An Examination of Athletic Trainers' Occupational Recovery Experiences During Time After Work

We used a cross-sectional survey design. To collect data from the collegiate work setting, we distributed electronic links to the Web-based surveys via e-mail to full-time ATs at NCAA institutions. Demographic survey items included self-identified gender, years of experience, work setting, salary, and hours worked per week. Follow-up reminder e-mails were sent 10 and 20 days after the initial invitation to participate. Recruitment was limited to the Midwestern region of the United States. From consent to survey completion, ATs dedicated approximately 5 to 7 minutes to study participation.

Participants were 144 ATs working in a variety of NCAA sports. Their demographic characteristics are displayed in Table 1. A survey response rate of 31.7% was achieved. All participants provided written informed consent, and the study was approved by the Drake University Institutional Review Board.

To assess stress, we administered the 10-item Perceived Stress Scale (PSS).^16,17  The PSS items are scored on a 5-point Likert scale, ranging from 0 (never) to 4 (very often), with stress scores computed as the sum of the item responses. The reliability (α = .84) and validity (⁠\(\def\upalpha{\unicode[Times]{x3B1}}\)\(\def\upbeta{\unicode[Times]{x3B2}}\)\(\def\upgamma{\unicode[Times]{x3B3}}\)\(\def\updelta{\unicode[Times]{x3B4}}\)\(\def\upvarepsilon{\unicode[Times]{x3B5}}\)\(\def\upzeta{\unicode[Times]{x3B6}}\)\(\def\upeta{\unicode[Times]{x3B7}}\)\(\def\uptheta{\unicode[Times]{x3B8}}\)\(\def\upiota{\unicode[Times]{x3B9}}\)\(\def\upkappa{\unicode[Times]{x3BA}}\)\(\def\uplambda{\unicode[Times]{x3BB}}\)\(\def\upmu{\unicode[Times]{x3BC}}\)\(\def\upnu{\unicode[Times]{x3BD}}\)\(\def\upxi{\unicode[Times]{x3BE}}\)\(\def\upomicron{\unicode[Times]{x3BF}}\)\(\def\uppi{\unicode[Times]{x3C0}}\)\(\def\uprho{\unicode[Times]{x3C1}}\)\(\def\upsigma{\unicode[Times]{x3C3}}\)\(\def\uptau{\unicode[Times]{x3C4}}\)\(\def\upupsilon{\unicode[Times]{x3C5}}\)\(\def\upphi{\unicode[Times]{x3C6}}\)\(\def\upchi{\unicode[Times]{x3C7}}\)\(\def\uppsy{\unicode[Times]{x3C8}}\)\(\def\upomega{\unicode[Times]{x3C9}}\)\(\def\bialpha{\boldsymbol{\alpha}}\)\(\def\bibeta{\boldsymbol{\beta}}\)\(\def\bigamma{\boldsymbol{\gamma}}\)\(\def\bidelta{\boldsymbol{\delta}}\)\(\def\bivarepsilon{\boldsymbol{\varepsilon}}\)\(\def\bizeta{\boldsymbol{\zeta}}\)\(\def\bieta{\boldsymbol{\eta}}\)\(\def\bitheta{\boldsymbol{\theta}}\)\(\def\biiota{\boldsymbol{\iota}}\)\(\def\bikappa{\boldsymbol{\kappa}}\)\(\def\bilambda{\boldsymbol{\lambda}}\)\(\def\bimu{\boldsymbol{\mu}}\)\(\def\binu{\boldsymbol{\nu}}\)\(\def\bixi{\boldsymbol{\xi}}\)\(\def\biomicron{\boldsymbol{\micron}}\)\(\def\bipi{\boldsymbol{\pi}}\)\(\def\birho{\boldsymbol{\rho}}\)\(\def\bisigma{\boldsymbol{\sigma}}\)\(\def\bitau{\boldsymbol{\tau}}\)\(\def\biupsilon{\boldsymbol{\upsilon}}\)\(\def\biphi{\boldsymbol{\phi}}\)\(\def\bichi{\boldsymbol{\chi}}\)\(\def\bipsy{\boldsymbol{\psy}}\)\(\def\biomega{\boldsymbol{\omega}}\)\(\def\bupalpha{\bf{\alpha}}\)\(\def\bupbeta{\bf{\beta}}\)\(\def\bupgamma{\bf{\gamma}}\)\(\def\bupdelta{\bf{\delta}}\)\(\def\bupvarepsilon{\bf{\varepsilon}}\)\(\def\bupzeta{\bf{\zeta}}\)\(\def\bupeta{\bf{\eta}}\)\(\def\buptheta{\bf{\theta}}\)\(\def\bupiota{\bf{\iota}}\)\(\def\bupkappa{\bf{\kappa}}\)\(\def\buplambda{\bf{\lambda}}\)\(\def\bupmu{\bf{\mu}}\)\(\def\bupnu{\bf{\nu}}\)\(\def\bupxi{\bf{\xi}}\)\(\def\bupomicron{\bf{\micron}}\)\(\def\buppi{\bf{\pi}}\)\(\def\buprho{\bf{\rho}}\)\(\def\bupsigma{\bf{\sigma}}\)\(\def\buptau{\bf{\tau}}\)\(\def\bupupsilon{\bf{\upsilon}}\)\(\def\bupphi{\bf{\phi}}\)\(\def\bupchi{\bf{\chi}}\)\(\def\buppsy{\bf{\psy}}\)\(\def\bupomega{\bf{\omega}}\)\(\def\bGamma{\bf{\Gamma}}\)\(\def\bDelta{\bf{\Delta}}\)\(\def\bTheta{\bf{\Theta}}\)\(\def\bLambda{\bf{\Lambda}}\)\(\def\bXi{\bf{\Xi}}\)\(\def\bPi{\bf{\Pi}}\)\(\def\bSigma{\bf{\Sigma}}\)\(\def\bPhi{\bf{\Phi}}\)\(\def\bPsi{\bf{\Psi}}\)\(\def\bOmega{\bf{\Omega}}\)\({\rm{\chi }}_{35}^2\) = 898.945, P < .001, root mean square error of approximation [RMSEA] = 0.141, [90% confidence interval {CI} = 0.134, 0.150], comparative fit index [CFI] = 0.932, Tucker-Lewis index [TLI] = 0.913) of the measure for use in adult populations have been established.¹⁷  Whereas Taylor¹⁷  concluded that the unidimensional model of the PSS may be inferior to the multidimensional model, model-fit indices generated from our data supported the adequacy of the unidimensional model (⁠\({\rm{\chi }}_{35}^2\) = 73.573, P < .001, RMSEA = 0.087 [90% CI = 0.059, 0.115], standardized root mean square residual [SRMR] = 0.062, CFI = 0.900, TLI = 0.872). Furthermore, Cohen and Janicki-Deverts¹⁸  also provided support for unidimensional model construct validity in US adult populations.

