Introduction

Exercise addiction is a behavioral disorder characterized by an uncontrollable desire to engage in exercise activities, even when these activities lead to negative consequences (1). While regular exercise is widely recognized as beneficial for physical and mental health, excessive and obsessive participation in these activities can lead to detrimental outcomes. As a behavioral phenomenon, sports addiction has numerous physical and psychological consequences that can endanger an individual's health. In Iran, with the spread of a fitness culture and social pressures to achieve ideal body standards, the prevalence of this behavioral disorder has increased. However, research conducted in this area is scattered and lacks a unified theoretical framework. Meta-analysis of previous research, by combining empirical findings, can provide deeper insight into the role of psychological predictors of this disorder. Given the large number of empirical studies in this area, a scientific meta-analysis is needed to combine and interpret the findings coherently. The main innovation of this study is in identifying and analyzing the role of psychological variables as predictors of sports addiction and providing an overall picture of the current situation in Iran. By integrating empirical evidence, this study highlights the gaps in the research literature and will pave the way for designing effective psychological interventions. Also, focusing on the Iranian cultural context provides added value for the local understanding of this phenomenon. On the other hand, contradictory results about the role of gender or existing psychological variables have created a kind of confusion and ambiguity in studies that a meta-analytic approach will be responsible for resolving. This in itself can lead to finding and identifying the psychological variables with the greatest impact. Therefore, the purpose of the present study is to meta-analyze studies related to sports addiction in Iran, focusing on the psychological determinants of this issue. In line with the purpose of the study, the following question can be raised. Which psychological variables have the greatest impact on sports addiction? Does gender play a role as a moderating variable?

Materials and Methods: This study was conducted in accordance with the guidelines provided by the "Preferred Reporting Items for Systematic Reviews and Meta-Analysis Protocol (PRISMA)" and the "Cochrane Guide for Systematic Reviews". In this study, the meta-analysis method was used in accordance with the purpose. The unit of analysis in meta-analysis is the quantitative findings of other studies. Therefore, the scope of this study included studies conducted on the relationship between psychological variables and exercise addiction during the past 13 years in Iran (1403-1390) that met the necessary methodological requirements. Accordingly, the search sources included all studies published in domestic scientific-research journals and theses (full-text studies that were accessible in the databases of Normags, Mag Iran, Comprehensive Portal of Humanities, Alamnet, Civica, Jihad Daneshgahi, and IranDoc). To select primary studies from the above sampling framework, first, valid keywords were determined based on a review of the research background to be used in searching for primary studies. After identifying the keywords for selecting primary studies, they were selected based on the inclusion and exclusion criteria. The keywords were: exercise addiction, exercise addiction, psychology and exercise addiction, psychological prediction of exercise addiction, psychology and exercise addiction. Studies that did not have sufficient information to calculate the effect size or were complete replications of other studies were excluded. First, qualitative content analysis techniques were used to select appropriate and relevant keywords to monitor the articles used. This method allows for the identification of central concepts and hidden conceptual relationships in the text. First, related topics in the articles were divided into smaller semantic units, then the main keywords were identified by coding key concepts. In the selection process, the researcher first searched for related studies and, after eliminating duplicates, examined the titles and abstracts. In case of ambiguity, more detailed assessments were performed, and finally, the studies eligible for selection and key findings were recorded. From each selected study, information such as the author's name, year of publication, research method, sample characteristics (size, gender, age, education, ethnicity, and location), and instruments measuring psychological variables and exercise addiction were extracted. Also, effect sizes, correlation coefficients, p-values, standard deviations, and 95% confidence intervals were entered into the software to make the analyses more accurate. To assess the quality of the studies, the risk of bias assessment tool in the CMA2 software was used. The researcher independently conducted this assessment and ensured that the selected studies met the necessary standards in terms of objectives, participant selection, sample size, management of missing data, and ethical considerations. In the present study, cma2 software and Hedges' g index were used to calculate the effect size and to perform further statistical operations related to the combination of results. Heterogeneity between studies was assessed using Tau2, I², and Cochran's Q statistics. I² values ​​of 25%, 50%, and 75% were considered to indicate low, moderate, and high levels of heterogeneity, respectively. Depending on the level of heterogeneity, random-effects and fixed-effects meta-analyses were performed. Specifically, when heterogeneity was greater than 50%, random-effects meta-analyses were performed, whereas when heterogeneity was less than 50%, fixed-effects meta-analyses were estimated

