The hidden cost of controlling teaching in Physical Education: A systematic review and meta-analysis of students’ Physical activity intention and levels

Abstract

Despite the benefits of physical activity (PA), most young people fail to meet the World Health Organization’s recommendations. Grounded in Self-Determination Theory (SDT), Physical Education (PE) teachers play a pivotal role in promoting PA through their motivational teaching style. While need-supportive behaviors show clear benefits, the effects of controlling teaching behaviors on PA-related outcomes remain unclear. This systematic review and meta-analysis examined the associations between students’ perceptions of controlling teaching in PE and their PA intention and PA levels, and explored whether these associations differ by age or sex. The review was registered in PROSPERO (CRD42022362459) and conducted following PRISMA 2020 guidelines. Searches were performed in PubMed, Scopus, Web of Science, and APA PsycNet up to 2025. Eligible studies included participants aged 5–18 years, assessed controlling teaching behaviors in PE based on SDT, and reported PA intention and/or PA levels. Random-effects meta-analyses were conducted, along with subgroup analyses by sex and a meta-regression on age. Study quality was assessed using the NIH Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies. Of 3,819 records, 16 studies (n = 9,925; ages 11–19 years) were included. Controlling teaching style was negatively associated with PA intention (r = –0.14; 95%CI: –0.22, –0.06) and PA levels (r = –0.18; 95%CI: –0.31, –0.06), with high between-study heterogeneity (PA intention: I² = 90.05%; PA levels: I² = 95.79%), based exclusively on cross-sectional studies. Subgroup analyses showed similar trends for boys and girls. Neither sex nor age moderated these associations. Controlling behaviors in PE were associated with students’ PA intention and PA levels, highlighting the importance of minimizing such practices in PE contexts.

Resumen

A pesar de los beneficios de la actividad física (AF), la mayoría de los jóvenes no cumplen con las recomendaciones de la Organización Mundial de la Salud. La Teoría de la Autodeterminación (TAD) sostiene que el profesorado de Educación Física (EF) juega un papel clave en la promoción de la AF a través de sus conductas motivacionales docentes. Mientras que las conductas de apoyo a las necesidades psicológicas básicas muestran beneficios claros en consecuencias relacionadas con la AF, los efectos de las conductas controladoras siguen siendo inciertos. Por ello, esta revisión sistemática y metaanálisis ha analizado las asociaciones entre las percepciones de los estudiantes sobre las conductas controladoras en EF y su intención y niveles de AF, así como posibles diferencias según la edad y el sexo. La revisión ha sido registrada en PROSPERO (CRD42022362459) y se ha realizado siguiendo las directrices PRISMA 2020. Se han realizado búsquedas en PubMed, Scopus, Web of Science y APA PsycNet hasta 2025. Los estudios elegibles han incluido participantes de entre 5 y 18 años, han evaluado conductas controladoras en EF basadas en la TAD y han medido la intención y/o los niveles de AF. Se han realizado metaanálisis de efectos aleatorios, junto con análisis de subgrupos por sexo y una metarregresión sobre la edad. La calidad de los estudios se ha evaluado utilizando la herramienta de Evaluación de Calidad para Estudios de Cohorte Observacionales y Transversales del NIH. De 3.819 registros, se han incluido 16 estudios (n = 9.925; edades de 11 a 19 años). El estilo controlador se ha asociado negativamente con la intención de AF (r = –0.14; IC 95%: –0.22, –0.06) y los niveles de AF (r = –0.18; IC 95%: –0.31, –0.06), con una heterogeneidad alta entre los estudios (intención de AF: I² = 90,05 %; niveles de AF: I² = 95,79 %), en base a estudios transversales. Los análisis por subgrupos han mostrado tendencias similares en chicos y chicas, sin que el sexo ni la edad hayan moderado estas asociaciones. Los comportamientos controladores en EF se han asociado con la intención y los niveles de AF de los estudiantes, destacando la importancia de minimizar estas prácticas en EF.

Introduction

A growing body of literature highlights the importance of physical activity (PA) for the physical, psychological, and cognitive health of young people (World Health Organization, 2020). To achieve these benefits, the World Health Organization (2020) recommends that young people aged 5 to 17 engage in at least 60 minutes of moderate-to-vigorous physical activity (MVPA) per day. However, Guthold et al. (2020) reported that 81% of adolescents worldwide do not meet these PA guidelines, with lower compliance observed in girls than in boys and a progressive decline in PA levels as age increases.

In this context of widespread physical inactivity, Physical Education (PE) is recognized as a key setting to promote PA, both during lessons and beyond, by providing students with the knowledge, skills, and values necessary for lifelong participation (van Sluijs et al., 2021). According to Self-Determination Theory (SDT; Ryan & Deci, 2017, 2020), PE teachers’ motivational teaching style is associated with students’ PA behaviors. While need-supportive teaching behaviors are consistently associated with adaptive outcomes, the impact of controlling teaching behaviors remains less clear, with previous research yielding mixed and inconclusive findings (Vasconcellos et al., 2020). Therefore, the present systematic review and meta-analysis aims to examine the relationship between students’ perceptions of PE teachers’ controlling behaviors and two PA-related outcomes (i.e., intention to be physically active and PA levels), as well as to investigate whether these associations are moderated by age and sex, given that both factors have been consistently linked to differences in PA engagement during adolescence (van Sluijs et al., 2021).

