Digital Well-Being and Problematic TikTok Use in Higher Education: Prevalence, Risk Factors, and Preliminary Psychometric Evaluation of the TikTok Addiction Scale among Vietnamese Students

Study Design

This study employed a cross-sectional quantitative research design to examine the patterns of TikTok use and levels of problematic engagement among Vietnamese university students, following established methodological recommendations for survey research (Albarello, 2012; Mace & Petry, 2021). A survey-based approach was used to assess demographic characteristics, platform usage patterns, and addiction-related symptoms. The cross-sectional design enabled the identification of associations between variables but did not permit causal inference.

Study participants

A convenience sampling method was used to recruit participants from five universities in Ho Chi Minh City, Vietnam (three public and two private). Eligible participants were required to (1) be enrolled in an undergraduate program and (2) have actively used TikTok during the past 12 months. Active TikTok use was determined via a platform-specific frequency item assessing use of major social media platforms (TikTok, Facebook, Twitter/X, and YouTube) on a 5-point scale ranging from 1 (never) to 5 (very frequently); participants reporting “never” for TikTok were routed out before the TikTok-specific items and thus excluded from the completed sample. A total of 400 eligible students completed the questionnaire. During data screening, 35 questionnaires were excluded due to substantial missing data or invalid response patterns (e.g., selecting the same response option across nearly all questionnaire items). The final analytic sample therefore comprised 365 participants (91.25% usable questionnaires). A post hoc power analysis was conducted using G*Power 3.1 to contextualize the statistical power of the final multiple linear regression model (α = .05, four predictors). Based on the observed effect size (f2=.088) and the final sample (N = 365), the achieved statistical power was 0.998, indicating that the study had sufficient power to detect small-to-moderate effects.

Instrument

The survey instrument comprised two sections. The first collected demographic information, including gender, age, field and year of study, living arrangements, and academic performance. Participants also reported their primary purposes for using TikTok (e.g., entertainment, education, passing time, social networking, and following trends), average daily TikTok use (hours/day), and the frequency of using major social media platforms (TikTok, Facebook, YouTube, and X). Social media use was rated on a 5-point Likert scale, ranging from 1 (Never) to 5 (Very Frequently).

The second section evaluated problematic TikTok use (PTU) using the 15-item Vietnamese-adapted TikTok Addiction Scale (TTAS) (Galanis et al., 2024). To ensure linguistic and conceptual equivalence, the scale underwent a structured cultural adaptation process. First, the original English version was translated into Vietnamese by a university lecturer specializing in English linguistics. This preliminary translation was then independently reviewed by a university lecturer in psychology to evaluate its conceptual equivalence, linguistic clarity, and cultural appropriateness. Discrepancies and ambiguous wording were resolved through consensus within the research team to maintain conceptual consistency with the original instrument. The scale assesses six behavioral addiction dimensions using a 5-point Likert scale (1 = very rarely, 5 = very often):

• Salience (two items)

• Tolerance (three items)

• Mood modification (two items)

• Relapse (two items)

• Withdrawal (two items)

• Conflict (four items).

Higher averaged scores indicate greater addiction severity. Following the original TTAS validation study by Galanis et al. (Galanis et al., 2024), a mean score of 3.23 was adopted as the recommended cut-off for identifying individuals at high risk of problematic TikTok use. Because this threshold was established using a Greek validation sample, it was used solely for descriptive classification in the present study and should not be interpreted as a clinically validated cut-off for Vietnamese university students. The original TTAS demonstrated high internal consistency, with overall Cronbach’s alpha and McDonald’s omega of 0.911 and 0.914, respectively, and subscale alphas ranging from 0.659 to 0.868 (Galanis et al., 2024).

Data collection

Primary data were collected via paper-based questionnaires administered to undergraduate students across five universities from September 10 to October 12, 2025. Eligible students received information about the study and provided written informed consent before participation. Participation was voluntary, responses were anonymous and confidential, and participants could withdraw at any time without penalty. The questionnaire required approximately 10 minutes to complete.

