Assessment of Self-Determined Motivation in Exercise: A Systematic Review and Meta-analysis

Self-determined motivation is measured on numerous scales as a predictor of long-term exercise. Our study applied a systematic review and meta-analysis to assess associations between types of self-determined motivation in exercise with selected parameters of studies using Behavioral Regulation in Exercise Questionnaires (BREQs). Following PRISMA 2020 guidelines, we screened 244 studies from PubMed, ScienceDirect, and Scopus against criteria and selected 43 articles for qualitative synthesis. Of those, 40 studies reporting mean scores and standard deviations of six regulations of BREQs, representing self-determined motivation types, were selected for meta-analysis. The pooled mean scores were the highest in intrinsic regulation at 4.00 (95%CI: 2.92–5.09), followed by identified, integrated, introjected, external, and motivation regulations at 3.65 (95%CI: 3.06–4.24), 3.11 (95%CI: 2.68–3.55), 2.21 (95%CI: 1.88–2.53), 1.42 (95%CI: 0.93–1.92), and 0.94 (95%CI: 0.67–1.12), respectively. Findings indicated significant associations between longer exercise duration and introjected (β = 0.014, p = 0.027) and identified (β = 0.014, p = 0.021) regulations. An inverse relationship was found with a higher female participation rate (β=-0.047, p=0.042), while exercise settings in sports and fitness centers exhibited a stronger association with intrinsic motivations (β=2.700, p=0.039). No significant differences were observed among the versions of BREQs in measuring self-determined motivation. This investigation of context invariance utilizing the particular validated scale contributes to furthering comprehension of the instrument in sports and fitness settings. Additionally, it is essential to take into account sex and the exercise environment concerning self-determined motivation when predicting long-term exercise adherence.


