Reliability and Validity of the Korean Version of the Internet Addiction Test among College Students (2013).

Internet addiction is a new clinical entity defined as a maladaptive pattern of internet use causing clinically significant impairment or distress to affected individuals (1). Official diagnostic criteria for internet addiction do not exist yet, however, and the disorder has been considered as either impulse control disorder (1) or behavior addiction (2). The upcoming Diagnostic and Statistical Manual of Mental Disorders, fifth edition (DSM-5) will include internet addiction in its appendix (3). The prevalence of internet addiction varies according to the methodology and population studied, but in some countries such as Korea, it is substantial; for example, it was estimated that 8.5% of the total population are currently affected by this disorder (4). It is therefore not difficult to understand why the Korean Government termed internet addiction a serious public health issue and established an independent government agency for policy making and for treatment of those suffering from the problem (5).

Internet addiction has been also designated pathologic internet use (6), compulsive internet use (7), and problematic internet use (8). Although there are some minor differences between the proposed diagnostic criteria, all share common elements such as excessive use of internet, withdrawal, tolerance, and negative consequences for interpersonal or personal well-being (9). Several tools have been developed and tested for their psychometric properties; these include the Internet Addiction Test (IAT) (10), Generalized Problematic Internet Use Scale (11), and Korea internet addiction scale (12). Among these, the IAT has been most widely used and well-tested for its psychometric properties (13). This 20-item Likert-type questionnaire was developed for screening and measuring levels of internet addiction. Each item is rated from 1 (rarely) to 5 (always) and total scores can range from 20 to 100. Although the norms and cut-off score of the IAT are not established, Young has suggested the score above 70 causes significant problems (10). Items of the IAT include compulsive behavior related to use of the internet, the occupational or academic difficulties, lack of competence at home, problems in interpersonal relations, and emotional problems (10).

The excellent psychometric properties of the original version are well documented in the literature (13), and good reliability and validity data have been reported for other language versions, thus suggesting the adaptability of the IAT to other cultures. These languages include Chinese (14), French (15), Italian (16), Portuguese (17), Finnish (18), German (19), and Malay (20). In Korea, two major translated versions have been used in (21,22), and studies used them often with minor modifications depending on populations targeted. Psychometric data of the Korean versions are available including good internal consistencies (Cronbach alpha 0.79-0.94) and mixed results for factor structure (23). Criterion validities have not reported and test-retest reliability was shown in only one study (24); furthermore, during the development no process of back-translation was conducted, which may limit the cross-cultural adaptability of the original scale (25). Therefore, in this study, we developed a Korean version of the IAT (KIAT) through a process of forward and back translation and examined its reliability and validity in a sample of college students.

Measures  

Translation and back-translation

Internet Addiction Diagnostic Questionnaire

The Korea Internet addiction scale

The Patient Health Questionnaire-9

Statistical analysis

Reliability

Factorial validity

Concurrent and convergent validity

In this study, we translated and adapted the IAT to the Korean language and found good reliability and validity of the translated version. First, the internal consistency was excellent (Cronbach’s alpha > 0.90), this value is better than those that have been reported for the original version (13) but similar to other language versions (15,17). And item-to-total correlations and Cronbach’s alpha values with deletion of individual items showed that the internal consistency was generally stable. However, one exception was item 4; it had a low correlation, and overall internal consistency exceeded that of total items when the item was deleted. We therefore had to exclude the item for the factor analysis. Item 4 concerns newly formed social relations on the internet: “How often do you form new relationships with fellow on-line users?” We believe that our result reflects recent change in the internet environment where many young people now build their social relationships through social-networking service such as Facebook (31). The issue of the validity issue of item 4 was also raised in two recent factor analytic studies: one of Korean college students (26) and the other of US students (32). Therefore, item 4 nowadays has more relevance to an average pattern of internet use rather than being a construct for internet addiction. In line with change in pattern of internet use, we propose that the item 4 needs to be revised.

