minimum obtainable score on the IAT is 18 and the maximum score is 90.
Range Mean (SD) Range Mean (SD)
(2.77) 6–18 16.41 (3.98) 8–26 (1.37) 4–10 8.88 (3.15) 4–17 (2.39) 5–18 11.63 (3.35) 6–21 (1.47) 3–10 7.88 (2.29) 3–15 (4.98) 18–35 44.80 (7.25) 36–61
Range Mean (SD)
22–24 9.28 12–16 5.05 17–20 7.80 11–13 4.79 67–68 26.93 (1.41) (2.83) (2.12) (1.41) (.71) 23.00 14.00 18.50 12.00 67.50
Range Mean (SD) Range Mean (SD)
23–23 10.19 (2.94) 6–18 16.47 (4.29) 8–26 12–12 5.12 (1.37) 4–9 8.44 (2.77) 4–17 18–18 8.08 (2.40) 5–15 11.65 (3.24) 6–21 12–12 4.85 (1.23) 3–8 7.80 (2.26) 3–15 65–65 28.23 (4.61) 18–35 44.35 (7.02) 36–61
Range Mean (SD) Range Mean (SD)
(3.13) 10–21 23.00 (–) (3.84) 4–16 12.00 (–) (3.72) 7–19 18.00 (–) (2.42) 3–12 12.00 (–) (7.93) 37–59 65.00 (–) 16.25 10.06 11.56 8.13 46.00
Moderate users (n = 16)
Range Mean (SD)
(2.33) 6–15 (1.39) 4–10 (2.38) 5–18 (1.63) 3–10 (4.88) 18–35
Mean (SD)
Neglect of duty 8.45 Online dependence 5.00 Virtual fantasies 7.55 Privacy and self-defense 4.74 IAT 25.76
Excessive users (n = 3) Moderate users (n = 59) Minimal users (n = 110)
Total (n = 172)
Excessive users (n = 2) Minimal users (n = 52)
Moderate users (n = 43) Male (n = 97)
Minimal users (n = 58)
Excessive users (n = 1) Female (n = 75) IAT factors/IAT
Table 4 Descriptive statistics for 18-item Bangla IAT and four IAT factors.
22–24 12–16 17–20 11–13 65–68
A.K.M. R. Karim, N. Nigar / Asian Journal of Psychiatry xxx (2013) xxx–xxx Range
6
items (2 items dropped; Table 2). Factor 1 (6 items) measures neglect of duty (e.g., avoiding household works, avoiding study or duty, and complaints of time spent online), Factor 2 (4 items) measures online dependence (e.g., cutting down online time, preferring time to spend with online friends, and feeling depressed offline), Factor 3 (5 items) measures virtual fantasies i.e., how addicted people tend to substitute the real world (e.g., preferring online relationships to spouse’s intimacy, thoughts of online excitements and new relationships), and Factor 4 (3 items) measures privacy and self-defense (e.g., being defensive and aggressive if asked about online activities, and hiding time spent online). These factors together accounted for 55.68% of the total variance, their individual contributions ranging from 10.93% to 17.44% (Table 2). The factors show good to high internal consistency (Cronbach’s a = .60 to .84), the first one being most reliable as indicated by its highest coefficient (Cronbach’s a = .84). As shown in Table 5, all the factors were significantly correlated with the whole IAT (r = .71–.88, p < .01). The moderate (linear) inter-correlations of the factors (r = .43–.58, p < .01) support the representation of the factors as unique, distinguishable components (Jelenchick et al., 2012). As mentioned earlier in this paper, previous analyses have shown between one- and six-factor solutions for the IAT. In spite of this wide disagreement, factor arrangements from prior analyses in Table 7 show striking resemblances to our results. An inspection of this table indicates that the greatest agreement was found between our four-factor model and the two-factor models of both the English (Jelenchick et al., 2012) and German (Barke et al., 2012) versions. That is, Factor 1 (Neglect of duty) of the Bangla version IAT most closely resembles, in content items (item no. 1, 2, 6, 8 and 14), the second factors of the two-factor structures of the English and German versions (English: Excessive use, German: Loss of control and interference with daily life). Factor 2 (Online dependence), Factor 3 (Virtual fantasies) and Factor 4 (Privacy and self-defense) together represent well the first factors of the English and German versions (English: Dependent use, German: Emotional and cognitive preoccupation). The items clustered within the first factor of the Bangla version also show broad overlap (item no. 1, 2, 6 and 8) with the second factors (English: Time management issues, Chinese: Time management and performance) of the three-factor structures of both the English (Widyanto et al., 2011) and Chinese (Chang and Law, 2008) versions. Thus items 1, 2, 6, 8 and possibly a few others tend to cluster together, forming a factor to be stable across cultures. This reveals that the most common symptoms of Internet addiction across cultures are unusual stay online, complaints of time spent online, avoidance of daily activities such as household works, poor school grades or performance and sleep disturbances. However, discrepancies still exist between the remaining three factors of our model and those of prior analyses. This was most likely due to differences inherent in the samples such as differences in sociocultural characteristics or cultural norms surrounding Internet use, rather than differing representations of the overall construct. One interesting and distinct aspect of this study is that we examined here the discriminant validity of the IAT and IAT dimensions/factors by estimating correlations with social intimacy. We hypothesized that social intimacy would be negatively related with the IAT and IAT dimensions. As hypothesized, we found social intimacy to be significantly and negatively correlated with the IAT (r = .25, p < .01) and IAT dimensions (r = .16 to .26, p < .05 or .01). This was also evident when the IAT and IAT dimensions were correlated with the intensity and frequency dimensions of social intimacy (Table 6). These findings are consistent with a previous finding that the IAT is significantly and negatively correlated with social support (Korkeila et al., 2010). Another important aspect is that unlike other studies that analyzed the IAT score or IAT item scores (e.g., Jelenchick et al.,
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011
G Model
AJP-502; No. of Pages 9 A.K.M. R. Karim, N. Nigar / Asian Journal of Psychiatry xxx (2013) xxx–xxx
7
Table 5 Correlation matrix for 18-item Bangla IAT and four IAT factors. IAT factors/IAT
F1: Neglect of duty
F2: Online dependence
F3: Virtual fantasies
F4: Privacy and self-defense
IAT
F1: Neglect of duty F2: Online dependence F3: Virtual fantasies F4: Privacy and self-defense IAT
1 .58** .57** .50** .88**
1 .51** .51** .79**
1 .43** .79**
1 .71**
1
Note: N = 172. ** p < .01 (one-tailed).
Table 6 Correlations of the 18-item Bangla IAT and IAT factors with the Bangla MSIS and MSIS dimensions. IAT factors/IAT
MSIS dimensions/MSIS MSIS: frequency
F1: Neglect of duty F2: Online dependence F3: Virtual fantasies F4: Privacy and self defense IAT
MSIS: intensity .21** .28** .22** .25** .29**
.06 .17* .11 .20** .15*
MSIS .16* .26** .19** .25** .25**
Note: N = 172. * p < .05 (one-tailed). ** p < .01 (one-tailed).
2012), we analyzed both the IAT score and IAT factor scores at each level of Internet use (minimal, moderate, and excessive) for male and female participants separately (Table 4). We argue that exploring each factor score is more important to locate the problematic area of the addicted individual than to look into item wise. The findings indicate that males outnumber females in the pathological (moderate/excessive) use of Internet, the ratio being roughly 2:1. However, females in the moderate category score higher than their male counterparts (46.00 7.93 vs 44.35 7.02) whereas females in the excessive category score less than their male counterparts (65.00 vs 67.50 0.71).
