Personality, Algorithmic Awareness, and Addictive Symptoms of TikTok Use in University Students

Abstract

(1) Background: Problematic social media use has increasingly been conceptualized as a non-clinical addictive-like behavior characterized by impaired control and negative functional consequences. Despite the rapid growth of TikTok and its algorithm-driven content delivery, the contribution of individual psychological factors and users’ awareness of algorithmic processes to addictive symptoms remains insufficiently understood, particularly in Latin American contexts. This study examined the associations between personality traits, algorithmic awareness, and addictive symptoms of TikTok use among university students. (2) Methods: A quantitative, cross-sectional design was conducted with a convenience sample of 238 university students from Ecuador. Participants completed self-report measures of social media addiction, algorithmic media content awareness, and Big Five personality traits. Spearman correlations and hierarchical multiple regression analyses were performed, controlling for age and sex. (3) Results: Algorithmic awareness dimensions were not significant predictors of addictive symptoms. Demographic variables explained minimal variance, whereas personality traits accounted for the largest increase in explained variance in the final model. Neuroticism and Extraversion were positively associated with addictive symptoms, while Conscientiousness and Openness to Experience were negatively associated. (4) Conclusions: Personality traits were more informative than algorithmic awareness in explaining addictive-like TikTok use among university students, underscoring the relevance of self-regulatory and affective dispositions for prevention and intervention strategies.

1. Introduction

The widespread use of social media in the daily lives of young people has been associated with a sustained increase in problematic behaviors related to these platforms (Dutot, 2020). This phenomenon has generated growing interest within psychological research aimed at identifying the factors that contribute to the development of addiction-like symptoms associated with social media use (Ji et al., 2023; Jo & Baek, 2023; Valkenburg, 2022; Zhao et al., 2022). Although this type of behavior is not recognized as an addictive disorder in either the DSM-5 (American Psychiatric Association, 2022) or the ICD-11 (World Health Organization, 2019)—unlike Gambling Disorder and Gaming Disorder—it has increasingly been conceptualized as a non-clinical addictive behavior characterized by excessive use, difficulties in behavioral control, and the emergence of negative consequences in interpersonal relationships and everyday functioning (Cataldo et al., 2021; Chen, 2023; Hussain & Starcevic, 2020). In this regard, problematic patterns of social media use show functional similarities with other addictive behaviors, which has led to their growing consideration as a phenomenon of relevance to public health (Ihssen & Wadsley, 2021; Moretta & Wegmann, 2025). Consequently, it is necessary to examine in an integrative manner the psychological variables that may contribute to the emergence of these behavioral patterns. The application of the ‘addiction’ construct to digital technologies remains a subject of intense debate (Baggio et al., 2018). Behavioral addiction in this context is often characterized by core components such as salience, mood modification, tolerance, withdrawal symptoms, conflict, and relapse (Abendroth et al., 2020; Griffiths et al., 2016; Karakose et al., 2023; Moretta et al., 2022). However, substantial controversy exists regarding the mechanical transposition of substance-use disorder (SUD) models to digital behaviors. Critics argue that such confirmatory approaches may lead to the over-pathologization of common activities, as certain criteria—such as salience and tolerance—might reflect high engagement or normative intensive use rather than true clinical dysfunction (Reinecke et al., 2018). Furthermore, it is suggested that excessive social media use may function as a symptom of underlying psychological difficulties rather than a primary disorder (Tullett-Prado et al., 2023). Consequently, this study adopts the term ‘addictive symptomatology’ to describe a maladaptive pattern of use defined by impaired control and significant functional impairment, without assuming a formal clinical diagnosis.

In addition, rapidly growing social media platforms such as TikTok—which currently has approximately 1.9 billion monthly active users (Singh, 2026)—facilitate the development of problematic use patterns through the provision of highly engaging and personalized content (Xiong et al., 2024). TikTok pioneered the short-form vertical video format—subsequently adopted by other major social platforms such as Instagram Reels and YouTube Shorts—and it remains the primary exponent of this medium due to its algorithm-driven “For You” feed architecture, which prioritizes algorithmic content curation and discovery over pre-existing social connections or networks (Baumann et al., 2026; Cervi & Marín-Lladó, 2021). These digital environments have been linked to an increased prevalence of psychological difficulties, including anxiety, depressive symptoms, and low self-esteem, largely driven by processes of social comparison and the internalization of unrealistic beauty standards (Ciacchini et al., 2023). Such effects appear to be particularly salient during emerging adulthood (Zhao et al., 2022).

