Evaluation of self-efficacy and psychological well-being among coaches of athletes with and without disabilities: a comparative cross-sectional study

Abstract

Background

Coaches influence athletes’ performance and psychosocial development, yet evidence comparing psychological well-being (PWB) and coach self-efficacy (CSE) between coaches working with athletes with disabilities and those coaching athletes without disabilities remains limited. This study aimed to compare PWB and CSE between these two coaching contexts, examine the association between CSE and PWB, and identify sociodemographic and professional predictors of both outcomes.

Methods

In this cross-sectional study, 312 coaches (mean age 35.79 ± 7.46 years; 50.0% female) were recruited through voluntary online participation. Of these, 49.4% coached athletes with disabilities and 50.6% coached athletes without disabilities. Data were collected using a structured sociodemographic form, the Psychological Well-Being scale, and the Coach Self-Efficacy scale. Pearson correlation analyses examined associations between psychological well-being and self-efficacy dimensions. Separate multiple linear regression models were constructed for each coaching context to identify independent predictors of psychological well-being and total coaching self-efficacy.

Results

Total PWB scores did not differ significantly between coaches working with athletes with disabilities and those coaching athletes without disabilities (41.85 ± 10.07 vs. 43.26 ± 10.61; p = 0.229). In contrast, CSE was consistently higher among coaches of athletes with disabilities across all domains: performance (p = 0.032), psychological (p = 0.006), teaching (p = 0.011), character-building (p = 0.027), and management competence (p = 0.016). Total CSE scores were also higher in this group (84.32 ± 18.13 vs. 78.92 ± 18.21; p = 0.009). PWB was strongly and positively correlated with CSE in both groups (r values > 0.60). In regression analyses, income was the strongest independent predictor of PWB in both groups (p < 0.05). In the non-disability group, sex was additionally associated with PWB (p = 0.034). For total CSE, income and coaching level independently predicted higher scores in the non-disability group (p < 0.05), whereas relationship status was the only significant predictor in the disability group (p = 0.036).

Conclusions

Coaches working with athletes with disabilities report higher self-efficacy across multiple domains, although overall psychological well-being is comparable across coaching contexts. The strong association between self-efficacy and well-being highlights perceived professional competence as a key correlate of coaches’ mental health. Sociodemographic and professional factors, particularly income, certification level, and relationship status, were associated with these outcomes. Given the cross-sectional design and voluntary online recruitment, the findings should be interpreted with caution.

Introduction

Individuals responsible for training and guiding athletes occupy pivotal leadership roles within sport systems, and meta-analytic evidence indicates that leadership behaviors in these roles are meaningfully associated with athlete satisfaction and group cohesion [1, 2]. The modern athletic landscape positions the sport professional as a critical determinant of both performance outcomes and the psychosocial development of athletes [3]. Within this landscape, coaching self-efficacy (often operationalized as coaching efficacy) reflects coaches’ confidence in their capacity to influence athletes’ learning and performance across domains such as motivation, game strategy, teaching technique, and character building [4, 5]. A meta-analysis of the coaching efficacy literature supports the importance of mastery-related sources as correlates of coaching efficacy dimensions, strengthening the case for efficacy as a core coach-level resource [6]. Accordingly, examining coaching self-efficacy provides a theoretically grounded pathway for explaining variability in coaching behaviors and related sport outcomes across contexts [7, 8].

The psychological well-being of sport professionals is increasingly recognized as an essential component of professional competence, characterized by dimensions such as self-acceptance, autonomy, and environmental mastery [9, 10]. Although the relationship between self-efficacy and psychological well-being has been primarily documented in student and general sport populations [9], theoretical frameworks and emerging coach-focused evidence suggest that analogous mechanisms operate in professional coaching contexts [11]. Elite coaches are exposed to significant performance pressures, organizational stressors, and excessive workloads that may compromise their mental health [12, 13]. Chronic exposure to these stressors can lead to elevated risks of burnout, anxiety, and depression, which in turn impair coaching effectiveness and decision-making [3, 12]. Empirical evidence suggests a reciprocal relationship where a coach’s psychological functioning directly influences the motivational climate and well-being of their athletes [8, 14]. Therefore, preserving the psychological health of the coach is paramount for sustaining a successful sports system [13].

Working with athletes with disabilities introduces unique pathophysiological and psychosocial challenges that require specialized adaptive expertise [15–17]. Coaches in the para-sport domain must navigate systemic barriers, including inadequate facilities, limited financial support, and a significant lack of specialized equipment [18]. Beyond physical constraints, societal attitudes such as ableism and stigmatization present continuous obstacles to the full participation and performance of athletes with disabilities [18, 19]. These external stressors often impose an additional emotional and advocacy burden, potentially influencing self-efficacy and well-being. Despite these challenges, there is a noted scarcity of formal disability-specific training within general coaching education [18, 20].

