Validity and usefulness of the student-athletes’ motivation toward sport and academics questionnaire: a Bayesian multilevel approach

Methods

Results

The SAMSAQ-PT presented a three-factor structure (Table 1). Due to low factor loadings (< 3), items 11 (2.74), 18 (1.92), 25 (2.60), and 30 (1.92) were excluded. In particular, seven items loaded the factor named “Sport Motivation” (SM), sixteen items loaded the factor named “Academic Motivation” (AM), and three items loaded the factor named “Career Motivation” (CM).

Discussion

Considering the translation and psychometric structure of the SAMSAQ-PT, we observed a three-factor structure. This structure is in line with the original American and the versions validated for Emirati, European, and Korean student-athletes (Gaston-Gayles, 2005; Fortes et al., 2010; Lupo et al., 2015; Park, Hong & Lee, 2015). In particular, SM items express the desire for sports performance and for continuing sports career (Keshtidar & Behzadnia, 2017), which is influenced by cultural and educational context for pursuing a sports performance (Fernandes, Moreira & Gonçalves, 2019; Lupo et al., 2015); the AM items mirror the academic commitment of student-athletes, measuring the Brazilian student-athletes’ engagement in an educational/ vocational path concerning a sport or a dual-career path (Cartigny et al., 2019), probably considering the academic degree relevant for future jobs and professional careers (Fortes et al., 2010); and the CM items present a desire for developing a professional sports career (Gaston-Gayles, 2005). Thus, it could be speculated that the Brazilian student-athletes may perceive sport and academic commitments separately.

Methods

Results

In the BCFA, items 1 (0.28), 10 (0.45), 12 (0.47), 14 (0.25) and 21 (0.38) presented posterior values lower than 0.5 and were excluded. Additionally, this first model did not present good evidence of model fit (BRMSEA = 0.08; BGammaHat = 0.87; adjBgammahat = 0.84; BMc = 0.43). The new model (21 items) was further tested and four items [items 5 (0.30), 17 (0.35), 24 (0.33), and 29 (0.39)] with low factor load values were excluded. In this model, fit indexes were superior, but not well adequate (BRMSEA = 0.07; BGammaHat = 0.93; adjBgammahat = 0.89; BMc = 0.67). Thus, the resulting 17-item model presented factor loadings close or above 0.5 for all items (Table S2, available at https://osf.io/cpwdv/). Only item 13 presented factor loading (0.49) bellow 0.5. Thus, the item was retained. Additionally, fit indexes showed satisfactory adequacy (BRMSEA = 0.06; BGammaHat = 0.96; adjBgammahat = 0.93; BMc = 0.84).

Discussion

In this study, we tested the factor structure of SAMSAQ-PT, which emerged from study one by applying BCFA. Our observations confirmed the three-factor structure but indicated a better structure composed of 17 items for SM (7 items), AM (7 items), and CM (3 items). The confirmatory factor analysis of the SAMSAQ-PT substantiated the three-factor structure (Fortes et al., 2010; Gaston-Gayles, 2005; Guidotti et al., 2013; Lupo et al., 2015; Park, Hong & Lee, 2015), even though to avoid misinterpretations, the Bayesian inference based on strong similarities determined the substantial reduction of the item’s distribution. Different factor structures can be expected due to dual-career policies and social-cultural contexts, as well as to the sport and educational levels of student-athletes (Fortes et al., 2010; Gaston-Gayles, 2005; Guidotti et al., 2013; Lupo et al., 2015; Lupo et al., 2017; Park, Hong & Lee, 2015; Stambulova & Alfermann, 2009). Given the Brazilian miscegenation and the potential influence of European and other cultures on Brazilian student-athletes, both similarities and differences with other cultures were expected.

Regarding the similarities, the AM dimension presented stable comparing to other SAMSAQ validations (Gaston-Gayles, 2005; Guidotti et al., 2013; Lupo et al., 2015; Lupo et al., 2017; Park, Hong & Lee, 2015), although in the present version we had a higher reduction of items. The results may indicate that the academic motivation of higher education student-athletes is a drive based on similar purposes, such as learning (item 3), get good grades (item 4), and the course content (item 28). There were divergences, such as item 16 (the achievement of a degree is important to enrich my knowledge) had loaded in a different dimension than the academic motivation (Lupo et al., 2017). However, this item has a clear academic motivation.

Our results were consistent with the other validations in the SM dimension, but there is less agreement about the items from this dimension across the validation studies (Lupo et al., 2015; Lupo et al., 2017; Gaston-Gayles, 2005; Park, Hong & Lee, 2015). The feeling of being better than other athletes (item 13) is an example of how it can vary between the validations. It represents a sport and career motivation (Lupo et al., 2017), sport motivation (federation structure; Lupo et al., 2015; Gaston-Gayles, 2005), or even academic motivation (Park, Hong & Lee, 2015). In summary, there was a shift of items in SM and CM dimensions in the factor structure of the SAMSAQ-PT. The items with explicit statements focused on professional or athletic careers loaded in the CM. The results suggest that Brazilian student-athletes interpret their motivation for a high level of performance or athletic career as a priority.