As demonstrated, two of the IAT items, i.e., item 07 and item 16, produced lower factor loadings (<.40) in our study. This leads us to suggest that the IAT should be refined in accordance with technological advancement and changes in social life. Since the development of the IAT, advancement in digital technology has offered easy access to Internet through wireless connections and mobile devices (e.g., tablets and smartphones) from anywhere at any time, even when users are flying, traveling or attending a party. This technological advancement has made significant changes in how people live, behave and interact. In recent days, for example, people prefer to get their daily share of news/ information by looking at stories and updates posted by friends or colleagues and relatives on different social media (e.g., Facebook and Twitter) instead of checking and reading emails. Item 7 is about the frequency of checking email before doing something which is important. People usually use Internet to connect with friends, colleagues and relatives who are far away, and form relationships with other people all over the vast world. And this can be easily done by using a variety of professional and social networking sites such as Academia.edu, Facebook, hi5, LinkedIn, Netlog, and Twitter. By using these sites people can instantly share their feelings, thoughts, ideas, interests, documents and even photos/pictures with other people who are far away geographically. As a result, frequent email checking might not be so important nowadays. Thus item 07 i.e., ‘‘How often do you check
Table 7 Comparison of the current IAT factor model with the previous multi-factor solutions. Current IAT factors
IAT Items in brief
Previous multi-factor solutions 2004a
2008b
2010c
2011d
2012e
2012f
F1: Neglect of duty
Item Item Item Item Item Item
01 02 05 06 08 14
Stays online longer than intended Neglects household chores to spend more time online Others in life complain about the time spent online Grade or school work suffers due to spending time online Performance/productivity suffers because of the Internet Loses sleep due to late night logins
2 2 5 3 3 2
2 2 1 2 2 3
2 2 1 1 1 1
2 2 1 2 1, 2m 3
2 2 1 2 2 2
2 2 1 2 2 2
F2: Online dependence
Item Item Item Item
15 17 19 20
Fantasizes about being online when actually offline Tries to cut down time to spend online, but fails Chooses time to spend more online than with others Feels depressed offline and it goes away once back online
1 5 1 2
1 2 1 1
1 2 1 1
3 1 1 3
1 2 1 1
1 2 1 1
F3: Virtual fantasies
Item Item Item Item Item
03 04 10 11 12
Prefers excitement of the Internet to close one’s intimacy Forms new relationships with fellow online users Blocks out disturbing thoughts with those of the Internet Finds anticipating when to go online again Fears of empty and joyless life without the Internet
6 6 1 4 1
1 1 3 – 3
1 1 1 1 1
1 3 1 1 3
1 1 1 1 1
1 1 1 1 1
F4: Privacy and self-defense
Item 09 Item 13 Item 18
Becomes defensive when asked about online activities Snaps, yells or acts annoyed if bothered while online Tries to hide the amount of time spent online
3 1 2
1 1 1
1 1 1
1 3 1
1 1 1
1 1 1
Note: Numbers in table indicate the factors each item belongs to (e.g., 1 indicates that the item belongs to Factor 1, 2 indicates that the item belongs to Factor 2, and so on). (–) item dropped. a Widyanto and McMurran (2004). b Chang and Law (2008). c Korkeila et al. (2010). d Widyanto et al. (2011). e Jelenchick et al. (2012). f Barke et al. (2012). m Item loaded equally on multiple factors.
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011
G Model
AJP-502; No. of Pages 9 A.K.M. R. Karim, N. Nigar / Asian Journal of Psychiatry xxx (2013) xxx–xxx
8
your e-mail before something else that you need to do’’ may no longer represent a symptom of Internet addiction, and therefore produced lower factor loading in our study. Furthermore, we identified item 16 i.e., ‘How often do you find yourself saying ‘‘just a few more minutes’’ when online?’ as a vague item, and therefore it may not be appropriate for assessing pathological use of Internet, particularly in Bangladesh. Consistent with our claim, item 07 has been identified as a problematic item in a couple of the previous studies (Chang and Law, 2008; Barke et al., 2012). However, it is not clear why item 16, which is appropriate in other cultures, is inappropriate in Bangladesh. This item is about the frequency somebody finds himself/herself to say ‘‘just a few more minutes’’ when online. We propose that people who are obsessed with using Internet are not aware of their time spending online. Addicted people get excited and fantasized about the colorful virtual world, and therefore may not possess the mental state represented by item 16. Thus being inappropriate to measure pathological use of Internet this item produced lower factor loading. Therefore, with the exclusion of item 07 and item 16 the IAT can be considered as a valid and reliable instrument for assessing Internet addiction in Bangladesh. Thus the Bangla version IAT comprises 18 items, with a minimum obtainable score of 18 and a maximum score of 90. As with many other studies, this study suffers from a number of limitations, suggesting avenues for future studies. The first limitation is the reliance on a sample of small size. Data from such a small sample might have led to the insufficient reliability of the factor ‘Privacy and self-defense’ (Cronbach’s a = .60). Though the minimum sample required for factor analysis was satisfied here future studies on larger samples can increase the reliability coefficients, confirm the factor structure and other psychometric properties of the IAT in Bangladeshi culture. A second shortcoming is the use of undergraduate/graduate students only. Such a sample of convenience facilitates the early phase of a test construction, but generalizing results to other young adults may not be warranted. So, future studies should address the question of validating the IAT on other young adults. A third limitation of this study is that it did not examine the association of the IAT score with frequency and duration of Internet usage, and did not examine its temporal stability. Future studies can consider these variables to further validate the IAT in Bangladesh and also examine its temporal stability. Finally, a fourth area that warrants further empirical investigation is the social desirability variable that might have affected the present findings. Despite these limitations, the present findings can serve as a base or open the door of further research on the IAT and Internet addiction in Bangladesh. 5. Conclusion This study equips us with a psychometric tool, the Bangla version IAT, to be useful to investigate the pathological or addictive use of Internet technology in Bangladesh and its impacts on people’s mental health. Research using this tool can help us to design Internet management strategies and aid the policy makers and administrators in prevention, diagnosis and treatment of Internet addiction. For example, policy makers and administrators can take appropriate measures to train the counselors, set up counseling centers and offer treatment at counseling centers, clinics, and hospitals. All this together will help maintain good mental health among people, particularly among the vulnerable student population of the country. Funding source None.
Contributors Both the authors significantly contributed in designing and conducting the study, analysis and interpretation of data, preparing first draft and critically appraised the final draft. Conflict of interest Authors have no conflict of interest to disclose with this publication. Acknowledgements Authors would like to express their special thanks to all the Dhaka University students who participated in this study. They are grateful to those experts who judged the accuracy of translation and relevance/suitability of each item for measuring Internet addiction in the socio-cultural context of Bangladesh. Authors are also happy to express their gratitude to the anonymous reviewers who made a number of constructive suggestions on a previous version of this paper. References Barke, A., Nyenhuis, N., Kro¨ner-Herwig, B., 2012. The German version of the Internet Addiction Test: a validation study. Cyberpsychology, Behavior and Social Networking 15 (10) 534–542. BTRC, 2013. Telco News Internet Subscribers in Bangladesh. Retrieved fromhttp:// www.btrc.gov.bd/. Cattell, R.B., 1966. The scree test for number of factors. Multivariate Behavioral Research 1 (2) 245–276. Cattell, R.B., 1978. The Scientific Use of Factor Analysis. Plenum, New York. Chang, M.K., Law, S.P.M., 2008. Factor structure for Young’s Internet Addiction Test: a confirmatory study. Computers in Human Behavior 24 (6) 2597–25619. Everitt, B.S., 1975. Multivariate analysis: the need for data, and other problems. British Journal of Psychiatry 126, 237–240. Ferraro, G., Caci, B., D’Amico, A., Di Blasi, M., 2007. Internet addiction disorder: an Italian study. Cyberpsychology and Behavior 10 (2) 170–175. Field, A.P., 2005. Discovering Statistics Using SPSS, 2nd ed. Sage, London. Floyd, F.J., Widaman, K.F., 1995. Factor analysis in the development and refinement of clinical assessment instruments. Psychological Assessment 7, 286–299. Gorsuch, R.L., 1983. Factor Analysis, 2nd ed. Erlbaum, Hillsdale. Gratz, K.L., Roemer, L., 2004. Multidimensional assessment of emotion regulation and dysregulation: development, factor structure, and initial validation of the difficulties in emotion regulation scale. Journal of Psychopathology and Behavioral Assessment 26 (1) 41–54. Guilford, J.P., 1956. Psychometric Methods. McGraw-Hill, New York. Hawi, N.S., 2013. Arabic validation of the Internet Addiction Test. Cyberpsychology, Behavior and Social Networking 16 (3) 200–204. Internet World Stats, 2013. Internet World Stats: Usage and Population Statistics. Retrieved fromhttp://www.Internetworldstats.com/stats.htm. Jelenchick, L.A., Becker, T., Moreno, M.A., 2012. Assessing the psychometric properties of the Internet Addiction Test (IAT) in US college students. Psychiatry Research 196 (2-3) 296–301. Kaiser, H.F., 1970. A second generation little jiffy. Psychometrika 35 (4) 401–415. Kendall, M.G., Stuart, A., 1958. The Advanced Theory of Statistics. Hafner, New York. Khazaal, Y., Billieux, J., Thorens, G., Khan, R., Louati, Y., Scarlatti, E., Theintz, F., Lederrey, J., Van Der Linden, M., Zullino, D., 2008. French validation of the Internet Addiction Test. Cyberpsychology and Behavior 11 (6) 703–706. Kline, P., 1979. Psychometrics and Psychology. Academic Press, London. Korkeila, J., Kaarlas, S., Ja¨a¨skela¨inen, M., Vahlberg, T., Taiminen, T., 2010. Attached to the web – harmful use of the Internet and its correlates. European Psychiatry 25, 236–241. Kunce, J.T., Cook, D.W., Miller, D.E., 1975. Random variables and correlational overkill. Educational and Psychological Measurement 35 (3) 529–534. Marascuilo, L.A., Levin, J.R., 1983. Multivariate Statistics in the Social Sciences. Brooks/Cole, Monterey. Miller, R.S., Lefcourt, H.M., 1982. The assessment of social intimacy. Journal of Personality Assessment 46 (5) 514–518. Muris, P., 2001. A brief questionnaire for measuring self-efficacy in youths. Journal of Psychopathology and Behavioral Assessment 23 (3) 145–149. Nigar, N., 2012. The impact of Internet addiction on social intimacy and lifestyle. In: The Unpublished Undergraduate Research Project, Department of Psychology. University of Dhaka, Dhaka, Bangladesh. Nunnally, J.C., 1978. Psychometric Theory, 2nd ed. McGraw-Hill, New York.