Introduction
Self-determined motivation is defined as "the process whereby goal-directed activities are instigated and sustained" [1].This definition highlights the integral role of motivation in initiating and maintaining activity to achieve a specific objective, with a particular emphasis on goal-oriented behavior [2].As exercise predictors, three main types of motivation from the self-determination theory, including extrinsic motivation, intrinsic motivation, and amotivation, were adapted to examine relationships with exercise maintenance via indirect or direct mechanisms [3].
The Behavioral Regulation in Exercise Questionnaires (BREQs) and recent amendments have emerged as the prevailing instruments in psychological research for self-determined exercise motivation measurement [4][5][6].Over the years, the BREQs have been restructured in various iterations, featuring four to six constructs, including amotivation, external regulation, introjected regulation, identified regulation, integrated regulation, and intrinsic regulation.Besides, the number of items within each construct has fluctuated, typically from three to four, consolidated into a single scale [7][8][9].Various participant response formats have been applied in exercise motivation research, predominantly using a 5point Likert scale to gauge the varying degree of agreement with the endorsement of specific statements [10][11][12].Nevertheless, more scale points may result in greater distinction between questions, leading to higher variability [13].Furthermore, previous studies focused on the validation of the BREQs to measure motivation across a spectrum of exercise settings, including public settings [9,14,15], educational institutions [16,17], healthcare settings [18,19], and sports and fitness centers [20][21][22].Various research methodologies, including cross-sectional design [9,15], longitudinal studies [23,24], randomized controlled trials [18,25], or mixed-method approaches [11,26], have been applied in different investigations.The existing versions of the BREQs demonstrated satisfactory reliability and validity across diverse participant groups, such as the elderly [27,28], students [16,17], middle-aged adults [10,12], and company workers [15].In addition, previous studies overlooked the full spectrum of constructs within the regulatory continuum of motivation.For example, the original BREQ and the BREQ-2 removed the amotivation or integrated regulation subscales.The absence of these constructs in exercise suggests their little importance when individuals decide about exercise participation.The exclusion of integrated regulation can be attributed to the difficulty in proving discriminant validity between this construct, identified regulation, and intrinsic motivation [29].Additionally, there is a lack of research examining the psychometric properties of the BREQ-3 on a 7-point scale [22].
As a part of behavioral regulations, intrinsic motivation, emerging as an important predictor, demonstrated a significant association with exercise habits.In contrast, factors related to extrinsic motivation, such as engagement in exercise-related social media activities and online social support, did not yield a significant effect [30].However, these external motivators may act to adopt better exercise behavior and to sustain these habits over an extended period.In this context, long-term exercisers are characterized by an adherence to exercise at least three times per week, with each exercise session lasting a minimum of thirty minutes and sustaining this routine exceeding six months [31][32][33].Furthermore, a positive correlation was observed between exercise behavior and autonomous types of self-determined motivation.Among these, intrinsic motivation exhibited a stronger association with long-term exercise adherence than identified regulation and integrated regulation.The identified regulation, on the other hand, demonstrated a more substantial influence on initial adoption and short-term adherence [3].Conversely, controlled motivation, including external regulation and introjected regulation, appeared to exert an adverse impact, as evidenced by higher rates of exercise discontinuation [3].For those who primarily work out due to external pressure, although this may bring about temporary changes in exercise habits, it frequently hinders the development of long-lasting exercise behaviors.Given the multifaceted nature of self-determined motivation underlying exercise maintenance, it is crucial to identify and advocate a variety of motivational styles [32].
Previous studies showed non-invariance between groups of athletes, gym users, and exercisers, which explained why participants' views of each motivation varied according to the characteristics of each environment [22,34].Sports competitors, for instance, were accustomed to training plans and aspects of competition that drove them to extreme exercise behaviors [34].Contrarily, habitual exercisers arrange their own flexible schedule of workouts with interest and enjoyment, without being influenced by rewards for achieving expected outcomes from managers or fitness professionals [22].These studies applied various measurements, not only the BREQs, that may lead to differences in terms of exercise behaviors such as frequency of exercise sessions, length of exercise for each session, and exercise commitment.These findings offered the first concrete proof of the distinctiveness of exercise domain measurements.
As far as the authors know, no study has yet compared exercisers' motivational types measured by the BREQs and other relevant characteristics that influence their long-term commitment to exercise [13,35,36].The results of the present study, therefore, advance current knowledge by filling a gap in the literature.On the assumption that exercise motivation differs among individuals, this systematic review and meta-analysis aims to evaluate whether disparities exist in self-determined exercise motivation when employing various versions of the BREQs, contingent upon factors such as the exerciser's age, sex, duration of exercise adherence, and different exercise settings.The main findings of this study are presented through associations between regulations of BREQ versions, length of exercise, and study settings.Multivariate meta-regressions of included studies are shown using selected characteristics of participants.This study is expected to contribute insights into the most appropriate application of the BREQs scale for a specific exercise context.

Study Protocol
According to the PRISMA 2020 guidelines [37], a systematic review was conducted, addressing all 27 items on the checklist.

Literature Review
The following databases were used to find research articles: PubMed, ScienceDirect, and Scopus.Boolean logic tools featuring the operators "AND" and "OR," as well as the asterisk (*) wildcard, were utilized during the search process.In PubMed, the advanced search employed the following keywords: ("exercise" AND "motiv*" AND ("BREQ" OR "Behavioral regulation") AND "self-determin*").Meanwhile, the search query consisted of "exercise" AND "motivation" AND ("BREQ" OR "Behavioral regulation") AND "self-determination" in ScienceDirect.Additional efforts were made by manually searching through Google Scholar and ResearchGate to expand the scope and encompass a broader range of eligible publications.