Coach self-efficacy and psychological well-being are increasingly regarded as linked determinants of coaching effectiveness and sustainability, yet coach-focused evidence remains less developed and methodologically heterogeneous than athlete-oriented research [12]. While the literature on coaching effectiveness in mainstream sports is extensive, research specifically comparing coaches of athletes with and without disabilities remains limited [15, 16, 18–20]. Moreover, methodological heterogeneity characterizes this literature, including differences in study populations (elite vs. recreational settings), measurement tools, outcome definitions, and analytical approaches, which complicates comparisons across studies and limits generalizable conclusions. Consequently, it remains unclear whether coaching athletes with disabilities is associated with differences in professional confidence and psychological well-being compared with coaching athletes without disabilities. Clarifying these issues is important for developing targeted support strategies and education programs that promote inclusivity and sustainability in coaching environments. Addressing these gaps, this cross-sectional study aimed to compare self-efficacy and psychological well-being between coaches working with athletes with and without disabilities, examine the association between coaching self-efficacy and psychological well-being, and identify sociodemographic and professional predictors of these outcomes. Based on this framework, the study tested the following hypotheses: (H1) there would be significant differences in self-efficacy levels between coaches working with disabled and non-disabled athletes; (H2) there would be significant differences in psychological well-being between these two groups; and (H3) coaching self-efficacy would be significantly associated with psychological well-being.

Methods

Study design, context, and setting, and ethical approval

This study employed a cross-sectional, comparative design to examine psychological well-being and coach self-efficacy among sports coaches working with two distinct athlete populations: (i) coaches training athletes with disability and (ii) coaches training athletes without disability. Data were collected using a structured, self-administered survey that captured participants’ sociodemographic and professional characteristics, alongside standardized measures of psychological well-being and coaching self-efficacy. The study was conducted within the organized sports coaching context in Türkiye, encompassing coaches from a range of sport branches and certification levels (Levels 1–5), thereby reflecting a broad spectrum of coaching environments and professional profiles. Group comparisons and stratified analyses were prespecified to allow evaluation of differences in outcomes and correlates by athlete group coached.

Ethical approval for this study was obtained from the Dokuz Eylül University Non-Interventional Research Ethics Committee (Decision date: 19 February 2025; Protocol code: 2025/06–08). All procedures were conducted in accordance with the ethical standards of the institutional review board and the principles of the Declaration of Helsinki. Participation was voluntary, and informed consent was obtained from all participants prior to data collection. Data were collected anonymously (de-identified) and used exclusively for scientific purposes, with confidentiality safeguarded throughout the study.

Sampling strategy and participant recruitment, and eligibility criteria

An a priori sample size calculation was performed using G*Power (version 3.1) to ensure adequate statistical precision for the primary between-group comparison. The calculation was based on an independent-samples t test (two-tailed), assuming a standardized mean difference of small-to-moderate effect, a type I error rate of α = 0.05, and desired power of 1 − β = 0.80. Under these assumptions, the minimum total sample required was 220 coaches, corresponding to at least 110 participants in each group under an equal allocation ratio (1:1). This target was selected to maintain acceptable control of type I and type II error while providing sufficient sensitivity to detect the prespecified effect size in the primary outcome analyses.

Coaches were eligible for inclusion if they (i) were aged ≥ 18 years, (ii) were actively working as a sports coach at the time of the study, (iii) had direct coaching responsibility for either athletes with disability or athletes without disability, and (iv) provided informed consent to participate. To ensure comparability between groups, participants were categorized according to the primary athlete group they coached (disability vs. no disability). Coaches were excluded if they (i) were not currently engaged in active coaching duties, (ii) did not clearly report the athlete group they predominantly coached, (iii) provided incomplete or inconsistent responses in key study measures (i.e., psychological well-being and/or Coaching Self-Efficacy Scale), or (iv) withdrew consent at any stage of the study. Coaches working simultaneously with both disabled and non-disabled athletes were classified according to the population they identified as their primary coaching group.

Participants were recruited using a convenience sampling approach. Invitations to participate in the study were distributed through national and regional sports federations, sports clubs, and professional coaching networks. Additionally, the survey link was shared via email and social media platforms commonly used by coaching communities. Participation was voluntary, and data were collected through an anonymous online questionnaire. Coaches who met the inclusion criteria and provided informed consent were included in the study. Coaches working with athletes with disabilities reported coaching individuals with physical, intellectual, or sensory impairments across recreational and competitive levels. Data collection was conducted between March 2025 and January 2026. Coaches working in both disability and non-disability contexts were asked to indicate their primary athlete population, and classification was based on this response. Participants were not recruited consecutively; therefore, the sample represents voluntary self-selection through convenience sampling.

Data collection tools and procedures

Study data were planned to be collected online using a structured questionnaire created in Google Documents (Google Forms), and a secure survey link was generated for access. To facilitate recruitment and maximize reach, the link was distributed via email lists of sports federations and clubs, as well as professional coaching networks and social media platforms (e.g., WhatsApp groups and coaching community pages). Prior to participation, all individuals were provided with a brief explanation of the study aims and procedures, and only those who agreed to take part proceeded to complete the online form.

