Abstract
Abstract
This study presents the profiles of heavy Internet users and provides empirical evidence that it is not how much time university students spend online but what they do online that is associated with academic grades and psychological adjustment. Using a nationally representative sample from Taiwan, we employed K-mean cluster analysis and identified profiles based on nine Internet practices in which users engaged. Female heavy users favoring information seeking and chatting had better academic performance but tended to feel more depressed than nonheavy users, while those favoring information seeking, chatting, and online games had lower academic grades and greater loneliness, physical illness, and depression scores than nonheavy users. In contrast, only male heavy users favoring online games had lower academic grades, whereas those who favored information seeking, chatting, and online games were more likely than nonheavy users to feel physically ill and depressed.
Introduction
Previous research has established associations between different types of Internet practices and college students' academic performance and psychosocial adjustment. For example, Kuh and Hu 2 examined associations between computer use and selected learning and development outcomes for 18,344 undergraduates from 71 four-year colleges and universities. They found the Internet practice of searching for information related to course material to be positively correlated with students' intellectual development and vocational preparation as well as personal development. Kubey et al. 3 explored Internet use and collegiate academic performance decrements with data from 572 students at a large public university. They reported heavier recreational Internet use, defined as usage of synchronous computer-mediated communication (CMC), which includes combined hours of usage of multiuser domains (MUDs) and Internet Relay Chat (IRC), to be highly correlated with impaired academic performance. Lanthier and Windham 4 investigated Internet use and college adjustment among 272 college students and found social use of the Internet to be positively associated with male students' college overall adjustment (e.g., academic, social, emotional/personal adjustment, and institutional attachment). In addition, several studies5–9 linked various Internet functions with college students' health problems, depression, and loneliness. Drawing on these works as a reference, we propose that it is not how much time university students spend on the Internet that is associated with academic grades and psychosocial adjustment, it is what they do online.
Method
This study is part of a project supported by the National Science Council in Taiwan, called A National Survey of College Student Life Experience in Taiwan. The data were collected between October 2005 and February 2006. Included in the survey were 156 universities and colleges in Taiwan. All juniors at each institution were stratified by major, and 25% of each group was selected randomly, resulting in a sample of 49,609 students from 17 academic majors. They were asked by e-mail to fill out an online questionnaire, and 26,307 (53.3%) valid questionnaires were obtained.
From this questionnaire, data regarding students' background information, weekly time spent on nine Internet activities, self-reported GPAs from the previous semester, and psychological adjustment were used in this study. Their psychological adjustment, addressing loneliness, physical illness, and depression, was measured by a battery of 24 items on 5-point rating scales (1, never; 5, = often). The Cronbach's alpha coefficients for loneliness, physical illness, and depression were 0.91, 0.91, and 0.90 respectively.
After deleting unreasonable data from respondents who reported spending more than 168 hours a week online, N equaled 24,596. We used the same criteria as Chen and Peng 1 and classified heavy Internet users as those who spend 33.81 hours or more online every week, which is one standard deviation above the mean. This criterion resulted in 2,784 (11.33%) heavy Internet users and 21,810 (88.67%) nonheavy users.
We then used cluster analyses to identify groups of students who shared homogeneous patterns of engagement in Internet practices and other characteristics. Since many previous studies have established that there are gender differences in how university students use the Internet,1,10–12 two sets of cluster analyses, one for female and the other for male heavy users, were performed by using their scores on the nine Internet practices to assign each student to a homogeneous group. Because of the large number of students in the study, we chose a K-mean rather than a hierarchical cluster analysis. 13 We selected the 3-cluster for both the female and the male solutions because the patterns of factor scores differed one from the other, yet the number of groups was not so great that the distinctions between the groups would be trivial. Finally, we used an analysis of variance (ANOVA) with post hoc pairwise contrasts to compare the nonheavy users and the three clusters for each gender on the four outcome measures: academic grades, loneliness, physical illness, and depression.