Study Selection
Initially, we applied the filtering functions provided by PubMed and ScienceDirect to select eligible articles.We excluded review articles, book chapters, conference abstracts, or letters, as well as those not written in English.Furthermore, only full-text articles published from 2003 to September 2023 were considered for inclusion.The search results were exported, and duplicates were removed using the Endnote reference manager program.Following the initial filtering, the title and abstract of each article were meticulously scrutinized to identify at least one aspect related to the regulations of the BREQs, which include amotivation, external regulation, introjected regulation, identified regulation, integrated regulation, and intrinsic motivation.After this preliminary screening, the full-text of the eligible articles was retrieved for comprehensive assessment and review.The systematic review included only those articles that contained the BREQs, utilized a Likert-pointing scale, and provided scores (mean and standard deviation) for each domain, encompassing amotivation, external regulation, introjection, identified regulation, integrated regulation, and intrinsic motivation.The Critical Appraisal Skills Programme (CASP) criteria for reporting quantitative data were applied to ensure the quality of the included articles and reduce the selection bias [38].This involved a detailed evaluation of eligible studies based on a set of standardized questions -12 questions for the CASP observational studies checklist and 11 questions for the CASP randomized controlled trial standard checklist-aligned with CASP guidelines.Each question elicited one of four responses: "yes," "no," "cannot tell," and "not applicable".Only studies that received a "yes" for every question were included.In cases of discrepancy, a consensus of "yes" from two out of three independent reviewers (T.T.N, D.H.P, and C.N.L) was required for the study inclusion.This approach ensured a rigorous and impartial evaluation of the selected studies.

Data Extraction
Each retrieved publication's data underwent a meticulous examination, and the following details were systematically extracted and documented within a Microsoft Excel 2019 spreadsheet: (1) author(s); (2) year of publication; (3) country; (4) study settings categorized into four groups, including educational institutions (students), healthcare settings (patients), sports and fitness centers (athletes, gym members), and public settings (general exercisers); (5) study design (observational studies/randomized control trials, mixed-methods); (6) sample size; (7) type of participants (adults/ adolescents and children); (8) participant's gender -female rate; (9) length of exercise (months); (10) versions of the BREQs, (11) type of Likert-pointing scale; ( 12) amotivation (mean score, SD); (13) external regulation (mean score, SD), ( 14) introjection (mean score, SD); (15) identified regulation (mean score, SD); ( 16) integrated regulation (mean score, SD); and (17) intrinsic motivation (mean score, SD).Two independent reviewers (T.T.N and D.H.P) were responsible for retrieving data from the included studies to minimize the potential for information bias.Subsequently, a third reviewer (C.N.L) conducted a meticulous cross-verification of the extracted data.Any disparities or inconsistencies were diligently resolved through consensus among all reviewers, ensuring the integrity and accuracy of the data before proceeding with the analysis.