Personal information form

A researcher-developed Personal Information Form, prepared based on the relevant literature, was used to capture key sociodemographic and background characteristics of the participating coaches. The form included items on age, sex, coaching experience (years), primary sport branch, athlete group coached, educational level, and graduation status. The form also included items on income level, coaching certification level, relationship status, employment status (full-time/part-time), and weekly coaching workload.

Coaching self-efficacy scale [21]

Coach self-efficacy was assessed using the Coaching Self-Efficacy Scale developed by Koçak [21], for which validity and reliability have been established. The instrument is a 21-item measure rated on a five-point Likert scale ranging from 1 (“Strongly disagree”) to 5 (“Strongly agree”). It comprises five subscales: Performance Competence (4 items), Psychological Competence (4 items), Teaching Competence (5 items), Character-Building Competence (4 items), and Management Competence (4 items). In the original validation study, internal consistency was reported as satisfactory, with Cronbach’s alpha of 0.86 for the total scale; and 0.73, 0.81, 0.75, 0.83, and 0.75 for the respective subscales. The internal consistency of the Coaching Self-Efficacy Scale was also evaluated using Cronbach’s alpha coefficients calculated for the current sample (N = 312). The overall scale demonstrated excellent reliability (Cronbach’s α = 0.97). Subscale reliability coefficients were acceptable to excellent: Performance Competence (α = 0.81), Psychological Competence (α = 0.90), Teaching Competence (α = 0.91), Character-Building Competence (α = 0.90), and Management Competence (α = 0.88). These findings indicate high internal consistency for both the total scale and its subdimensions.

Psychological well-being scale

Psychological well-being was measured using the Psychological Well-Being Scale originally developed by Diener and colleagues (2010) [22] to capture socio-psychological aspects of well-being beyond traditional indicators. The Turkish adaptation and psychometric evaluation were conducted by Telef (2013) [23]. In the validation study, exploratory factor analysis supported the construct structure, with the scale explaining 42% of the total variance and item factor loadings ranging from 0.54 to 0.76. The instrument demonstrated satisfactory internal consistency (Cronbach’s α = 0.80) and strong temporal stability, with test–retest reliability showing a high positive correlation between administrations (r = 0.86, p < 0.001). The scale consists of 8 positively worded items rated on a 7-point Likert format from 1 (“strongly disagree”) to 7 (“strongly agree”), yielding total scores between 8 and 56; higher scores reflect greater psychological resources and strengths. The Psychological Well-Being Scale also demonstrated good internal consistency in the current sample (Cronbach’s α = 92). Missing data were minimal (< 5%) and handled using listwise deletion in regression analyses.

The primary outcomes were total psychological well-being and total coaching self-efficacy scores. Subscale-level analyses were considered secondary outcomes.

Statistical analysis

All analyses were conducted on the full sample of coaches (n = 312), with results additionally stratified by the athlete group coached. Normality, homoscedasticity, and linearity assumptions were assessed using residual plots, Kolmogorov-Smirnov tests for normality, and inspection of standardized residuals and scatterplots for homoscedasticity. Descriptive statistics were used to summarize participant characteristics and study variables, reporting means with standard deviations (SD) for continuous variables (e.g., age, years of coaching) and frequencies with percentages for categorical variables. Between-group comparisons of PWB total and coach self-efficacy (CSE total and subscales) were performed using two-tailed independent-samples t tests. Mean differences (MD) with 95% confidence intervals (CI) were reported to quantify the magnitude and direction of group differences. The assumption of homogeneity of variances was evaluated using Levene’s test; as no evidence of variance inequality was observed across outcomes (all p > 0.05), the “equal variances assumed” results were presented. In addition to p-values, standardized effect sizes were calculated using Cohen’s d and interpreted as small (0.20), medium (0.50), and large (0.80) [24]. Associations between PWB and CSE were examined using Pearson correlation coefficients (r), computed separately within each coaching context and visualized as correlation heatmaps. To examine whether the correlation coefficients between PWB and CSE subscales differed significantly across the two groups, Fisher’s z-transformation tests were conducted for each pair of correlations. Correlation matrices included PWB total, and the CSE subdomains (performance, psychological, teaching, character-building, and management competence), enabling inspection of both cross-construct associations (PWB–CSE) and within-construct interrelationships among CSE subscales. To identify independent predictors of outcomes, multiple linear regression models were fitted separately by athlete group coached. Separate regression models were conducted to explore context-specific predictors within each coaching group. PWB total score was modeled as the dependent variable; the CSE total score was the dependent variable. Predictors entered simultaneously in each model were age (years), sex, education level, relationship status, income level, years of coaching, and coaching level (grade). Categorical variables were dummy-coded using the following reference categories: sex = female; relationship status = in a relationship; income level = low income, and education = Secondary school. Coaching level was entered as an ordinal variable (Levels 1–5), with higher values indicating higher coaching grade. For each predictor, unstandardized regression coefficients (B) with standard errors (SE), standardized coefficients (β), t statistics, p values, and 95% CIs were reported, alongside overall model fit indices (R, R², adjusted R², and the model F test). Model assumptions were assessed using standard regression diagnostics: linearity and homoscedasticity were evaluated via residual plots; normality was examined using the distribution of standardized residuals; and influential observations were checked using leverage and Cook’s distance. Independence of residuals was evaluated using the Durbin–Watson statistic, which was reported for each model. Multicollinearity among independent variables was assessed using variance inflation factor (VIF) values. No influential outliers were detected based on standardized residuals (± 3). As analyses were exploratory, no formal adjustment for multiple comparisons was applied. All statistical analyses were performed using STATA (StataCorp, version 18.0, College Station, TX) and IBM SPSS Statistics for Mac, Version 30.0 (IBM Corp., Armonk, NY, USA). All hypothesis tests were conducted using two-tailed procedures, with statistical significance set a priori at p < 0.05.