Results: Initially, a total of 40 records were reviewed. After eliminating 10 studies by title, 30 articles remained. Of the remaining 30 articles, full assessments were performed to determine their suitability for inclusion in the study. Finally, 14 studies met all inclusion criteria. The remaining studies were excluded for the following reasons: 11 studies did not meet the inclusion criteria such as appropriate scientific content and journal credibility, 3 had incomplete methodological information, and 2 were excluded due to statistical deficiencies and gaps or because they did not provide sufficient data to report effect sizes. The meta-analysis sample included a total of 3331 participants, with the number of subjects ranging from 40 to 391. Participants in all studies included both men and women. Geographically, all studies were conducted in Iran.

The results show that in order to eliminate the level of significance obtained in the study (to reach the alpha significance level), 8491 more studies must be examined, which indicates that there is no publication bias in these studies. In other words, after adding 8491 more studies, the estimated effect sizes become insignificant. For greater certainty, the Begg and Mazumdar rank correlation test was used to examine publication bias. The Kendall tau value was estimated to be -0.158, which was confirmed with a significance level of 0.074 for single domain and 0.148 for two domains. The null hypothesis of symmetry of the graph and lack of bias was also confirmed. For greater certainty, the Egger's test was used. The value of the Egger's test statistic is 5.785. Given that the p-value of the significance level of 0.055 for the single-domain and 0.110 for the two-domain, which is greater than 0.05, it can be concluded that there is no strong evidence of publication bias in the studies. The calculated effect sizes for the studies are reported below. Also, in order to assess the possibility of publication bias, the Trim and Fill method presented by Duval and Tweedie was used under both fixed-effect and random-effect models. In the fixed-effect model, no studies were estimated as missing studies and the observed effect size remained unchanged at r = 0.302, indicating no evidence of publication bias. However, in the random-effect model, the adjusted effect size decreased slightly from r = 0.302 to r = 0.273 after including the hypothetical studies. This small decrease indicates a small amount of asymmetry in the funnel plot and a low probability of publication bias. However, the impact of this correction on the final result was insignificant and the adjusted effect size remained in the medium range; therefore, the results of the meta-analysis have appropriate stability and reliability.

Effect size of social media addiction (r = 0.265, p < 0.05), harmful eating behaviors (r = 0.475, p < 0.05), negative body image (r = 0.195, p < 0.05), perfectionism (r = 0.216, p < 0.05), psychological needs (r = -0.060, p > 0.05), neuroticism (r = -0.360, p < 0.05), extraversion (r = 0.309, p < 0.05), openness (r = 0.253, p < 0.05), conscientiousness (r = 0.317, p < 0.05), agreeableness (r = 0.196, p < 0.05), narcissism (r = 0.131, p > 0.05), stress (r = 0.400, p < 0.05), self-handicapping (r = 0.450, p < 0.05), depression (r = -0.123, p > 0.05), competitiveness (r = 0.460, p < 0.05), triumphalism (r = 0.210, p < 0.05), goal orientation (r = 0.300, p < 0.05), behavioral activation (r = 0.470, p < 0.05), emotional regulation problems (r = 0.740, p < 0.05), sports commitment (r = 0.444, p < 0.05), self-compassion (r = -0.560, p < 0.05), body image concerns (r = 0.360, p < 0.05), dysfunctional attitudes (r = 0.845, p < 0.05), eating disorder (r = 0.294, p < 0.05), novelty seeking (r = 0.198, p < 0.05), harm avoidance (r = -0.196, p < 0.05), reward seeking (r = 0.214, p < 0.05), perseverance (r = 0.171, p < 0.05), self-directedness (r = -0.115, p > 0.05), cooperation (r = 0.423, p < 0.05), self-transcendence (r = 0.579, p < 0.05), amotivation (r = -0.054, p > 0.05), intrinsic motivation (r = 0.324, p < 0.05), external regulator (r = 0.251, p < 0.05), mixed regulation (r = 0.325, p < 0.05), self-accepting regulation (r = 0.322, p < 0.05), introverted regulation (r = 0.328, p < 0.05), bystander stress (r = 0.110, p < 0.05), sensation seeking (r = 0.472, p < 0.05), social support (g = 0.364, p < 0.05) were estimated. As can be seen, the observed effect size covers a distribution from -0.060 to 0.845. The variables that did not have a significant effect on sports addiction include: psychological needs, narcissism, depression, self-strategy, and amotivation. These variables did not show a significant effect on sports addiction in statistical analysis. The rest of the effect sizes of the study were significant (P < 0.05).