Data Analysis
The synthesis and evaluation of the extracted data from the selected studies were conducted through descriptive analyses, encompassing proportions, medians, interquartile ranges (IQR), means, and standard deviation [39].Furthermore, Spearman's correlation coefficient was employed to explore the association between all domains of BREQ and long-term exercise.Non-parametric statistical methods and Kruskal-Wallis tests were applied to examine the associations between the scores of all BREQ domains and various study settings, including four groups representing healthcare settings (patients), educational institutions (students), public settings (general exercisers), and sports and fitness centers (athletes, gym members).
Given that the data collected from the retrieved studies included mean scores and standard deviations for each BREQ domain, we conducted a meta-analysis of means utilizing the generic inverse variance approach following the guidelines proposed by Harrer et al. (2021) [40].The meta-analysis parameters encompassed the number of observations (i.e., participants) (n), the mean score for each BREQ domain (mean), the standard deviation of the mean [39], and the type of summary measure for raw means that were pooled (sm).A random-effects model was utilized to estimate the withinstudy (VYi) and between-studies (τ 2 ) variance.Heterogeneity among the selected studies for meta-analysis was assessed using the inverse variance index (I 2 ), where an index exceeding 75.0% and a p-value <0.05 indicated significant heterogeneity, as described by previous studies [41,42].
Factors associated with each BREQ domain were identified utilizing multiple meta-regression models.The explanatory variables considered included age (in years), duration of exercise (in months), the percentage of females, the type of BREQ version (e.g., BREQ-2, BREQ-3, BREQ-4), study settings (e.g., educational institutions, healthcare settings, sports and fitness centers, public settings), and geographic regions (e.g., Asia, Americas, Europe, Oceania).Initially, univariable meta-regression was employed to screen potential variables associated with each BREQ domain, selecting only those with a p-value <0.2.A stepwise approach was then applied to explore potential pairs of variables and build multivariable meta-regression models, with a p-value <0.05 indicating a significant association with the BREQ domain.The outcomes of the meta-analysis were visualized using forest plots.
Two methods were applied to evaluate publication bias in the meta-analysis.Contour-enhanced funnel plots were used to visualize any asymmetrical patterns that could indicate publication bias.In addition to assessing funnel plot asymmetry, Egger's regression test was employed based on a linear regression model to quantitatively determine publication bias by testing for asymmetry in the funnel plots.If the intercept (βo) significantly deviated from zero, it indicated funnel plot asymmetry, suggesting the presence of publication bias.
All statistical analyses were conducted using R software, with the 'meta' and 'metafor' packages utilized for metaanalysis and multiple meta-regression, respectively.The 'metamean' function was employed for data pooling of means, while the 'tidyverse' package facilitated the assessment of publication bias.Figures depicting the study results were generated using the 'ggplot2' package.

Characteristics of the Selected Studies
A total of 448 articles were initially retrieved from all databases.After eliminating duplicate entries (n = 199) and excluding non-research papers (n = 5), 244 studies remained.The title and abstract of each article were carefully assessed, leading to the exclusion of articles that were not available in full-text (n=25), those written in languages other than English (n=12), and those lacking sufficient information concerning the regulatory characteristics of the BREQs (n=164).This screening process yielded 43 full-text studies, with a total of 27,421 participants, for the systematic review.However, upon reviewing these full texts, three more papers were excluded since they did not provide the mean scores and SD for measuring Behavioral BREQ regulations required to perform the meta-analysis [18,27,43].Figure 1 presents the PRISMA flow diagram of study selection.

Correlations Between the Regulations of BREQ Versions and Long-Term Exercise
Spearman's correlation coefficients were employed to investigate the relationship between the duration of exercise commitment among study participants and the six regulations characteristics assessed by the BREQs (Figure 2).A positive correlation was observed between exercise duration and integrated regulation (R = 0.27), followed by amotivation and intrinsic motivation (all R = 0.21), external (R = 0.17), introjected (R = 0.12), and identified (R = 0.08) regulations.However, none of these correlations reached statistical significance.

Associations between the Regulations of BREQ Versions and Study Settings
Among the selected studies, the domain of identified regulation had the highest median score, at 3.32 (IQR 2.92-4.22),closely followed by intrinsic motivation, with a median score of 3.20 (IQR 2.79-4.13),and then integrated regulation, introjected regulation, external regulation, and amotivation at 2.77 (IQR 2.43-3.71),2.02 (IQR 1.43-2.83),1.12 (IQR 0.57-1.93),and 0.60 (IQR 0.31-1.40),respectively.We explored the connections between the scores of the BREQ domains and various study settings, including four distinct categories representing healthcare settings, educational institutions, public settings, and sports and fitness centers.Although there were no significant associations between most BREQ domains and study settings, a significant difference was found in intrinsic motivation across different study settings (p = 0.010).The highest scores were observed in the sports and fitness groups, with a median score of 3.50 (IQR 3.31-4.93),which was higher than the median scores of other groups, including educational institutions, healthcare settings, and public settings at 2.65 (IQR 2.50-3.31),3.10 (IQR 2.89-6.07),and 2.93 (IQR 2.82-3.37),respectively.The boxplots in Figure 3 provide an overview of the mean scores across all BREQ domains and their distribution among the different study settings.