Results

The sociodemographic and professional profiles of the participating coaches were summarized in Table 1 (n = 312). Overall, the sample was relatively young-to-middle adulthood (mean age 35.79 ± 7.46 years) with substantial practical exposure to coaching (mean experience 9.17 ± 6.28 years). Sex distribution was balanced, with equal representation of female and male coaches (each 50.0%), and coaches were similarly distributed across the athlete groups they trained, with 50.6% working with athletes without disabilities and 49.4% coaching athletes with disabilities. The cohort was highly educated, as most participants reported a university degree (80.4%) and a meaningful proportion held postgraduate qualifications (16.0%), while only a small minority had secondary school education (3.5%). Most coaches were in a relationship (73.1%), and income levels were predominantly middle (51.9%) or high (34.9%), with a smaller low-income segment (13.1%). Coaching grades clustered mainly in Levels 1–3 (23.7%, 34.9%, and 31.1%, respectively), with fewer coaches in advanced Levels 4–5 (8.7% and 1.6%). With respect to sport branches, team ball sports constituted the largest group (27.9%), followed by athletics/track & field (23.1%), swimming/aquatic sports (17.9%), and racket sports (17.6%), with smaller representation in combat sports (6.1%), fitness/conditioning (1.0%), and other disciplines (6.4%).

Table 1.

Sociodemographic and professional characteristics of the participants (n = 312)

Characteristic	n (%) Mean ± SD
Age (years)	35.79 ± 7.46 (median: 35, IQR: 30–41)
Coaching experience (years)	9.17 ± 6.28
Sex
Female	156 (50.0)
Male	156 (50.0)
Athlete group coached
No disability	158 (50.6)
Disability	154 (49.4)
Education level
Secondary school	11 (3.5)
University	251 (80.4)
Postgraduate	50 (16.0)
Relationship status
In a relationship	228 (73.1)
Not in a relationship	84 (26.9)
Income level
Low	41 (13.1)
Middle	162 (51.9)
High	109 (34.9)
Coaching level (grade)
Level 1	74 (23.7)
Level 2	109 (34.9)
Level 3	97 (31.1)
Level 4	27 (8.7)
Level 5	5 (1.6)
Primary sport branch
Team ball sports (basketball/football/futsal/volleyball/handball)	87 (27.9)
Athletics/track & field (incl. mixed entries with athletics)	72 (23.1)
Swimming/aquatic sports	56 (17.9)
Racket sports (table tennis/tennis/badminton; incl. mixed entries with table tennis)	55 (17.6)
Combat sports (judo/taekwondo/wrestling)	19 (6.1)
Fitness/conditioning (fitness/pilates/bodybuilding)	3 (1.0)
Other sports (e.g., archery, gymnastics, skiing, bocce, floor curling, movement education, etc.)	20 (6.4)

Values are presented as n (%) unless otherwise indicated. Mean ± SD is reported for continuous variables. Percentages may not sum to 100% due to rounding and/or missing data. For selected variables, the denominator reflects the number of participants with available responses (valid n)

IQR Interquartile range

Table 2 compares psychological well-being and coach self-efficacy outcomes between coaches working with athletes with disability and those working with athletes without disability. Overall psychological well-being scores did not differ significantly between groups (41.85 ± 10.07 vs. 43.26 ± 10.61; p = 0.229, Cohen’s d = 0.14). In contrast, all domains of coach self-efficacy were significantly higher among coaches working with athletes with disability, including performance competence (MD = − 0.91, 95% CI − 1.74 to − 0.08; p = 0.032, Cohen’s d = 0.24), psychological competence (MD = − 1.27, 95% CI − 2.17 to − 0.36; p = 0.006, Cohen’s d = 0.31), teaching competence (MD = − 1.36, 95% CI − 2.41 to − 0.31; p = 0.011, Cohen’s d = 0.29), character-building (MD = − 0.88, 95% CI − 1.66 to − 0.10; p = 0.027, Cohen’s d = 0.25), and management competence (MD = − 0.98, 95% CI − 1.78 to − 0.18; p = 0.016, Cohen’s d = 0.27), culminating in a higher total self-efficacy score in the disability group (78.92 ± 18.21 vs. 84.32 ± 18.13; p = 0.009, Cohen’s d = 0.30).

Table 2.

