Psychometric Validation of the Compulsive Internet Use Scale

Abstract

Compulsive Internet usage is on the rise in developing countries such as India. To date, no tested, validated, and verified instrument for measuring compulsive Internet use among Indian adolescents has been made available. In order to bridge this gap, our present study has examined the psychometric properties of the English version of the Compulsive Internet Use Scale (CIUS) with 2,381 adolescent Internet users (aged 12–19) in India. Exploratory and confirmatory factor analyses were conducted to examine the factorial and construct validity, reliability, and homogeneity of the English version of the CIUS. Relationships between adolescents’ CIUS scores, demographics, Information and communication technologies (ICT) accessibility, and problematic ICT use were also examined. The study results confirm that the CIUS has good psychometric properties, high internal reliability, and homogeneity and is a valid self-reporting instrument for measuring compulsive Internet use. The present study reveals the same factor structure as the earlier studies using the CIUS in other languages. Finally, we found that male and older adolescents experience higher compulsive Internet use compared to female and younger adolescents, while compulsive Internet users experience lower life satisfaction, lower academic performance, and problematic ICT use including Internet, mobile, and online gaming.

Keywords

Adolescents compulsive Internet use empirical study ICT accessibility psychometric validation problematic ICT use

Introduction

Internet usage has increased tremendously in the recent years. The Internet provides endless opportunities to its users, including communication, social network building, online gaming, and so on. But the continuous use of the Internet can also become excessive and uncontrollable in some cases. Excessive use of the Internet might lead to various problems such as loss of sleep, preoccupation with the Internet, and poor social skills (Griffiths & Wood, 2000; Liu & Potenza, 2007; Young, 1996; Young & Case, 2004). Earlier research examining the impact of Internet addiction on real-life well-being found that preoccupation with the Internet can impact work and academic performance as well as one’s sexual, marriage, and social life (Krajewska-Kulak et al., 2011; Young, 1999). Internet addiction, in some cases, can even lead to psychiatric disorders (Yen et al., 2008) such as social phobias and depression (Yen, Ko, Yen, Wu, & Yang, 2007) and substance misuse (Batthyany, Muller, Benker, & Wolfling, 2009). As a result, researchers have pointed out the need for verified and validated Internet addiction measurement instruments to better understand this phenomenon (Chang & Law, 2008).

India is one of the fastest growing developing economies, with the second largest Internet user base. In December 2013, there were nearly 213 million Internet users in India, with a growth of over 40% from the previous year (India Internet Users, 2013). The Internet user base will reach 243 million by June 2014 (India Internet Users, 2013), and mobile Internet users will cross the 155 million mark by March 2014 (India Internet Users, 2013). India has already overtaken the United States, which accounts for 207 million Internet users. Currently, India is second after China, which is leading the world with over 300 million Internet users. The sharp increase in adoption and use of the Internet among the Indian population has fueled growing concern about compulsive Internet use among its population. It should be noted that the majority of existing Internet users in India are youths, aged below 24 years, who have been found to be spending an increasing amount of time daily on the Internet (Youth Internet Users, 2013). Unfortunately, despite this sharp increase in Internet usage among the Indian population, there is, to the best of our knowledge, no psychometrically validated instrument for screening compulsive use of the Internet in the Indian population. Psychometrically valid and verified instruments for measuring compulsive Internet use are needed in order to gain better understanding of this quite recent phenomenon. Therefore, in order to address this gap in the existing literature, the main goal of the present study is to examine the psychometric validity of the English version of a 14-item Compulsive Internet Use Scale (CIUS) with a sample of Indian adolescents aged between 12 and 19 years. In addition to this, the present study also examined the relationship between compulsive Internet use, the adolescents’ demographic profiles, ICT accessibility, and problematic ICT use among adolescents.

Background Literature

Psychometric Validations

Researchers have emphasized the need to conduct short questionnaire-based studies with larger samples, in order to measure the severity of core elements of compulsive Internet use behavior (Meerkerk, Van Den Eijnden, Vermulst, & Garretsen, 2009). To date, several diagnostic instruments for compulsive Internet use are available but most of them lack psychometric validation (Wartberg, Petersen, Kammerl, Rosenkranz, & Thomasius, 2014). The two most commonly used psychometrically validated instruments are Kimberly Young’s Internet Addiction Test (IAT; Young, 1998) and the CIUS (Meerkerk et al., 2009).

The purpose of this study was to validate an instrument for fast screening compulsive Internet use based on consensus mental disorder behaviors. Therefore, the brief and concise 14-item CIUS was selected. The CIUS provides a one-dimensional score, that is, severity of compulsive Internet use. The CIUS items were developed based on the 10 criteria for pathological gambling and 7 criteria for substance dependence in the Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition; American Psychological Association, 1994). In addition, 6 criteria for behavioral addiction formulated by Griffiths (1999), based on the earlier work by Brown (1993) and Marks (1990), were also considered. The CIUS covers the core elements of addiction behavior, namely, loss of control (4 items), withdrawal symptoms (1 item), conflict (4 items), coping, escape and mood modification (2 items), and preoccupation (3 items; Meerkerk et al., 2009). The composite score is a determinant of compulsive Internet use (Wartberg et al., 2014).

Empirical Findings on the CIUS

At present, seven psychometric validations of the CIUS exist, and six of the seven have resulted in a one-dimensional structure (Guertler et al., 2014; Khazaal et al., 2011; Khazaal et al., 2012; Meerkerk et al., 2009; Peukert et al., 2012; Wartberg et al., 2014; Table 1). However, the CIUS validation in Persian suggested a three-factor solution (Alavi, Jannatifard, Eslami, & Rezapour, 2011). Five of the seven existing CIUS validations were conducted in Europe (Guertler et al., 2014; Khazaal et al., 2012; Meerkerk et al., 2009; Peukert et al., 2012; Wartberg et al., 2014), and two in the Middle East (Alavi et al., 2011; Khazaal et al., 2011). There is a need to validate the CIUS with Asian Internet users, since Asian countries account for 42.86% of the world’s total Internet users, with over 1.2 billion Internet users dwelling in Asia (Internet Usage in Asia, 2013). It has also been noted that all existing CIUS validations were performed with wide age-groups (e.g., 11–80 years), but specific age-groups such as adolescents aged 12 to 19 have not yet been taken into account (Table 1).

Table 1.

Review of Previous Validations of the Compulsive Internet Use Scale.

