Out of the total number of students (N = 2214), 24.5% were classified as No-risk Users (NRU), 63% as Low-risk Users (LRU), and 12.5% as Risk and High-risk Users (RHU). As shown in Table 1, those students in the NRU group presented a significantly higher mean age (M= 22.2 years, SD= 1.9) than those in the other groups. Most participants were women (78.6%), and the proportion of women was near 83% in the NRU and 70% in the RHU. The NRU group presented a lower income than the other groups (Scheffé's test, p < 0.01). In all groups, most of the students were from the Southeast region of Brazil, followed by the Northeast. Although only near 8% were married, the proportion of married students was three times higher in the NRU students. Most of the participants were from private institutions (64.6%) and no differences were found in students’ frequency from private and public Universities classified as RHU.

Table 1

Note. NRU= No-risk Users (n = 543); LRU = Low-risk Users (n = 1395), RHU = Risk and High-risk Users (n = 276). *p < 0.05, ** p < 0.01, *** p < 0.001 a1$= 3.22 BRL (Brazilian Real currency- average quotation in June/2016).

78% of those from the RHU group reported having problems with their smartphone use, whereas this proportion was 30% in the LRU (Table 2). Additionally, the number of participants who reported driving while using their mobile was three times higher in the RHU group than in the NRU. We also detected that those students with severe psychiatric symptoms were significantly higher in the RHU. In the LRU, the proportion of participants with severe symptoms of depression, anxiety, and stress was 6.8%, 9.2%, and 8.8%, respectively, whereas, in the RHU group, this proportion was 35.1%, 34.8%, and 33.3%, respectively.

Table 2

Note. NRU= No-risk Users (n = 543); LRU = Low-risk Users (n = 1395), RHU = Risk and High-risk Users (n = 276). **p < 0.01, ***p < 0.001, ^aA total of 905 students reported they did not drive and were excluded from this analysis [No-risk Users (n = 205); LRU = Low-risk Users (n = 586), RHU = Risk and High-risk Users (n = 205)].

Figure 1 shows the different app and smartphone functionalities among students from the NRU, LRU, and RHU groups. The RHU group showed an average daily time of 446 minutes (7.4 hours), almost twice that of the NRU participants (F= 56.4, p < 0.001, η² = 0.05). As regards the use of WhatsApp (Figure 1B), the RHU group also presented a longer time than the other groups (F = 38.9, p < 0.001, η² = 0.03). Additionally, we detected that users from the RHU group spent more time than the other participants on Facebook (Figure 1C) (F= 50.6, p < 0.001, η² = 0.04); using websites (Figure 1D) (F = 11.2, p < 0.001, η² = 0.01); on Instagram (Figure 1E) (F = 19.6 , p < .001, η² = 0.01); and playing games (Figure 1F) (F = 5.6, p < 0.001, η² = 0.00). Those students from the RHU group reported receiving (Figure 1G) and sending (Figure 1H) a significantly higher number of messages than those from the other groups. Students from the NRU group received a significantly higher average of messages than the other participants (Scheffé’s test, p < 0.001). Lastly (Figure 1I), students from the RHU reported checking their smartphones on average 32 times a day (F = 27.4, p < 0.001, η² = 0.04).

Figure 1

Note. 1A. Time spent using smartphone daily (M and SE) among students classified as No-risk Users (NRU; n = 543); Low-risk Users (LRU; n = 1395) and Risk and High-risk Users (RHU; n = 276) 1B. Time spent using WhatsApp (Mean and S.E.) 1C. Time spent using Facebook (M and SE). 1D. Time spent navigating websites, blogs, and others (M and SE). 2E. Time spent using Instagram (M and SE). 1F. Time spent playing games (M and SE). 1G. Messages received daily (M and SE) from any app and website. 1H. Some messages are sent daily from any app and website (M and SE). 1I. Some smartphone checks daily (M and SE). * Differs from the NRU group (p < 0.05), *** Differs from the NRU groups (p < 0.01), ** Differs from the LRU and NRU groups (p < 0.001).

We aimed to evaluate Internet addiction prevalence among college students and its association with emotional problems. Our main findings showed that 12.5% of the students were classified as I.A. (RHU Group). The prevalence of severe depression, anxiety, and stress symptoms in this group were five, four, and four times higher than students with no problematic internet use. Additionally, the perception of problematic use and the psychological problems was the main predictors of I.A. On the other hand, we did not observe significant differences between students from private and public universities.

Some authors detected a higher prevalence of I.A.s observed with Chilean students (11.5%), compared with our data by Berner et al. (2014). On the other hand, in a Brazilian study, some authors lower the prevalence of I.A. (7.3%) in a sample with medical students with Brazilian students (Marin et al., 2016). Besides, the prevalence of I.A. in our study is lower than observed in most Latin American studies, such as Ecuador (20.3% and 51.5%, Ramos-Galarza et al., 2018; Rueda et al., 2017, respectively), Colombia (25%, Cañón et al., 2016) and Brazil (62% and 21.2%, Della-Méa et al., 2016; Terres-Trindade & Mosmann, 2016, respectively). In Europe, 98% of Swiss students (12 to 19 years old) had smartphones, and 17% were considered dependent. Additionally, the prevalence of I.A. was more frequent among younger students (15-16 years old) than among older ones (older than 19 years old) (Haug et al., 2015). Some authors observed that in Lithuania, there was a higher prevalence of internet-dependent youngsters (from 11 to 18 years old) (nearly 11% of the boys and 8% of the girls), where 67% of the males spent at least five hours a day playing on the internet, while 19% of the girls did so (Ustinaviciene et al., 2016). These findings are very heterogeneous because they depend not only on the studies’ methodological characteristics but also on each country’s cultural aspects.