To personalize content and increase user engagement in everyday life, TikTok—and other social media platforms that have subsequently adopted its short-form vertical video model—relies on sophisticated artificial intelligence-based algorithms that differ from traditional social networks by focusing on real-time engagement patterns rather than social connections (Y. Huang & Liu, 2025; Klug et al., 2021). By maximizing user interaction through continuous personalization, these systems foster persistent feedback loops that reinforce repetitive usage patterns and hinder voluntary disengagement (Montag et al., 2021). Within this context, algorithmic awareness—defined as the extent to which users recognize and understand the functioning of recommendation systems—has begun to be conceptualized as a relevant psychological variable, potentially linked to metacognitive processes and self-regulation (Siles et al., 2024). This approach is supported by contemporary Media Literacy frameworks, which suggest that a technical understanding of digital platforms acts as a fundamental tool to mitigate and prevent the negative impacts of social media (Perez-Lozano & Saucedo Espinosa, 2024). Theoretically, fostering critical awareness of how algorithms and design principles operate—such as filter bubbles and content curation—should empower users to regain agency and resist impulsive engagement (Shanmugasundaram & Tamilarasu, 2023). Examining how algorithmic awareness interacts with individual characteristics, such as personality traits and demographic variables, is essential for understanding the psychological mechanisms underlying problematic social media use (Thomas et al., 2022), particularly in data-driven environments such as TikTok where algorithmic mediation is the central axis of the user experience.

From an individual differences perspective, personality traits represent a key dispositional factor for understanding variability in how users interact with highly stimulating digital environments (Hadlington & Scase, 2018). Previous research has consistently shown that traits such as impulsivity (Ferracci et al., 2024; Morales Cevallos et al., 2025), neuroticism (Izhar et al., 2022), and sensation seeking (Chase & Ghane, 2023) are associated with greater difficulties in behavioral self-regulation and heightened sensitivity to immediate rewards—characteristics that increase vulnerability to the development of problematic patterns of digital technology and social media use (Zahrai et al., 2022). In the context of platforms driven by algorithmic recommendation systems, these traits may amplify prolonged exposure to content, promote repetitive behaviors, and reduce voluntary control over use, thereby contributing to the emergence of addiction-like symptoms (Yan & Chen, 2023). Accordingly, personality functions not only as an individual risk or protective factor but also as a moderator of the relationship between algorithmic dynamics and digital consumption behaviors.

In the Ecuadorian context, recent studies have examined the consequences of problematic social media use among young populations. Research conducted with adolescents has shown that time spent on social media platforms predicts higher levels of body dissatisfaction, along with indicators of emotional habituation or dissociation among individuals exhibiting excessive use (Oleas Rodríguez et al., 2025). Complementarily, studies involving university students have found that symptoms of addictive social media use are negatively associated with self-esteem, particularly through problems related to compulsive use and persistent preoccupation with online connectivity (Oleas Rodríguez & López-Barranco Pardo, 2024).

Within this framework, the present study aims to examine the association between personality traits, algorithmic awareness, and symptoms of addictive TikTok use among university students. The analysis focuses on characterizing patterns of platform use and their relationship with levels of addictive symptomatology. This approach contributes to a more nuanced understanding of the psychological mechanisms involved in problematic social media use and provides an empirical foundation for the development of future preventive strategies targeting young populations.

2. Materials and Methods

2.1. Design

A quantitative, cross-sectional, correlational design was employed, with a single administration of self-report questionnaires. The aim was to analyze the psychological and motivational predictors of TikTok addiction, including personality traits and algorithmic awareness as relevant variables.

2.2. Participants

The sample was obtained through non-probabilistic convenience sampling, recruiting participants from a university located in the canton of Samborondón (Ecuador). Inclusion criteria were: (a) being 18 years of age or older, (b) regular class attendance, and (c) being an active TikTok user, defined as having an account on the platform and reporting any level of use, including less than once per week.