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011
G Model
AJP-502; No. of Pages 9 A.K.M. R. Karim, N. Nigar / Asian Journal of Psychiatry xxx (2013) xxx–xxx Smith, G.T., McCarthy, D.M., 1995. Methodological considerations in the refinement of clinical assessment instruments. Psychological Assessment 7 (3) 300–308. Widyanto, L., Griffiths, M.D., Brunsden, V., 2011. A psychometric comparison of the Internet addiction test, the Internet-related problem scale, and self-diagnosis. Cyberpsychology, Behavior and Social Networking 14 (3) 141–149.
9
Widyanto, L., McMurran, M., 2004. The psychometric properties of the Internet Addiction Test. Cyberpsychology and Behavior 7 (4) 443–450. Young, K.S., 1996. Internet addiction: the emergence of a new clinical disorder. Cyberpsychology and Behaviour 1 (3) 237–244. Young, K.S., 1998. Caught in the Net. John Wiley and Sons, New York.
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011
AJP-502; No. of Pages 9 Asian Journal of Psychiatry xxx (2013) xxx–xxx
Contents lists available at ScienceDirect
Asian Journal of Psychiatry journal homepage: www.elsevier.com/locate/ajp
The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture A.K.M. Rezaul Karim *, Naima Nigar Department of Psychology, University of Dhaka, 1000 Dhaka, Bangladesh
A R T I C L E I N F O
A B S T R A C T
Article history: Received 18 September 2013 Received in revised form 16 October 2013 Accepted 28 October 2013 Available online xxx
There is growing importance of the Internet Addiction Test (IAT) in Internet addiction research around the world. Since the development of the IAT (Young, 1996, 1998), a number of validation studies have been done in various cultures. The aim of this study was to translate the instrument into Bangla and validate in Bangladeshi culture, a culture vulnerable to Internet addiction. A total of 177 Internet users (77 females and 100 males) participated in the study. Exploratory factor analysis (EFA) of the data from 172 participants (who provided complete responses) identified a four factor structure of the IAT with 18 items. The four factors namely ‘Neglect of duty’, ‘Online dependence’, ‘Virtual fantasies’, and ‘Privacy and self-defense’ together explained 55.68% of the total variance. Problematic (moderate/excessive) users on the IAT scored, on average, higher on each of the four IAT factors as compared to average or nonproblematic (minimal) users consistently across genders. The IAT and its factors showed good internal consistency (Cronbach’s a = .89 for the IAT, and .60–.84 for the factors), strong convergent and discriminant validity. Thus, the Bangla version IAT appears to be valid and reliable and therefore may be used in further research on Internet addiction in the country. ß 2013 Elsevier B.V. All rights reserved.
Keywords: IAT Internet Addiction Bangla Validation
1. Introduction Addictive use of the Internet has emerged as a new and rapidly developing area of addiction research. In researching this topic, a valid and reliable tool is essential. Young (1996, 1998) designed such a research and diagnostic tool, based on DSM-IV criteria for pathological gambling, called the Internet Addiction Test (IAT). The IAT is a widely used tool for assessing the symptoms of Internet addiction defined as any online-related, compulsive behavior which causes impairment of normal social interactions/intimacy (with friends and relatives), and increases daily stress, feeling of loneliness, anxiety and depression. The IAT results can help decide whether someone needs to see a doctor or other mental health professional to discuss his/her addictive use of the Internet and its negative consequences. Because of its growing importance in clinical practice and addiction research, a number of validation studies have been attempted on the English (Widyanto and McMurran, 2004; Widyanto et al., 2011; Jelenchick et al., 2012), Italian (Ferraro et al., 2007), Chinese (Chang and Law, 2008), French (Khazaal et al., 2008), Finnish (Korkeila et al., 2010), German (Barke et al., 2012) and Arabic (Hawi, 2013) versions of the test. These studies have shown moderate-to high-internal consistency of the
* Corresponding author. Fax: +880 2 8615583. E-mail addresses: [email protected], [email protected] (A.K.M. Rezaul Karim).