Physical Education teachers’ (de)motivational style and physical activity-related outcomes

SDT (Ryan & Deci, 2017, 2020) is a leading framework for understanding human motivation, emphasizing the role of the social context in supporting or thwarting three basic psychological needs (i.e., autonomy, competence, and relatedness). In education, the teacher’s interpersonal style is a key factor associated with students’ motivation and engagement (Ryan & Deci, 2017, 2020). In this field, SDT distinguishes between two opposing interpersonal styles: need-supportive and need-thwarting. Need-supportive behaviors in PE are teaching practices that nurture students’ basic psychological needs. For example, they may include providing meaningful choices to support autonomy, offering clear guidance and constructive feedback to build competence, and showing care and empathy to strengthen relatedness. Extensive research in PE has consistently linked need-supportive teaching to adaptive outcomes, including PA behavior, through the fulfillment of students’ basic psychological needs and motivation (Tian & Shen, 2023; Vasconcellos et al., 2020).

By contrast, need-thwarting behaviors have been identified as correlates of basic psychological need frustration and maladaptive outcomes in PE (Burgueño et al., 2024; Haerens et al., 2015). Among these, controlling behaviors are the most widely studied form in PE (Vasconcellos et al., 2020). They refer to teachers’ deliberate attempts to pressure students to think, feel, or act in specific ways, often prioritizing compliance and control over personal initiative. Such behaviors may involve direct strategies (e.g., threats, conditional regard, excessive criticism) as well as more subtle forms of control (e.g., inducing guilt or shame, limiting opportunities for choice, disregarding students’ perspectives) (Bartholomew et al., 2018; Reeve, 2009). Consistent with SDT (Ryan & Deci, 2017; Vansteenkiste et al., 2020), such behaviors have been directly related to the frustration of students’ basic psychological needs, which in turn has been associated with controlled motivation, amotivation, and other maladaptive outcomes (Burgueño et al., 2023; Haerens et al., 2015).

While the systematic review and meta-analysis by Vasconcellos et al. (2020) identified negative correlations between controlling behaviors and several maladaptive outcomes, it did not examine their specific association with PA intention or PA levels. Evidence on these PA-related outcomes remains inconsistent. Some studies report negative associations between controlling teaching behaviors and students’ intention to be physically active (Koka et al., 2020; Sevil-Serrano et al., 2022) as well as PA levels (Koka et al., 2020; Trigueros et al., 2020), whereas others have found no significant relationships with either intention (Moreno-Murcia et al., 2018; Viksi & Tilga, 2022) or PA levels (Sevil-Serrano et al., 2022). Notably, one study has even reported a positive association (Diloy-Peña et al., 2024). The findings, which are inconsistent with one another, highlight the necessity for a systematic review and meta-analysis with a specific focus on the associations between controlling teaching behaviors in PE and PA-related outcomes, namely PA intention and PA levels. Unlike previous reviews, the present study provides a quantitative synthesis specifically focused on these behavioral outcomes, thereby offering a more precise account of the association between perceived controlling teaching behaviors and adolescents’ PA. Moreover, given that age and sex are known to shape PA patterns during adolescence (van Sluijs et al., 2021), it is important to consider whether these sociodemographic factors may moderate the associations under examination.

The moderating role of sex and age in the impact of controlling behaviors

Evidence suggests that the impact of controlling behaviors in PE may vary across sociodemographic variables such as sex and age. Regarding sex, Abós et al. (2022) analyzed the effects of internal and external controlling behaviors on motivational outcomes, including basic psychological need frustration, controlled motivation, amotivation, and oppositional defiance. Their results showed that internal controlling behaviors (e.g., inducing guilt or shame) were negatively associated with all these outcomes in both sex, but with greater intensity among girls. In contrast, external controlling behaviors (e.g., threats or imposed deadlines) predicted need frustration in boys and oppositional defiance in girls, suggesting a differential pattern by sex. Complementarily, Koka and Sildala (2018) found that girls were more strongly affected by teachers’ perceived conditional regard and intimidating behaviors in relation to amotivation, whereas boys were more strongly influenced by teachers’ controlling use of praise. Taken together, these findings suggest that controlling teaching behaviors may differentially shape the motivational processes of boys and girls, reinforcing the need to further investigate these differences.

With regard to age or grade level, research has mainly examined differences in students’ perceptions of teachers’ (de)motivating styles rather than whether age moderates the consequences of controlling behaviors. Several studies suggest that perceptions of controlling teaching behaviors decrease as students grow older (Burgueño and Medina-Casaubón, 2021; Diloy-Peña et al., 2024; Jang et al., 2020). Nevertheless, other studies have found no significant associations between age and perceptions of controlling teaching behaviors (De Meyer et al., 2016; Tilga et al., 2020). Overall, these findings suggest that although age may be linked to differences in how students perceive controlling behaviors in PE, the evidence remains scarce and inconsistent, and no studies to date have examined whether age moderates the associations between such behaviors and PA-related outcomes.

Examining whether sex and age moderate the associations between controlling teaching behaviors and PA-related outcomes is important to better understand potential differences across student groups. Such knowledge is essential to clarify whether controlling behaviors affect boys and girls, or younger and older students, in distinct ways. From a practical perspective, this line of research may help PE teachers reduce the use of controlling behaviors and foster more equitable and motivating environments that support students’ long-term engagement in PA.

The present study

To date, the evidence regarding the relationship between controlling teaching behaviours in PE and PA-related behaviours (intention to engage in PA and PA levels) is limited and inconsistent (Koka et al., 2020; Moreno-Murcia et al., 2018; Sevil-Serrano et al., 2022; Trigueros et al., 2020; Viksi & Tilga, 2022). Similarly, although some studies have suggested that age may influence controlling behaviours (Burgueño & Medina-Casaubón, 2021; Diloy-Peña et al., 2024; Jang et al., 2020), the available evidence remains scarce and inconclusive. Together, these limitations highlight the need for a systematic examination of these associations. In line with the above, this systematic review and meta-analysis aimed to (a) analyse the association between PE teachers’ controlling teaching style and students’ intention to engage in PA and their levels of PA, and (b) investigate whether age and gender moderate this relationship.