To assess the recovery experience, we administered the 16-item REQ.¹⁴  The REQ items are scored on a 5-point Likert scale, ranging from 1 (I do not agree at all) to 5 (I fully agree), with each of the 4 subscale scores (ie, psychological detachment, mastery, control, relaxation) calculated as an average of 4 items. For each of the 16 items, the stem was “During time after work. . .” Sample items included “. . .I forget about work” (psychological detachment), “. . .I seek out intellectual challenges” (mastery), “. . .I decide my own schedule” (control), and “. . .I kick back and relax” (relaxation). All 16 items were listed and fully described in the original validation study.¹⁴  The reliability (Cronbach α range = 0.79–0.85) and validity (⁠\({\rm{\chi }}_{98}^2\) = 403.60, RMSEA = 0.08, SRMR = 0.05, CFI = 0.96, nonnormed fit index = 0.96) of the measure for use in adult populations have been established.¹⁴ 

To examine the validity of the REQ measurement structure, we performed a confirmatory factor analysis (CFA) using Mplus software (version 8.0; Muthén and Muthén, Los Angeles, CA).¹⁹  The measurement model was defined as 4 related latent variables (ie, psychological detachment, mastery, control, relaxation) explained by 4 items per factor. The covariance matrix was analyzed using maximum likelihood estimation procedures (Table 2). Simulation studies^20,21  have demonstrated acceptable model rejection rates for maximum likelihood estimation methods when items are evaluated in 5 or more categories, the data are normally distributed, and adequate sample size is achieved. Unstandardized loadings for 1 item from each latent variable were constrained to 1.0 to define units within the model tested.

A χ² test of fit, residuals-based indices (ie, RMSEA, SRMR), and incremental fit indices (ie, CFI, TLI) were calculated and reported to determine the goodness of model fit. Model-fit indices were assessed collectively relative to recommended cutoff values.^22–25  Excellent model-fit indices were designated as follows: χ² test statistic that was not different, RMSEA < 0.050, SRMR < 0.050, CFI > 0.950, and TLI > 0.950. Minimally acceptable model-fit indices were as follows: χ² test statistic that was not different, RMSEA < 0.080, SRMR < 0.060, CFI > 0.900, and TLI > 0.900.^22,25 

To verify the reliability of each REQ factor, McDonald Ω coefficients were computed from standardized factor loadings and residual variances.²⁶  We considered Ω coefficients of >0.70 to be acceptable; >0.80, very good; and >0.90, excellent.²⁷ 

Descriptive statistics and bivariate correlations were computed using SPSS (version 25; IBM Corp, Armonk, NY). Correlation coefficients with magnitudes of <0.30 were interpreted as weak; between 0.30 and 0.70, moderate; and >0.70, strong. Data were also reviewed to confirm assumptions of univariate normality (ie, significance of skewness and kurtosis, visual analysis of histograms).