Comparison of total and subscale scores among coaches working with athletes with disability versus athletes without disability (n = 312)

Outcome	Total sample Mean ± SD	No disability (Mean ± SD) (n = 158)	Disability (Mean ± SD) (n = 154)	Cohen’s d	p-value	MD, 95% CI
Psychological well-being (total)	42.54 ± 10.35	41.85 ± 10.07	43.26 ± 10.61	0.14	0.229	−1.41, − 3.71 to 0.89
Coach self-efficacy: Performance competence	14.90 ± 3.75	14.45 ± 3.75	15.36 ± 3.71	0.24	0.032†	−0.91, − 1.74 to − 0.08
Coach self-efficacy: Psychological competence	15.37 ± 4.11	14.74 ± 4.19	16.01 ± 3.93	0.31	0.006†	−1.27, − 2.17 to − 0.36
Coach self-efficacy: Teaching competence	19.03 ± 4.75	18.36 ± 4.87	19.72 ± 4.53	0.29	0.011†	−1.36, − 2.41 to − 0.31
Coach self-efficacy: Character-building	16.36 ± 3.53	15.92 ± 3.42	16.81 ± 3.60	0.25	0.027†	−0.88, − 1.66 to − 0.10
Coach self-efficacy: Management competence	15.93 ± 3.60	15.45 ± 3.65	16.43 ± 3.49	0.27	0.016†	−0.98, − 1.78 to − 0.18
Coach self-efficacy (total)	81.59 ± 18.34	78.92 ± 18.21	84.32 ± 18.13	0.30	0.009†	−5.39, − 9.44 to − 1.35

Values are presented as mean ± standard deviation. Group comparisons were performed using independent-samples t tests (two-tailed). Levene’s test indicated no evidence of variance inequality across outcomes (all p > 0.05), therefore the “equal variances assumed” results are reported

CI Confidence interval, M mean, n number of participants, p probability value, SD Standard deviation, MD Mean differences

^†Statistically significant. Effect sizes were calculated using Cohen’s d based on pooled standard deviation. Effect sizes were calculated using Cohen’s d based on pooled standard deviation. Values of 0.20, 0.50, and 0.80 were interpreted as small, medium, and large effects, respectively

The Pearson correlation structure between overall psychological well-being (PWB) and coach self-efficacy (CSE) dimensions across the two coaching contexts is presented in Fig. 1. In both panels, all coefficients are positive and predominantly large in magnitude, indicating that higher perceived coaching competence co-occurs with higher psychological well-being and that the CSE subdomains share substantial common variance. In panel A, correlations between PWB total and the CSE subscales range from moderate to strong (r = 0.586–0.739, p < 0.001), with the weakest association observed for character-building (r = 0.586). The internal coherence of the CSE construct is also evident, as intercorrelations among CSE subscales are consistently high (approximately r = 0.693–0.849, p < 0.001). Panel B shows a broadly similar pattern but with uniformly stronger linkages between PWB and CSE (PWB total with CSE subscales r = 0.744–0.767, p < 0.001). The results revealed that only the correlation between psychological well-being and the character-building subscale differed significantly between the two groups (z = − 2.99, p = 0.003), whereas no statistically significant differences were observed for the remaining subscales (all ps > 0.05).

Fig. 1.

Side-by-side correlation heatmaps of psychological well-being and coach self-efficacy scores, stratified by athlete group. (n = 312). Panel A shows Pearson correlation coefficients (r) among study variables for coaches working with athletes without disability (n = 158), and panel B shows the same correlations for coaches working with athletes with disability (n = 154). All correlations were statistically significant (two-tailed, p < 0.001). PWB: psychological well-being; CSE: coach self-efficacy; r: Pearson correlation coefficient; n: number of participants

Table 3 presents separate multiple linear regression models examining sociodemographic and professional predictors of PWB among coaches working with athletes without disabilities versus those coaching athletes with disabilities. In the no-disability group, the model explained a moderate proportion of variance in PWB (R² = 0.275; adjusted R² = 0.241; F(7,149) = 8.059; p < 0.001), with income emerging as the strongest independent predictor (B = 5.111, SE = 1.276; β = 0.339; p < 0.001), indicating higher PWB among coaches in higher income categories relative to the low-income reference. Sex was also a significant predictor (B = − 3.100, SE = 1.448; β = −0.154; p = 0.034), suggesting lower PWB for males compared with females (reference), while age showed a borderline association (B = 0.255; p = 0.058). Other covariates—including education, relationship status, years of coaching, and coaching level—were not statistically significant, implying limited incremental explanatory value beyond income and sex in this subgroup. For coaches working with non-disabled athletes, VIF values ranged between 1.06 and 1.73, indicating no evidence of multicollinearity. In the disability group, overall explanatory power was smaller (R² = 0.176; adjusted R² = 0.134; F(7,137) = 4.184; p < 0.001), but income again remained a significant predictor (B = 3.126, SE = 1.347; β = 0.191; p = 0.022). In contrast to the no-disability group, sex was not associated with PWB (p = 0.338), and coaching level showed only a weak, non-significant positive trend (B = 1.617; p = 0.099). For coaches working with athletes with disabilities, VIF values ranged between 1.03 and 1.95, also suggesting that multicollinearity was not a concern.