Results
Among the heavy users (n = 2,784), 50% (n = 1,389) were male and 50% (n = 1,395) female. Heavy users' average total weekly time on the Internet was 53.49 hours (SD = 22.98), with an average of 12.73 hours a week spent making friends and chatting, 9.96 searching for academic information, 9.83 playing online games, 9.31 searching for nonacademic information, 3.31 checking e-mails, 3.20 shopping, 2.23 visiting sex sites, 2.21 browsing school news, and 0.71 browsing the stock market. In contrast, nonheavy users' average total weekly time on the Internet was 12.30 hours (SD = 7.77), with an average of 2.91 hours a week spent making friends and chatting, 2.80 searching for academic information, 2.32 searching for nonacademic information, 1.28 playing online games, 1.04 checking e-mails, 0.83 shopping, 0.73 browsing school news, 0.26 visiting sex sites, and 0.13 browsing the stock market. Significant differences were found between heavy users' and nonheavy users' time spent on all nine types of practices.
Using K-mean cluster analyses, we identified three profiles for the female heavy Internet users and three profiles for the male and assigned labels based on the nature of the nine Internet practices in which they engaged. Table 1 presents the named profiles and the mean scores for the nine Internet practices. As expected, ANOVAs revealed significant mean differences across profiles on all of the Internet practices with only two exceptions. Another way to depict the profiles is the percentages of heavy Internet users' hours on nine Internet practices by profile, as shown in Figure 1. Table 2 shows the means and standard deviations for academic grades, loneliness, physical illness, and depression for female nonheavy users (FNH) and the three profiles of female heavy users, as well as the results of the ANOVA and post hoc pairwise contrasts. Table 3 shows the same results for male nonheavy users and the three profiles of male heavy users.

Percentage of heavy Internet users' hours on nine Internet practices by profile cluster.
FH1, female heavy users favoring information seeking and chatting; FH2, female heavy users favoring information seeking, chatting and shopping; FH3, female heavy users favoring information seeking, chatting and online games.
MH1, male heavy users favoring online games; MH2, male heavy users favoring information seeking and chatting, MH3, male heavy users favoring information seeking, chatting and online games.
p < 0.05; **p < 0.01; ***p < 0.001.
p < 0.05; **p < 0.01; ***p < 0.001.
p < 0.05; **p < 0.01; ***p < 0.001.
FH1s (n = 459) are female heavy users who especially favor information seeking and chatting. They spent the fewest total hours online (M = 50.02, SD = 20.57) among the three female profiles (see Table 1), but more than two thirds of them spent more than 7 hours a week on seeking academic as well as nonacademic information and also on making friends and chatting (see Fig. 1). As a group, they had the highest average academic grade, which was significantly higher than that of FNHs; however, their average score on depression was also significantly higher than that of FNHs (see Table 2).
FH2s (n = 632) had the most students among the three female heavy user profiles and consisted of those who not only favored information seeking and chatting but also liked online shopping, with around 40% spending more than 7 hours a week shopping online. Compared with the FNHs, they had significantly higher average depression scores, but no significant difference was found for academic grades, loneliness, or physical illness.
FH3s (n = 304) are those female students who not only favor information seeking and chatting but also like online games; they spent the most total hours online (M = 55.17, SD = 25.86) among the three female profiles, with two thirds spending more than 7 hours a week playing games. As a group, they had the lowest average academic grades, and their average grade was not only significantly lower than that of the FNHs but also lower than those of the FH1s and FH2s. At the same time, they had the highest average scores on loneliness, physical illness, and depression, and these scores were all significantly higher than those of the FNHs.
In terms of the male students, MH1 (n = 578) is the only profile that consists of those who do not prefer information seeking and chatting; instead, playing online games is the only distinguishing feature of this group, with 92% of them spending more than 7 hours a week on games (see Fig. 1). Even though as a group they spent the fewest total hours online among the three male profiles (M = 52.22, SD = 22.23) (see Table 1), they had the lowest average academic grade among groups (see Table 3). Their average grade was significantly lower not only than that of the male nonheavy users but also than that of the MH2s. Yet no significant difference was found between the MH1s and the male nonheavy users (MNH) for any of the three psychological adjustment variables.
MH2 (n = 410) is somewhat similar to FM1 in that it consists of those who favor information seeking and chatting. As a group, they also have the highest average academic grade among the male profiles, but no significant difference was found between their average academic grade and that of the MNHs. No significant difference was found between the MH2s and MNHs for any of the three psychological adjustment variables.
Likewise, MH3 (n = 401) is somewhat similar to FH3 in that it consists of those who favor information seeking, chatting, and also online games, and they spent the most total hours online (M = 58.81, SD = 24.12) among the three male profiles. As a group, no significant difference was found between their academic grades and those of the MNHs, but their average scores on physical illness and depression were both significantly higher than those of the MNHs.