Data Analysis

Data were analyzed using IBM SPSS Statistics version 25.0, IBM SPSS Amos version 20.0 (IBM Corp., Armonk, NY, USA), Stata version 15.1 (StataCorp LLC, College Station, TX, USA), and G*Power version 3.1. Descriptive statistics were used to summarize participants’ demographic characteristics and TikTok use patterns. Internal consistency was assessed using Cronbach’s alpha (α) and the Spearman-Brown coefficient where appropriate. An initial confirmatory factor analysis (CFA) was conducted using IBM Amos. The final CFA was then estimated in Stata 15.1 using the Full Information Maximum Likelihood (FIML) estimator to accommodate missing data without listwise deletion. Statistical significance was set at p < .05. A post-hoc power analysis for the multiple linear regression model, based on the observed effect size, confirmed that the sample provided adequate statistical power.

Demographic Characteristics of the Sample

The study comprised 365 students recruited from five universities in Ho Chi Minh City, with 54.8% (n = 200) attending public universities and 45.2% (n = 165) attending private universities. Participants had a mean age of 19.73 years (SD = 1.28; range, 18–23), and 53.2% (n = 194) were female. The largest proportion majored in Economics and Finance (34.0%), followed by Foreign Languages (25.5%) and Social Sciences and Humanities (22.5%). Most participants reported Good (42.5%) or Excellent (32.3%) academic performance, and 56.4% lived in rented accommodation or dormitories. Detailed demographic characteristics are presented in Table 1.

Social Media and TikTok Usage

As shown in Table 2, TikTok was the most frequently used platform (M = 3.92, SD = 1.021), followed by Facebook (M = 3.72, SD = 1.006) and YouTube (M = 3.27, SD = 1.155), whereas X (formerly Twitter) showed the lowest level of use (M = 1.87, SD = 1.168).

Regarding TikTok use duration, 54.0% of participants (n = 197) had used the platform for more than three years, whereas 26.6% (n = 97) and 19.4% (n = 71) reported one to two years and two to three years of use. Overall, 73.4% had used TikTok at least two years.

Furthermore, participants reported a mean TikTok usage of 3.81 hours per day (SD = 1.76). Public university students spent significantly more time on TikTok each day than private university students (3.99 vs. 3.58 hours, t(363) = 2.213, p = .027). Daily TikTok usage also differed significantly according to living arrangements, F (2, 362) = 4.205, p = .016. Students living in rented accommodation or dormitories reported the highest average daily usage (M = 4.03, SD = 1.73), followed by those living with family (M = 3.55, SD = 1.79) and those living with relatives (M = 3.37, SD = 1.73).

Table 3 shows that entertainment (90.1%), keeping up with trends (76.2%), and passing time (62.7%) were the most common reasons for using TikTok. Educational use was reported by 49.3% of participants, whereas content creation (18.6%) and monetization content (11.0%) were less common.

Psychometric properties of the TikTok Addiction Scale (TTAS)

Regarding internal consistency, Cronbach’s alpha (α) values above 0.80 are generally considered indicative of good reliability (Hair et al., 2019). The overall scale demonstrated good internal consistency (α = 0.879). Given that several subscales comprise only two items, the Spearman-Brown coefficient was additionally reported for these subscales. The internal consistency estimates for the overall scale and each subscale are presented in Table 4. Notably, the mood modification subscale showed low internal consistency (α = .362; Spearman-Brown = .364), substantially below the subscale-level reliability observed for the other five dimensions (ranging from .617 to .766) and below conventional acceptability thresholds (≥ .70) (Hair et al., 2019).

To further examine the construct validity of the Vietnamese version of the TTAS, a confirmatory factor analysis (CFA) was conducted based on the original six-factor model proposed by Galanis et al. To accommodate missing data configurations rigorously without sample depletion, the model was estimated using the Full Information Maximum Likelihood (FIML) approach. The model showed a marginal fit to the data (χ² (75)=317.39, p<0.001; χ²/df=4.23; CFI=0.878; TLI=0.804; SRMR=0.088; RMSEA=0.094, 90% CI [0.084, 0.105]). Although the comparative fit indices (CFI, TLI) and RMSEA did not meet commonly recommended thresholds for good model fit, the SRMR of .088 was close to, but slightly above, the conventional .08 criterion, indicating that the overall model fit remained marginal (Hu & Bentler, 1999). Standardized factor loadings ranged from 0.279 to 0.843. While most items loaded satisfactorily (> 0.50) on their intended factors, item 3 (mood modification) and item 7 (tolerance) showed relatively low loadings (0.306 and 0.279, respectively). These findings indicate that the original six-factor structure received only marginal empirical support in the Vietnamese sample. Although several items demonstrated satisfactory standardized loadings, the overall fit indices suggest that further psychometric evaluation is needed before the Vietnamese TTAS can be considered a fully validated adaptation of the original instrument.