Table 3.

Multiple linear regression models predicting psychological well-being total score among coaches working with athletes without disabilities vs. athletes with disabilities

Predictor	No disability group (n = 157)					Disability group (n = 145)
	B (SE)	β	t	p	95% CI for B	B (SE)	β	t	p	95% CI for B
Intercept	21.347 (6.641)	—	3.214	0.002**	8.225 to 34.470	27.386 (7.895)	—	3.469	< 0.001***	11.775 to 42.998
Age (years)	0.255 (0.134)	0.175	1.909	0.058	−0.009 to 0.519	0.179 (0.139)	0.134	1.289	0.200	−0.096 to 0.454
Sex	−3.100 (1.448)	−0.154	−2.141	0.034**	−5.962 to − 0.239	−1.599 (1.663)	−0.076	−0.962	0.338	−4.887 to 1.689
Education level	2.099 (1.695)	0.093	1.239	0.217	−1.250 to 5.448	1.831 (2.217)	0.067	0.826	0.410	−2.553 to 6.214
Relationship status	−0.704 (1.596)	−0.032	−0.441	0.660	−3.858 to 2.450	−2.996 (2.032)	−0.118	−1.475	0.143	−7.014 to 1.021
Income status	5.111 (1.276)	0.339	4.006	< 0.001***	2.590 to 7.633	3.126 (1.347)	0.191	2.321	0.022**	0.463 to 5.789
Years of coaching	0.038 (0.156)	0.022	0.245	0.807	−0.270 to 0.347	0.127 (0.173)	0.080	0.734	0.464	−0.215 to 0.469
Coaching level (rank)	0.365 (0.875)	0.035	0.417	0.678	−1.365 to 2.094	1.617 (0.972)	0.144	1.662	0.099	−0.306 to 3.540

Model fit (No disability): R = 0.524; R²=0.275; Adj. R²=0.241; F(7,149) = 8.059; p < 0.001; Durbin–Watson = 1.325; For coaches working with non-disabled athletes, variance inflation factor (VIF) values ranged between 1.06 and 1.73, indicating no evidence of multicollinearity. Model fit (disability): R = 0.420; R²=0.176; Adj. R²=0.134; F(7,137) = 4.184; p < 0.001; Durbin–Watson = 1.330; For coaches working with athletes with disabilities, VIF values ranged between 1.03 and 1.95, also suggesting that multicollinearity was not a concern

B  unstandardized regression coefficient, SE  Standard error, β  standardized coefficient, CI  Confidence interval, Adj  Adjusted, CI Confidence interval, DW Durbin–Watson, SE Standard error. Categorical predictors were dummy-coded using the following reference categories: sex = female, relationship status = in a relationship, and income level = low income, and education = Secondary school. Coaching level (grade) was entered as an ordinal variable with categories 1–5 (higher values indicate higher coaching grade)

p values: p < 0.05 **; p < 0.001 ***

Table 4 also reports multiple linear regression models predicting the total coach self-efficacy score, stratified by coaches working with athletes without disabilities versus those working with athletes with disabilities. In the no-disability group, the model showed modest explanatory power (R² = 0.214; adjusted R² = 0.177; F(7,149) = 5.786; p < 0.001), and two predictors independently emerged: income level (B = 5.176, SE = 2.398; β = 0.190; p = 0.032) and coaching level/grade (B = 3.477, SE = 1.645; β = 0.185; p = 0.036), indicating higher self-efficacy among coaches with higher income and higher certification level after adjustment for other covariates. Sex demonstrated a borderline association (B = − 5.306; p = 0.053), suggesting a possible tendency toward lower self-efficacy among male coaches compared with females, although this did not reach conventional statistical significance. Age, education, relationship status, and years of coaching were not significant contributors in this subgroup. For coaches working with non-disabled athletes, VIF values ranged between 1.07 and 1.72, indicating no evidence of multicollinearity In the disability group, overall model fit was comparable but slightly lower (R² = 0.174; adjusted R² = 0.132; F(7,137) = 4.126; p < 0.001), and relationship status was the only statistically significant predictor (B = − 7.229, SE = 3.408; β = −0.169; p = 0.036), implying lower self-efficacy among coaches who were not in a relationship relative to those who were (reference category). Income (B = 4.324; p = 0.058) and coaching level (B = 2.855; p = 0.082) showed positive but non-significant trends, while age, sex, education, and years of coaching again did not contribute independently. For coaches working with athletes with disabilities, VIF values ranged between 1.03 and 1.98, also suggesting that multicollinearity was not a concern.

Table 4.