Discussion
With a nationally representative sample of university students in Taiwan, we identified around 11% of students who spent 53.49 hours online every week as heavy users. When we grouped the students according to how often they engaged in nine types of Internet practices, three distinct profiles of female and male heavy Internet users were identified. Around 33% of the female heavy users (FH1) liked to spend time on the Internet searching for both academic and nonacademic information and at the same time making friends and chatting; 45% of them (FH2) liked to do the same things as FH1 plus engage in online shopping, and 22% of them (FH3) liked to do the same things as FH1 plus playing online games. The sums of the three profiles' time online ranged from 50 to 55 hours per week, about 7 to 8 hours per day on average. However, it is interesting that the FH1s had better academic performance than the FNHs, even though they also tended to feel more depressed than the FNHs. In contrast, the FH3s' academic grades were significantly lower than those of the FNHs, and at the same time, their scores on the three variables of psychosocial adjustment were all significantly lower than those of the FNHs.
The profiles of the male heavy users are quite different from those of the female heavy users. Around 42% of the male heavy users (MH1) liked to spend a lot of time online for games only; 30% of them (MH2) liked to spend time on the Internet searching for both academic and nonacademic information and at the same time making friends and chatting, and 29% of them (MH3) liked to do the same things as MH2 plus play online games. Playing online games seemed to be a more popular practice for male heavy users than for female heavy users. The sums of the three profiles' time online ranged from 52 to 59 hours per week, about 7.5 to 8.5 hours per day on average. Only the MH1s' average academic grade was found to be significantly lower than that of the MNHs. Among the three male heavy user profiles, the MH2s' academic performance was found to be significantly better than that of the MH1s and MH3s. No significant difference was found for loneliness scores across groups, while the MH3s reported feeling more physically ill and depressed than the MNHs.
In other words, while Chen and Peng 1 found that heavy Internet users as a group had significantly lower academic grades than nonheavy users, we found only two profiles, FH3 and MH1, to have significantly lower academic grades than nonheavy users. One profile, FH1, female heavy Internet users favoring information seeking and chatting, even had significantly better grades than female nonheavy users. These results suggest that spending a lot of time on information seeking and chatting on the Internet does not necessarily impair students' academic achievement; rather, sometimes it helps students' academic performance, with FH1 being the most obvious example. These results are consistent with Kuh and Hu's findings. 2 These results also suggest that spending a lot of time playing online games seems to damage students' academic performance, as in the cases of FH3, MH1, and MH3.
Chen and Peng 1 found that heavy Internet users as a group have significantly higher loneliness, physical illness, and depression scores than nonheavy users; however, when we further grouped students into homogeneous profiles based on their Internet practices, we detected more subtle differences. For females, all three heavy user profiles had significantly lower average scores than those of the female nonheavy users on only one of the psychosocial adjustment variables: depression. And among the three profiles, FH3 had the worst psychosocial adjustment, reporting significantly higher scores than those of the female nonheavy users on all three variables: loneliness, physical illness, and depression. The associations between patterns of Internet practice and psychological adjustment seem to be less significant for male heavy users. Only MH3s reported feeling more physically ill and depressed than the MNHs. Overall, for psychological adjustment, the FH3s and MH3s, who shared the characteristics of spending the most time online and favoring information seeking, making friends, and chatting, while at the same time playing online games, were the most likely to feel physically ill, depressed, and for the females, also lonely.
This study helps advance the understanding of the relationships between heavy Internet users' patterns of practices and their college lives. Three limitations need to be acknowledged. First, because the data were cross-sectional in nature, only associations, but not casual relations, could be achieved from the analyses. Second, this study did not address how college students get involved in problematic Internet use and how such involvement influences their lives, because these issues exceeded the scope of this study. Future research on these issues would be a great extension to this study. And finally, because the data were obtained from the participants' self-reporting, inaccuracy or ambiguity is inevitable in the participants' estimation of their Internet behavior.
Footnotes
Acknowledgments
The Taiwan Higher Education Survey reported in this study was supported by a grant from the National Science Council of Taiwan (NSC 93-2413-H-007-002). The authors would also like to thank National Tsing Hua University for financially supporting this research (97N 25577 E1).
Disclosure Statement
No competing financial interests exist.