Validation and Authors	Sample Size, Age-Group	Results
Dutch CIUS^a(Meerkerk et al., 2009)	Study A composed of two samples A1 (n = 447) and A2(n = 229) with mean age (SD) = 38.5 (12.5) years and Study B (n = 16,925) aged 11–80 years (M = 25.3, SD = 10.0)	Model A1: RMSEA = 0.053, CFI = 0.984; Model A2: RMSEA = 0.084, CFI = 0.966; Model B: RMSEA = 0.054, CFI = 0.986. Items 1–2, 6–7, 8–9, 10–11, 12–13 were correlated. One factor based on CFA, α = .89
French CIUS^b(Khazaal et al., 2012)	126 users (84.3% females) aged 16–45 years. Mean age 23.3 (SD = 5.9)	Model F1: RMSEA = 0.11, CFI = 0.84; Model F2: RMSEA = 0.08, CFI = 0.92. Items 1–2, 12–13 were correlated. EFA, MAP, PA, CFA, α = .91, one factor
Arabic CIUS^b (Khazaal et al., 2011)	186 users (51% female) aged 15–25 years, Mean age (SD) = 17.6 (2.9) years	Model Ar1: RMSEA = 0.09, CFI = 0.78; Model Ar2: RMSEA = 0.06, CFI = 0.90. Items 4–7, 4–11, 5–7, 5–11, 8–9, 12–13 were correlated. EFA, MAP, PA, CFA, α = .70, one factor
Persian CIUS (Alavi et al., 2011)	400 Persian students (age was not specified)	Only EFA, α = .89, three-factor solution
German CIUS (Peukert et al., 2012)	2,506 students where age was 18–54 years	Item 1–2, 6–7, 8–9, 10–11, 12–13 were correlated. Only CFA, α = .89, one factor
German CIUS^b (Wartberg et al., 2014)	1,744 adolescents aged 14–17 years, mean age 15.5 (SD = 1.12) years old. Male–female ratio 1:1	Model G1: RMSEA = 0.122, CFI = 0.856; Model G2: RMSEA = 0.060, CFI = 0.968. Items 1–2, 6–7, 8–9, 10–11, 12–13 were correlated. Only CFA, α = .93, one factor solution
German CIUS^b (Guertler et al., 2014)	8,132 participants (mean age = 35.24 (SD = 13.79) years, 51.1% were male	Model Ga1: CFI = 0.861, RMSEA = 0.052; Model Ga2: CFI = 0.965, RMSEA = 0.026. Item 1–2, 6–7, 8–9, and 12–13 were correlated. Only CFA, one-factor solution

Note. MAP = Velicer’s minimum average partial; PA = parallel analysis; CIUS = Compulsive Internet Use Scale; CFA = confirmatory factor analysis; CFI, comparitive fit index; EFA = exploratory factor analysis; RMSEA = root mean squared error of approximation.

^aIn the Dutch CIUS validation, all three models were hypothesized as single factor models with five correlated errors. ^bIn all these cases, the first model does not and the second model does include correlated error terms.

The CIUS instrument offers various benefits over other instruments, including concise structure and ease of use for screening in clinical settings and for online studies (Meerkerk et al., 2009). The CIUS has high internal consistency and test–retest validity; a single-factor structure was found in different independent samples (Meerkerk et al., 2009; Wartberg et al., 2014; see Table 1). Previous studies have revealed that the CIUS score was correlated with time spent online (Khazaal et al., 2011; Meerkerk et al., 2009), compulsive Internet use (Meerkerk et al., 2009; Wartberg et al., 2014), scores on the Online Cognition Scale (Meerkerk et al., 2009), diminished impulse control (Meerkerk et al., 2009), parents’ perception of frequency of excessive media use, and gaming console use (Wartberg et al., 2014). Similarly, significant positive correlations were found between adolescents’ CIUS scores, frequency of excessive media use, and frequency of conflicts at home (Wartberg et al., 2014). In addition to these, negative significant correlations between the CIUS scores and satisfaction with life were found (Wartberg et al., 2014). In the Arabic (Khazaal et al., 2011) and French validations (Khazaal et al., 2012), no significant relationship was found between the CIUS scores and gender or age. This contradicts earlier research, where gender differences in compulsive Internet use were prominent. Male adolescents were at greater risk (Choi et al., 2009; Ferraro, Caci, D’Amico, & Blasi, 2007; Ko et al., 2005), tended toward escapism or avoidance (Billieux et al., 2011), and played more online games than their female counterparts (Ha et al., 2007). Furthermore, earlier studies have also suggested that younger Internet users face greater risk of Internet addiction (Ferraro et al., 2007; Widyanto & McMurran, 2004).

Methods

Aims and Research Questions of the Study

In the present study, the following research questions were investigated:

Research Question 1: Do exploratory and confirmatory factor analyses of the English version of the CIUS reveal a one-factor model as shown by other validations in Dutch, French, Arabic, and German?

Research Question 2: How reliable (measured by α) and homogeneous (index of measurement error and value of alpha if item deleted) is the English version of the CIUS?

Research Question 3: Does the English version of the CIUS fulfill different criteria for instrument validity and reliability?

Research Question 4: What are the relationships between the CIUS scores, demographics, ICT accessibility, and problematic ICT use?

Research Question 5: How can adolescents’ demographics, ICT accessibility, and problematic ICT use predict compulsive Internet use among adolescents?

Data Collection

A sample of 2,383 Indian adolescents (aged 12–19 years) was randomly drawn from a total of 10 junior and senior high schools. A paper-and-pencil survey was organized in English during November 2013–January 2014. The first author recruited the participants after giving a workshop entitled, “The incidence of Internet addiction among Indian adolescents” in the participating schools. Participation in the study was voluntary and anonymous. Participating schools were private and English was the language of instruction and communication. Participants were given an incentive of learning their personal CIUS score and the associated score interpretation after completing the survey. After completing the survey, they were informed how to calculate their own CIUS score. Participants calculated it and returned the actual survey copy to the researchers. All required ethical clearances and permissions were obtained from the participating schools before conducting this research. Before the actual study, a pilot test of the overall survey was performed with 25 students (representing target users aged between 12 and 19 years). Pilot testing and follow-up informal interviews were performed in order to understand if the target population found any of the survey items to be unclear, confusing, or difficult to understand and interpret. Simplifications of a few difficult English words were carried out so that the study participants could easily understand all items of the survey. Oral informed consent was obtained from the participating schools and students. The administration of participating schools and students were clearly informed about the study’s aim, objectives, and outcomes. The data screening revealed that a total of 12 cases had more than 50% missing data, so they were removed, resulting in an effective sample size of 2,369 adolescents. A total of 40.1% (n = 950) were female and 59.6% (n = 1,412) were male. The mean age of the study participants was 14.5 years (SD = 1.26), and the mean CIUS score for the sample was 32.5 (SD = 10.2).

Measures

CIUS

A total of 14 items of the CIUS were scored on a 5-point Likert-type scale, coded as never = 0, rarely = 1, occasionally = 2, frequently = 3, and very frequently = 4. Three adaptations of the original CIUS version were deliberately made in order to address the target user group, that is, school-going adolescents. These changes were “Item 3: Do others (e.g., friends and family) say you should use the Internet less?” “Item 4: Do you prefer to use the Internet instead of spending time with others (e.g., friends and family)?” and “Item 10: Do you rush through your homework/schoolwork in order to go on the Internet?” These changes are in line with the recommendations of earlier research studies (Khazaal et al., 2011; Van der Aa et al., 2009). Before answering the CIUS, the study participants were asked to take into account their overall Internet usage inside and outside for either leisure or study. There were two main reasons behind this choice: (1) During our field studies in the participating schools, several students expressed that on many occasions, Internet use for academic purposes turn into leisure use. For example Student A planned to utilize Internet for completing a school assignment but later he or she started using Internet for nonacademic reasons, for example chatting with friends, accessing social media pages, listening to music, and accessing other entertainment content on Internet. (2) It was found that most participating schools were having 3–6 hr per week of Internet computer classes. We learned from observation exercises that although teachers keep strict vigilance that students must utilize Internet solely for academic reasons; however, many times, students use Internet for leisure purposes.