Multiple linear regression models predicting total coach self-efficacy score among coaches working with athletes without disabilities vs. athletes with disabilities

Predictor	Coaches of athletes without disabilities (n = 157)					Coaches of athletes with disability (n = 145)
	B (SE)	β	t	p	95% CI for B	B (SE)	β	t	p	95% CI for B
Intercept	65.125 (12.481)	—	5.218	< 0.001***	40.462 to 89.788	67.361 (13.244)	—	5.086	< 0.001***	41.173 to 93.550
Age (years)	0.167 (0.251)	0.064	0.666	0.506	−0.329 to 0.664	0.168 (0.233)	0.075	0.722	0.471	−0.293 to 0.629
Sex	−5.306 (2.722)	−0.146	−1.950	0.053	−10.684 to 0.072	−0.077 (2.789)	−0.002	−0.028	0.978	−5.593 to 5.439
Education level	−1.561 (3.185)	−0.038	−0.490	0.625	−7.854 to 4.733	0.221 (3.719)	0.005	0.059	0.953	−7.132 to 7.575
Relationship status	−3.353 (3.000)	−0.086	−1.118	0.265	−9.281 to 2.574	−7.229 (3.408)	−0.169	−2.121	0.036*	−13.968 to − 0.489
Income level	5.176 (2.398)	0.190	2.158	0.032*	0.438 to 9.914	4.324 (2.259)	0.158	1.914	0.058	−0.143 to 8.791
Years of coaching	0.424 (0.293)	0.138	1.446	0.150	−0.156 to 1.004	0.386 (0.290)	0.144	1.328	0.186	−0.188 to 0.960
Coaching level (grade)	3.477 (1.645)	0.185	2.114	0.036*	0.227 to 6.728	2.855 (1.631)	0.151	1.750	0.082	−0.371 to 6.081

Model fit (no disabilities): R = 0.462; R²=0.214; Adj. R²=0.177; F(7,149) = 5.786; p < 0.001; Durbin–Watson = 1.519; For coaches working with non-disabled athletes, variance inflation factor (VIF) values ranged between 1.07 and 1.72, indicating no evidence of multicollinearity. Model fit (disability): R = 0.417; R²=0.174; Adj. R²=0.132; F(7,137) = 4.126; p < 0.001; Durbin–Watson = 1.234; For coaches working with athletes with disabilities, VIF values ranged between 1.03 and 1.98, also suggesting that multicollinearity was not a concern

B unstandardized regression coefficient, SE Standard error, β  standardized coefficient, CI Confidence interval, Adj. adjusted, CI Confidence interval, DW Durbin–Watson, SE Standard error. Categorical predictors were dummy-coded using the following reference categories: sex = female, relationship status = in a relationship, and income level = low income, and education = Secondary school. Coaching level (grade) was entered as an ordinal variable with categories 1–5 (higher values indicate higher coaching grade)

p values: p < 0.05 **; p < 0.001 ***

Discussion

The present cross-sectional investigation provides a comparative evaluation of coach self-efficacy and psychological well-being across two cohorts, namely coaches working with athletes with disabilities and those training athletes without disabilities. The sex distribution was balanced, and participants were similarly distributed across athlete groups, with 50.6% coaching athletes without disabilities and 49.4% coaching athletes with disabilities. Educational attainment within the sample was relatively high, with over 80% of participants holding university degrees and 16% possessing postgraduate qualifications, indicating a trend toward professionalization in the coaching sector. Coaches play a critical role in shaping athletes’ performance and psychosocial outcomes [16], making it important to understand their own well-being and sense of efficacy. In this sample of 312 coaches, overall psychological well-being did not differ between groups, whereas coaches of athletes with disabilities reported significantly higher self-efficacy across all domains and in the total score. The present data do not permit identification of the mechanisms underlying this group difference, as coaching experience profiles, contextual demands, and skill deployment were not directly measured. This finding, therefore, remains descriptive, and its interpretation is deferred to future studies that include direct assessment of coaching process variables. The correlation coefficients were large in magnitude, indicating an association between perceived coaching competence and psychological well-being across both coaching contexts. These findings are consistent with the literature suggesting that higher perceived competence is associated with better psychological functioning among coaches [16].

Contrary to expectations, coach psychological well-being was comparable across the two groups, suggesting that psychological well-being was comparable across coaching contexts did not differentially affect well-being [11]. Psychological well-being appeared more closely related to contextual resources, with higher income emerging as the strongest predictor in both groups, consistent with evidence indicating that socioeconomic status strongly influences quality of life [25]. Among coaches working with non-disabled athletes, male coaches reported slightly lower psychological well-being. The mechanisms underlying this sex-related difference cannot be determined from the present data, as variables pertaining to stress exposure, coping, or social support were not assessed. This finding should therefore be regarded as an incidental observation requiring targeted investigation in future studies. Importantly, psychological well-being showed strong positive correlations with coach self-efficacy in both samples. In other words, coaches who felt more competent in their role also reported better psychological well-being, consistent with the association between self-efficacy and mental health that has been documented across sport and physical activity contexts [9], and with evidence specifically linking coaching efficacy to coaches’ hedonic and eudaimonic well-being [26].

Of note, the observed between-group difference in self-efficacy is a descriptive finding of the present study. As coaching process variables, contextual demands, and skill use were not measured, no mechanistic interpretation is warranted. Future studies incorporating direct assessment of coaching practices across disability and non-disability contexts would be better positioned to explain this difference.