IAT, with concurrent validity in British culture (Widyanto and McMurran, 2004), convergent and discriminant validity in Chinese (Chang and Law, 2008) and Finish (Korkeila et al., 2010) cultures, and construct validity in American culture (Jelenchick et al., 2012). All these versions have further demonstrated factorial validity. However, there is a wide variation in its factor structure. As reported, it has both the uni-factorial and multi-factorial structures. Specifically, the French and the Arabic versions have shown to comprise only one factor. The study of Korkeila et al. (2010) supported for both a single-factor model (dependent use) and a two-factor model (salient use and loss of control) of the Finnish version. On the other hand, the English version has a 6factor structure (factors: salience, excessive use, neglect of work, anticipation, lack of control, and neglect of social life; Widyanto and McMurran, 2004) and a 3-factor structure (factors: emotional/ psychological conflict, time management issues and mood modification; Widyanto et al., 2011) in British culture as well as a 2-factor structure (factors: dependent use and excessive use; Jelenchick et al., 2012) in American culture. The ‘dependent use’ and ‘excessive use’ of the 2-factor structure of the English version resembles the ‘emotional and cognitive preoccupation’ and ‘loss of control and interference with daily life’ of the 2-factor structure of the German version respectively. However, other versions of the IAT are widely different from each other even when they comprise the same number of factors. Thus factor structure of the IAT varies from study to study, from culture to culture and even within the same culture, and therefore it requires a validation study to be
1876-2018/$ – see front matter ß 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ajp.2013.10.011
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011
G Model
AJP-502; No. of Pages 9 2
A.K.M. R. Karim, N. Nigar / Asian Journal of Psychiatry xxx (2013) xxx–xxx
usable in a new culture. Examining factor structure or dimensionality of an instrument in a new culture is important for accurate specifications of theories (Smith and McCarthy, 1995) and theorydriven research. The dimensions can provide a greater detail of the nature of culture-based addiction than just using Internet addiction as an overall concept. For example, addicts may suffer different levels of negative impacts in different aspects of Internet addiction. Knowing the levels can help us to focus remedy and treatment service on the areas that require special attention. The Internet has broken down geographic and cultural boundaries, expanding from the highly developed to the less- or under-developed countries. However, since its development with the work of Young (1996, 1998) research on Internet addiction has been confined to a few developed countries only. Thus it has been partially and poorly understood. In order to advance our understanding of the phenomenon, research in other countries, especially in the less- and under-developed countries should be initiated. Bangladesh is such a country where the first connectivity of Internet became a reality in 1996. Though it was somehow late, over the last few years the growth has been rapid. The number of Internet users in the country grew from 100,000 in 2000 to 450,000 in 2007, to 556,000 in 2009, to 617,300 in 2010, to 5,501,609 in 2011, and to 8,054,190 in 2012 (Internet World Stats, 2013). According to latest reports of Bangladesh Telecommunication Regulatory Commission (BTRC, 2013), there are about 34,711,101 mobile Internet users, 1,222,620 ISP (Internet Service Provider) plus PSTN (Public circuit-Switched Telephone Networks) users and 315,297 WiMAX (Worldwide Interoperability for Microwave Access) users in Bangladesh. The Internet provides a constant, ever-changing source of information and entertainment, and can be accessed from most smartphones, tablets, laptops and desktop computers. Email, blogs, social networks, instant messaging, and message boards allow for both public and anonymous communication about any topics. While time spent online can be hugely productive, compulsive Internet use (e.g., compulsively checking smartphones, or other mobile devices) can interfere with daily life, work, and relationships. However, due to a lack of scientific enquiry we still do not know so much about this negative side, i.e., the presence and severity of addictive use of Internet and its impact on personal, academic, professional and social life in Bangladesh. We have therefore developed active interest to create a field of addiction research in the country, aiming to focus on pathological use of Internet among the people. As a first step toward this end, we validated the IAT within the socio-cultural context of Bangladesh. We used an exploratory factor analysis (EFA) to examine the psychometric properties and establish our culture-based factor structure of the IAT for university students. University students have easy access to Internet through their own computers (desktops, laptops, and tablets), and/or via mobile phones. We chose university students because we understand from the broadcast of different media (e.g., newspapers, Internet blogs, and T.V.) that they are vulnerable to a variety of social networks such as Facebook, hi5, and Twitter. They are vulnerable because of their age characterized by psychological and developmental dynamics, ready access to the Internet, and an expectation of computer/Internet use. However, future research using this tool will advance our understanding of the motives underlying Internet addiction among university students and help design effective recovery programs for them. 2. Method 2.1. Participants A total of 177 students (77 females and 100 males) studying at the University of Dhaka voluntarily participated in this study. All
the participants had connections to Internet through their personal computers and/or via mobile phones. However, because of incomplete responses 5 participants (2 females and 3 males) were dropped. The age of the remaining 172 participants ranged from 18 to 25 years with a mean of 22.33 and standard deviation (SD) of 2.01. Of the participants, 43.6% were females and 56.4% were males. The female participants’ age ranged from 18 to 25 years with a mean of 22.11 and SD of 1.89 and the male participants’ age ranged from 18 to 25 years with a mean of 22.49 and SD of 2.09. 83.1% of the participants were single, 7.6% were married, 6.4% were in a relationship, 1.2% were divorced, and 1.7% were separated. 2.2. Measures 2.2.1. Internet Addiction Test Developed by Young (1996), Internet Addiction Test (IAT) is the first validated and reliable measure of addictive use of the Internet. It is a 20-item questionnaire that measures psychological dependence, compulsive use, and withdrawal, as well as related problems of school, sleep, family, and time management. The IAT is a 5-point Likert type scale ranging from 1 (rarely) to 5 (always). The minimum obtainable score on the IAT is 20 and the maximum is 100. The higher the score the greater is the level of Internet addiction. Based on the severity of addictive behavior individuals on the IAT can be classified as minimal users (20–39), moderate users (40–69) or excessive users (70–100) (Widyanto and McMurran, 2004; Widyanto et al., 2011). Minimal users are the average online users who have complete control over their Internet usage. Moderate users are those experiencing occasional or frequent problems due to Internet usage. Excessive users are those having significant problems caused by Internet usage. As stated earlier in this article, the English version IAT has exhibited good psychometric properties in various studies (Widyanto and McMurran, 2004; Jelenchick et al., 2012). 2.2.2. Miller Social Intimacy Scale The Bangla version of the Miller Social Intimacy Scale (MSIS) was used to examine the discriminant validity of the Bangla version IAT. The original MSIS is a 17-item tool designed to assess intimacy in a variety of relationships such as friendships, family and spouse (Miller and Lefcourt, 1982). The Bangla version instrument comprises 14 items; 5 items to measure the frequency of intimate contacts and 9 items to assess the intensity of intimate relations (Nigar, 2012). Each of the items is rated on a 10-point scale, ranging from 1 (very rarely) to 10 (almost always). The higher the score on MSIS the higher is the intimacy. Internal consistency (Cronbach’s a) of the original MSIS ranges from .86 to .91, the test–retest reliability coefficients being .84 for one-month interval and .96 for a twomonth interval (Miller and Lefcourt, 1982). The psychometric properties of the Bangla version MSIS also support that it is a reliable and valid measure of social intimacy (Nigar, 2012). 2.3. Procedures 2.3.1. Translating the IAT into Bangla The IAT items were first translated into Bangla, called the first draft. It was then given to four judges, including one expert in Bangla, one expert in English and two experts in Psychology/ Psychometrics. Their native language was Bangla, but being professors/lecturers of a university or college they had also very good command in English. Their task was to judge the accuracy of translation and relevance/suitability of each item for measuring Internet addiction in the socio-cultural context of Bangladesh. Each expert independently rated the translation using a 2-point scale (0 = Not correct, 1 = Correct) and the relevance of each item using
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011
G Model
AJP-502; No. of Pages 9 A.K.M. R. Karim, N. Nigar / Asian Journal of Psychiatry xxx (2013) xxx–xxx
3
responses and were therefore excluded from further processing. Data for the remaining 172 participants were fed into computer for factor analysis on IBM SPSS Statistics 20. According to standard textbook authors and researchers, the minimum sample size for factor analysis varies from 100 (e.g., Kline, 1979; Gorsuch, 1983) to 250 (e.g., Cattell, 1978), and there is practice of applying factor analysis even to the data for less than 100 participants (e.g., Widyanto and McMurran, 2004). There is another set of recommendations varying from a minimum SV (subjects-tovariables) ratio of 2:1 (e.g., Guilford, 1956; Kline, 1979) to 10:1 (e.g., Everitt, 1975; Kunce et al., 1975; Nunnally, 1978; Marascuilo and Levin, 1983). The number of participants in this study was more than 8 times the number of IAT items/variables (20). Thus the sample size required for factor analysis was satisfied. However, before carrying out factor analysis we examined the response distributions of all IAT items and estimated their internal consistency by investigating inter-item correlations and itemtotal correlations. Then we analyzed the data in EFA, a method widely used to uncover the underlying structure of a relatively large set of variables (e.g., Gratz and Roemer, 2004; Hawi, 2013; Korkeila et al., 2010; Muris, 2001; Widyanto et al., 2011; Widyanto and McMurran, 2004). We also examined the discriminant validity by correlating the IAT and IAT subtests with the MSIS and the reliability by estimating internal consistency (Cronbach’s a) of the full test and subtests as well.
another 2-point scale (0 = Not relevant, 1 = Relevant). Following their evaluation, accuracy of the translation was examined by calculating for each item the Accuracy Index (AI = Number of rating at 1/Number of experts). The item yielding an AI of 1 (AI = 4/4) was considered to be correctly and reliably translated. All the four experts rated 16 items’ translation at 1, the AI for each of them becoming 1. The remaining 4 items’ yielded an AI of less than 1. The experts suggested some corrections to the clarity, wording and organization of these items. By reviewing those items in the light of their comments and suggestions the accuracy of translation was ensured. The relevance/suitability of the items in Bangladeshi culture was examined by calculating for each item the Relevance Index (RI = Number of rating at 1/Number of experts). The item yielding an RI of 1 or .75 (RI = 4/4 or RI = 3/4) was considered to be relevant or suitable. All the four experts rated the relevance of each item at 1, the RI for them becoming 1. Thus the second draft of the Bangla version IAT was finalized to administer on the selected participants. 2.3.2. Data acquisition Each participant was briefed about the general study purpose and requested to volunteer in the study. Participants were assured that their responses would be kept confidential and used only for research purposes. After getting their verbal consent the paperbased survey (2nd draft) was administered in person. The survey components included an informed consent statement, sociodemographic section, the IAT and the MSIS. Participants were asked to sign on the consent paper, record the socio-demographic information (age, gender, education, marital status and SES), and read carefully the standard instructions of how to respond before going through the items or questions of the test/scale. Thus the surveys were administered and data were collected over a 6-week period from all the participants.