Participants

The whole sample of 862 student-athletes (females: 56.8%; males: 43.4%) enrolled in public (n = 401) and private (n = 461) Brazilian universities and competing at international (n = 59), national (n = 270), state (n = 160) and university (n = 373) levels participated in this study.

Data analysis

Multilevel regression models were used to estimate SAMSAQ-PT and its dimensions among Brazilian student-athletes when grouped by sex (e.g., female and male), sports level (e.g., international, national, state and university), type of university (e.g., public and private) and student-athletes’ status (e.g., yes and no). The multilevel models estimate the individual scores relying on the available information of individuals characteristics and using additional “random” predictors such as group or context characteristics (for individual i, with indexes, s, a, l, and u for, sex, student-athlete status, sport level, and type of university, respectively). In Bayesian terms, these “random” or “group-level” effects are related to each other by their grouping structure, and the individuals’ responses are partially pooled towards the group mean (Gelman & Hill, 2007), as follows: ${\displaystyle y_i={\beta }^0+{\alpha }_{s\left[i\right]}^{sex}+{\alpha }_{a\left[i\right]}^{student-athletestatus}+{\alpha }_{l\left[i\right]}^{sportlevel}+{\alpha }_{u\left[i\right]}^{typeofuniversity}+{ϵ}_i}$ ${\alpha }_{s\left[i\right]}^{sex}\sim N\left(0,{\sigma }_{sex}^2\right)$, for s = 1, 2.

${\alpha }_{a\left[i\right]}^{student-athletestatus}\sim N\left(0,{\sigma }_{student-athletestatus}^2\right)$, for a = 1, 2.

${\alpha }_{l\left[i\right]}^{sportlevel}\sim N\left(0,{\sigma }_{sportlevel}^2\right)$, for l = 1, 2, 3, 4.

${\alpha }_{u\left[i\right]}^{typeofuniversity}\sim N\left(0,{\sigma }_{typeofuniversity}^2\right)$), for u = 1, 2.

${ϵ}_i\sim N\left(0,{\sigma }_{y_i}^2\right)$)

Weakly informative prior distributions, normal prior (0, 10) for population-level effect (intercept), and normal priors (0,1) for group-level effects (i.e., the standard deviations of varying intercepts) to regularize the multilevel model estimations were used. Two chains for 4,000 iterations with a warm-up length of 1,000 iterations to ensure convergence of the Markov chains were run. To check the models and estimations, trace plots to examine the convergence of Markov chains and posterior predictive checks to validate the models were used (Gelman et al., 2013). The Bayesian multilevel models were fitted with the “brms” package (Bürkner, 2017), available as a package in the R statistical language (RCoreTeam, 2018). The brms package implements Bayesian multilevel models using the probabilistic language Stan (Carpenter et al., 2017). For computational and interpretation convenience, the outcomes were standardized (z-scores). This methodology is also described in previous studies (Quinaud et al., 2020; Quinaud et al., 2020b).

Results

Figures 1–3 present the standardized values of the SAMSAQ-PT’s dimensions related to the respondents’ sex, type of universities, sport levels, and student-athletes’ status. Male student-athletes presented higher effect scores for SM and CM dimensions. On the other hand, female student-athletes presented higher scores for AM, albeit with a small magnitude. Student-athletes from private universities had substantially higher scores than student-athletes from public universities for SM, AM, and CM dimensions, but at best, the magnitude was small for AM. Considering sport level variation, student-athletes who competed at higher level of performance showed higher scores than those of lower competitive levels for SM and CM dimensions. Lastly, there was no substantial variation in the motivation scores by student-athlete status. Table S3 (available at https://osf.io/cpwdv/) presents the SAMSAQ-PT estimates and uncertainty (90% confidence intervals).

Discussion

We tested the SAMSAQ-PT construct validity by exploring the variation in the motivation scores considering student-athletes’ individual, sport, and academic characteristics. There was substantial variation in the SAMSAQ-PT dimension scores by sex, university type, and sport level. The observed substantial gender-related variation in SM and CM could be due to lower dual career support for Brazilian female athletes and limited economic opportunities for top-level women’s sport (Harrison et al., 2020). Coherently, during the academic path, the highest effect scores for AM emerged for female student-athletes, who might emphasize more their academic career concerning their sports career (Tekavc, Wylleman & Cecić-Erpič, 2015). These findings urge implementing sex-related dual careers at the sport and public education level, as envisaged by the European-funded Collaborative Partnership “Dual Career for Women Athletes” (DONA). Consistent with previous studies (Guidotti et al., 2013; Lupo et al., 2017), the sport level resulted as a predictor of CM and SM scores. It seems reasonable to assume that student-athletes competing at a higher level might be more motivated to pursue their sports career compared to their pairs competing at lower levels, whereas it is plausible to assume that AM is linked to the academic motivation of the athlete (Gaston-Gayles, 2005; Guidotti et al., 2013; Lupo et al., 2015; Lupo et al., 2017).