Drawing on previous studies grounded in SDT (Koka et al., 2020; Sevil-Serrano et al., 2022; Trigueros et al., 2020), and in line with the first objective, it was hypothesized that perceived controlling teaching behaviours would be negatively associated with pupils’ intention to engage in PA and their overall levels of PA. Regarding the second objective, although previous research has suggested possible differences based on gender (Abós et al., 2022; Koka & Sildala, 2018) and age (Burgueño and Medina-Casaubón, 2021; Diloy-Peña et al., 2024; Jang et al., 2020), the available evidence is limited and inconsistent.

Method

This systematic review and meta-analysis was registered in PROSPERO (CRD42022362459) and conducted in accordance with PRISMA 2020 (Page et al., 2021). No deviations from the registered protocol were made.

Eligibility criteria

Inclusion criteria were defined a priori using the PICOS framework (Participants, Intervention/Exposure, Comparison, Outcomes, Study design). Participants: healthy children and adolescents aged 5–18 years enrolled in school-based PE. Intervention/Exposure: students’ perceptions of PE teachers’ controlling behaviors grounded in SDT, assessed with validated SDT-consistent instruments. Comparison: not applicable. Outcomes: intention to be physically active and regular PA levels assessed by self-report and/or device-based measures. Study design: quantitative studies (cross-sectional, longitudinal, or intervention) reporting either correlation coefficients (r) or statistics convertible to r at the student level. Only peer-reviewed journal articles published in English or Spanish were included.

Exclusion criteria were studies with samples outside the 5–18-year range or involving diagnosed disease/disorder; studies conducted exclusively outside the school PE setting (e.g., university courses, extracurricular sport); outcomes not aligned with the review scope; designs not eligible for quantitative synthesis (systematic reviews, meta-analyses, qualitative studies); literature grey (books, book chapters, editorials, letters, expert opinions, conference abstracts, protocols); and reports lacking sufficient data to compute or convert effect sizes.

Information sources

Two reviewers (J.L.P.-L. and J.S.-S.) searched four bibliographic databases from inception to 1 April 2025: PubMed, Scopus, Web of Science Core Collection, and APA PsycNet. To ensure currency, we continued to update the search after that date using the same strategy; all databases were last searched on 1 September 2025. We also screened the reference lists of included studies and relevant reviews and contacted corresponding authors when full texts or key data were unavailable. No trial registers, preprint servers, or institutional repositories were searched, and no automation tools were used for study selection.

Search strategy

Search strategies combined controlled vocabulary (when available) and free-text terms for four domains: (1) controlling/need-thwarting teaching (e.g., “control* style/behav/teach*”, “autonomy thwart*”, “internal/external control*”, “domineering”, “need thwart*”), (2) Physical Education (“physical education”), (3) physical activity and intention (“physical activ*”, exercis*, sport*, movement*, intention*), and (4) youth (student*, child*, adolescent*, teen*, youth). Syntax and field tags were adapted for each database; truncation and Boolean operators were applied consistently. No date, design, or population filters were applied at the database level. Full, reproducible strategies for each database are provided in Supplementary Table 1. Language limits (English/Spanish) were not applied in the databases and were enforced during eligibility screening. The top-up searches conducted after 1 April 2025 used the same strategies and sources and are documented with timestamps in Supplementary Table 1.

Selection process

Two reviewers (J.L.P.-L. and J.S.-S.) independently screened records in two stages against the prespecified eligibility criteria. First, titles and abstracts were screened in duplicate; second, potentially eligible reports were assessed in duplicate at full text. Before screening, duplicate records were identified and removed in a shared Excel spreadsheet (Microsoft Excel, Microsoft, Redmond, WA, USA). Reviewers worked independently at both stages and were not masked to authors, journals, or institutions. Disagreements were resolved by discussion; unresolved cases were adjudicated by a third reviewer (L.G-G.). Reference lists of included studies and relevant reviews were hand-searched to identify additional reports. Corresponding authors were contacted when full texts or essential data were unavailable. Reasons for full-text exclusions were documented and are presented in Supplementary Table 2, alongside the PRISMA 2021 flow diagram. All exclusion decisions were based on the predefined eligibility criteria, ensuring consistency and transparency in the selection process. No automation tools (e.g., machine learning, text mining) were used at any stage.

Data collection process

Two reviewers (J.L.P.-L., and M.A.T.-S.) independently collected data from each included report using a piloted extraction form; one reviewer extracted all items and the other verified accuracy. Discrepancies were resolved by discussion and, when needed, by a third reviewer (J.S.-S.). When full texts were not accessible, corresponding authors were contacted to obtain the report. When a study reported multiple correlations for separate controlling teaching practices (e.g., intimidating, conditional regard, guilt induction), we extracted the overall controlling teaching style score provided in the article; if a global score was not reported, we requested it from the authors. When both self-reported and device-based PA levels were available, we extracted the device-based estimate by prespecified priority. No automation tools were used in the data collection process. When studies reported sex-specific correlations based on non-overlapping subsamples (boys and girls), these estimates were extracted separately. As these estimates were derived from independent subsamples, no statistical correction for dependence was required. Finally, when multiple studies conducted by the same research group were identified (e.g., Trigueros et al., 2020), these studies were treated as independent effect sizes because they were based on different samples and datasets.

Data items

For each study we extracted: authors and year, study design, sample size, mean, standard deviation, and range, instrument used to assess intention to be physically active and/or PA levels -including whether PA was measured using self-reported or device-based-, the instrument used to assess perceived controlling teaching behaviors, and the bivariate association (Pearson’s r) between controlling behaviors and each outcome. When r is not directly reported but convertible statistics are available (e.g., regression coefficients), they were converted to r. The specific items extracted for each study are summarized in Tables 1 and 2.

Table 1.