Participants were recruited during regular class sessions in transversal undergraduate courses, which include students from a wide range of academic programs and different stages of their academic progression. This approach allowed access to a heterogeneous sample across different fields of study. Participation was voluntary, and no financial incentives or course credits were provided. To enhance data quality, attention-check items were included in the questionnaire (e.g., simple arithmetic questions) to identify inattentive or random responding.

A total of 238 participants took part in the study, with a mean age of 20.6 years (SD = 3.2; range = 18–42). The majority were women (68.1%; n = 162) and single (93.7%; n = 223). Nearly half of the sample reported being unemployed (47.9%; n = 114).

Regarding TikTok use, 98.3% (n = 234) reported having an active profile on the platform. Daily use was reported by 63.9% (n = 152), with most participants indicating an average daily usage of 1 to 3 h (53.8%; n = 128). Session duration was primarily concentrated between 10 and 30 min (49.2%; n = 117). In terms of content creation, most participants reported not posting videos (65.6%; n = 156). With respect to social reach, 80.3% (n = 191) reported having between 0 and 500 followers, and 78.6% (n = 187) reported following between 0 and 500 accounts. Participants selected predefined categorical ranges to report their number of followers and accounts followed. These values were obtained using predefined categorical ranges selected by participants and were designed to reflect the typical distribution of user reach on the platform. This distribution indicates that the sample is primarily composed of typical users rather than high-reach content creators. A more detailed description of consumption patterns and usage characteristics is provided in

Table 1. TikTok Consumption Patterns.

Variable	Category	N	%
Personal profile	Yes	234	98.32
	No	4	1.68
Frequency of use	Less than once a week	11	4.62
	1–3 days per week	19	7.98
	3–5 days per week	22	9.24
	More than 5 days per week, but not daily	34	14.29
	Every day	152	63.87
Weekly hours of use	Less than 1 h	48	20.17
	1–3 h	128	53.78
	3–5 h	45	18.91
	More than 5 h	17	7.14
Session duration	Less than 10 min	18	7.56
	10–30 min	117	49.16
	30–60 min	66	27.73
	More than 1 h	37	15.55
Videos posted per week	No videos posted	156	65.55
	0–5	71	29.83
	5–10	2	0.84
	10–20	5	2.10
	More than 20	4	1.68
Number of followers	0–500	191	80.25
	500–1000	16	6.72
	1000–5000	20	8.40
	More than 5000	11	4.62
Accounts followed	0–500	187	78.57
	500–1000	27	11.34
	1000–5000	17	7.14
	More than 5000	7	2.94

.

2.3. Instruments

Data were collected using an online questionnaire developed in Microsoft Forms, structured into four main sections.

In the first section, together with informed consent, an ad hoc questionnaire was administered to collect sociodemographic information and data on platform use. Sociodemographic variables included age, sex, marital status, and employment status.

TikTok consumption patterns were assessed using single-item questions with predefined categorical response options. These variables included frequency of use (“How often do you use TikTok?”), daily hours of use (“On a typical day, approximately how many hours do you spend on TikTok?”), average session duration (“Each time you access TikTok, how long do you usually remain connected?”), number of videos posted per week (“Approximately how many videos do you post per week on TikTok?”), approximate number of followers (“How many followers do you currently have?”), and number of accounts followed (“How many accounts do you currently follow?”) (see Table 1). Participants selected the category that best represented their usage patterns. These variables were used for descriptive purposes.

In the second section, social media addiction was assessed using the Social Media Addiction Questionnaire (ARS) (Mayaute & Blas, 2014). This instrument comprises a 24-item general social media addiction scale, which assesses addiction levels across several dimensions (e.g., obsession with social media, lack of control, and excessive use). Example items include: “I feel anxious when I cannot access social media” and “I spend more time on social media than I initially intended.” Responses are recorded on a 7-point Likert scale ranging from 1 = strongly disagree to 7 = strongly agree. In the original validation study, the questionnaire showed high internal consistency (α = 0.92). In the present study, reliability was also high (α = 0.94; ω = 0.94).