3. Results 3.1. Factor structure 3.1.1. Item analysis Response distributions of the IAT items indicated that none of these variables/items were excessively skewed or kurtotic (Kendall and Stuart, 1958). The correlation matrix (R-matrix, Table 1) contained no negative values and out of 190 inter-item correlation coefficients 179 were significant, the average inter-item coefficient being .29. The item-total correlations were significant and ranged from .33 to .68 with a mean of .57. So, no item was excluded on the basis of the item response distributions or correlation matrix.
2.4. Data analyses Each participant’s responses to the test items were scored according to the scoring principles of the IAT and the MSIS. Five participants left a few IAT items and/or MSIS items with missing
Table 1 Correlation matrix (R-matrix) for IAT. Item
01
01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20
1 .60** 1 .32** .32** 1 .26** .15* .31** 1 .40** .45** .25** .19** 1 .48** .62** .33** .19** .45** 1 .27** .11 .15* .06 .15* .07 1 .35** .51** .31** .20** .42** .58** .17* 1 .22** .17* .25** .20** .30** .24** .13 .25** 1 .21** .26** .40** .28** .19** .36** .05 .35** .25** 1 .35** .39** .34** .33** .26** .40** .06 .44** .19** .35** 1 .33** .26** .32** .30** .27** .32** .15* .27** .19** .24** .43** 1 .29** .29** .16* .17* .18** .24** .10 .25** .33** .24** .19** .30** 1 .48** .46** .22** .32** .36** .47** .18** .38** .31** .26** .33** .38** .51** 1 .36** .45** .24** .20** .24** .32** .17* .46** .17* .40** .29** .42** .27** .33** 1 .39** .38** .23** .24** .38** .39** .15* .38** .28** .30** .41** .42** .36** .41** .45** 1 .22** .33** .11 .15* .30** .32** .15* .44** .39** .17* .20** .22** .27** .30** .29** .43** 1 .26** .30** .19** .17* .17* .36** .08 .40** .35** .20** .27** .27** .31** .26** .24** .21** .38** 1 .21** .26** .28** .09 .34** .29** .14* .29** .17* .26** .28** .34** .27** .23** .25** .33** .30** .21** 1 .28** .39** .30** .15* .22** .35** .08 .44** .14* .29** .29** .33** .34** .22** .47** .43** .42** .51** .48** 1
IAT total
.65**
02
.67**
03
.50**
04
.43**
05
.58**
06
.67**
07
.33**
08
.68**
09
.49**
10
11
.52**
.59**
12
.60**
13
.55**
14
.67**
15
.62**
16
.67**
17
.56**
18
.53**
19
.52**
20
IAT Total
.61** 1
Note. N = 172; average inter-item correlation = .29; average item-total correlation = .57. * p < .05 (one-tailed). ** p < .01 (one-tailed). Determinant = .001.
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011
G Model
AJP-502; No. of Pages 9 4
A.K.M. R. Karim, N. Nigar / Asian Journal of Psychiatry xxx (2013) xxx–xxx
3.1.2. Factor analysis First, in order to examine whether data were suitable for factor analysis, measures of sampling adequacy were carried out on the 20-item IAT. Inspection of the R-matrix revealed a substantial number (46.84%) of coefficients .30 and above. The determinant of the R-matrix was .001 (>.00001, Field, 2005), indicating that there was no multicolinearity (very highly correlated variables) or singularity (perfectly correlated variables) problem. The Kaiser– Meyer–Olkin (KMO) measure indicated a value of .87 which exceeded the recommended value of .60 (Kaiser, 1970) and Bartlett’s test of sphericity indicated a x2 value of 1197.83 (p < .001). All this together supports the factorability of the Rmatrix. Data for the full set of IAT items were therefore subjected to EFA. Method of principal component (PC) with varimax rotation was used. The initial analysis with eigenvalue > 1.00 (the Kaiser– Guttman criterion) extracted 5 factors, accounting for 57.91% of the total variance (data not shown). However, Floyd and Widaman (1995) suggested that the scree test (Cattell, 1966) is a more accurate method for retaining factors than the Kaiser–Guttman criterion. Inspection of the scree plot revealed a clear break after the 4th component (Fig. 1a), leading us to retain 4 components. Therefore, we analyzed the data in another EFA, limiting the number of factors to 4 with all factor loadings < .40 suppressed. The four factors together accounted for 52.56% of the total variance (data not shown). Results further revealed that two items (item 07and item 16) loaded at < .40. These low factor loadings indicated that perhaps the items do not represent a symptom of problematic Internet use in Bangladeshi culture. After discarding these two items, data were subjected to a final EFA with all factor loadings < .40 suppressed. Though the EFA was run this time without specifying the number of factors a four-factor structure of the IAT with 18 items was identified, consistently with the scree plot (Fig. 1b). These four factors, which were rotated to position of maximum orthogonality in eight iterations, explained together 55.68% of the total variance (Table 2). Factor 1 accounts for 17.44% of the variance, Factor 2 accounts for 13.73% of the variance, Factor 3 accounts for 13.58% of the variance, and Factor 4 accounts for
10.93% of the variance. Before labeling the factors we identified three pairs of cross-loadings between the factors. Specifically, item 08 was cross loaded on Factor 1 and Factor 2 with the loadings of .54 and .49 respectively; item 14 was cross-loaded on Factor 1 and Factor 4 with the loadings of .55 and .52 respectively; and item 17 was cross-loaded on Factor 2 and Factor 4 with the loadings of .52 and .49 respectively. We grouped both item 08 and item 14 under Factor 1, the factor of their greater loadings and best conceptual fit; and item 18 under Factor 4, the factor of its smaller loading but best conceptual fit. Thus Factor 1 comprises items 1, 2, 5, 6, 8 and 14 which we termed as ‘Neglect of duty’, Factor 2 comprises items 15, 17, 19 and 20 which we termed as ‘Online dependence’, Factor 3 comprises items 3, 4, 10, 11 and 12 which we termed as ‘Virtual fantasies’, and Factor 4 comprises items 9, 13 and 18 which we termed as ‘Privacy and self-defense’. 3.2. IAT scores and IAT factor scores We proposed for the Bangla version 18-item IAT a cutoff scores system equivalent to that of the English version 20-item IAT (Table 3). Using this cutoff scores system, the majority of participants i.e., 63.95% were categorized as ‘minimal users’ (IAT score < 36), 34.30% were categorized as moderate users (62 IAT score 36) and only 1.74% as excessive users (IAT score > 62). Of the female participants, 77.33% were categorized as ‘minimal users’, 21.33% as moderate users and only 1.33% as excessive users. Of the male participants, 53.61% were categorized as ‘minimal users’, 44.33% as moderate users and only 2.06% as excessive users. Thus the percentage of males is about double the percentage of females falling in the moderate/higher level of Internet addiction. The mean IAT score calculated over all participants was 33.75 11.14. Table 4 provides detailed descriptive statistics for the full IAT and four IAT factors. Regardless of participant’s sex, the mean IAT score was 26.93 4.89 for minimal users, 44.80 7.25 for moderate users and 66.67 1.53 for excessive users (Table 4). The mean IAT score was 25.76 4.88 for female minimal users, 46.00 7.93 for female moderate users and 65.00 for female
Fig. 1. The scree plots generated in EFA: (a) for 20 items, and (b) for 18 items.