To our knowledge, this is the first study investigating the student-athletes’ motivation related to university type (e.g., public and private). Indeed, the social context of universities shapes the opinions of student-athletes (Druckman et al., 2014). The present findings highlight that Brazilian student-athletes’ sports and education commitments vary between public and private universities. Concerning their peers from public universities, in the present models, student-athletes from private universities have a high probability of scoring higher values for CM and SM, probably due to dual-career financial and logistic support (Aquilina, 2013; Aquilina & Henry, 2010). Enrolling 75% of the total undergraduate students (Instituto Nacional de Estudos e Pesquisas Educacionais Anísio Teixeira, 2018), private Brazilian universities might employ marketing strategies incorporating student-athletes to increase the media attention leveraging the universities’ profile (Harrison et al., 2010; Teixeira, 2010). Unlike private universities, Brazilian public universities do not require tuition fees and offer academic undergraduate and graduate degrees (e.g., master’s and Ph.D.), opportunities for research, and internships, which might contribute to future professional advantages (Cecić Erpič, Wylleman & Zupančič, 2004). Thus, compared to their counterparts enrolled in private universities, student-athletes attending public universities could show lower SM when prioritizing their academic careers to prepare for future job opportunities (Amara et al., 2004; McKenna & Dunstan-Lewis, 2004). Additionally, the Brazilian university sports facilitating sub-elite athletes’ participation in public institutions are still in a developmental phase (Starepravo et al., 2010).

At the international level, there is a call for awareness of the role and responsibilities of dual-career actors (Condello et al., 2019), and the implementation of support policies for student-athletes is a priority of the European Union (Aquilina & Henry, 2010; European Commission, 2012; EuropeanParliament, 2015; European Parliament, 2017). Indeed, student-athletes should be informed about their rights, and the implementation of dual-career counseling has been strongly recommended to help them manage their academic and sports commitment (Hansen & Sackett, 1993; Martin, 2005; López de Subijana, Barriopedro & Conde, 2015). During the academic path, a recognized student-athlete status could allow the monitoring of the academic and sport progresses to individualize necessary dual-career support. Although this status has been considered crucial to influence substantially the athlete’s motivation to achieve an academic degree, especially for those determined to pursue a sports career, the present models do not support this hypothesis. Thus, it is likely needed to take a step back and review the conceptions of support based on interactions at and across environmental levels (Knight, Hardwood & Sellars, 2018).

Methods

Results

Predicted changes for SM, AM, and CM, after an academic year and adjusted for sex, type of university, sport level, and student-athletes’ status are summarized in Fig. 4. The influence of the individual and context characteristics was similar at the baseline and after an academic year. The post-stratified predictions for the total sample showed a trend of stability for all student-athlete motivation scores after an academic year, although there was a slight decrease in SM.

Discussion

We explored the student-athletes’ motivation scores’ sensitiveness across an academic year adjusted for sex, sport level, student-athlete status, and type of university. The results showed that student-athletes’ motivation was stable across an academic year. In addition to limited variations observed across an academic year, uncertainty estimates were narrow after adjusting for sex, type of university, sport level, and student-athletes’ status. These findings suggest that SAMSAQ-PT may be sensitive to track individuals changes over time. Although sports burnout was beyond the aim of the present study, decreased SM over time could be considered an indicator in interpreting dual career paths (Sorkkila et al., 2018), monitoring university dual-career social contexts, or highlighting the needs of the dual-career implementation. The present predictive models indicate that the student-athletes’ motivation could be influenced by the university context and how it cares about the requirements necessary to combine sport and academic commitments, especially when the academic requirements increase towards graduation. Based on the present longitudinal observations, the university sports contexts appear not to motivate student-athletes in their dual-career. Future studies focusing on support dual-career policies could contribute to a sound interpretation of how institutions and managers deal with dual-career needs and services.

While in some countries, a well-systematized relationship between education and high-performance sport seems to determine the high motivation for an athletic career (Fortes et al., 2010; Simons, Van Rheenen & Covington, 1999), in European countries with distinct academic and sports systems, athletes are confronted with the choice of one of the two paths (European Commission, 2016). In Brazil, sports and educational systems are also distinct. In line with the development of university sports in Brazil, the awareness of the importance of supporting student-athletes dual-career increases (Guidotti, Cortis & Capranica, 2015; Stambulova & Wylleman, 2019). While the present results showed that student-athletes’ motivation does not vary substantially across an academic year, further longitudinal studies are needed to verify whether student-athletes are motivated towards a dual-career untill their graduation. Indeed, cooperation between the academic and sports sectors needs to be implemented for providing a supportive entourage for student-athletes.