The third section assessed awareness of algorithmically mediated content using the Algorithmic Media Content Awareness Scale (AMCA) (Zarouali et al., 2021). This scale comprises 13 items distributed across four dimensions: (a) awareness of content filtering, (b) awareness of automated decision-making, (c) awareness of human–algorithm interaction, and (d) awareness of ethical considerations Items are phrased to assess the extent to which users are aware of how algorithms operate, rather than their evaluation of these processes. For example, participants indicate their level of awareness of statements such as “Algorithms are used to recommend content to me on TikTok” and “The content that algorithms recommend to me depends on my online behavior.” More broadly, items address whether algorithms are used to recommend or prioritize content, whether content selection is based on automated decisions, and whether recommendations depend on users’ online behavior and personal data. The scale was adapted into Spanish using a back-translation procedure and contextualized for TikTok by replacing the platform name in the items and adjusting the referenced content, in accordance with the authors’ recommendations. Responses are recorded on a 5-point Likert scale (1 = not aware at all, 5 = fully aware). In the original study, the scale demonstrated high reliability across dimensions (α ranging from 0.86 to 0.94). In the present study, reliability indices were high (α ranging from 0.84 to 0.92; ω ranging from 0.85 to 0.92). To further examine the structural validity of the adapted version of the Algorithmic Media Content Awareness Scale (AMCA), a confirmatory factor analysis (CFA) was conducted to test the original four-factor structure. The model was estimated using a diagonally weighted least squares (DWLS), given the ordinal nature of the Likert-type items. The results indicated an excellent fit to the data (CFI = 0.997, TLI = 0.997, RMSEA = 0.074, SRMR = 0.044). All items loaded significantly on their respective factors, with standardized loadings ranging from 0.76 to 0.95, providing strong evidence for the structural validity of the scale in this sample.

The fourth section included the assessment of personality using the 10-item version of the Big Five Inventory (BFI-10) (Gosling et al., 2003). This instrument provides a brief measure of the five major personality dimensions: Extraversion, Agreeableness, Conscientiousness, Neuroticism, and Openness to Experience. Each dimension is assessed by two items, one positively worded and one reverse-coded. Example items include: “Is reserved” (Extraversion), “Does a thorough job” (Conscientiousness), and “Has an active imagination” (Openness to Experience). Responses are recorded on a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree). The Spanish version of the BFI-10 has demonstrated adequate psychometric properties for research contexts with time constraints (Romero et al., 2012). In the present study, internal consistency indices were acceptable for an ultra-brief measure (α = 0.65; ω = 0.64), supporting its use in this research context.

2.4. Procedure

Data collection took place in August 2025 in classrooms of a private university in Ecuador, using an online questionnaire developed in Microsoft Forms. Informed consent was presented at the beginning of the questionnaire, and acceptance was mandatory in order to proceed to the remaining sections. The questionnaire was designed such that participants could not continue or submit their responses without completing each prior section, thereby ensuring data completeness and integrity.

The study complied with the ethical principles of the Declaration of Helsinki and was approved by the Research Committee of the Psychology program at Universidad ECOTEC. Participation was voluntary, anonymous, and uncompensated, and data confidentiality was guaranteed.

2.5. Data Analysis

Statistical analyses were conducted using JASP software (version 0.95.4) (JASP Team, 2025). First, descriptive statistics (means, standard deviations, frequencies, and percentages) were computed to characterize the sample and the main study variables. Assumptions of normality were then evaluated using distributional tests as well as skewness and kurtosis values. Given observed violations of normality criteria, associations between variables were examined using Spearman’s rank-order correlation (ρ).

To examine predictors of TikTok addiction, a hierarchical multiple linear regression analysis was performed using standardized z scores to facilitate interpretation and comparison of coefficients across variables measured on different scales. The regression model was structured in three steps. Predictors within each block were entered simultaneously using the enter method:

• Model 1: included sociodemographic variables (sex and age) as controls.
• Model 2: added algorithmic awareness factors (content filtering, automated decision-making, human–algorithm interaction, and ethical considerations).
• Model 3: incorporated personality dimensions assessed with the BFI-10.

The incremental contribution of each block was evaluated using changes in explained variance (ΔR²), and statistical significance was assessed using the F-test for change in R², which accounts for the increase in the number of predictors across models.

A post hoc power analysis conducted using G*Power (version 3.1.9.7) (Faul et al., 2020) indicated that the final regression model achieved a statistical power of approximately 0.90, based on the observed effect size (R² = 0.15), the sample size (N = 238), and an alpha level of 0.05.

3. Results

3.1. Descriptive Analysis

Table 2. Descriptive statistics and normality tests for study variables.