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011
G Model
AJP-502; No. of Pages 9 A.K.M. R. Karim, N. Nigar / Asian Journal of Psychiatry xxx (2013) xxx–xxx
5
Table 2 Rotated factor matrix for a reduced set of IAT items (item 07 and item 16 discarded). IAT items in brief
Factor loadings F1
Item Item Item Item Item Item Item Item Item Item Item Item Item Item Item Item Item Item
01 02 05 06 08 14 15 17 19 20 03 04 10 11 12 09 13 18
Stays online longer than intended Neglects household chores to spend more time online Others in life complain about the time spent online Grade or school work suffers due to spending time online Performance/productivity suffers because of the Internet Loses sleep due to late night logins Fantasizes about being online when actually offline Tries to cut down time to spend online, but fails Chooses time to spend more online than with others Feels depressed offline and it goes away once back online Prefers excitement of the Internet to close one’s intimacy Forms new relationships with fellow online users Blocks out disturbing thoughts with those of the Internet Finds anticipating when to go online again Fears of empty and joyless life without the Internet Becomes defensive when asked about online activities Snaps, yells or acts annoyed if bothered while online Tries to hide the amount of time spent online
F2 .72 .80 .64 .72 .54 .55
F4
(.49) (.52) .47 .52 .60 .81
(.49)
.66 .66 .62 .58 .56 .74 .65 .47
(.53) 3.14 17.44% .84
Eigenvalue Variance explained Cronbach’s (standardized) a
F3
2.47 13.73% .70
2.45 13.58% .71
1.97 10.93% .60
Note. N = 172. Factor loadings < .40 were suppressed. Items corresponding to the parenthesized loadings did not conceptually fit with the corresponding factors. Extraction method: principal component analysis. Rotation method: varimax with Kaiser normalization. Rotation converged in 8 iterations.
excessive users. The mean IAT score was 28.23 4.61 for male minimal users, 44.35 7.02 for male moderate users and 67.50 0.71 for male excessive users. Thus males, on average, scored higher in the higher level whereas females scored higher in the moderate level of Internet addiction. Figures in Table 4 further indicate that problematic (both moderate and excessive) users on the IAT scored, on average, higher on each of the four IAT factors consistently across sexes as compared to average or non-problematic (minimal) users. 3.3. Validity 3.3.1. Convergent validity Convergent validity of the Bangla version IAT was examined by estimating inter-factor correlations and the factor-IAT total correlations. The inter-factor correlations (Pearson’s r) were all significant, the coefficient ranging from .43 to .58 (Table 5). The strongest correlation was found between F1 (Neglect of duty) and F2 (Online dependence) and the weakest between F3 (Virtual fantasies) and F4 (Privacy and self-defense). The four IAT factors also significantly correlated with the full IAT, with the coefficients ranging from .71 to .88. The IAT had strongest correlation with F1 Table 3 Cutoff scores for categorizing Internet users on the IAT. User category
20-Item English IATa,b
18-Item Bangla IAT
Minimal users Moderate users Excessive users
20–39 40–69 70–100
18–35 36–62 63–90
Minimal users: Lower score = Number of items 1.0, Upper score = Number of items 2.0–1.0. Moderate users: Lower score = Number of items 2.0, Upper score = Number of items 3.5–1.0. Excessive users: Lower score = Number of items 3.5, Upper score = Number of items 5.0. a Widyanto et al. (2011). b Widyanto and McMurran (2004).
(Neglect of duty) and the weakest with F4 (Privacy and selfdefense). 3.3.2. Discriminant validity Discriminant validity of the Bangla version IAT was examined by correlating the IAT and IAT factors with the Bangla version MSIS and MSIS dimensions. Negative or non-significant correlations were hypothesized between the measures in order for them to have discriminant validity. As hypothesized, the IAT and IAT factors were correlated significantly and negatively with the MSIS (r = .16 to .25) and MSIS: Intensity (r = .21 to .29, Table 6). The IAT and two of the IAT factors (F2: Online dependence, F4: Privacy and selfdefense) significantly and negatively correlated with the MSIS: Frequency (r = .15 to .20), and the other two Factors (F1: Neglect of duty, F3: Virtual fantasies) non-significantly correlated with the MSIS: Frequency (r = .06 to .11). All these correlations strongly support the discriminant validity of the measures. 3.4. Reliability 3.4.1. Internal consistency The interitem correlation matrix (Table 1) contained no negative values, indicating that the items were measuring the same characteristic. The reliability of the Bangla version IAT was further examined by estimating internal consistency. The coefficients of Cronbach’s a were calculated. Cronbach’s a (standardized) for the full IAT was .89. The coefficients for the ‘Neglect of duty’, ‘Online dependence’, ‘Virtual fantasies’, and ‘Privacy and self- defense’ subtests were .84, .70, .71, and .60 respectively (Table 2). 4. Discussion This study was designed to investigate the psychometric properties of a Bangla version of the IAT. Analyzing the data in EFA we identified a four factor model for the IAT comprising 18
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011
G Model
AJP-502; No. of Pages 9
(1.00) (2.31) (1.53) (1.00) (1.53) 23.00 13.33 18.33 12.00 66.67 Note. (–) SD was not possible to determine as n = 1. The minimum obtainable score on the ‘Neglect of duty’ is 6 and the maximum score is 30. The minimum obtainable score on the ‘Online dependence’ is 4 and the maximum score is 20. The minimum obtainable score on the ‘Virtual fantasies’ is 5 and the maximum score is 25. The minimum obtainable score on the ‘Privacy and self-defense’ is 3 and the maximum score is 15. The minimum obtainable score on the IAT is 18 and the maximum score is 90.
Range Mean (SD) Range Mean (SD)
(2.77) 6–18 16.41 (3.98) 8–26 (1.37) 4–10 8.88 (3.15) 4–17 (2.39) 5–18 11.63 (3.35) 6–21 (1.47) 3–10 7.88 (2.29) 3–15 (4.98) 18–35 44.80 (7.25) 36–61
Range Mean (SD)
22–24 9.28 12–16 5.05 17–20 7.80 11–13 4.79 67–68 26.93 (1.41) (2.83) (2.12) (1.41) (.71) 23.00 14.00 18.50 12.00 67.50
Range Mean (SD) Range Mean (SD)
23–23 10.19 (2.94) 6–18 16.47 (4.29) 8–26 12–12 5.12 (1.37) 4–9 8.44 (2.77) 4–17 18–18 8.08 (2.40) 5–15 11.65 (3.24) 6–21 12–12 4.85 (1.23) 3–8 7.80 (2.26) 3–15 65–65 28.23 (4.61) 18–35 44.35 (7.02) 36–61
Range Mean (SD) Range Mean (SD)
(3.13) 10–21 23.00 (–) (3.84) 4–16 12.00 (–) (3.72) 7–19 18.00 (–) (2.42) 3–12 12.00 (–) (7.93) 37–59 65.00 (–) 16.25 10.06 11.56 8.13 46.00
Moderate users (n = 16)
Range Mean (SD)
(2.33) 6–15 (1.39) 4–10 (2.38) 5–18 (1.63) 3–10 (4.88) 18–35
Mean (SD)
Neglect of duty 8.45 Online dependence 5.00 Virtual fantasies 7.55 Privacy and self-defense 4.74 IAT 25.76
Excessive users (n = 3) Moderate users (n = 59) Minimal users (n = 110)
Total (n = 172)
Excessive users (n = 2) Minimal users (n = 52)
Moderate users (n = 43) Male (n = 97)
Minimal users (n = 58)
Excessive users (n = 1) Female (n = 75) IAT factors/IAT
Table 4 Descriptive statistics for 18-item Bangla IAT and four IAT factors.