Meta-analysis of the Selected Studies
Among the 40 studies selected for meta-analysis, thirty-seven provided data on mean scores and standard deviations related to amotivation.These mean scores ranged from 0.1 to 3.3, with an unadjusted mean (crude mean) of 0.95 (SD ± 0.83).The findings from the meta-analyses revealed a pooled mean of 0.94 (95% CI: 0.67-1.12)(Figure 4-a).Regarding external regulation, all 40 selected studies reported mean scores ranging from 0.21 to 9.78, with a crude mean of 1.43 (SD ± 1.56).The pooled mean of 1.42 (95% CI: 0.93-1.92)was obtained after conducting meta-analyses (Figure 4-b).Thirty-nine studies provided mean scores for introjected regulation, ranging from 0.8 to 5.68, with a crude mean of 2.21 (SD ± 2.21).The pooled mean remained consistent at 2.21 (95% CI: 1.88-2.53)(Figure 4-c).Regarding identified regulation, 38 studies reported mean scores, varying from 2.1 to 13.3.The crude mean was calculated at 3.65 (SD ± 1.79).The meta-analysis yielded a pooled mean of 3.65 (95% CI: 3.06-4.24)(Figure 4-d).Twenty studies provided data on integrated regulation, with scores ranging from 2.16 to 5.65 and a crude mean of 3.11 (SD ± 0.93).The pooled mean closely mirrored the crude mean, measuring 3.11 (95% CI 2.68-3.55)(Figure 4-e).All forty selected studies reported data on intrinsic motivation, with results at a range of 1.68 to 21.70.The crude mean was determined to be 4.01 (SD ± 3.39), and the means of all studies' pooled intrinsic motivation converged at 4.00 (95% CI 2.92-5.09)(Figure 4-f).Moreover, findings from random effect models of meta-analysis revealed that the absolute heterogeneity of the selected studies with the inverse variance index was almost 100.0%(p < 0.001).The disparity among the articles selected for the meta-analysis was discovered during analyses for publication bias.Each scatter of contour-enhanced funnel plots illustrates for each article in our meta-analysis that scatters are unevenly distributed, distant from the pooled effect size (vertical line), and mostly plotted on the darkened regions of p < 0.05 and p < 0.01.The results of Egger's regression tests support the conclusions.The papers selected for meta-analyses of amotivation, external regulation, introjected regulation, identified regulation, integrated regulation, and intrinsic motivation demonstrated that the intercept (βo) was different from zero value (all βo < -0.41), revealing publication bias.Figure 5 displays each contour-enhanced funnel plot, highlighting the asymmetry between the selected studies and the results of Egger's regression tests.