Variable	Mean	SD	Min	Max	Skewness (g1)	Kurtosis (g2)	KS 1
Addiction	86.20	27.27	26.00	168.00	0.26	−0.06	0.05
Content filtering awareness	16.10	3.22	8.00	20.00	−0.44	−0.79	0.12 *
Automated decision-making awareness	11.18	2.71	4.00	15.00	−0.12	−0.79	0.13 *
Human–algorithm interaction awareness	12.34	2.68	5.00	15.00	−0.64	−0.66	0.21 *
Ethical considerations awareness	10.92	2.81	3.00	15.00	−0.14	−0.70	0.12 *
Neuroticism (BFI-10)	5.79	1.50	2.00	10.00	−0.24	0.70	0.22 *
Openness to experience (BFI-10)	6.82	1.42	2.00	10.00	0.20	0.37	0.18 *
Agreeableness (BFI-10)	5.84	1.41	2.00	10.00	0.31	1.11	0.20 *
Extraversion (BFI-10)	5.84	1.55	2.00	10.00	−0.09	0.74	0.19 *
Conscientiousness (BFI-10)	6.85	1.48	3.00	10.00	0.22	−0.27	0.18 *

¹ Kolmogorov–Smirnov test * p < 0.05.

presents the descriptive statistics for the main study variables. Kolmogorov–Smirnov normality tests indicated that most variables did not meet the assumption of normality (p < 0.05). Accordingly, nonparametric statistical procedures and regression analyses with standardized coefficients were used in subsequent analyses.

3.2. Correlational Analysis

Table 3. Spearman correlations among study variables.

Variable	1.	2.	3.	4.	5.	6.	7.	8.	9.	10.
1. Age	—
2. Addiction	−0.11	—
3. Content filtering awareness	−0.003	0.08	—
4. Automated decision-making awareness	0.14 *	0.08	0.66 ***	—
5. Human–algorithm interaction awareness	0.04	−0.02	0.79 ***	0.62 ***	—
6. Ethical considerations awareness	0.09	0.02	0.59 ***	0.62 ***	0.65 ***	—
7. Neuroticism (BFI-10)	−0.10	0.24 ***	0.09	0.08	0.06	0.05	—
8. Openness to experience (BFI-10)	0.03	−0.19 **	0.11	0.07	0.11	0.08	−0.21 ***	—
9. Agreeableness (BFI-10)	0.04	−0.07	−0.15 *	−0.13	−0.03	−0.01	−0.08	0.07	—
10. Extraversion (BFI-10)	0.07	0.14 *	0.03	0.04	−0.02	0.003	−0.23 ***	0.13 *	0.20 **	—
11. Conscientiousness (BFI-10)	0.14 *	−0.25 ***	0.03	0.02	0.05	−0.01	−0.23 ***	0.27 ***	0.02	0.03

* p < 0.05, ** p < 0.01, *** p < 0.001.

shows the correlations among the study variables. Spearman’s rho (ρ) was used due to violations of normality assumptions, as it provides a more robust estimate of association under non-normal distributions.

3.3. Hierarchical Regression Analysis

Hierarchical regression analyses were conducted to examine the relative contribution of demographic variables, algorithmic awareness, and personality traits in predicting addiction scores (

Table 4. Hierarchical regression analysis predicting addiction.

Predictor	M1 β [95% CI B]	M2 β [95% CI B]	M3 β [95% CI B]
Age	−0.14 [−2.30, −0.14] *	−0.16 [−2.44, −0.26] *	−0.10 [−1.89, 0.25]
Sex (male)	0.14 [−6.90, 7.95]	≈0.00 [−7.42, 7.36]	−0.32 [−8.49, 6.10]
Content filtering awareness	—	0.17 [−0.39, 3.35]	0.16 [−0.44, 3.21]
Automated decision-making awareness	—	0.12 [−0.59, 3.10]	0.10 [−0.80, 2.74]
Human–algorithm interaction awareness	—	−0.22 [−4.54, 0.02]	−0.18 [−4.04, 0.37]
Ethical considerations awareness	—	0.03 [−1.42, 1.92]	0.03 [−1.36, 1.85]
Neuroticism (BFI-10)	—	—	0.18 [0.81, 5.70] **
Openness to experience (BFI-10)	—	—	−0.14 [−5.10, −0.10] *
Agreeableness (BFI-10)	—	—	−0.02 [−2.76, 2.18]
Extraversion (BFI-10)	—	—	0.19 [0.97, 5.57] **
Conscientiousness (BFI-10)	—	—	−0.15 [−5.25, −0.37] *
R2	0.02	0.05	0.15
ΔR2	—	0.03	0.10

* p < 0.05, ** p < 0.01.