IAT

Twenty items of the IAT were scored on a 5-point Likert-type scale, coded as strongly disagree = 1, disagree = 2, neutral = 3, agree = 4, and strongly agree= 5. The IAT was utilized for examining the concurrent validity of the CIUS.

Demographics

A total of 5 demographic items were included in the survey, namely, age (12–19 years), gender (Male = 1, Female = 0), economic condition (evaluated using 4-point Likert-type scale namely, difficult = 1, so-so = 2, middle class = 3, rich = 4), academic performance (percentage marks received in the last annual exams examined using a 4-point Likert-type scale of below 40% = 1, 41–60% = 2, 61–80% = 3, above 80% = 4), and satisfaction with life (refers to the current level of satisfaction with life as a whole assessed using a 5-point Likert-type scale of completely dissatisfied = 1, somewhat dissatisfied = 2, don’t know = 3, somewhat satisfied = 4, completely satisfied = 5; see Table 2).

Table 2.

Descriptive Statistics for the Demographic Information.

Demographic Variables	Category	Percentage (Frequency)
Age	12–13	23.0 (541)
	14	28.3 (663)
	15	29.5 (692)
	16	12.9 (304)
	17–19	6.4 (151)
Gender	Male	59.6 (1412)
Gender	Female	40.1 (950)
Economic Condition	Difficult	0.3 (8)
	So-So	2.6 (61)
	Middle class	84.2 (1995)
	Rich	11.5 (272)
Academic performance	Below 40%	1.3 (31)
	41–60%	16.6 (394)
	61–80%	41.7 (988)
	Above 80%	38.5 (911)
Satisfaction with life	Completely dissatisfied	3.5 (82)
	Somewhat dissatisfied	8.4 (198)
	Don’t know	23.4 (555)
	Somewhat satisfied	28.4 (673)
	Completely satisfied	34.0 (806)

ICT accessibility

A total of 7 items covering different aspects of ICT accessibility were included in the survey. These included Internet at home (Yes = 1, No = 0), mobile phone ownership (Yes = 1, No = 0), and mobile phone Internet (Yes = 1, No = 0). Another four variables, namely, daily Internet use, number of years of Internet use, Internet activity, and frequency of excessive Internet use, were assessed using a 5-point Likert-type scale (Table 3). Mobile phone Internet includes the use of the mobile Internet on the respondent’s own personal phone as well as on their parents’ mobile phone.

Table 3.

Descriptive Statistics of ICT Accessibility.

ICT Accessibility	Category	Percentage (Frequency)
Internet at Home	Yes	78.0 (1849)
Internet at Home	No	21.4 (506)
Personal mobile phone	Yes	50.1 (1186)
Personal mobile phone	No	48.6 (1152)
Mobile phone Internet	Yes	50.5 (1197)
Mobile phone Internet	No	47.8 (1133)
Number of years of Internet use	<1 year	24.4 (579)
	1–2 years	27.8 (659)
	2–4 years	28.2 (667)
	4–6 years	14.0 (331)
	More than 7 years	4.1 (98)
Daily Internet usage	<1 hr	61.3 (1453)
	1–3 hr	28.5 (676)
	3–5 hr	5.5 (130)
	5–7 hr	1.7 (40)
	>7 hr	1.0 (24)
Excessive Internet use	Never	22.2 (525)
	Rarely	25.7 (610)
	Sometimes	36.6 (868)
	Very often	8.7 (205)
	Always	4.9 (115)
Overall Internet activity	Don’t use much	13.8 (326)
	Less active	19.7 (466)
	Neutral	30.7 (727)
	Active	23.8 (565)
	Very active	10.0 (237)

Problematic ICT use

A total of 3 items covering problematic ICT use among adolescents were included. These were problematic use of the Internet, mobile phones, and online gaming, evaluated using a 4-point Likert-type scale (see Table 4). Participants were clearly informed that the 3 items were enquiring about their personal experience with the problematic use of the Internet, mobile phones, and online gaming, for example, do they consider that the Internet, mobile phone, and online gaming use causes problems for their personal and social well-being?

Table 4.

Descriptive Statistics of Problematic ICT Use.

Problematic ICT Use	Category	Percentage (Frequency)
Problematic Internet use?	Unproblematic	45.6 (1080)
	Low problem	30.9 (732)
	Medium problem	18.6 (440)
	High-level problem	3.1 (73)
Problematic Mobile phone use?	Unproblematic	54.4 (1288)
	Low problem	24.4 (579)
	Medium problem	13.9 (330)
	High-level problem	4.4 (104)
Problematic Online gaming use?	Unproblematic	50.1 (1188)
	Low problem	25.5 (603)
	Medium problem	14.3 (339)
	High-level problem	7.3 (174)

Statistical Analyses

In this study, IBM SPSS 21.0 and AMOS 21.0 for Windows, and Factor¹ were used for performing the statistical analyses and treatments. First, the data were screened using a missing value analysis procedure, and the range of minimum and maximum missing data points was 0.4–1.7%. In order to address the problem of missing data, we consulted a considerable amount of methodological literature on how to handle the missing data. Traditional methods of handling missing data, such as listwise deletion, pairwise deletion, and mean substitution, introduce Type II errors and tend to underestimate the correlations and regression weights of the sample data (Acock, 2005; Fichman & Cummings, 2003). In order to examine if there is a specific pattern of missing values (Acock, 2005; Little & Rubin, 1987), Little’s Missing Completely at Random (MCAR) test (Little & Rubin, 1987) was used to find if the data were missing completely at random. The results of Little’s MCAR test resulted in a statistically significant value, (χ² = 810.0, df = 740, and p = .04). This shows that our data is not missing completely at random. In other words, the data is either missing at random (MAR) or missing not at random (MNAR). Most researchers agree that the Expectation–Maximization (EM) algorithm is an appropriate technique to impute missing values compared to all other available techniques (Schafer & Graham, 2002). Furthermore, EM is often unbiased with MNAR data hence data imputation using EM is not a mater of concern (Schafer & Graham, 2002). Therefore, the missing values from the survey items were imputed using the EM algorithm of SPSS (Allison, 2001; Pigott, 2001).

Second, the normal distribution of the variables was examined by calculating skewness and kurtosis for all the variables, which were found to be in the acceptable range, that is ±1 for a symmetrical or normal distribution (Byrne, 2001; George & Mallery, 2003; Hair, Anderson, Tatham, & Black, 1998). Moreover, according to Tabachnick and Fidell (1996), outliers have a deleterious impact on the cross-validation procedures and the output of statistical tests. The presence of outliers in the sample data was examined through the calculation of the Z-score for all the survey items. The present study data sample did not contain any outliers based on the Z-score criteria recommended by Tabachnick and Fidell (1996) and Stevens (1996).