Characteristics and results of studies on controlling teaching behaviors and students’ physical activity intention

Study	Study design	Sample				Assessment of controlling behaviors	Assessment of PA intention	Correlation coefficient
		n	Boys	Girls	Mean ± SD (range)
Moreno-Murcia et al. (2018)	CS	416	–	–	16.71 ± 0.73 (16–18)	CCBS (adapted for PE teachers)	IPA Scale	r = 0.04; p > .05
Trigueros, Aguilar-Parra, Cangas, López-Liria et al. (2019)^a	CS	604	321	283	15.73 ± 1.30 (13-19)	CCBS (adapted for PE teachers)	IPA Scale	r = −0.18; p < .05
Trigueros, Aguilar-Parra, Cangas, Fernández-Batanero et al. (2019)^a	CS	653	311	342	15.56 ± 1.23 (13–19)	SPSEC-T Scale	TPB Questionnaire	r = −0.19; p < .05
Trigueros, Cangas, Aguilar-Parra, Álvarez et al. (2019)^a	CS	606	323	283	15.70 ± 1.43 (13–19)	PCTs modified	IPA Scale	r = −0.54; p < .05
Trigueros, Navarro-Gómez, Aguilar-Parra et al. (2019)^a	CS	602	320	282	15.23 ± 1.13 (13–19)	PCTs modified	IPA Scale	r = −0.36; p < .05
Koka et al. (2020)	CS	234	104	130	15.29 ± 2.06 (11–19)	CCBS (adapted for PE teachers)	TPB Questionnaire	r = −0.10; p > .05
Leo et al. (2022)	CS	1120	559	561	11.70 ± 1.63 (10–17)	TISQ-PE	One 5-point Likert item	r = −0.04; p > .05
Leo et al. (2023)	CS	654	317	337	11.96 ± 1.95 (10–16)	TISQ-PE	One 5-point Likert item	r = −0.04; p > .05
Sevil-Serrano et al. (2022)	CS	210	99	111	13.06 ± 0.61 (12–14)	CTS-PE	TPB Questionnaire	r = −0.21; p < .01
Viksi and Tilga (2022)	CS	299	164	135	16.10 ± 1.86 (13–18)	CCBS (adapted for PE teachers)	TPB Questionnaire	r = −0.04; p > .05
Diloy-Peña et al. (2024)	CS	669	322	347	16.65 ± 1.47 (12–17)	SIS-PE	TPB Questionnaire	r = 0.12; p < .001
Stefanos et al. (2025)	CS	250	116	134	11.66 ± 0.6 (11–13)	CCBS (adapted for PE teachers)	TPB Questionnaire	r = 0.10; p < .001

Notes. n = sample size; SD = Standard Deviation; CCBS = Controlling Coach Behaviors Scale (adapted for PE teachers); SPSEC-T = Scale of Perceived Support in Exercise Contexts on the part of the Teacher; PCTs = Psychologically Controlling Teaching Scale; TISQ-PE = Teaching Interpersonal Style Questionnaire in Physical Education; CTS-PE = Controlling Teaching Scale for Physical Education; SIS-PE = Situations-in-School in Physical Education; IPA = Intention Physical Activity; TPB = Theory of Planned Behavior; PE = Physical Education; PA = Physical Activity.

a

The studies by Trigueros et al. (2019) were treated as independent effect sizes because they were derived from different samples and datasets.

Table 2.

Characteristics and results of studies on controlling teaching behaviors and students’ physical activity levels

Study	Study design	Sample				Assessment of controlling behaviors	Assessment of PA	Main results
		n	Boys	Girls	Mean ± SD (range)			Variables	Correlation coefficient
Moreno-Murcia et al. (2018)	CS	416	–	–	16.71 ± 0.73 (16–18)	CCBS (adapted for PE teachers)	Habitual PA questionnaire	Controlling style and PA	r = 0.06; p > .05
Koka et al. (2019)	CS	159	64	95	14.94 ± 2.11 (11–19)	CCBS (adapted for PE teachers)	Accelerometry	Controlling style and MVPA	r = −0.09; p > .05
Trigueros, Mínguez, González-Bernal, Jahouh et al. (2019)^a	CS	1127	653	474	15.27 ± 1.35 (13–18)	PCTs	WHO HBSC Survey	Controlling style and PA	r = −0.36; p < .001
Koka et al. (2020)	CS	234	104	130	15.29 ± 2.06 (11–19)	CCBS (adapted for PE teachers)	IPAQ-SF	Controlling style and MVPA	r = −0.05; p > .05
Tilga et al. (2020)	CS	381	157	224	13.64 ± 1.19 (12–15)	CCBS (adapted for PE teachers)	LTEQ	Controlling style and LTPA	r = 0.04; p > .05
Trigueros et al. (2020)^a	CS	1941	1031	910	15.34 ± 1.22 (13–18)	PCTs	WHO HBSC Survey	Controlling style and PA levels	r = −0.52; p < .001
Sevil-Serrano et al. (2022)	CS	163	99	111	13.06 ± 0.61 (12–14)	CTS-PE	Accelerometry	Controlling style and PA levels	r = −0.06; p > .05
Viksi and Tilga (2022)	CS	299	164	135	16.10 ± 1.86 (13–18)	CCBS (adapted for PE teachers)	IPAQ-SF	Controlling style and PA levels	r = 0.09; p > .05

Notes. WHO: World Health Organization; n = sample size; SD = Standard Deviation; PA = Physical Activity; PE = Physical Education; CCBS = Controlling Coach Behaviors Scale (adapted for PE teachers); PCTs = Psychologically Controlling Teaching Scale; CTS-PE = Controlling Teaching Scale for Physical Education; MVPA = Moderate to Vigorous Physical Activity; IPAQ-SF = International Physical Activity Questionnaire–Short Form; LTEQ = Leisure-Time Exercise Questionnaire; LTPA = Leisure-Time Physical Activity.

a

The studies by Trigueros et al. were treated as independent effect sizes because they were derived from different samples and datasets.