22–24 12–16 17–20 11–13 65–68
A.K.M. R. Karim, N. Nigar / Asian Journal of Psychiatry xxx (2013) xxx–xxx Range
6
items (2 items dropped; Table 2). Factor 1 (6 items) measures neglect of duty (e.g., avoiding household works, avoiding study or duty, and complaints of time spent online), Factor 2 (4 items) measures online dependence (e.g., cutting down online time, preferring time to spend with online friends, and feeling depressed offline), Factor 3 (5 items) measures virtual fantasies i.e., how addicted people tend to substitute the real world (e.g., preferring online relationships to spouse’s intimacy, thoughts of online excitements and new relationships), and Factor 4 (3 items) measures privacy and self-defense (e.g., being defensive and aggressive if asked about online activities, and hiding time spent online). These factors together accounted for 55.68% of the total variance, their individual contributions ranging from 10.93% to 17.44% (Table 2). The factors show good to high internal consistency (Cronbach’s a = .60 to .84), the first one being most reliable as indicated by its highest coefficient (Cronbach’s a = .84). As shown in Table 5, all the factors were significantly correlated with the whole IAT (r = .71–.88, p < .01). The moderate (linear) inter-correlations of the factors (r = .43–.58, p < .01) support the representation of the factors as unique, distinguishable components (Jelenchick et al., 2012). As mentioned earlier in this paper, previous analyses have shown between one- and six-factor solutions for the IAT. In spite of this wide disagreement, factor arrangements from prior analyses in Table 7 show striking resemblances to our results. An inspection of this table indicates that the greatest agreement was found between our four-factor model and the two-factor models of both the English (Jelenchick et al., 2012) and German (Barke et al., 2012) versions. That is, Factor 1 (Neglect of duty) of the Bangla version IAT most closely resembles, in content items (item no. 1, 2, 6, 8 and 14), the second factors of the two-factor structures of the English and German versions (English: Excessive use, German: Loss of control and interference with daily life). Factor 2 (Online dependence), Factor 3 (Virtual fantasies) and Factor 4 (Privacy and self-defense) together represent well the first factors of the English and German versions (English: Dependent use, German: Emotional and cognitive preoccupation). The items clustered within the first factor of the Bangla version also show broad overlap (item no. 1, 2, 6 and 8) with the second factors (English: Time management issues, Chinese: Time management and performance) of the three-factor structures of both the English (Widyanto et al., 2011) and Chinese (Chang and Law, 2008) versions. Thus items 1, 2, 6, 8 and possibly a few others tend to cluster together, forming a factor to be stable across cultures. This reveals that the most common symptoms of Internet addiction across cultures are unusual stay online, complaints of time spent online, avoidance of daily activities such as household works, poor school grades or performance and sleep disturbances. However, discrepancies still exist between the remaining three factors of our model and those of prior analyses. This was most likely due to differences inherent in the samples such as differences in sociocultural characteristics or cultural norms surrounding Internet use, rather than differing representations of the overall construct. One interesting and distinct aspect of this study is that we examined here the discriminant validity of the IAT and IAT dimensions/factors by estimating correlations with social intimacy. We hypothesized that social intimacy would be negatively related with the IAT and IAT dimensions. As hypothesized, we found social intimacy to be significantly and negatively correlated with the IAT (r = .25, p < .01) and IAT dimensions (r = .16 to .26, p < .05 or .01). This was also evident when the IAT and IAT dimensions were correlated with the intensity and frequency dimensions of social intimacy (Table 6). These findings are consistent with a previous finding that the IAT is significantly and negatively correlated with social support (Korkeila et al., 2010). Another important aspect is that unlike other studies that analyzed the IAT score or IAT item scores (e.g., Jelenchick et al.,
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011
G Model
AJP-502; No. of Pages 9 A.K.M. R. Karim, N. Nigar / Asian Journal of Psychiatry xxx (2013) xxx–xxx
7
Table 5 Correlation matrix for 18-item Bangla IAT and four IAT factors. IAT factors/IAT
F1: Neglect of duty
F2: Online dependence
F3: Virtual fantasies
F4: Privacy and self-defense
IAT
F1: Neglect of duty F2: Online dependence F3: Virtual fantasies F4: Privacy and self-defense IAT
1 .58** .57** .50** .88**
1 .51** .51** .79**
1 .43** .79**
1 .71**
1
Note: N = 172. ** p < .01 (one-tailed).
Table 6 Correlations of the 18-item Bangla IAT and IAT factors with the Bangla MSIS and MSIS dimensions. IAT factors/IAT
MSIS dimensions/MSIS MSIS: frequency
F1: Neglect of duty F2: Online dependence F3: Virtual fantasies F4: Privacy and self defense IAT
MSIS: intensity .21** .28** .22** .25** .29**
.06 .17* .11 .20** .15*
MSIS .16* .26** .19** .25** .25**
Note: N = 172. * p < .05 (one-tailed). ** p < .01 (one-tailed).
2012), we analyzed both the IAT score and IAT factor scores at each level of Internet use (minimal, moderate, and excessive) for male and female participants separately (Table 4). We argue that exploring each factor score is more important to locate the problematic area of the addicted individual than to look into item wise. The findings indicate that males outnumber females in the pathological (moderate/excessive) use of Internet, the ratio being roughly 2:1. However, females in the moderate category score higher than their male counterparts (46.00 7.93 vs 44.35 7.02) whereas females in the excessive category score less than their male counterparts (65.00 vs 67.50 0.71).
As demonstrated, two of the IAT items, i.e., item 07 and item 16, produced lower factor loadings (<.40) in our study. This leads us to suggest that the IAT should be refined in accordance with technological advancement and changes in social life. Since the development of the IAT, advancement in digital technology has offered easy access to Internet through wireless connections and mobile devices (e.g., tablets and smartphones) from anywhere at any time, even when users are flying, traveling or attending a party. This technological advancement has made significant changes in how people live, behave and interact. In recent days, for example, people prefer to get their daily share of news/ information by looking at stories and updates posted by friends or colleagues and relatives on different social media (e.g., Facebook and Twitter) instead of checking and reading emails. Item 7 is about the frequency of checking email before doing something which is important. People usually use Internet to connect with friends, colleagues and relatives who are far away, and form relationships with other people all over the vast world. And this can be easily done by using a variety of professional and social networking sites such as Academia.edu, Facebook, hi5, LinkedIn, Netlog, and Twitter. By using these sites people can instantly share their feelings, thoughts, ideas, interests, documents and even photos/pictures with other people who are far away geographically. As a result, frequent email checking might not be so important nowadays. Thus item 07 i.e., ‘‘How often do you check
Table 7 Comparison of the current IAT factor model with the previous multi-factor solutions. Current IAT factors
IAT Items in brief
Previous multi-factor solutions 2004a
2008b
2010c
2011d
2012e
2012f
F1: Neglect of duty
Item Item Item Item Item Item
01 02 05 06 08 14
Stays online longer than intended Neglects household chores to spend more time online Others in life complain about the time spent online Grade or school work suffers due to spending time online Performance/productivity suffers because of the Internet Loses sleep due to late night logins
2 2 5 3 3 2
2 2 1 2 2 3
2 2 1 1 1 1
2 2 1 2 1, 2m 3
2 2 1 2 2 2
2 2 1 2 2 2
F2: Online dependence
Item Item Item Item
15 17 19 20
Fantasizes about being online when actually offline Tries to cut down time to spend online, but fails Chooses time to spend more online than with others Feels depressed offline and it goes away once back online
1 5 1 2
1 2 1 1
1 2 1 1
3 1 1 3
1 2 1 1
1 2 1 1
F3: Virtual fantasies
Item Item Item Item Item
03 04 10 11 12
Prefers excitement of the Internet to close one’s intimacy Forms new relationships with fellow online users Blocks out disturbing thoughts with those of the Internet Finds anticipating when to go online again Fears of empty and joyless life without the Internet
6 6 1 4 1
1 1 3 – 3
1 1 1 1 1
1 3 1 1 3
1 1 1 1 1
1 1 1 1 1
F4: Privacy and self-defense
Item 09 Item 13 Item 18
Becomes defensive when asked about online activities Snaps, yells or acts annoyed if bothered while online Tries to hide the amount of time spent online
3 1 2
1 1 1
1 1 1
1 3 1
1 1 1
1 1 1
Note: Numbers in table indicate the factors each item belongs to (e.g., 1 indicates that the item belongs to Factor 1, 2 indicates that the item belongs to Factor 2, and so on). (–) item dropped. a Widyanto and McMurran (2004). b Chang and Law (2008). c Korkeila et al. (2010). d Widyanto et al. (2011). e Jelenchick et al. (2012). f Barke et al. (2012). m Item loaded equally on multiple factors.