Both coaching groups showed strong positive correlations between self-efficacy and psychological well-being. Our data revealed moderate-to-large positive associations between overall well-being and each self-efficacy subscale, aligning with broad evidence that self-efficacy is linked to higher psychological well-being [9]. Thus, coaches who perceived themselves as more effective in their role also reported greater well-being, indicating a positive association between coaching competence and psychological well-being in this sample.

Sociodemographic analyses indicated that contextual resources were associated with outcomes. Higher income was the strongest predictor of psychological well-being in both groups, consistent with previous research linking socioeconomic status with quality of life [25]. Although this evidence derives from a general, non-coaching population, the fundamental role of financial security in buffering occupational stress is likely to generalize to coaching professionals, particularly given evidence that economic constraints represent a recognized barrier to sustainable coaching careers. In the non-disabled group, higher coaching certification level predicted greater self-efficacy. In the disability group, relationship status was also associated with self-efficacy. The literature on social support emphasizes that a spouse or partner is often the primary source of support outside of the sport, and satisfaction with this relationship is a robust correlate of mental health [27, 28]. The present study did not include measures of social support or relationship quality; therefore, the association between relationship status and self-efficacy cannot be attributed to any specific supportive mechanism. This finding is noted as a hypothesis-generating observation, with the recommendation that future studies incorporate validated measures of social support to clarify this association.

Limitations

Despite the insights provided by this study, several methodological limitations must be acknowledged. First, the cross-sectional nature of the data measurement precludes the establishment of temporal ordering or causal inference between self-efficacy, income, and well-being. It is possible that individuals with higher baseline well-being are more likely to seek out partnerships and achieve the professional success required to earn a higher income, suggesting a reciprocal relationship that needs to be explored through longitudinal tracking. Furthermore, since participation was voluntary and conducted through an online survey, the study may be subject to self-selection bias, as individuals with a greater interest in psychological well-being or coaching-related issues may have been more likely to participate. Moreover, the absence of objective coaching performance indicators limits the interpretation of the relationship between psychological variables and real-world coaching effectiveness. Additionally, cultural characteristics of the Turkish sport system may influence the observed associations, and therefore caution is warranted when generalizing these findings to different sociocultural contexts. Therefore, the findings should be interpreted with caution in terms of generalizability. Second, the reliance on self-report questionnaires introduces the risk of social desirability bias, particularly regarding sensitive domains like character-building and psychological competence. Third, the study concentrated specifically on athletes with disabilities, and the results may not be generalizable to coaches working with specific physical, sensory, or multiple impairments, as each disability type imposes unique pathophysiological and structural demands. Fourth, the sample was primarily composed of university-educated individuals in middle-to-high income categories, which may limit the applicability of the findings to volunteer or part-time coaches in low-resource community settings who often face greater systemic barriers. Fifth, the study did not measure objective performance outcomes, such as athlete success rates or physiological improvements, which are critical mediators of a coach’s efficacy over time. Future research should utilize mixed-methods approaches and cross-lagged panel models to further elucidate the complex psychological dynamics that underpin coaching effectiveness in both inclusive and segregated environments.

The absence of formal correction for multiple comparisons constitutes a methodological limitation that warrants explicit acknowledgment. Given the number of statistical tests conducted across t-tests, Fisher’s z-transformation comparisons, and regression models, the probability of at least one false-positive finding across the full set of analyses is non-trivial. These findings should be treated as preliminary observations requiring confirmatory replication rather than established effects. In addition, the modest explanatory power of the regression models indicates that the examined variables account for only part of the variability in the outcomes (R² values ranging from approximately 0.17 to 0.27), and other contextual or individual factors not included in the present study may also play a role.

Conclusions

In conclusion, psychological well-being did not differ significantly between coaches working with athletes with and without disabilities, whereas coaches in the disability context reported higher levels of self-efficacy across all domains. A strong positive association between self-efficacy and psychological well-being was observed in both groups, indicating that higher perceived coaching competence was related to better psychological well-being. Income emerged as a consistent predictor of psychological well-being across contexts. In addition, coaching level and relationship status showed context-specific associations with self-efficacy. These findings highlight the importance of perceived professional competence and selected sociodemographic factors in relation to coaches’ psychological well-being. However, given the cross-sectional design, causal inferences cannot be made, and further longitudinal research is needed to clarify the direction of these associations.

Authors’ contributions

LE, SGA, and MEA: Conceptualization, Methodology, Software, Writing-Original draft preparation, Writing-Reviewing and Editing, Critical Review, and data collection. MEA: Data analysis. LE, SGA, and MEA: Writing-Original draft preparation, Writing-Reviewing and Editing. MEA: Critical Review. The final manuscript was reviewed and approved by all authors.

Funding

None.

Data availability

The datasets used and/or analyzed in this study are available upon reasonable request from the corresponding author.

Declarations

Ethics approval and consent to participate

All study procedures complied with the principles of the Declaration of Helsinki and received ethical approval from the Dokuz Eylul University Non-Interventional Research Ethics Committee (Decision date: 19 February 2025; Protocol no: 2025/06–08). Written/electronic informed consent was obtained from all participants before data collection.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.