Quality assessment

Two reviewers (J.L.P.-L. y M.A.T.-S.) independently assessed quality assessment for all included studies using the National Heart Lung and Blood Institute (2013) Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies. Disagreements were resolved by discussion, and when consensus could not be reached, a third reviewer (J.S.-S.) made the final decision. The NHLBI tool contains 14 items for cohort/longitudinal designs; for cross-sectional studies, Items 7, 10, and 13 are not applicable, yielding 11 applicable items. Each item was rated Yes, No, Not reported (NR), or Not applicable (NA); for summary scoring, Yes = 1 and No/NR = 0, with NA items excluded from the denominator. Maximum attainable scores were 14 (longitudinal) and 11 (cross-sectional). Following predefined thresholds (Álvarez-Bueno et al., 2020), studies were classified as Good (>75% Yes), Fair (50–75% Yes), or Poor (<50% Yes), calculated against the design-specific maximum.

Because the present review focuses on associations between variables rather than on the effectiveness of interventions, intervention studies were evaluated using the same NHLBI criteria as observational studies. In other words, their methodological quality was judged on the basis of the observational components relevant to exposure–outcome associations, rather than on the quality of the intervention itself. No automation tools were used at any stage of quality assessment.

Effects measures

For both outcomes (i.e., intention to be physically active and PA levels), the primary effect measure was Pearson’s correlation coefficient (r) with 95% confidence intervals. Correlations were interpreted using conventional benchmarks (small = 0.10, moderate = 0.30, strong = 0.50) to aid comparability across studies (Cohen, 2013).

Synthesis method

Two independent meta-analyses were conducted: one for PA intention and one for PA levels. Before synthesis, data were prepared to maximize comparability across instruments and avoid multiplicity. When Pearson’s r was unavailable, convertible statistics (e.g., standardized or unstandardized regression coefficients) were transformed into r using established formulae (Bring, 1994; Peterson & Brown, 2005). Results of individual studies were tabulated and visually displayed with forest plots for each outcome. Effect sizes were synthesized using random-effects meta-analysis (DerSimonian–Laird), given the expected methodological heterogeneity across settings, instruments, and samples (Cooper et al., 2019; Hedges & Vevea, 1998). All analyses were performed in Stata v17 (StataCorp, College Station, TX, USA). Between-study heterogeneity was quantified with Cochran’s Q and I² statistics; I² values were interpreted as low (<50%), moderate (50%–75%), or high (>75%), acknowledging these thresholds as heuristic (Higgins, 2003; Higgins & Thompson, 2002). Prespecified moderators included subgroup analyses by sex (boys, girls, mixed-sex samples) and a study-level meta-regression on mean age for each outcome. Subgroup analyses comparing self-reported and device-based PA measures were planned; however, they were not conducted because the number of studies using objective measures was insufficient (fewer than 10), in line with methodological recommendations (Higgins et al., 2021). Finally, leave-one-out analyses were conducted to assess the robustness of pooled estimates and the influence of individual studies on summary effects and heterogeneity indices (Cooper et al., 2019).

Reporting bias assessment

Risk of bias due to missing results was examined using funnel plots and Egger’s regression test for the overall analyses of each outcome. Following current guidance (Higgins et al., 2021), a two-tailed p < .10 was considered indicative of asymmetry. Egger’s tests were not applied to subgroups with a low number of studies.

Results

Study selection

As shown in Figure 1, the database search identified 3,819 records, of which 17 duplicates were removed, leaving 3,802 records for screening. After title and abstract screening, 3,778 records were excluded, and 24 full-text articles were assessed for eligibility. Of these, eight articles were excluded (three did not include the target population, two did not examine the relationship between the variables of interest, two assessed only PA during PE classes; and one presented overlapping data; detailed reasons for exclusion are provided in Supplementary Table 2). Ultimately, 16 studies met the inclusion criteria and were included in both the systematic review and the meta-analysis.

Figure 1.

Preferred reporting items for systematic reviews and meta-analyses (PRISMA) flow diagram.

Study characteristics

Tables 1 and 2 summarize the characteristics of the included studies for PA intention and PA levels, respectively. A total of 9,925 participants were included, of whom 4,861 were boys (51%) and 4,648 were girls (47%). The remaining 416 participants (2%) did not report their sex. Although the predefined eligibility criteria allowed for samples aged 5–18 years, all studies that were ultimately included involved adolescents aged 11–19 year. All studies (n = 16) employed a cross-sectional design.

Regarding controlling behaviors, all studies used self-report questionnaires, of which seven studies (Koka et al., 2019, 2020; Moreno-Murcia et al., 2018; Stefanos et al., 2025; Tilga et al., 2020; Trigueros, Aguilar-Parra, Cangas, López-Liria et al., 2019; Viksi & Tilga, 2022) used an adaptation for PE teachers of the Controlling Coach Behaviors Scale, four studies (Trigueros, Cangas, Aguilar-Parra, Álvarez et al., 2019; Trigueros et al., 2020; Trigueros, Mínguez, González-Bernal, Jahouh et al., 2019; Trigueros, Navarro-Gómez, Aguilar-Parra et al., 2019) used the Psychologically Controlling Teaching Scale, two studies used the Teaching Interpersonal Style Questionnaire in PE (Leo et al., 2022, 2023) one study (Trigueros, Aguilar-Parra, Cangas, Fernández-Batanero et al., 2019) used the Scale of Perceived Support in Exercise Contexts on the part of the Teacher, one study used the Controlling Teaching Scale for PE (Sevil-Serrano et al., 2022), and one study used the Situations-in-School in PE (Diloy-Peña et al., 2024).