). In Model 1, which included only demographic covariates (age and sex), the overall model was not statistically significant, F(2, 235) = 2.48, p = 0.09, and the explained variance was minimal (R² = 0.02). Although age showed a significant coefficient at the individual level, this effect should be interpreted with caution given the non-significant overall model.

In Model 2, algorithmic awareness variables (content filtering, automated decision-making, human–algorithm interaction, and ethical considerations) were added. This model showed a slight increase in explained variance (R² = 0.05; ΔR² = 0.03); however, the overall model fit did not reach statistical significance, F(6, 231) = 2.03, p = 0.06.

In Model 3, personality traits (neuroticism, openness to experience, agreeableness, extraversion, and conscientiousness) were incorporated. This model resulted in a substantial increase in explained variance (R² = 0.15; ΔR² = 0.10) and reached statistical significance, F(11, 226) = 3.76, p < 0.001.

In the final model, Neuroticism and Extraversion were positively associated with addictive symptoms, indicating that higher levels of these traits were related to higher levels of TikTok addiction. In contrast, Conscientiousness and Openness to Experience were negatively associated with addiction, suggesting that higher levels of these traits were linked to lower levels of addictive symptoms. No significant associations were observed for the algorithmic awareness dimensions.

4. Discussion

The primary aim of the present study was to examine the psychological and motivational predictors of social media addiction, specifically TikTok addiction, in a sample of university students. This research provides novel evidence from a Latin American context on how stable personality dispositions and technological understanding—operationalized as algorithmic awareness—interact in the development of addictive behaviors.

Our findings indicate that personality traits constitute the most robust predictors of TikTok addiction, surpassing the explanatory power of demographic and cognitive variables. Specifically, Neuroticism and Extraversion emerged as significant risk factors. This pattern is consistent with the meta-analysis conducted by C. Huang (2022), which identified neuroticism as a risk factor for social media addiction, suggesting that individuals high in neuroticism may resort to these platforms as a coping mechanism to regulate negative emotional states. Similarly, Extraversion has been consistently associated with intensive social media use, as extraverted individuals tend to seek social interaction and self-presentation opportunities, thereby increasing their susceptibility to addictive use patterns (Blackwell et al., 2017; Dilawar et al., 2022; Santos & Alves, 2025).

In contrast, Conscientiousness and Openness to Experience functioned as protective factors. The negative association between addiction and Conscientiousness aligns with prior research indicating that self-control, planning, and discipline protect against excessive and dysregulated use of substances (Escamilla et al., 2024). Of particular interest is the finding related to Openness to Experience. Although the literature often reports a positive association between this trait and general social internet addiction (Öztürk et al., 2015), our results converge with studies that conceptualize openness as a protective factor (Rajesh & Rangaiah, 2022). One possible explanation lies in the qualitative nature of the stimulation involved. While platforms such as TikTok increasingly provide novel and even informational content, their algorithm-driven structure promotes rapid, fragmented, and highly repetitive exposure, which may limit sustained cognitive engagement. Individuals high in openness typically seek novelty that is complex, exploratory, and cognitively demanding. As such, they may be less attracted to passive, continuous consumption patterns, even when content appears diverse or stimulating at a superficial level. This distinction may help explain why openness to experience operates as a protective factor in the present study, despite mixed findings in the broader literature.