The results for the TTAS are presented in Table 4. The overall mean TTAS score was 2.58 (SD = .773). The cut-off mean score of 3.23 was adopted from the original TTAS validation study by Galanis et al. (Galanis et al., 2024) and was solely used for descriptive classification. Based on this criterion, 17.5% (n = 64) of the participants scored ≥ 3.23 and exceeded the recommended threshold for high-risk problematic TikTok use, whereas 82.5% (n = 301) scored below the cut-off. Regarding the six factors of the scale, mood modification had the highest mean score, followed by tolerance and withdrawal symptoms, whereas salience had the lowest mean score. However, given the low reliability of the mood modification subscale noted above, this particular ranking should be interpreted with caution. Descriptive statistics and internal consistency estimate for the TTAS are presented in Table 4.

Regarding the relationship between gender, year of study, and academic performance and overall TTAS scores, no significant differences were observed. Initial univariate analyses suggested that students attending private universities reported significantly higher TTAS scores than those enrolled in public universities. Although differences across academic disciplines were observed in the univariate analysis, these associations were no longer statistically significant after adjusting for potential confounding variables.

To further examine the independent associations between participant characteristics and problematic TikTok use, a multiple linear regression analysis was conducted with TTAS score as the dependent variable and university type, field of study, current living arrangement, and daily TikTok usage (hours/day) as predictors. The overall model was statistically significant, F(4,360) = 7.98, p <.001, explaining 8.1% of the variance in TTAS scores (R2 = .081; adjusted R2 = .071). After controlling for the other variables, university type remained the only significant independent predictor of TTAS scores (B = .422, β = .272, p < .001). In contrast, field of study (p = .087), current living arrangement (p = .961), and daily TikTok usage (p = .156) were not significantly associated with TTAS scores.

Prevalence and Platform Dominance

The results indicate that TikTok was the most frequently used social media platform among the sampled students, both in terms of frequency of use and average daily engagement. This finding is consistent with the descriptive statistics presented in Table 2, in which TikTok had the highest mean frequency of use. This pattern aligns with recent international trends documenting TikTok’s rapid growth and intensive engagement among young adult (Jain et al., 2025).

Using the cut-off score proposed in the original TTAS validation study (Galanis et al., 2024), 17.5% of participants exceeded the recommended threshold for high-risk PTU. Because this threshold has not yet been calibrated for the Vietnamese population, this proportion should be interpreted as a descriptive estimate rather than a definitive prevalence of PTU. Nevertheless, the findings suggest that a substantial subgroup of students exhibited elevated PTU symptoms according to the TTAS, a pattern that is broadly consistent with previous international studies (Katsiroumpa et al., 2025; Li & Wang, 2025).

Mood Modification: A Preliminary Observation

Among the six TTAS dimensions, mood modification had the highest mean score, which may be broadly consistent with Compensatory Internet Use Theory, suggesting that individuals utilize digital platforms to regulate negative emotions (Kardefelt-Winther, 2014). This descriptive pattern is consistent with prior research linking escapism and emotional relief to problematic use (Ghadban et al., 2025; Yang et al., 2023), and may reflect the platform’s short-form, highly stimulating content, which could facilitate rapid emotional shifts and reinforce repeated engagement.

However, this finding warrants considerable interpretive caution. As reported above, the mood modification subscale demonstrated low internal reliability in the present sample, well below commonly accepted thresholds and substantially lower than the other five TTAS subscales. One of its two items also showed a weak standardized factor loading in the CFA (.306). Taken together, these psychometric limitations indicate that the two items comprising this subscale may not have consistently captured a single underlying construct among Vietnamese university students, in contrast to the original validation sample (Galanis et al., 2024). Therefore, the elevated mean score for mood modification should be interpreted as a descriptive observation rather than robust evidence that emotional regulation is a well-measured or central mechanism underlying problematic TikTok use in this population. Previous research has suggested that cultural context may shape digital engagement patterns (Cheng et al., 2021); however, because cultural factors were not directly assessed in the present study, no conclusions can be drawn regarding their contribution to the observed findings. Furthermore, because of the cross-sectional design, causal relationships cannot be inferred from the observed associations.