Our study is one of a few studies to investigate the test-retest reliability of the IAT. One Korean study using a different translation of the IAT reported two-week correlation of r = 0.85 among high school students (23). A recent German study reported similar two-week reliability of r = 0.83 among college students (19). Our study also confirmed the temporal stability of the KIAT among college students.

In our exploratory factor analysis, four factors were extracted. Others have proposed various factor solutions: one factor (15,18), two factor (19,31), three (33,34), five (20), and six factors (13,16,17). These variations may be explained by differences in language versions (culture or translation), population studied (online sample or college students), and methods of factor extraction. Our finding of five factors is new but is in line with common elements in the instruments measuring internet addiction: 1) compulsive internet use and excessive time spent; 2) withdrawal symptoms; 3) using the internet for social comfort; 4) negative consequences (34).

The six-factor structure found in the first factor analytic study of the IAT by Widyanto and McMurran (13) is of limited significance as these authors recruit a small online sample of 86 participants of diverse backgrounds and nationality. Further studies failed to replicate this factor solution, although one Portuguese study (17) extract six factors from a group of university students, but the items clustered in each domain coincided only partially with the original version. Recent studies on larger samples of students support a fewer factors: Jelenchick et al. (32) identified two factors (dependent use and excessive use) among 215 US college students; Korkeila et al. (18) and Barkes et al. (19) supported two factor solution among university students. A recent study on Korean university students also found two-factor solution as the best fitting model for the IAT (34). This two-factor structure was similar to that identified in the US and Finnish study (18,31). The items clustered as Factor 1 in our study are identical to “Excessive Use” and Factor 2, 3, 4 are items in “Dependent Use” in the study of Jelenchick et al. (32). Thus, although the number of factors in our exploratory factor analysis is larger than in these studies, our finding point to similarity to different language versions in factorial validity of the IAT.

Convergent validity of the KIAT was demonstrated by the significant correlation with depression, which is one of the most commonly reported symptomatic correlates of internet addiction (35). Other studies have reported convergent validity of the IAT with time of internet use and specific online activities (14), and with frequency of internet use (35). The concurrent validity of the KIAT was shown by demonstrating significant correlation with other established measures of internet addiction. Studies reported significant correlations of the IAT with the Compulsive Internet Use Scale and Chen Internet Addiction Scale (36).

Limitations of this study were as follows. First, participants in this study were students from a single university who volunteered through school advertisements. There needs to be a careful consideration for representativeness of this sample for the method of sampling was not randomized. Second, we did not investigate the detailed activities undertaken on the internet, which may have yielded insight into aspects of internet overuse. Third, as the KIAT is a self-administered scale, we cannot rule out effects of denial or minimization on the part of the respondents (37). Future study may benefit from combined use of questionnaires by spouses or parents. Finally, our study did not investigate the discriminant validity and diagnostic utility of the KIAT; for example, cut-off scores between normal and pathological internet users and comparison with clinical interviews for internet addiction disorder will be necessary. Our results need to be replicated with other populations including adolescents, community population, and those seeking mental health services. And to shed more light on factor structure of the KIAT, confirmatory factor analysis is required to confirm our finding and to compare with other factor solutions suggested from previous studies.

The significance of this study is as follows: first, we confirmed the test-retest reliability and concurrent validity of the KIAT, which has hardly examined in the literature. Second, although there existed two older Korean versions of the IAT, only our version was produced by backward translation, which is an important procedural element when one requires cross-cultural adaptation of a scale. Third, by changing the item 7 we were able to extract a more stable factor structure and achieve better construct validity. Thus, with respect to the revised version of the IAT, we recommend that “email” in item 7 should be reworded as “the internet” and that item 4 should be deleted or altered to reflect recent changes in the significance of social networks in the medium of the internet.

In conclusion, the KIAT had an excellent internal consistency and high test-retest reliability. It also has concurrent validity as shown by the significant correlation with other scales reflecting internet addiction. A four-factor structure, comparable to the original version, suggests adequate factorial validity of the KIAT. The KIAT is a sound psychometric measure that can be used for screening for, and research on, internet addiction among the Korean-speaking population.