The rest of the statistical analysis was performed in six stages, namely, (1) examining the reliability and homogeneity, (2) exploratory factor analysis (EFA) to reveal the internal structure of the CIUS data, (3) confirmatory factor analysis (CFA) with Maximum Likelihood Estimation (MLE) to evaluate the model fit of the factorial structure, (4) different types of instrument validity and reliability, (5) Pearson’s correlation and t-test to understand the relationships between the CIUS scores, demographic variables, ICT accessibility, and problematic ICT use, and (6) multiple hierarchical regression to predict the CIUS scores among adolescent Internet users based on demographics, ICT accessibility, and problematic ICT use.

Results

Reliability and Homogeneity

The 14-item CIUS returned a Cronbach’s α value of .87 (N = 2,361), which shows high internal consistency among the CIUS items (Cronbach & Meehl, 1955; DeVellis, 2003; Khazaal et al., 2011; Streiner, 2003). The α value was used to calculate the index of measurement error for the 14-item CIUS, using the formula (1 − [Square of (α)]; Tavakol & Dennick, 2011). The CIUS returned a very small measurement error value of .23.

EFA

The process of EFA is usually followed when there do not exist any a priori factorial structures for the instrument being considered. However, in the case of the CIUS, discrepancy still exists in the correct factorial structure since the Persian CIUS (Alavi et al., 2011) concluded a three-factor structure. In addition to this, unlike previous CIUS validations, the present study involved adolescents aged 12–19 years from India. Therefore, we decided to examine the factorial structure for the CIUS by first utilizing EFA of the Indian CIUS sample.

The entire sample was randomly split into two nearly equal data sets, namely, Sample A (N = 1161) and Sample B (N = 1208). Sample A was utilized for performing EFA while Sample B was used for CFA. Before performing EFA, the suitability of the CIUS data for EFA was examined by inspecting the values of the correlation matrix for communality, the Kaiser-Meyer-Olkin (KMO; Kaiser, 1970) test for sampling adequacy, and the Bartlett’s test for sphericity (Bartlett, 1954). The correlation matrix returned minimum and maximum values of .10 and .55, and the KMO was very good (.91), well over the required threshold of .60. It also passed the Bartlett’s test (χ² = 5193.13, df = 91, p < .01). Additionally, an interitem correlation matrix of 14 CIUS items returned no negative values, which indicates that all the items were measuring similar behavior, phenomena, or characteristics of the participants. These results supported the appropriateness of EFA.

Later, the EFA of 14-item CIUS using Maximum Likelihood was performed using SPSS 22.0. The EFA returned a two-factor solution, and loadings for all 14 items were above the minimum threshold of 0.40, except item 8 (see Table 5). The two-factor solution was based on the Kaiser criterion which states that eigenvalue >1.0. However, previous CIUS validations have clearly shown that Kaiser criterion tend to overestimate the number of factors, hence other techniques should be practiced for deciding the number of factorial solution. The factorial solution for CIUS was examined using Velicer’s Minimum average partial (O’Connor, 2000) which recommended one-factor solution where averaged partial value was .01124. Later, Parallel Analysis with optimal implementation (Velicer, 1976) and Catell’s scree test (Catell, 1966) were also examined which also returned a single-factor solution for the CIUS.

Table 5.

Exploratory Factor Analysis of the 14-Item CIUS.

Item	English CIUS (How often … )	EFA (N = 1,161)
Item	English CIUS (How often … )	Factor 1	Factor 2
Item 1	Do you find it difficult to stop using the INTERNET when you are online?	.49	.21
Item 2	Do you often continue to use the INTERNET despite your intention to stop?	.60	.21
Item 3	Do others (e.g., friends and family) say you should use the INTERNET less?	.48	.47
Item 4	Do you prefer to use the INTERNET instead of spending time with others (e.g., friends and family)?	.55	−.01
Item 5	Are you short of sleep because of the INTERNET?	.63	−.03
Item 6	Do you think about the INTERNET, even when not online?	.69	−.10
Item 7	Do you look forward to your next INTERNET session?	.65	−.05
Item 8	Do you think you should use the INTERNET less often?	.31	.50
Item 9	Have you unsuccessfully tried to spend less time on the INTERNET?	.41	.34
Item 10	Do you rush through your (homework)/(schoolwork) in order to go on the INTERNET?	.60	−.03
Item 11	How often do you neglect your daily obligations (work, school, or family life) because you prefer to go on the INTERNET?	.69	−.15
Item 12	How often do you go on the INTERNET when you are feeling down?	.62	−.25
Item 13	How often do you use the INTERNET to escape from your sorrows or get relief from negative feelings?	.64	−.20
Item 14	How often do you feel restless, frustrated, or irritated when you cannot use the INTERNET?	.66	−.18
Eigen Value		5.33	1.46
% Variance explained		38.06	10.39

Note. CIUS = Compulsive Internet Use Scale; EFA = exploratory factor analysis. Total variance explained by the two-factor solution was 48.45% and Cronbach’s α value for 14-item CIUS was .87.

CFA

Because the EFA of the CIUS concluded that a one-factor model was the best fit, it was assumed that only a single latent factor entitled “compulsive Internet use” exists. In order to assess the adequacy of the one-factor model, various goodness-of fit-indices, in addition to other statistical measures, were utilized using Sample B (Table 6). The CFA involved testing two models, namely, Model A, where correlation between residual error terms was not permitted, and Model B, where correlation was permitted, similar to previous CIUS validations (Khazaal et al., 2011; Khazaal et al., 2012; Meerkerk et al., 2009; Wartberg et al., 2014). Model A returned a χ² ratio df value of 10.61 where p < .01. The acceptable range for a good fit between the research model and the collected empirical data is between 2 and 1, while a reasonable fit is between 3 and 1 (Byrne, 1989; Carmines & McIver, 1981). The χ² ratio df for Model A was too high, but according to Pratarelli and Browne (2002), the χ² ratio df test is also dependent on the sample size. Instead, the suitability of the model fit was evaluated based on the goodness-of-fit indices. Model A returns an root mean square error of approximation value of .09 which is considered mediocre fit (Browne & Cudeck, 1993), comparitive fit index = .86 and Tucker–Lewis index = .84, which shows that Model A has a mediocre fit (Hu & Bentler, 1999). The results suggest that a discrepancy between the hypothetical model and the empirical data exists, since all indices were not good, although Model A presented a mediocre fit. In structural equation modelling, not only identification of the model is important but also the quality of the model fit should be ensured (Khazaal et al., 2011). Considering the cutoff criteria suggested by Kline (2011, p. 27), Model A was rejected.

Table 6.

Confirmatory Factor Analysis of the 14-item CIUS.