One of the most revealing findings of this study was that none of the dimensions of algorithmic awareness significantly predicted addictive symptoms. This contradicts prevailing Media Literacy models, which posit that understanding the underlying mechanisms and design principles of digital platforms serves as a primary protective factor against social media addiction (Perez-Lozano & Saucedo Espinosa, 2024). Theoretically, being aware of algorithmic tactics—such as content personalization and filter bubbles—should activate critical thinking and empower users to regain agency, allowing them to resist immediate rewards and manage their screen time more deliberately (Shanmugasundaram & Tamilarasu, 2023). However, our results indicate that, in the specific context of TikTok, such technical knowledge is insufficient to counteract the platform’s reinforcing architecture. While these educational frameworks assume that cognitive empowerment facilitates self-regulation, the ‘gamified’ and immersive nature of TikTok’s ‘For You’ feed appears to bypass these rational defenses. This pattern can be explained through the incentive sensitization theory proposed by Ihssen and Wadsley (2021), which posits that compulsive social media use involves a dissociation between wanting—a motivational drive—and cognitive evaluation or liking. Accordingly, even when users possess high levels of algorithmic awareness at a cognitive level, this awareness fails to inhibit impulsive responses triggered by a sensitized reward system. In addition, Nie (2025) argues that social media platforms deliberately employ addictive design features and so-called dark patterns aimed at reducing users’ agency and autonomy. By removing natural decision points (e.g., through infinite scrolling), application architecture undermines users’ capacity to exert critical judgment, rendering digital literacy or algorithmic awareness largely irrelevant in the face of addictive behavior. Furthermore, it is important to note that the AMCA scale assesses the cognitive recognition of algorithmic mediation but does not evaluate the user’s affective or evaluative stance toward these processes. This absence of attitudinal measurement might further explain why awareness alone, despite its theoretical importance in Media Literacy, is insufficient to counteract the impaired control that characterizes addictive symptomatology.

Regarding demographic variables, age showed a significant association at the individual level in the baseline model; however, this effect should be interpreted with caution given that the overall model was not statistically significant; however, its relation disappeared once personality traits were included in the model. This finding suggests that age may operate as an indirect marker of underlying psychological dispositions. Similar patterns have been reported in studies on problematic smartphone use (Pera, 2020) and substance use (Zilberman et al., 2020), where sociodemographic effects diminish once psychological factors are accounted for. The final model explained 15% of the total variance (R² = 0.15). Although this proportion is statistically meaningful, a substantial amount of unexplained variance remains, highlighting the need to consider contextual, environmental, and interface-design factors that extend beyond individual-level variables.

These findings should be interpreted in light of several limitations. First, the cross-sectional design precludes causal inferences. Second, the use of the BFI-10, an ultra-brief personality measure, entails lower reliability compared to longer inventories, although its psychometric performance was acceptable for the purposes of this study. Third, the sample was obtained through convenience sampling from a private university, which limits generalizability. Despite these limitations, the study offers a valuable contribution by prioritizing psychological predictors in the explanation of this emerging behavioral phenomenon.

5. Conclusions

The results of this study indicate that TikTok addiction among university students is primarily driven by personality traits rather than by demographic factors or the level of technical knowledge about the platform. A risk profile characterized by high levels of Neuroticism and Extraversion, together with low levels of Conscientiousness and Openness to Experience, was identified. These findings tentatively suggest that individual differences in emotional self-regulation and impulse control may be relevant targets for future research and potential intervention strategies. However, given the cross-sectional nature of the data, further longitudinal and experimental studies are needed to establish causal relationships before drawing firm conclusions for prevention or intervention.

Moreover, this study provides critical evidence regarding the insufficiency of algorithmic awareness as a protective factor. The fact that understanding how content filtering systems operate does not reduce addictive behavior challenges traditional digital literacy approaches. This suggests that educational strategies based solely on informing users about technological risks or AI-driven mechanisms are limited when confronted with the persuasive architecture and reward-based mechanisms embedded in the application. Accordingly, future public health and educational policies should integrate not only cognitive training but also behavioral strategies aimed at counteracting addictive design features that constrain user agency.

Beyond the specificities of the platform, it must be considered whether TikTok represents a specific case study or a proxy for contemporary social media engagement. While the present research highlights the platform’s unique algorithmic loop, the functional convergence across digital environments suggests that the identified psychological vulnerabilities may reflect a generalized response to persuasive design and constant content personalization. Consequently, although the intensification of addictive symptoms is linked to TikTok’s specific architecture, these findings provide a foundation for understanding broader patterns of problematic use in an increasingly data-driven digital ecosystem.

Finally, although age appears to be associated with problematic use, its effect is secondary to underlying psychological dispositions as reflected in personality traits. Future research should employ longitudinal designs to clarify the causal direction between these traits and addictive behaviors, as well as to explore the role of contextual and design-related variables (e.g., dark patterns) that may account for the unexplained variance beyond individual-level factors.