Based on PA-related outcomes, 12 studies measured PA intention (Diloy-Peña et al., 2024; Koka et al., 2020; Leo et al., 2022, 2023; Moreno-Murcia et al., 2018; Sevil-Serrano et al., 2022; Stefanos et al., 2025; Trigueros, Aguilar-Parra, Cangas, Fernández-Batanero et al., 2019; Trigueros, Aguilar-Parra, Cangas, López-Liria et al., 2019; Trigueros, Cangas, Aguilar-Parra, Álvarez et al., 2019; Trigueros, Mínguez, González-Bernal, Jahouh et al., 2019; Viksi & Tilga, 2022) and eight studies measured PA levels (Koka et al., 2019, 2020; Moreno-Murcia et al., 2018; Sevil-Serrano et al., 2022; Tilga et al., 2020; Trigueros, Aguilar-Parra, Cangas, López-Liria et al., 2019; Trigueros, Mínguez, González-Bernal, Jahouh et al., 2019; Viksi & Tilga, 2022). In all studies, self-reported measures were used to assess PA intention and PA levels, except for two studies that used accelerometers to evaluate PA levels (Koka et al., 2019; Sevil-Serrano et al., 2022).

Methodological quality assessment

The results of the methodological quality assessment are presented in Table 3. In this sample, 100% (n = 16) were classified as having medium quality. All studies clearly stated the research objective (item 1), defined the study population (item 2), and specified the inclusion criteria (item 4). Additionally, 53.3% reported an acceptable minimum participation rate (item 3). None of the studies provided a sample size justification (item 5) or measured exposures prior to outcomes (item 6). As shown in the table, the scores for items 7, 10, and 13 were 0%, consistent with their non-applicability to cross-sectional designs. Regarding independent variables, all studies examined different levels of exposure and clearly defined the variables (items 8 and 9), while 75% clearly defined outcome measures (item 11). None of the studies reported blinding of outcome assessors (item 12). Finally, 25% included covariates (item 14) in their analyses, such as sex and age.

Table 3.

Quality assessment of cross-sectional studies

Author(s)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	Total score	Quality
Moreno-Murcia et al. (2018)	✓	✓		✓			NA	✓	✓	NA	✓	NR	NA		6	Fair
Koka et al. (2019)	✓	✓	✓	✓			NA	✓	✓	NA	✓	NR	NA		7	Fair
Trigueros, Aguilar-Parra, Cangas, López-Liria et al. (2019)	✓	✓		✓			NA	✓	✓	NA	✓	NR	NA		6	Fair
Trigueros, Aguilar-Parra, Cangas, Fernández-Batanero et al. (2019)	✓	✓	✓	✓			NA	✓	✓	NA	✓	NR	NA		7	Fair
Trigueros, Cangas, Aguilar-Parra, Álvarez et al. (2019)	✓	✓	✓	✓			NA	✓	✓	NA	✓	NR	NA		7	Fair
Trigueros, Mínguez, González-Bernal, Jahouh et al. (2019)	✓	✓		✓			NA	✓	✓	NA	✓	NR	NA		6	Fair
Trigueros, Navarro-Gómez, Aguilar-Parra et al. (2019)	✓	✓		✓			NA	✓	✓	NA		NR	NA		5	Fair
Koka et al. (2020)	✓	✓	✓	✓			NA	✓	✓	NA	✓	NR	NA		7	Fair
Tilga et al. (2020)	✓	✓		✓			NA	✓	✓	NA	✓	NR	NA		6	Fair
Trigueros et al. (2020)	✓	✓		✓			NA	✓	✓	NA		NR	NA		5	Fair
Leo et al. (2022)	✓	✓	✓	✓			NA	✓	✓	NA		NR	NA	✓	7	Fair
Leo et al. (2023)	✓	✓	✓	✓			NA	✓	✓	NA		NR	NA	✓	7	Fair
Sevil-Serrano et al. (2022)	✓	✓	✓	✓			NA	✓	✓	NA	✓	NR	NA	✓	8	Fair
Viksi and Tilga (2022)	✓	✓		✓			NA	✓	✓	NA	✓	NR	NA		6	Fair
Diloy-Peña et al. (2024)	✓	✓		✓			NA	✓	✓	NA	✓	NR	NA	✓	7	Fair
Stefanos et al. (2025)	✓	✓	✓	✓			NA	✓	✓	NA	✓	NR	NA		7	Fair
Average (%)	100.0	100.0	53.3	100.0	0.0	0.0	NA	100.0	100.0	0.0	75.0	0.0	0.0	25.0

Note. ✓ = Yes;

 = No; NA = Not applicable; NR = Not reported. Item 1: Was the research question or objective clearly stated? Item 2: Was the study population clearly specified and defined? Item 3: Was the participation rate of eligible persons at least 50%? Item 4: Were all subjects recruited from the same or similar populations (including the same time period) with uniform eligibility criteria? Item 5: Was a sample size justification, power description, or variance/effect estimates provided? Item 6: Was the exposure measured prior to the outcome? Item 7: Was the timeframe sufficient to expect an association (if it existed)? Item 8: Did the study examine different levels of the exposure? Item 9: Were exposure measures clearly defined, valid, reliable, and implemented consistently? Item 10: Was the exposure assessed more than once over time? Item 11: Were outcome measures clearly defined, valid, reliable, and implemented consistently? Item 12: Were outcome assessors blinded to exposure status? Item 13: Was loss to follow-up after baseline 20% or less? Item 14: Were key potential confounders measured and statistically adjusted? Items 7, 10, and 13 are NA for cross-sectional designs.