Item	English CIUS (How often … )	CFA (N = 1,208)
Item	English CIUS (How often … )	Model A	Model B
Item 1	Do you find it difficult to stop using the INTERNET when you are online?	.50	.45
Item 2	Do you often continue to use the INTERNET despite your intention to stop?	.61	.58
Item 3	Do others (e.g., friends and family) say you should use the INTERNET less?	.46	.43
Item 4	Do you prefer to use the INTERNET instead of spending time with others (e.g., friends and family)?	.53	.53
Item 5	Are you short of sleep because of the INTERNET?	.64	.65
Item 6	Do you think about the INTERNET, even when not online?	.70	.72
Item 7	Do you look forward to your next INTERNET session?	.63	.63
Item 8	Do you think you should use the INTERNET less often?	.33	.29
Item 9	Have you unsuccessfully tried to spend less time on the INTERNET?	.40	.37
Item 10	Do you rush through your (homework)/ (schoolwork) in order to go on the INTERNET?	.61	.61
Item 11	How often do you neglect your daily obligations (work, school, or family life) because you prefer to go on the INTERNET?	.70	.71
Item 12	How often do you go on the INTERNET when you are feeling down?	.61	.59
Item 13	How often do you use the INTERNET to escape from your sorrows or get relief from negative feelings?	.63	.60
Item 14	How often do you feel restless, frustrated, or irritated when you cannot use the INTERNET?	.66	.67
Fit indices	Recommended values
χ²ratio df	≤3.0	10.61	4.37
CFI	≥.95	.86	.95
GFI	≥.95	.90	.96
TFI	≥.92	.84	.94
RMSEA	≤.05	.09	.05

Note. CIUS = Compulsive Internet Use Scale; CFA = confirmatory factor analysis; CFI, comparitive fit index; EFA = exploratory factor analysis; GFI = goodness-of-fit index; RMSEA = root mean squared error of approximation. The cronbach’s α value for 14-item CIUS (N = 1,208) was .87.

In the next step, Model B was tested. Standardized residuals are fitted residuals that are divided with their own asymptotic standard error (Khazaal et al., 2011). If the standardized residuals are too large, it shows that the model under consideration has a certain misspecification. Different researchers have given different cutoffs for the standardized residuals, for example, Jöreskog and Sörbom (1999) emphasized that standardized residuals between items pairs should be <2.58, while Schumacker and Lomax (2004) favored a value <1.96 for good fit. Since our sample size is large, a cutoff value of 2.58 was considered so as to discover potential misspecifications and constraints in the correlation measurement errors. The correlation of the error variances is only performed when they meet the two criteria given by Kline (2011), namely, (a) the Pearson correlation value between error variances should be greater than .3 and (b) the connecting residual errors should be theoretically grounded. The standard residual covariances exceeding cutoff values and meeting the two criteria (Kline, 2011) were relaxed by correlating them. Similar to Meerkerk et al. (2009) and Wartberg, Petersen, Kammerl, Rosenkranz, and Thomasius (2014), the error variances between the five pairs were correlated (e1–e2, e3–e8, e3–e9, e8–e9, and e12–e13). The Pearson’s correlation for all pairs was higher than .30. Previous CIUS validations have justified correlating error variances of item pairs because items show overlap in their content. Model B returned improved goodness-of-fit indices after the correlation of error variances (see Table 6). All the goodness-of-fit indices indicate that Model B has good fit (Hu & Bentler, 1999; Kline, 2011).

CIUS Validity and Reliability

Prior research has indicated that the CIUS has shown sufficient convergent and criterion validity (Meerkerk et al., 2009). Moreover, this study has examined four types of scale validities and reliabilities. These are (1) Content validity: This ensured that the utilized instrument is correctly representing the different aspects of the underlying phenomena. Since the CIUS is a well-known compulsive Internet use instrument and it has already received five psychometric validations, its content validity is ensured. (2) Construct reliability: This refers to the internal consistency of the utilized instrument. Our present study revealed that the CIUS has α value of .88 (very good), and thus construct reliability is ensured. (3) Face validity: This ensures that the utilized instrument is easy to comprehend and appears valid to the study participants. Face validity for the CIUS was ensured through a short pilot study and later an informal interview with the study participants. (4) Concurrent validity: This evaluates if the relationship shared between the utilized instrument and other variables is consistent with earlier findings. It was found that the CIUS score was highly correlated with Young’s IAT score where r = .75, p < .01. Furthermore, the CIUS had a significant positive correlation with daily Internet use time (r = .35, p < .01). Both these findings confirm the presence of the concurrent validity of the English version of the CIUS.

Relationship Between CIUS and Other Variables

Pearson’s correlation was utilized to examine the relationship between the CIUS, demographics, ICT accessibility, and Problematic ICT use. The CIUS scores were found to be in positive agreement with respondents’ age (r = .17, p < .01, CI₉₅ = [.13, .21]), but relationship was very weak agreement with economic condition (r = .05, p < .05, CI₉₅ = [.01, .09]) of the Internet users. In comparison, CIUS scores were found to be in negative agreement with satisfaction with life (r = −.15, p < .01, CI₉₅ = [−.19, −.11]) and academic performance (r = −.17, p < .01, CI₉₅ = [−.21, −.13]). An independent sample t-test revealed that male adolescents (t = −11.09, p <. 01, Cohen’s d = .47, effect size = .23) scored higher on the CIUS, mean (SD) = 34.37 (10.01), compared to female adolescents, mean (SD) = 29.72 (9.97).

For the ICT accessibility and CIUS relationship, the CIUS scores shared a significant positive relationship with mobile phone ownership, r = .25, p < .01, CI₉₅ = [.21, .29], Internet connectivity on personal mobile phones, r = .24, p < .01, CI₉₅ = [.20, .27], time spent on daily Internet use, r = .35, p < .01, CI₉₅ = [.32, .39], number of years of Internet use, r = .19, p < .01, CI₉₅ = [.15, .23], excessive Internet use, r = −.35, p < .01, CI₉₅ = [−.38, −.30], and overall Internet activity, r = −.31, p < .01, CI₉₅ = [−.34, −.26]. Furthermore, CIUS score shared was very weak agreement with Internet connectivity at home, r = .08, p < .01, CI₉₅ = [.04, .12].

The correlations between the CIUS scores and problematic Internet use, r = .27, p < .01, CI₉₅ = [.20, .28], problematic mobile phone use, r = .14, p < .01, CI₉₅ = [.10, .18] and problematic online gaming, r = .11, p < .01, CI₉₅ = [.07, .15], were significant and positive.

Predicting CIUS Among Adolescents

Multiple hierarchical regressions were performed in order to examine the relative influence of demographics, ICT accessibility, and problematic ICT use on predicting compulsive Internet use among adolescents (see Table 7). It was found that demographic variables, ICT accessibility, and problematic ICT use accounted for 11.1%, 17.7%, and 2.3%, respectively, of variance explained in the CIUS score. Overall, the model was able to predict 31.1% variance in the compulsive Internet use among adolescents. The significant positive predictors of the CIUS were age, gender (male), excessive Internet use, overall Internet activity, daily Internet use, and problematic Internet use. In comparison, the significant negative predictors were academic performance and life satisfaction.

Table 7.

Predicting the Influence of Demographics, ICT Accessibility, and Problematic ICT Use.

	Compulsive Internet Use Scale (CIUS) Score
Predictor Variables	β^a	t value ^a
Block 1: Demographics
Age	.15**	7.05
Gender (Male =1, Female = 0)	.19**	8.70
Economic condition	.06	2.82
Academic performance	−.11**	−4.92
Life satisfaction (completely satisfied = 5)	−.13**	5.97
R² value	11.1%
Block 2: ICT accessibility
Internet at home	.04	2.04
Personal Mobile phone	.04	1.78
Mobile Internet	.04	1.69
Internet use experience	.01	.59
Daily Internet use	.21**	9.56
Excessive Internet use	.24**	−11.80
Overall Internet activity	.12**	−5.41
R² value	28.8%
Block 3: Problematic ICT use
Problematic Internet use	.13**	5.56
Problematic Mobile phone use	.04	1.85
Problematic Online gaming use	.02	.79
R² value	31.1%

Note. ^aIt represents standardized beta coefficient and t-value from the final hierarchical regression.