Interrelationship Among Addiction Dimensions

The positive correlations observed among tolerance, withdrawal, and conflict indicate that these dimensions tend to co-occur among students reporting higher levels of problematic TikTok use. Higher tolerance, manifested as difficulty limiting usage, was associated with increased life interference and emotional discomfort when access was restricted. This pattern is consistent with the behavioral addiction literature, where escalating use, impaired control, and functional conflict tend to co-occur. Nevertheless, caution should be exercised when interpreting these results. The observed associations do not indicate irreversible progression but rather suggest patterns of problematic use. Longitudinal studies are required to understand how these patterns develop over time.

Institutional Differences

After adjusting for field of study, current living arrangement, and daily TikTok usage, university type remained the only significant independent predictor of TTAS scores. Students attending private universities consistently reported higher levels of problematic TikTok use than those attending public universities. This finding suggests that university type may act as a proxy for unmeasured institutional or socioeconomic characteristics. Although the mechanisms underlying this association remain unclear, one possible explanation is that university type may partially reflect socioeconomic differences between student populations. Tuition fees at private universities in Vietnam are generally higher than those at public universities (Welch, 2007), suggesting that students attending private universities may, on average, come from households with greater access to digital devices and internet connectivity. However, because socioeconomic status and digital devices were not directly measured, this explanation remains speculative. Future research should incorporate direct measures of socioeconomic status, digital access, and academic workload to clarify this association.

Interestingly, although public university students reported significantly longer daily TikTok usage, they exhibited lower PTU scores than students attending private universities. These findings suggest that problematic TikTok use cannot be explained solely by the amount of time spent on the platform. Rather, PTU may reflect qualitative aspects of engagement, including impaired control, compulsive use, emotional dependence, and functional interference, beyond mere usage duration. This interpretation is consistent with previous research, indicating that problematic social media use involves more than excessive screen time and that the duration of platform use alone is insufficient to explain problematic engagement or well-being outcomes (Amirthalingam & Khera, 2024; Hendrikse & Limniou, 2024).

Living Arrangements and TikTok Use

Although students living in rented accommodation or dormitories reported longer daily TikTok use, living arrangement was not independently associated with TTAS scores after adjusting for covariates, suggesting that residential status alone does not explain problematic TikTok use. Overall, these findings suggest that problematic TikTok use cannot be explained solely by demographic characteristics or usage duration, and that additional individual, psychosocial, or contextual factors may also contribute. The present findings should also be interpreted cautiously because, although the Vietnamese TTAS demonstrated acceptable internal consistency, confirmatory factor analysis provided only marginal support for the original six-factor structure, particularly for the mood modification subscale. These limitations are discussed further in the Limitations section.

Limitations

Several limitations should be acknowledged. First, the cross-sectional design prevents causal inference; future longitudinal studies are needed to examine the developmental trajectories of problematic TikTok use (PTU) over time. Second, the use of convenience sampling with 365 students from five universities in Ho Chi Minh City limits the generalizability of the findings. Future studies should employ multi-regional probability sampling to improve representativeness and generalizability of the findings. Third, restricting participation to active TikTok users may have introduced selection bias regarding reported prevalence and usage patterns. Fourth, all measures were self-reported and may therefore be subject to recall or social desirability bias. Fifth, although the present study did not assess mental health outcomes such as anxiety or depression, the observed demographic correlates may provide a useful basis for future research examining the psychological correlates of PTU. Finally, although the Vietnamese TTAS showed acceptable overall internal consistency, the confirmatory factor analysis provided only partial support for the original six-factor structure, and the mood modification subscale exhibited relatively low reliability. Taken together, these findings, along with the absence of back-translation, pilot testing, and cognitive interviewing, indicate that further psychometric refinement and validation of the Vietnamese TTAS are warranted in larger and more diverse Vietnamese samples before it can be recommended for routine screening in educational settings.

Strengths

Despite these limitations, the present study has several notable strengths. First, the study is among the first to examine problematic TikTok use (PTU) among Vietnamese university students using the TikTok Addiction Scale (TTAS), thereby providing empirical evidence from a Southeast Asian context where research on short-form video addiction remains limited. Second, the study offers preliminary evidence regarding the reliability and construct validity of the Vietnamese TTAS, while also identifying areas requiring further psychometric refinement and cross-cultural validation. Third, by recruiting participants across both public and private universities and various academic disciplines, the study captured a relatively heterogeneous sample, providing broader contextual representation within the sampled population. Finally, by identifying demographic correlates of PTU, the study provides preliminary evidence that may help inform university administrators and student support services in developing digital well-being initiatives and preventive strategies for Vietnamese university students.