Regarding socio-demographic characteristics, we did not observe any association of gender with I.A., as reported by some authors (Moromizato et al., 2017; Marin et al., 2016). Although male students used the internet for more extended periods than females In Canada, the authors did not detect significant differences in the prevalence of I.A. regarding gender (Dufour et al., 2016). The full internet addiction scores and frequency of internet problems were similar in male and female Pakistani medical students (Khan et al., 2017). On the other hand, some authors observed a gender difference in which I.A.’s prevalence was higher among male participants (Cañón et al., 2016; Demirer & Bozoglan, 2016; Ak et al., 2013) than female participants (Andrade, Scatena, Bedendo et al., 2020). In one of these studies, males were more interested in playing online games and reading newspapers and magazines, whereas females used the internet, mainly for communication (Ak et al., 2013). In this sense, it is essential to consider considering these data in specific cultures (Shadzi et al., 2020). In Demirer and Bozoglan (2016), the authors pointed out that male students spent more time on the internet than female students because boys have more cultural freedom than girls and more social opportunities.

We observed that those students from the NRU group had the lowest income and those from the LRU the highest. In a study conducted with students from Hong Kong, those with a family income level below average had a chance almost twice as high to have I.A. (Wu et al., 2016). Some authors detected that unemployed internet users were more likely to use the internet for private purposes than working ones (Duplaga, 2017; Rumpf et al., 2014). The authors pointed out that free time may increase the time spent on the internet. Interestingly, in a study conducted with Polish students, those from urban areas presented a lower prevalence of IA (0.45%) compared with those who are living in rural areas (2.9%) (Pawlowska et al., 2015).

The new configuration of social relations has influenced the overuse of digital media (smartphones) and the internet in the last few years, and these technologies are becoming easily accessible. However, the impact of this use on students' quality of life is a little-explored issue (Wang et al., 2020). In our study, the higher the severity of internet use, the longer the daily time using smartphones and computers, and the RHU used their smartphones 7.4 hours daily on average. Wu et al. (2016) reported that 57% of Chinese students with I.A. spent more than six hours a day on it. In South Korea, about 70% of the children aged 11 and 12 have a Smartphone, and 25% meet the criteria for smartphone addiction, using it 5.4 hours a day, on average (Jeong et al., 2016).

When the first studies on I.A. were developed in the 1990s, researcher Mark Griffiths (1999) created the term "digital addiction" to define characteristics these individuals shared, including detachment and mood swings. The association between digital addiction and these symptoms was reinforced in the last decade, notably regarding anxiety and mood disorders (Baturay & Toker, 2019; Tsai et al., 2020). In our study, nearly 35% of the RHU group students presented more severe depression, anxiety, and stress symptoms. Therefore, the depletion of interpersonal relationships could be triggered by the increase of virtual relationships, distorting the individual's perception of the physical world (Tsai et al., 2020).

Rosenthal et al. (2016) study is an example of this process, as researchers found that youth's depressive symptoms were worsened by experiences and time spent on social networks such as Facebook. According to the authors, a hypothesis could be that as most people post photos/texts only in situations of happiness and well-being, users experience reduced self-esteem and worsen depressive symptoms. In the study of Berner et al. (2014), almost 19% of college students with I.A. showed significant scores to symptoms of depression, but in both Brazilian studies that evaluated the association of I.A. with psychological problems, such association was not found (Moromizato et al., 2017; Della-Méa et al., 2016). The association between I.A. and some mental disorders was deeply discussed by some authors who proposed a theoretical model to explain this relationship (see Brand et al., 2019; Brand et al., 2016).

In our study, those students from public institutions presented IAT and anxiety scores slightly higher than those from private universities. However, we did not detect a robust effect size in both cases. This relationship should be better investigated, considering the restricted number of studies in the literature comparing students from different institutions. In Colombia, Cañón et al. (2016) reported no differences in I.A. prevalence between students from private and public universities, but these last groups showed scores slightly higher than the first ones. In another report, 8.9% of Lebanese public schools and 2.6% of private school students were classified as internet addicted (Hawi, 2012). On the other hand, some authors did not find differences in the prevalence of I.A. between students from the public (17.7%) and private (14.1%) schools (Seyrek et al., 2016).

We observed that depression, anxiety, and stress were all associated with the risk of I.A. Koo and Kwon (2014) conducted a meta-analysis to evaluate the main risky and protective factors to I.A. only in studies using the South Korean sample (N = 59,283). The magnitude of the effect size was significantly higher for intrapersonal variables than interpersonal ones. Regarding the intrapersonal variables, the most substantial risk factor was the "escape from self" variable followed by "attention problem" and "addictive/absorption traits." The authors also observed a significant effect size in the depression/anxiety category. The most substantial risk factor from interpersonal variables was "family functioning" and "parent relationship." Another study conducted with Jordanian college students, university year level, age, and depression were predictors of I.A. (Alzayyat et al., 2015).

This study has limitations that have been considered. Because an online survey collected our data to expand the sample size, there could be a bias regarding the students’ profiles who participated in the study. College students’ use of the internet in Brazil is even more expected from their smartphones, and internet addiction might have been mistaken for smartphone addiction. We did not evaluate smartphone addiction because there were no validated instruments available in Brazil during the period that our study was performed.

Our findings suggest a prevalence of IA in 12.5 % of the sample, and these students more often presented symptoms of anxiety, stress, and depression. We did not observe significant differences between students from public and private universities. Moreover, several variables were identified as predictors of the risk of internet addiction. Taken together, these data are relevant as they improve the understanding of the characteristics of this population and hence could help the development of mental health prevention actions focused on youngsters.