**p < .01.

Discussion

The study results indicate good psychometric properties for the English version of the CIUS for a sample of Indian adolescents (N = 2,369). This psychometric validation resulted in a single-factor solution; hence it confirms the results of previous CIUS validations. This is not the case with the factor structure of the IAT, where previous studies have heterogeneous results. The internal reliability of the English CIUS was comparable to the Dutch, French, and German CIUS validations, but the model fit of the English CIUS validation was superior when compared to all previous CIUS validations.

The factor loading for CIUS Item 8 (Do you think you should use Internet less often?) was low, indicating that this item has a small effect on the total CIUS score. A similar problem was also noticed in the Arabic (Khazaal et al., 2011) and German CIUS validations (Wartberg et al., 2014). Khazaal et al. (2011) argued that the low score of Item 8 reflects that the sample population and its society are more open to Internet usage because the Internet is still in its developmental stages in that region. This argument also holds weight in our case, since India has recently witnessed exponential growth in Internet connectivity and related infrastructure. Therefore, it is quite plausible to assume that Indian society is currently more open toward Internet use due to the developmental stage and promotion of the Internet there.

The age structures of the previous CIUS validation samples were either limited or too broad. The German CIUS validation by Wartberg et al. (2014) focused only on adolescents between 14 and 17 years, German CIUS validation by Guertler et al. (2014) considered mean age of 35.24 years, and the Arabic validation considered 15- to 25-year-old users, while the Dutch CIUS and French CIUS validations considered participants aged 11–80 years and 16–45 years, respectively. In contrast, this study involved adolescents between 12 and 19 years, so our study results are transferable to the population of all adolescents in this age-group.

The study results concluded that male adolescents scored higher on the CIUS than female adolescents. This supports the findings of earlier Internet addiction research that male Internet users were more addicted to the Internet compared to female Internet users (Choi et al., 2009; Ferraro et al., 2007; Ko et al., 2005), that male adolescents prefer escapism or avoidance (Billieux et al., 2011), and that they engage more in online games than their female counterparts (Ha et al., 2007). However, this contradicts the findings of the Arabic CIUS validation (Khazaal et al., 2011), which found no significant relationship between CIUS and gender.

The CIUS scores were positively correlated with the age of the adolescents. This suggests that older adolescents are more susceptible to compulsive Internet use compared to younger adolescents. The possible reasons could be (1) older adolescents experience greater freedom in ICT access and use than younger adolescents, making them prone to a higher level of compulsive Internet use and (2) due to the integration of Internet use in the educational curriculum, most private English schools in India are utilizing more and more Internet in their day-to-day instruction.

The CIUS scores shared very weak correlation with the economic condition of the adolescents. This suggests that due to the availability of low-cost computing devices and affordable Internet data usage plans, adolescents from different economic conditions are equally susceptible to compulsive Internet use. Furthermore, most of the study participants were from low- to middle-income families due to which, significant differences on the CIUS score were not visible.

The CIUS scores negatively correlated with the academic performance of the adolescents. This suggests that students with lower academic grades tend to experience higher compulsive Internet use. The possible reason could be that when adolescents tend to use the Internet more and more, even at the cost of their schoolwork, it results in lower academic performance.

The CIUS scores were negatively correlated with the respondents’ satisfaction with life (1 = completely dissatisfied, 5 = completely satisfied). This suggests that adolescents who are exposed to compulsive Internet use are likely to experience lesser life satisfaction. This is consistent with the findings of the Wartberg et al. (2014).

With regard to the issue of ICT accessibility, this study has found a positive correlation between CIUS scores, mobile phone ownership, Internet connectivity on personal mobile phones, daily time spent on the Internet, and number of years of Internet use. These findings clearly suggest that adolescents with anytime and anywhere access to ICT use are more susceptible to compulsive Internet use. Furthermore, it also shows that when adolescents get anytime and anywhere Internet use possibilities, they tend to utilize it more and finally start experiencing compulsive Internet use.

Finally, compulsive Internet users who tend to experience excessive Internet use consider themselves more active on the Internet and experience problematic ICT use (mobile phone, Internet, and online gaming). These findings suggest that (1) adolescents utilizing the Internet excessively and who remain active in their Internet use tend to showcase compulsive Internet use symptoms and (2) adolescents experiencing problematic ICT use are in fact experiencing compulsive Internet use.

The result of the multiple hierarchical regression has also confirmed the findings revealed by the relationship shared between the CIUS, demographics, ICT accessibility, and problematic ICT use.

Study Implications

This study has theoretical and practical implications for Internet addiction research in general and also for various concerned stakeholders (adolescents, parents, teachers, researchers, and practitioners). Some of the notable study implications are as follows: (1) this study is the first of its kind, that is, a CIUS validation study performed with Asian Internet users, particularly Indian adolescents. Furthermore, it also confirmed the findings of the earlier CIUS validations. Therefore, it shows that the CIUS is a valid and reliable instrument. The CIUS proved to have a fairly stable and homogeneous factorial structure across different languages, cultures, and countries. (2) Indian schools are currently in need of a tested, validated, and brief instrument to measure compulsive Internet use among their young populations. This very practical need is fulfilled to some extent by providing a valid and reliable self-reporting instrument to measure compulsive Internet use. Schools, social workers, child psychologists, clinical practitioners, researchers, educators, and parents can utilize the CIUS instrument to quickly screen compulsive Internet use among adolescents. The CIUS has many advantages over other instruments for screening Internet addiction, such as suitability for research and clinical applications, economy due to few survey items, and ease of use for online studies (Khazaal et al., 2012). In addition to this, Internet users without any working knowledge of clinical testing, for example, adolescents and parents, can also utilize this instrument to screen themselves if provided with sufficient explanation of the CIUS score. Due to the ease of using CIUS, adolescents can screen themselves on a regular basis, for example, after each quarter. This will enable them to receive rapid results regarding their compulsive Internet use behavior. (3) The study findings on the relationship shared between compulsive Internet use, demographic profile, ICT accessibility, and problematic ICT use shed light on the adolescents’ well-being. (4) Media researchers interested in understanding the use of the Internet in the developing world can certainly benefit from this study, which also reflects the Internet use and related attributes relating to India. (5) Policy makers can utilize the study findings to reflect on, build, and even change the existing policies governing Internet use inside and outside schools.