Multivariate Meta-regressions of the Selected Studies
Six independent variables, including (1) types of the BREQs, (2) study region, (3) study setting, (4) length of exercise by month, (5) age by year, and (6) sex, in univariable models were selected as prospective inputs for multivariable metaregression models.Univariable models indicated no significant associations between the tested independent variables with domains of amotivation, external regulation, and integrated regulation.Therefore, multivariable meta-regression models could not be performed for modelling these domains.The remaining domains included introjected regulation, identified regulation, and intrinsic motivation, with several potential variables selected in multivariable models.As shown, the longer length of exercise had a slightly significant association with introjected regulation (β = 0.014, p = 0.027) and identified regulation (β = 0.014, p = 0.021).Regarding intrinsic motivation, the higher female rate of participants had a slightly reverse association with self-determined motivation in exercise (β = -0.047,p = 0.042).In contrast, the study setting of those workouts in sports and fitness centers had a higher association with intrinsic motivations (β = 2.700, p = 0.039).Table 2 shows meta-regression models to examine the variables associated with the pooled mean scores of six Behavioral regulations.This study aimed to systematically assess relevant studies and illustrate the variations in the relationships among the six behavioral regulations assessed by the BREQs in diverse contexts.A meta-analytical method was applied to compute the overall mean scores of self-determined motivations as measured by different versions of the BREQs.Furthermore, potential biases were considered carefully in our review, striving to minimize their impact.Our review revealed several key findings from analyzing the selected studies.First, autonomous forms of motivation, such as intrinsic motivation, identified regulation, and integrated regulation, exhibited the highest mean scores of self-determined motivations, while controlled motivations, encompassing external regulation and introjected regulation, received intermediate scores, and amotivation garnered the lowest score.Second, our analysis found no significant differences among the various versions of the BREQs in measuring self-determined motivation.Third, autonomous forms of motivation, including identified regulation and intrinsic motivation, exhibited significant associations with factors such as sex, longer exercise duration, and the study settings of participants, whereas controlled regulation, specifically introjected regulation, was linked to longer-term exercise commitment.
In our review, intrinsic motivation exhibited the highest mean score, followed by extrinsic motivation.This suggests that most self-determined behaviors are primarily driven by intrinsic motivation, while extrinsic motivation plays a secondary role.These findings align with previous studies applying the BREQs to assess self-determined exercise motivation [22,51,62].Individuals who score higher on intrinsic regulation and lower on amotivation tend to perceive exercise as socially acceptable and enjoyable [70].Consequently, having a solid sense of control in situations where decisions about exercising are made becomes pivotal [70].In addition, our review revealed a factor contributing to the higher scores in intrinsic regulation.Specifically, a limited number of studies reported mean scores, representing unexpected outliers in the overall mean scores [18,27,49].Notably, among these studies, the participants were either athletes in sports and fitness centers or older cancer patients in hospitals, often exhibiting higher intrinsic motivation levels due to a passion for fitness competition or the need to overcome critical health challenges, including psychological issues.
Various versions of BREQs were performed in diverse populations and study settings without disparities, given that the reviewed studies were validated and adapted to regional contexts and cultural situations regarding exercise behaviors [20,21,45,46,51].However, the BREQ-2 lacked integrated regulation in most studies, and several others had modified the scale to add this domain [20,45,60,61].Both the BREQ-3 and BREQ-4 measure six self-determined motivational styles, but only one study applied the updated BREQ-4 version [43].In general, the numbers of domains and items for each domain were not considerably different among the included studies using the BREQs.
Regarding exercise duration, our findings revealed that individuals engaged in longer exercise sessions exhibited higher mean scores in both identified and introjected regulation.These findings have practical implications, suggesting that exercisers who maintain longer and autonomous exercise commitments, perceiving exercise as personally significant, are more likely to sustain regular exercise over an extended period.Such individuals aim to attain desired health benefits and enhance their physical appearance, which are primary motivations for long-term exercise adherence [71].Specific motivations such as "weight loss," "health," "aesthetics," and "hypertrophy" are also significantly linked to participants' commitment to ongoing exercise at fitness centers [72].This finding was in line with prior research in sports and fitness settings, where the most influential motivation for exercise participation is the perceived benefits in terms of health, physical function, and appearance.Factors such as experiencing physical changes and feeling in control further enhance commitment [73].Our review also found a significant association between introjected regulation and extended exercise duration.However, previous studies have shown that individuals with less controlled exercise motivations, such as introjected regulation oriented toward sport or exercise practice, are more susceptible to regressive tendencies, increasing the likelihood of exercise termination [3].Additionally, while this review did not find a significant association between intrinsic motivation scores and exercise duration, other studies have shown that regular, long-term exercisers experience higher levels of enjoyment and challenge, particularly during the first year of their membership in sports and fitness settings [74].
Concerning age and sex, the findings of this review did not reveal any significant age-related differences in selfdetermined motivation in most domains, except for the inverse association between female gender and intrinsic motivation.Another study implied that females engaged in exercise not primarily out of interest or enjoyment but tended to place more emphasis on appearance-related concerns, which was particularly prominent among young and middleaged individuals, with women showing more concern than men [73].Regarding study settings in our analyses, the results of this study align with previous empirical research, indicating that higher levels of intrinsic motivation are typically associated with individuals who exercise at sports and fitness centers [21].These individuals also tend to experience greater enjoyment, hold more favorable perceptions of their exercise behavior [75], exhibit higher persistence rates [76], and report greater life satisfaction [77].
Our study acknowledges several limitations.First, the vast majority of BREQ studies employed the two versions (BREQ-2 and BREQ-3), while research using the BREQ-4 version was methodologically necessary.Second, there were too few studies with BREQ estimations in samples representing children and adolescents, necessitating further investigations of BREQs in these age groups to enable the exploration of age as a potential moderator of the impact of exercise maintenance.Third, in terms of representativeness, although the included studies were from countries across 4 continents, most were from Europe and the Americas, highlighting the need for further studies in Asia and Africa to ensure better generalizability of BREQs.Finally, our analysis encompassed global research, spanning various geographical regions and study settings.Motivational styles can substantially differ due to exercise culture and behavioral patterns, which contribute to the notable heterogeneity observed in our meta-analysis results.This heterogeneity may be attributed to publication bias from including studies with limited sample sizes.Consequently, we recommend conducting additional studies to unearth novel motivational styles specific to particular populations, such as those in sports and fitness centers, where individuals demonstrate significantly higher levels of autonomous motivation.

