Literature searches were performed using PubMed, Web of Science, PsychInfo, SportDiscuss, Scientific Electronic Library Online (SciELO), Literatura Latino-Americana e do Caribe em Ciências da Saúde (Lilacs) and Periódico Capes databases. All searches were conducted in December 2021 using combinations of the following keywords “Sport Motivation Scale II” in English, Portuguese and Spanish. We did not restrict the date of publication as an inclusion or exclusion criteria. The systematic review was performed following the Prisma guideline. Two of the authors (JB and AS) independently analyzed study titles and abstracts. Subsequently, they read the full text of studies deemed potentially eligible for inclusion. Furthermore, reference lists of articles were reviewed to determine the relevance of any additional studies on the topic.

Studies had to meet the following criteria for inclusion in our review: 1) used the SMS-II as an instrument for data collection; 2) was published in a peer reviewed journal; 3) collected and analyzed the six constructs of the SMS-II questionnaire. As an exclusion criterion we did not analyze studies (i) published in books or conference annals, (ii) that used an adapted version of the SMS-II for the physical education context.

Initially, 134 studies were identified and after deduplication, 24 were excluded from the reviewing processes and 71 were excluded after title and/or abstract analysis. In total, 39 studies were selected for the full text reading. Two authors read these studies and, according to the inclusion and exclusion criteria, 28 studies were selected for our review.

We included in this review studies with different methods, such as validation and translation, intervention, and observation. Therefore, we could not find a specific tool that could evaluate the quality of the studies with such different methodologies. We report the impact factor of their scientific journals as an indication of their quality (Table 1).

Two independent researchers extracted data from each study. Variables were collected according two categories: methods and results. For the methods section we collected the following variables: number of participants, age, sex, experimental design, and sport. In the results section, we collected the absolute fit indexes, such as Chi-square (χ²), Akaike information criterion (AIC), Browne-Cudeck criterion (BCC), Bayes information criterion (BIC), Consistent AIC (CAIC) and Expected cross validation index (EVCI). The Comparative fit indexes were also collected, such as Normed fit index (NFI), Incremental fit index (IFI), Tucker-Lewis index (TLI), Comparative fit index (CFI) and Relative noncentrality fit index (RNI). The collected Parsimonious fit indexes were: Parsimony-adjusted NFI (PNFI), Parsimony-adjusted CFI (PCFI) and Parsimony-adjusted GFI (PGFI). For other indexes we found Goodness-of-fit index (GFI) and Adjusted GFI (AGFI), Root mean square residual (RMR), Standardized RMR (SRMR), Weighted root mean residual (WRMR) and Root mean square error of approximation (RMSEA). The parameters collected for internal reliability were Cronbach’s alpha and McDonald’s omega.

The descriptive statistics were used to summarize and present the collected data. Absolute and relative frequencies were used for that purpose. The one sample t-test was used to compare the collected Cronbach’s alpha of each dimension with an expected alpha of 0.7 considered as a good reliability (Agbo, 2010; Hayes & Coutts, 2020). The Cronbach’s alpha was the main reliability measure presented in the studies and, for that reason, was used in our analysis. Other parameters of validity and reliability are also presented in the results section. All the analysis was performed using SPSS software and the statistical significance was set at p < 0.05.

Table 1 shows the characteristics of the included studies in the current systematic review. Most of the studies were published in the year of 2018 and presents an exploratory design.

We found four studies that involved the translation and validation the SMS-II in different countries. The psychometric properties of each questionnaire are shown in Table 2. We presented the indexes that were reported at least for two studies and used the following recommended cutoff: GFI >0.9, RMSEA <0.05, TLI and CFI >0.9, and SRMR <0.09 (Schreiber, 2017). All the studies showed high values of validity.

Although all studies showed a good psychometric property, the study of Viciana et al. (2017) validated the instrument with only five dimensions for the Spanish context. The authors compared four models employing the maximum likelihood estimation (with six, five, three, and two factors). The five-factor S-SMS-II, without the introjected regulation, showed the best indexes of fit (Table 2). Similarly, the study of Pineda-Espejel et al. (2016) analyzed the psychometric properties of the Mexican Spanish version of the SMS-II. The authors excluded item 16 of the questionnaire given its low reliability. It is worth noting that this item is part of the introjected regulation.

It is important to note that the psychometric properties were mainly presented by the studies that translated and validated the instrument. Among the studies that used the SMS-II only for data collection, few of them calculated and presented the instrument’s properties (Table 3). The Cronbach’s alpha was the most used measure among them (Table 3).

Three studies cited the Cronbach’s alpha reported by Pelletier et al. (2013) in the original study, but did not present the reliability measure of their own data (Gonzalez, 2021; Piasecka et al., 2018; Russell & Molina, 2018). Two studies presented a general alpha based on all domains and did not discriminated by each dimension (Bingol & Yildiz, 2021; Fiorese et al., 2017; Lucas et al., 2020; Roberto et al., 2020). Four studies did not report the alpha based on Pelletier et al. (2013) neither based on their data (Castro et al., 2020; Jordalen, 2016; Jordalen et al., 2018; Menegassi et al., 2018; Stenling et al., 2015; Wahl-Alexander & Chomentowski, 2018).

Nine studies calculated and reported the alpha for each dimension. The frequency of dimensions that showed alpha above or below 0.7 are presented in Table 4. We found that most of the dimensions presented alpha above 0.7. However, most of the studies showed a value below 0.7 for the introjected regulation.