Study Limitations and Future Work

This study has several limitations. These include (1) Adolescent Internet users were recruited from only four cities in India, and participants represented only private schools. Due to this, the generalizability of this study findings to all adolescent Internet users in India is a matter of concern. We therefore suggest that Internet addiction researchers from India should organize a similar study with adolescent users from other cities. This will bring some elements of generalizability to the study findings. (2) Some of the study variables were unitary constructs (single-item variables), namely, economic condition, excessive Internet use, Internet activity, and problematic ICT use (mobile, games, and Internet). Consequently, it is likely that there exists some bias in the study findings due to the fact that unitary variables are prone to higher measurement error. Therefore, we recommend that other researchers to utilize multiple item constructs in a similar study and later compare their findings with the present study. (3) In the present study, participants were asked to take into account their overall Internet usage inside and outside the schools. This includes both types of Internet usage, that is, for academic and nonacademic reasons. However, the study by Montag, Jurkiewicz, and Reuter (2010) has shown that only private Internet use is linked to Internet addiction (IA).

The future research directions include (1) conducting panel, longitudinal, and cross-cultural studies with adolescents by utilizing similar study instruments. This will enable researchers to get a deeper insight into the concept of compulsive Internet use. Furthermore, it will enable the Internet addiction research community to examine the factorial structure of the CIUS using longitudinal, panel, and cross-cultural data. (2) The CIUS possesses good psychometric properties, so other researchers should utilize it and adapt it in order to understand compulsive use of specific Internet activities such as instant messaging, social networking, blogging, and online gaming. (3) Future research should examine the relationship between the CIUS and different motives and gratifications underlying the Internet use. Furthermore, the relationship between the CIUS and adolescents’ risky online behavior is worth examining. (4) At the time of preparing of this article, there was no statistically defined cutoff score for the CIUS except a study in Iran with university students (Alavi et al., 2011). However, Alavi, Jannatifard, Eslami, and Rezapour’s (2011) study was available only in Persian. Only recently, Guertler et al. (2013) determined statistical cutoff score for German CIUS. However, the work concerning cutoff score for CIUS is still in its early stages. Therefore, future studies should examine the cutoff score for CIUS with different user groups and cultures. (5) Internet researchers should examine the CIUS items and determine if any adaptations to this instrument are needed to address specific Internet user groups. Furthermore, it is also worth examining if there is a need to reduce the number of items of the CIUS. (6) Finally, it will be interesting to investigate in future that do findings presented in this study changes or remain same if the study participants are instructed to take into account the Internet usage for private reasons only, for example, for leisure purposes as suggested by Montag et al. (2010).

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was conducted in the Future Industrial Services (FutIS) research program (Project No 2113194), managed by the Finnish Metals and Engineering Competence Cluster (FIMECC), and funded by the Finnish Funding Agency for Technology and Innovation (TEKES), research institutes and companies. Their support is gratefully acknowledged. The support received from Academy of Finland in the form of researcher’s mobility grant to Taiwan (Decision No. 265969) and South Africa (Decision No. 277571) is acknowledged. Additionally, we would like to acknowledge the support received from TEKES funded research project namely Data to Intelligence (D2I) (Project No 21143201) and Mobile Financial Services (MoFS) (Project No 211440).

Note

References

Acock

A. C.

(2005). Working with missing values. Journal of Marriage and Family, 67, 1012–1028.

Alavi

S. S.

Jannatifard

Eslami

Rezapour

(2011). Validity, reliability and factor analysis of compulsive internet use scale in students of Isfahan’s universities. Health Information Management, 7, 724.

Allison

P. D.

(2001). Missing data. Thousand Oaks, CA: Sage.

American Psychological Association. (1994). Diagnostic and statistical manual of mental disorders. Washington, DC: American Psychiatric.

Bartlett

M. S.

(1954). A note on multiplying factors for various chi-square approximations. Journal of the Royal Statistical Society, 16, 296–298.

Batthyany

Muller

K. W.

Benker

Wolfling

(2009). Computer game playing: Clinical characteristics of dependence and abuse among adolescents. Wien Klin Wochenschr, 121, 502–509.

Billieux

Chanal

Khazaal

Rochat

Gay

Zullino

Van der Linden

(2011). Psychological predictors of problematic involvement in massively multiplayer online role-playing games: Illustration in a sample of male cybercafé players. Psychopathology, 44, 165–171.

Brown

R. I. F.

(1993). Some contributions of the study of gambling to the study of other addictions. In Eadington

W. R.

Cornelius

J. A.

(Eds.), Gambling behavior and problem gambling (pp. 241–272). Reno, NV: Institute for the Study of Gambling and Commercial Gaming, College of Business Administration, University of Nevada.

Browne

M. W.

Cudeck

(1993). Alternative ways of assessing model fit. In Bollen

K. A.

Long

J. S.

(Eds.), Testing structural equation models (pp. 136–162). Newbury Park, CA: Sage.

10.

Byrne

B. M.

(1989). A primer of LISREL: Basic applications and programming for confirmatory factor analytic models. New York, NY: Springer-Verlag.

11.

Byrne

M. B.

(2001). Structural equation modeling with AMOS. Basic concepts, applications, and programming. New York, NY: Taylor & Francis.

12.

Carmines

E. G.

McIver

J. P.

(1981). Analyzing models with unobserved variables: Analysis of covariance structures. In Bohrnstedt

G. W.

Borgotta

E. F.

(Eds.), Social measurement: Current issues (pp. 65–115). Thousand Oaks, CA: Sage.

13.

Catell

R. B.

(1966). The Scree test for number of factors. Multivariate Behavioral Research, 1, 245–276.

14.

Chang

M. K.

Law

S. P. M.

(2008). Factor structure for young’s internet addiction test: A confirmatory study. Computers in Human Behavior, 24, 2597–2619.

15.

Choi

Son

Park

Han

Kim

Lee

Gwak

(2009). Internet overuse and excessive daytime sleepiness in adolescents. Psychiatry and Clinical Neurosciences, 63, 455–462.

16.

Cronbach

Meehl

(1955). Construct validity in psychological tests. Psychological Bulletin, 52, 281–302.

17.

DeVellis

R. F.

(2003). Scale development: Theory and applications, applied social research methods (pp. 1–216). Thousand Oaks, CA: Sage.

18.

Ferraro

Caci

D’Amico

Blasi

M. D.

(2007). Internet addiction disorder: An Italian study. Cyberpsychology & Behavior, 10, 170–175.

19.

Fichman

Cummings

J. N.

(2003). Multiple imputation for missing data: Making the most of what you know. Organizational Research Methods, 6, 282–295.

20.

George

Mallery

(2003). SPSS for windows step by step: A simple guide and reference. 11.0 update (4th ed.). Boston, MA: Allyn & Bacon.

21.

Griffiths

(1999). Internet addiction: Fact or fiction? Psychologist, 12, 246–250.

22.

Griffiths

Wood

R. T. A.

(2000). Risk factors in adolescence: The case of gambling, videogame playing, and the Internet. Journal of Gambling Studies, 16, 199–225.

23.

Guertler

Broda

Bischof

Kastirke

Meerkerk

G.-J.

Ulrich

… Rumpf

H.-J.

(2014). Factor structure of the compulsive internet use scale. Cyberpsychology, Behavior, and Social Networking, 17, 46–51.

24.

Guertler

Rumpf

H.-J.

Bischof

Kastirke

Petersen

K. U.

Petersen

K. U.

… Meyer

(2013). Assessment of problematic internet use by the compulsive internet use scale and the internet addiction test: A sample of problematic and pathological gamblers. European Addiction Research, 20, 75–81.