Conclusion
Our investigation indicated the diverse measurements provided by BREQ modifications in assessing self-determined motivation among individuals engaged in exercise across various study settings.Our review highlighted the importance of employing context-specific measures to consistently evaluate how individuals regulate their behavior within a particular setting.Although both intrinsic and extrinsic motivation exhibited significant correlations with long-term exercise outcomes, intrinsic motivation emerged as the more crucial factor in self-determined motivation.We observed that the sports and fitness study setting displayed notably higher levels of intrinsic motivation than other groups.However, the current BREQs have limitations in fully and accurately capturing this form of motivation.Thus, the development of new instruments is warranted to comprehensively assess the motivation of individuals in fitness centers, allowing for a precise understanding of the reasons for maintaining their exercise routines.

Data Availability Statement
The data presented in the study are available in the article.

Funding
This research work was financially supported by Walailak University Graduate Research Fund (No. CGS-RF-2023/08).

Acknowledgements
The author acknowledges to the support of Walailak University PhD scholarships for High Potential Candidates to Enroll in Doctoral Programs (Contract No. HP015/2021), Walailak University Graduate Research Fund (No. CGS-RF-2023/08), and Can Tho University of Medicine and Pharmacy.

Institutional Review Board Statement
Not applicable.

Informed Consent Statement
Not applicable.

Declaration of Competing Interest
The authors declare that there is no conflict of interests regarding the publication of this manuscript.In addition, the ethical issues, including plagiarism, informed consent, misconduct, data fabrication and/or falsification, double publication and/or submission, and redundancies have been completely observed by the authors.

Figure 2 .
Figure 2. The correlation of regulations of BREQ domains and long-term exercise

Figure 3 .
Figure 3.The scores of regulations measured by BREQs in comparison among study settings

Figure 4 .
Figure 4. a &b) Forest plot for scores attributing to amotivation (in dark green) and external regulation (in purple), c & d) Forest plot for scores attributing to introjected regulation (in dark blue) and identified regulation (in orange), e & f) Forest plot for scores attributing to integrated regulation (in gray) and intrinsic motivation (in cyan).

Figure 5 .
Figure 5.The contour-enhanced funnel plots highlight the asymmetries between particular researches.The dotted vertical lines display the average effect sizes, each scatter displays the selected papers, and the colored areas highlight the statistical significance.