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011
G Model
AJP-502; No. of Pages 9 A.K.M. R. Karim, N. Nigar / Asian Journal of Psychiatry xxx (2013) xxx–xxx
8
your e-mail before something else that you need to do’’ may no longer represent a symptom of Internet addiction, and therefore produced lower factor loading in our study. Furthermore, we identified item 16 i.e., ‘How often do you find yourself saying ‘‘just a few more minutes’’ when online?’ as a vague item, and therefore it may not be appropriate for assessing pathological use of Internet, particularly in Bangladesh. Consistent with our claim, item 07 has been identified as a problematic item in a couple of the previous studies (Chang and Law, 2008; Barke et al., 2012). However, it is not clear why item 16, which is appropriate in other cultures, is inappropriate in Bangladesh. This item is about the frequency somebody finds himself/herself to say ‘‘just a few more minutes’’ when online. We propose that people who are obsessed with using Internet are not aware of their time spending online. Addicted people get excited and fantasized about the colorful virtual world, and therefore may not possess the mental state represented by item 16. Thus being inappropriate to measure pathological use of Internet this item produced lower factor loading. Therefore, with the exclusion of item 07 and item 16 the IAT can be considered as a valid and reliable instrument for assessing Internet addiction in Bangladesh. Thus the Bangla version IAT comprises 18 items, with a minimum obtainable score of 18 and a maximum score of 90. As with many other studies, this study suffers from a number of limitations, suggesting avenues for future studies. The first limitation is the reliance on a sample of small size. Data from such a small sample might have led to the insufficient reliability of the factor ‘Privacy and self-defense’ (Cronbach’s a = .60). Though the minimum sample required for factor analysis was satisfied here future studies on larger samples can increase the reliability coefficients, confirm the factor structure and other psychometric properties of the IAT in Bangladeshi culture. A second shortcoming is the use of undergraduate/graduate students only. Such a sample of convenience facilitates the early phase of a test construction, but generalizing results to other young adults may not be warranted. So, future studies should address the question of validating the IAT on other young adults. A third limitation of this study is that it did not examine the association of the IAT score with frequency and duration of Internet usage, and did not examine its temporal stability. Future studies can consider these variables to further validate the IAT in Bangladesh and also examine its temporal stability. Finally, a fourth area that warrants further empirical investigation is the social desirability variable that might have affected the present findings. Despite these limitations, the present findings can serve as a base or open the door of further research on the IAT and Internet addiction in Bangladesh. 5. Conclusion This study equips us with a psychometric tool, the Bangla version IAT, to be useful to investigate the pathological or addictive use of Internet technology in Bangladesh and its impacts on people’s mental health. Research using this tool can help us to design Internet management strategies and aid the policy makers and administrators in prevention, diagnosis and treatment of Internet addiction. For example, policy makers and administrators can take appropriate measures to train the counselors, set up counseling centers and offer treatment at counseling centers, clinics, and hospitals. All this together will help maintain good mental health among people, particularly among the vulnerable student population of the country. Funding source None.
Contributors Both the authors significantly contributed in designing and conducting the study, analysis and interpretation of data, preparing first draft and critically appraised the final draft. Conflict of interest Authors have no conflict of interest to disclose with this publication. Acknowledgements Authors would like to express their special thanks to all the Dhaka University students who participated in this study. They are grateful to those experts who judged the accuracy of translation and relevance/suitability of each item for measuring Internet addiction in the socio-cultural context of Bangladesh. Authors are also happy to express their gratitude to the anonymous reviewers who made a number of constructive suggestions on a previous version of this paper. References Barke, A., Nyenhuis, N., Kro¨ner-Herwig, B., 2012. The German version of the Internet Addiction Test: a validation study. Cyberpsychology, Behavior and Social Networking 15 (10) 534–542. BTRC, 2013. Telco News Internet Subscribers in Bangladesh. Retrieved fromhttp:// www.btrc.gov.bd/. Cattell, R.B., 1966. The scree test for number of factors. Multivariate Behavioral Research 1 (2) 245–276. Cattell, R.B., 1978. The Scientific Use of Factor Analysis. Plenum, New York. Chang, M.K., Law, S.P.M., 2008. Factor structure for Young’s Internet Addiction Test: a confirmatory study. Computers in Human Behavior 24 (6) 2597–25619. Everitt, B.S., 1975. Multivariate analysis: the need for data, and other problems. British Journal of Psychiatry 126, 237–240. Ferraro, G., Caci, B., D’Amico, A., Di Blasi, M., 2007. Internet addiction disorder: an Italian study. Cyberpsychology and Behavior 10 (2) 170–175. Field, A.P., 2005. Discovering Statistics Using SPSS, 2nd ed. Sage, London. Floyd, F.J., Widaman, K.F., 1995. Factor analysis in the development and refinement of clinical assessment instruments. Psychological Assessment 7, 286–299. Gorsuch, R.L., 1983. Factor Analysis, 2nd ed. Erlbaum, Hillsdale. Gratz, K.L., Roemer, L., 2004. Multidimensional assessment of emotion regulation and dysregulation: development, factor structure, and initial validation of the difficulties in emotion regulation scale. Journal of Psychopathology and Behavioral Assessment 26 (1) 41–54. Guilford, J.P., 1956. Psychometric Methods. McGraw-Hill, New York. Hawi, N.S., 2013. Arabic validation of the Internet Addiction Test. Cyberpsychology, Behavior and Social Networking 16 (3) 200–204. Internet World Stats, 2013. Internet World Stats: Usage and Population Statistics. Retrieved fromhttp://www.Internetworldstats.com/stats.htm. Jelenchick, L.A., Becker, T., Moreno, M.A., 2012. Assessing the psychometric properties of the Internet Addiction Test (IAT) in US college students. Psychiatry Research 196 (2-3) 296–301. Kaiser, H.F., 1970. A second generation little jiffy. Psychometrika 35 (4) 401–415. Kendall, M.G., Stuart, A., 1958. The Advanced Theory of Statistics. Hafner, New York. Khazaal, Y., Billieux, J., Thorens, G., Khan, R., Louati, Y., Scarlatti, E., Theintz, F., Lederrey, J., Van Der Linden, M., Zullino, D., 2008. French validation of the Internet Addiction Test. Cyberpsychology and Behavior 11 (6) 703–706. Kline, P., 1979. Psychometrics and Psychology. Academic Press, London. Korkeila, J., Kaarlas, S., Ja¨a¨skela¨inen, M., Vahlberg, T., Taiminen, T., 2010. Attached to the web – harmful use of the Internet and its correlates. European Psychiatry 25, 236–241. Kunce, J.T., Cook, D.W., Miller, D.E., 1975. Random variables and correlational overkill. Educational and Psychological Measurement 35 (3) 529–534. Marascuilo, L.A., Levin, J.R., 1983. Multivariate Statistics in the Social Sciences. Brooks/Cole, Monterey. Miller, R.S., Lefcourt, H.M., 1982. The assessment of social intimacy. Journal of Personality Assessment 46 (5) 514–518. Muris, P., 2001. A brief questionnaire for measuring self-efficacy in youths. Journal of Psychopathology and Behavioral Assessment 23 (3) 145–149. Nigar, N., 2012. The impact of Internet addiction on social intimacy and lifestyle. In: The Unpublished Undergraduate Research Project, Department of Psychology. University of Dhaka, Dhaka, Bangladesh. Nunnally, J.C., 1978. Psychometric Theory, 2nd ed. McGraw-Hill, New York.
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011
G Model
AJP-502; No. of Pages 9 A.K.M. R. Karim, N. Nigar / Asian Journal of Psychiatry xxx (2013) xxx–xxx Smith, G.T., McCarthy, D.M., 1995. Methodological considerations in the refinement of clinical assessment instruments. Psychological Assessment 7 (3) 300–308. Widyanto, L., Griffiths, M.D., Brunsden, V., 2011. A psychometric comparison of the Internet addiction test, the Internet-related problem scale, and self-diagnosis. Cyberpsychology, Behavior and Social Networking 14 (3) 141–149.
9
Widyanto, L., McMurran, M., 2004. The psychometric properties of the Internet Addiction Test. Cyberpsychology and Behavior 7 (4) 443–450. Young, K.S., 1996. Internet addiction: the emergence of a new clinical disorder. Cyberpsychology and Behaviour 1 (3) 237–244. Young, K.S., 1998. Caught in the Net. John Wiley and Sons, New York.
Please cite this article in press as: Karim, A.K.M. R., Nigar, N., The Internet Addiction Test: Assessing its psychometric properties in Bangladeshi culture. Asian J. Psychiatry (2013), http://dx.doi.org/10.1016/j.ajp.2013.10.011