25.

J. H.

Kim

S. Y.

Bae

S. C.

Bae

Kim

Sim

… Cho

S. C.

(2007). Depression and internet addiction in adolescents. Psychopathology, 40, 424–430.

26.

Hair

J. F.

Anderson

R. E.

Tatham

R. L.

Black

W. C.

(1998). Multivariate data analysis. London, United Kingdom: Prentice Hall.

27.

L. T.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55.

28.

India Internet Users. (2013). Internet and mobile association of India. Retrieved November 10, 2013, from http://www.iamai.in/PRelease_detail.aspx?nid=3222&NMonth=11&NYear=2013

29.

Internet Usage in Asia. (2013). Internet penetration in Asia. Retrieved November 10, 2013, from www.internetworldstats.com/stats3.htm#asia

30.

Jöreskog

K. G.

Sörbom

(1999). LISREL 7: A guide to the program and application. Chicago, IL: SPSS.

31.

Kaiser

H. A.

(1970). A second generation little Jiffy. Psychometrika, 35, 401–415.

32.

Khazaal

Chatton

Atwi

Zullino

Khan

Billieux

(2011). Arabic validation of the compulsive internet use scale (CIUS). Substance Abuse Treatment, Prevention, and Policy, 6, 32.

33.

Khazaal

Chatton

Horn

Achab

Thorens

Zullino

Billieux

(2012). French validation of the compulsive internet use scale (CIUS). Psychiatric Quarterly, 83, 397–405.

34.

Kline

R. B.

(2011). Principles and practice of structural equation modeling (3rd ed.). New York, NY: Guilford Press.

35.

C. H.

Yen

J. Y.

Yen

C. F.

Chen

C. C.

Yen

C. N.

Chen

S. H.

(2005). Screening for internet addiction: An empirical study on cut-off points for the Chen internet addiction scale. Kaohsiung Journal of Medical Sciences, 21, 545–551.

36.

Krajewska-Kulak

Kulak

Marcinkowski

J. T.

Damme-Ostapowicz

K. V.

Lewko

Lankau

… Rozwadowska

(2011). Internet addiction among students of the Medical University of Bialystok. Computers Informatics Nursing, 29, 657–661.

37.

Little

R. J. A.

Rubin

D. B.

(1987). Statistical analysis with missing data. New York, NY: Wiley.

38.

Liu

Potenza

M. N.

(2007). Problematic internet use: Clinical implications. CNS Spectrums, 12, 453–466.

39.

Marks

(1990). Behavioural (non-chemical) addictions. British Journal of Addiction, 85, 1389–1394.

40.

Meerkerk

G. J.

Van Den Eijnden

R. J.

Vermulst

A. A.

Garretsen

H. F.

(2009). The compulsive internet use scale (CIUS): Some psychometric properties. Cyberpsychology Behavior, 12, 1–6.

41.

Montag

C. M.

Jurkiewicz

Reuter

(2010). Low self-directedness is a better predictor for problematic Internet use than high neuroticism. Computers in Human Behavior, 26, 1531–1535.

42.

O’Connor

(2000). SPSS and SAS programs for determining the number of components using parallel analysis and Velicer’s MAP test. Behavior Research Methods, Instruments, & Computers, 32, 396–402.

43.

Peukert

Steffen

ElKasmi

Barth

G. M.

Meerkerk

G.-J.

Batra

(2012). Faktorielle Struktur der deutschen Version der Compulsive Internet Use Scale (CIUS) nach konfirmatorischer Faktorenanalyse [Factorial structure of the German version of the Compulsive Internet Use Scale (CIUS) after confirmatory factor analysis]. Zeits chrift für Klinische Psychologie und Psychotherapie, 41, 101–108.

44.

Pigott

T. D.

(2001). A review of methods for missing data. Educational Research and Evaluation, 7, 353–383.

45.

Pratarelli

M. E.

Browne

B. L.

(2002). Confirmatory factor analysis of internet use and addiction. Cyberpsychology and Behavior, 5, 53–64.

46.

Schafer

J. L.

Graham

J. W.

(2002). Missing data: Our view of the state of the art. Psychological Methods, 7, 147–177.

47.

Schumacker

R. E.

Lomax

R. G.

(2004). A beginner’s guide to structural equation modeling. New York, NY: Routledge.

48.

Stevens

(1996). Applied multivariate statistics for the social sciences. Mahwah, NJ: Lawrence Erlbaum.

49.

Streiner

(2003). Starting at the beginning: An introduction to co-efficient alpha and internal consistency. Journal of Personality Assessment, 80, 99–103.

50.

Tabachnick

B. G.

Fidell

L. S.

(1996). Using multivariate statistics (pp. 66–70). New York, NY: Harper Collins.

51.

Tavakol

Dennick

(2011). Making sense of Cronbach’s alpha. International Journal of Medical Education, 2, 53–55.

52.

Van der Aa

Overbeek

Engels

R. C. M. E.

Scholte

R. H. J.

Meerkerk

G. J.

Vanden Eijnden

R. J. J. M.

(2009). Daily and compulsive internet use and well-being in adolescence: A diathesis-stress model based on big five personality traits. Journal of Youth Adolescence, 38, 765–776.

53.

Velicer

W. F.

(1976). Determining the number of components from the matrix of partial correlations. Psychometrika, 41, 321–327.

54.

Wartberg

Petersen

K. U.

Kammerl

Rosenkranz

Thomasius

(2014). Psychometric validation of a German version of the compulsive internet use scale. Cyberpsychology Behavior & Social Networking, 17, 99–103.

55.

Widyanto

McMurran

(2004). The psychometric properties of the Internet addiction test. Cyberpsychology & Behavior, 7, 443–450.

56.

Yen

J. Y.

Yen

Chen

Chung

Chen

(2008). Psychiatric symptoms in adolescents with internet addiction: Comparison with substance use. Psychiatry and Clinical Neurosciences, 62, 9–16.

57.

Yen

J. Y.

C. H.

Yen

C. F.

H. Y.

Yang

M. J.

(2007). The comorbid psychiatric symptoms of internet addiction: Attention deficit and hyperactivity disorder (ADHD), depression, social phobia, and hostility. Journal of Adolescent Health, 41, 93–98.

58.

Young

K. S.

(1996). Pathological internet use: A case that breaks the stereotype. Psychological Reports, 79, 899–902.

59.

Young

K. S.

(1998). Caught in the net: How to recognize the signs of internet addiction—and a winning strategy for recovery. New York, NY: John Wiley.

60.

Young

K. S.

(1999). Internet addiction: Symptoms, evaluation, and treatment. In VandeCreek

Jackson

T. L.

(Eds.), Innovations in clinical practice: A source book (Vol. 17, pp. 19–31). Sarasota, FL: Professional Resource Press.

61.

Young

K. S.

Case

C. J.

(2004). Internet abuse in the workplace: New trends in risk management. CyberPsychology & Behavior, 7, 105–111.

62.

Youth Internet Users. (2013). More than 90 percent mobile internet users in India are men. Retrieved November 10, 2013, from http://tech2.in.com/news/mobile-phones/more-than-90-percent-mobile-internet-users-in-india-are-men/914080