The 4-factor measurement model demonstrated good fit (⁠\({\rm{\chi }}_{98}^2\) = 151.613, P < .001, RMSEA = 0.062 [90% CI = 0.041, 0.080], SRMR = 0.069, CFI = 0.946, TLI = 0.934). The 4-factor model demonstrated an equivalent fit to the hierarchical model (⁠\({\rm{\chi }}_{100}^2\) = 155.499, P < .001, RMSEA = 0.062 [90% CI = 0.042, 0.081], SRMR = 0.076, CFI = 0.944, TLI = 0.933). Therefore, the 4-factor model was deemed the more parsimonious model. Parameter estimates of factor loadings and residual variances are presented in Table 3. The REQ factors were deemed reliable scales of measurement based on computed McDonald Ω coefficients (Table 3). Correlations between latent variable factors are shown in Table 4.

Descriptive statistics for the stress and recovery experience variables are provided in Table 5. To further describe the data, we present the recovery experience scores across relevant demographic variables in the Figure. Weak correlations were identified between stress and perceptions of psychological detachment (r = −0.314, P < .001), mastery (r = −0.179, P = .32), control (r = −0.284, P = .001), and relaxation (r = −0.157, P = .06).

The purpose of our study was to examine the reliability and validity of the REQ for use in AT populations. Results of the CFA indicated preliminary support for the reliability and validity of the REQ in AT populations. Perceived stress scores among ATs were slightly lower than the most recent normative data available among US adults.¹⁸  Results of the descriptive statistics indicated that ATs scored lowest in psychological detachment and highest in relaxation. All factors of the recovery experience were weakly and negatively correlated with perceptions of stress.

Compared with previous US-based studies conducted across a range of professions, ATs in our study scored slightly lower in psychological detachment and control but slightly higher in mastery and relaxation than other workers.^14,28–31  Collectively, our data indicated that ATs may be better able to engage in leisure or relaxation and learn or take on new challenges than they can psychologically distance themselves from work or display autonomy over their nonwork time. These data are consistent with previous reports^1,7,12  of ATs who perceived that they were constantly in work mode and had little control over their work schedules.

Our results supported the scoring of both the 4 REQ subscales and the overall recovery experience. Given the equivalence observed between the fit of the 4-factor model and hierarchical models, we recommend that researchers use subscale scores as opposed to a general recovery experience score alone. This recommendation is further reinforced by the descriptive statistics presented in the Figure and Table 5, which demonstrate substantial variability in the multifaceted nature of ATs' recovery experiences.

The correlations between stress and recovery experience variables in our study highlighted the relationship between perceived stress and recovery, as well as the construct distinction between them. Based solely on these correlation data, only 2.5% to 9.9% of the variance in recovery experience was explained by perceived stress. Whereas correlation provided a limited understanding of the influence of stress on the multifaceted recovery experience among ATs, we suggested that the mere removal of stressors from work or life, or both, might marginally affect the recovery experience. Furthermore, this reinforced the original premise for our study: that occupational recovery is a unique and worthwhile variable to consider in the AT work-life balance literature.

Our study had limitations that prompt specific directions for future research. First, our sample size was small relative to the best practices in CFA and other such structural equation modeling procedures. Insufficient sample sizes in structural equation models can lead to model errors or biased parameter estimation, or both.^27,32  The lack of missing data, lack of concerns regarding model convergence, and strength of the CFI and TLI model-fit indices were evidence that the sample size was sufficient for this preliminary analysis of validity. Second, our study design did not specifically account for perceptions of burnout, life satisfaction, or measures of work-life balance. Furthermore, data were not collected about the recovery activities pursued during nonwork time.¹³  Future research is needed to clarify the role of the recovery experience and recovery activities in preventing burnout and improving life satisfaction and work-life balance. Third, our study was limited to ATs working in the Midwestern region of the United States. In future studies, investigators should examine the recovery experiences of ATs working in other geographic regions of the United States, as well as different conferences.

Based on our data, occupational recovery appeared to be a construct relevant to the scholarly conversation surrounding AT stress and work-life balance. We provided preliminary evidence to support the reliability and validity of the REQ measure for use in future athletic training research. Our data suggested that an NCAA AT's recovery experience may be unique compared with other professions in the United States but may be quite similar to that of ATs in various work settings, regardless of personal or occupational factors.