Results of syntheses

Controlling teaching style and physical activity intention

Figure 2 presents the relationship between controlling teaching style and students’ PA intention. The results showed a significant negative association, indicating that higher levels of controlling teaching were related to lower PA intention (r = –0.14; 95% CI: –0.22, –0.06; p < .001; I² = 90.05%). As shown in Figure 2, most individual study estimates were negative, although the magnitude of effects varied and several studies reported non-significant or small positive correlations. The sensitivity analyses indicated no substantial changes after removing one study at a time (Supplementary Figure 1). The funnel plot and Egger’s test indicated no evidence of publication bias (Supplementary Figure 2; Z = −0.41, p = .678). Meta-regression analyses showed no evidence that age was associated with variability in this association (Supplementary Figure 3; β = –0.26, p = .471).

Figure 2.

Forest plot of the association between controlling teaching behaviors and physical activity intention.

Controlling teaching style and physical activity levels

Figure 3 shows the relationship between controlling teaching style and students’ PA levels. The results showed a significant negative association, indicating that higher levels of controlling teaching were related to lower PA levels (r = –0.18; 95% CI: –0.31, –0.06; p < .01; I² = 95.79%). As shown in Figure 3, most individual study estimates were negative, although effect sizes varied considerably, and a small number of studies reported weak or non-significant positive correlations. Sensitivity analyses revealed no substantial modifications when each study was removed one at a time (Supplementary Figure 4). The funnel plot and Egger’s test indicated evidence of publication bias for this outcome (Supplementary Figure 5; Z = 2.97, p < .01). Meta-regression analyses revealed no evidence of an association between age and variability, nor between controlling teaching styles and PA levels (Supplementary Figure 6; β = –0.43, p = .563).

Figure 3.

Forest plot of the association between controlling teaching behaviors and physical activity levels.

Note. PA levels were assessed using self-report (*) and device-based (‡) measures; symbols next to study labels indicate method.

Subgroups analyses

Table 4 presents the sex-based subgroup analyses for the associations between controlling teaching style and both PA intention and PA levels. For PA intention, the strength of the association was similar for boys (r = –0.18; 95% CI: –0.31 to –0.04) and girls (r = –0.14; 95% CI: –0.27 to –0.01), and the association was statistically significant in both groups. No significant differences were found between subgroups (p =  .704) For PA levels, boys (r = –0.20; 95% CI: –0.40 to 0.01) and girls (r = –0.20; 95% CI: –0.37 to –0.04) also showed comparable associations, with no significant subgroup differences (p =  .951), although the association reached statistical significance only in girls.

Table 4.

Subgroup meta-analyses by sex for the association between controlling teaching behaviors and PA-related outcomes

Variables	Effect size statistics				Heterogeneity statistic		Publication bias
	k	Correlation (r)	95% CI	p	Q	I^2 (%)	p (Egger)
Physical activity intention
Overall studies	12	−0.14	−0.22 to −0.06	<.001	200.96	90.05	.785
Boys and girls	3	−0.05	−0.14 to 0.05	–	6.98	71.34	.392
Boys	9	−0.18	−0.31 to −0.04	<.001	88.62	90.97	.346
Girls	9	−0.14	−0.27 to −0.01	<.001	79.28	89.91	.846
Physical activity levels
Overall studies	8	−0.18	−0.31 to −0.06	<.01	332.51	95.79	<.001
Boys and girls	1	0.06	−0.04 to 0.16	–	–	–	–
Boys	7	−0.20	−0.40 to 0.01	.069	155.09	96.13	<.01
Girls	7	−0.20	−0.37 to −0.04	<.05	90.96	93.40	<.001

Note. k = number of effect sizes; CI = confidence intervals; Q = total variance; I² = percentage of heterogeneity; p (Egger) = two-tailed p-value from Egger’s test. Dashes indicate values not applicable.

Discussion

This systematic review and meta-analysis examined how students’ perceptions of controlling teaching behaviors in PE relate to two key PA-related outcomes (i.e., intention to be physically active and overall PA levels) and whether these associations differ according to sex and age.

Effects of controlling teaching on PA intention and PA levels

The meta-analysis revealed small negative effects on students’ intention to engage in PA and their overall PA levels. Although motivational processes were not measured in the included studies, the pattern observed is consistent with theoretical perspectives derived from SDT. These findings are in line with the a priori hypothesis that higher levels of perceived controlling teaching behaviors would be associated with lower PA intention and PA levels. Prior research shows that controlling teaching behaviors—whether expressed through overt strategies such as threats or excessive criticism, or through more subtle forms of pressure such as inducing guilt or disregarding students’ perspectives—aim to secure compliance at the expense of students’ personal initiative (Bartholomew et al., 2018; Reeve, 2009; Vasconcellos et al., 2020). These behaviors are typically accompanied by feelings of pressure, tension, and reduced volition (Haerens et al., 2016), creating conditions that, from an SDT perspective, are likely to frustrate students’ basic psychological needs. Restrictions on choice may undermine autonomy, persistent criticism may erode competence, and pressuring interpersonal interactions may weaken relatedness. Consistent with this reasoning, controlling teaching has been repeatedly linked to need frustration and to less adaptive motivational outcomes in PE (Vasconcellos et al., 2020). While the present review cannot speak directly to these mechanisms, SDT provides a coherent interpretive framework for understanding why adolescents who perceive their teachers as more controlling also tend to report lower intention to be active and reduced participation in PA.

In addition to SDT-based interpretations, evidence outside the scope of this meta-analysis demonstrates that controlling teaching behaviors are frequently accompanied by negative affective experiences—such as stress, anxiety, or reduced enjoyment—that relate to less favorable attitudes toward PA (Haerens et al., 2016). Qualitative research further illustrates this pattern, showing that students who perceive their teachers as controlling often describe PE as emotionally pressuring, evaluative, or discouraging, reporting feelings of being judged or reluctant to participate and, in some cases, a gradual disengagement from PE and from PA more generally (White et al., 2021). Although such findings do not establish causal pathways, they provide valuable insight into how controlling behaviors may shape students’ affective experiences and broader evaluations of PA, thereby offering a coherent backdrop for understanding why lower PA intention and reduced PA engagement tend to co-occur with perceptions of a controlling PE climate. Nevertheless, the magnitude of the observed heterogeneity suggests that these associations should be interpreted with caution, as substantial between-study variability indicates that additional contextual and methodological factors may influence the strength of these relationships.