The combined effect size in the fixed effects model was 0.302 and in the random effects model it was 0.273 (P<0.05). As can be seen, the value of the Q index for 40 effect sizes and 39 degrees of freedom is 716.749, which is statistically significant. The significance of the Q index indicates the presence of heterogeneity in the effect sizes of the primary studies. Since the Q value was statistically significant, it means that significant heterogeneity was observed in the effect sizes of the studies. This means that in most studies, different results were obtained regarding psychological variables and exercise addiction in Iranian evidence. Another index used for this purpose is the I square. This square has a value from zero to 100 percent and in fact shows the amount of heterogeneity as a percentage. The closer this value is to 100, the greater the heterogeneity of the effect sizes of the primary studies. The I square shows that 94.798 percent of the dispersion in the effect sizes of the primary studies indicates high heterogeneity in these studies. In other words, about 94.7% of the differences in effect sizes are due to various reasons such as the specific characteristics of each study. Accordingly, the random model was selected as the meta-analysis model and the combined effect size was considered to be the same value of 0.590. The Tau² value (0.276) indicates that there is significant true variance between studies. If Tau² is close to zero, the true heterogeneity between studies is very low. If Tau² is larger, it indicates greater true heterogeneity between studies. Next, meta-regression was used to examine the role of publication year. The meta-regression plot showed that there was no significant relationship between year of publication and Fisher's effect size (Z); the slope of the line was approximately zero and without a clear direction. This indicates a relative stability of the effect over time and a reduced likelihood of time bias, which strengthens the validity and generalizability of the results.

Conclusion: The present meta-analysis, for the first time in Iran, calculated the effect size of psychological variables on sports addiction and provided a quantitative synthesis of the recent research background. For this purpose, data from 14 studies were recorded in the meta-analysis worksheet, and the average effect size was 0.302. The effect size obtained (0.302) is in the medium to high range based on common criteria such as Cohen's (1988) classification. This value indicates that psychological variables play a significant and significant role in predicting sports addiction. From a clinical perspective, this finding indicates that therapists, sports psychologists, and counselors can identify at-risk individuals by assessing psychological components such as perfectionism, anxiety, intrinsic motivation, or emotion regulation.

Based on these findings, it can be concluded that a wide range of psychological variables are associated with sports addiction. The findings also highlight the importance of psychological variables in sports addiction and provide implications for future research. Therefore, it is essential to pay attention to this issue in educational and therapeutic interventions. The limitations of this study are related to several aspects. First, most of the existing studies have focused on the relationship between psychological variables and sports addiction in specific populations such as professional athletes, and therefore the results may be limited in generalizing to other population groups. In addition, the studies examined in this meta-analysis were often cross-sectional, which cannot clarify the causal relationships between variables and evaluate their long-term effects. Also, since the data used only include evidence available in Iran, the results of this study may vary significantly in other cultures or different countries and the results may not be the same. It can be analyzed that these limitations are the result of a lack of resources, practical difficulties in conducting longitudinal research, or the focus of the scientific community on specific groups.

Keywords: Exercise addiction, exercise dependence, psychology and exercise addiction, educational meta-analysis, sports meta-analysis.

Ethical Considerations

Compliance with ethical guidelines

The cooperation of the participants in the present study was voluntary and accompanied by their consent.

Funding

No funding.

Authors' contribution

All authors contributed equally to the article.

Conflict of interest

The authors declared no conflict of interest.