The moderating role of sex and age

Subgroup meta-analyses and meta-regressions revealed that neither sex nor age moderated the associations between students’ perceptions of controlling teaching and PA-related variables, indicating that the detrimental effects of controlling teaching are comparable for boys and girls and across the adolescent age range represented in the included studies. Although effect sizes were virtually identical across sexes, the association for PA levels reached statistical significance only in girls due to wider confidence intervals in boys; this reflects differences in statistical precision rather than true subgroup effects.

This finding contrasts with evidence from motivational research showing that boys and girls may differ in their sensitivity to particular forms of controlling behaviors (e.g., Abós et al., 2022; Koka & Sildala, 2018). Importantly, however, such studies examined internal motivational processes—such as need frustration, controlled motivation, or amotivation—rather than behavioral outcomes. The absence of moderation in the present review therefore suggests that, even if boys and girls react differently to specific controlling strategies at a motivational level, these differences may not translate into meaningful variations in their intention to be active or in their overall PA engagement.

A similar interpretation applies to age. Although developmental research indicates that adolescents’ motivational orientations and PE experiences may change across schooling years, the present results show that the negative associations between controlling behaviors and PA-related outcomes remained stable across ages. This supports the SDT proposition that the harmful effects of controlling climates may be relatively universal, as the frustration of autonomy, competence, or relatedness is theorized to be detrimental regardless of developmental period (Ryan & Deci, 2017).

Taken together, these findings suggest that individual differences based on sex and age do not appear to substantially modify the associations between controlling teaching behaviors and PA intention or PA levels. However, these conclusions should be considered cautiously, as they are based on exploratory analyses and cross-sectional evidence. This uniformity underscores the broad and generalized negative impact of controlling teaching, reinforcing the view that such practices are detrimental across adolescent subgroups rather than affecting only specific profiles of students.

Limitations and future directions

This review and meta-analysis have several limitations that open new avenues for research. First, as the data synthesized were drawn exclusively from cross-sectional studies, no causal inferences can be made about the relationships between controlling teaching style and the PA related outcomes examined. Longitudinal and experimental designs are therefore needed to establish temporal ordering and causality. Second, the search strategy included only studies published in English and Spanish, meaning that relevant studies in other languages may have been missed. Third, the included studies showed substantial variability in the methods used to assess PA, ranging from self-reported questionnaires to device-based measures. This methodological heterogeneity may introduce measurement bias and likely contributed to the high heterogeneity observed across effect sizes; therefore, caution is required when interpreting the magnitude of the effects on PA levels. In particular, the extremely high levels of heterogeneity suggest that the pooled estimates should be interpreted as average effects across highly diverse study contexts and may not reflect a uniform association across all settings. However, sensitivity analyses indicated that removing studies one at a time did not meaningfully alter the results. Fourth, although the findings for PA intention did not indicate publication bias, the presence of significant publication bias for PA levels suggests that these results should be interpreted with caution, as it may affect the precision and validity of the pooled estimates for this outcome. This finding may suggest that studies reporting significant or negative associations are overrepresented, potentially leading to an overestimation of effects. In contrast, the absence of publication bias for PA intention supports the robustness of those results. However, given the limited number of studies, findings should be interpreted with caution. Fifth, controlling teaching style was assessed exclusively through students’ self-reports, which may introduce perceptual or common-method bias; future studies should incorporate observational or teacher-reported measures. Additionally, most included samples were composed of adolescents, limiting the generalizability of the findings to younger children; further research in child populations is needed to understand how controlling teaching affects PA earlier in development. Finally, this review did not examine basic psychological needs or motivational processes, which are central to SDT and may help explain how controlling teaching exerts its effects. Future research should incorporate these constructs and explore contextual moderators such as educational level, school characteristics, and cultural background. Additionally, more work is needed to examine how controlling teaching practices in PE transfer to students’ PA behaviors beyond the school setting.

Conclusions

This systematic review and meta-analysis shows that controlling teaching behaviors in PE are associated with lower PA intention and reduced PA levels among adolescents, although the magnitude of these associations is small. Although age and sex did not moderate these associations, the negative link with PA levels was more clearly supported in girls. These findings should be interpreted with caution, given the cross-sectional nature of the evidence and the high between-study heterogeneity. From a practical perspective, minimizing controlling behaviors in PE may contribute to more supportive learning environments, although further longitudinal and experimental research is needed to clarify their role in students’ PA engagement.

CRediT authorship contribution statement

Miguel Ángel Tapia-Serrano: conceptualization, formal analysis, methodology, software, resources, writing – original draft. Javier Sevil-Serrano: conceptualization, methodology, resources, supervision, writing – original draft, writing – review & editing. José Luis Pérez-Lasierra: conceptualization, research, methodology, writing – original draft. Luis García-González: research, resources, supervisión, writing – review & editing.

Funding

This study was supported by the Regional Ministry of Education, Science and Vocational Training of the Government of Extremadura, with 85% co-financing from the European Regional Development Fund (ERDF; Ref. IB24203). Funding was also received from regional and European research programs (Ref. GR24155; Research Groups SEJ048 and SEJ059), also co-financed by the European Regional Development Fund. Finally, MATS is supported by the INVESTIGO program (Ref. INVESTIGO 048-17), included within the framework of the Recovery, Transformation and Sustainability Plan.