Physical activity is a well-documented method to reduce a number of diseases such as cardiovascular disease (Carnethon, 2009) and a widening variety of chronic diseases, including diabetes mellitus (Colberg et al., 2010; Burr et al., 2012b), cancer (Shi et al., 2015), obesity (Lakka and Bouchard, 2005), hypertension (Diaz and Shimbo, 2013), bone and joint diseases (Burr et al., 2012a). Physical activity is also a protective factor or useful intervention to reduce psychiatric symptoms in cognitive disorders (e.g., depression) (Lee et al., 2012) and improve cognitive functioning (Bamidis et al., 2014). Individuals of all ages, including healthy subjects, can achieve a number of benefits in different dimensions (physical, psychological, social, and emotional) through physical activity (Penedo and Dahn, 2005; Warburton et al., 2006; Vogel et al., 2009; Janssen and Leblanc, 2010). Nevertheless, Hallal et al. (2012) suggest that approximately 31% of adults and 80% of adolescents worldwide do not reach recommended levels of daily physical activity. Moreover, studies such as Lee et al. (2012) demonstrated that if inactivity decreased by 10%, half a million deaths could be prevented every year. Therefore, it seems crucial to understanding why some people are active and others are not.

Although the underlying causes of why some people are active and others not are highly complex, motivation is a possible explanatory factor. Motivation is a key process that influences individuals’ initiation and maintenance of behavior (Murcia et al., 2007a). The Self-Determination Theory (SDT) seems to be the most contemporary framework used to understand physical activity motivation and adherence (Hagger and Chatzisarantis, 2008; Molanorouzi et al., 2015). Based on SDT, motivation to engage and adherence to physical activity need to be distinguished, and the most basic distinction is between intrinsic and extrinsic motivation (Ryan and Deci, 2000a). Intrinsic motivation refers to doing something because it is inherently interesting or enjoyable (Ryan and Deci, 2000a,b). For instance, a Taekwondo practitioner who practices for enjoyment and challenge involved in the sport is said to be intrinsically motivated (Ryan et al., 1997). In the other hand, extrinsic motivation refers to engaging in an activity for instrumental reasons, such as rewards (Ryan and Deci, 2000a,b). For example, an aerobics practitioner who practices your activity for improving his appearance is said to be extrinsically motivated (Ryan et al., 1997).

In order to understand the motives for physical activity, Ryan et al. (1997) developed a scale named Motives for Physical Activity Measure-Revised (MPAM-R) where used as the theoretical background of the SDT. The scale was composed by 30 items divided into five factors: (1) Fitness; (2) Appearance; (3) Competence/Challenge; (4) Social; and (5) Enjoyment. In summary, the original scale showed acceptable psychometric properties by Exploratory Factor Analysis (EFA) and Cronbach’s alpha ranging between 0.78 to 0.92.

To the best of our knowledge, this scale was translated and cross-culturally adapted in three studies and two languages. Celis–Merchán (2006) performed the adaptation of the Spanish version of the MPAM-R on a sample of 120 Colombian subjects. Cronbach’s alpha of the Celis–Merchán (2006) study range between 0.75 to 0.91. In addition, the authors performed an EFA that showed several problems (Celis–Merchán, 2006). Murcia et al. (2007b) adapted MPAM-R for the Spanish population in a large sample size (464 subjects). The final version of the scale purpose by the Murcia et al. (2007b) was composed by the 28 items with acceptable results by the EFA. In addition, the Cronbach’s alpha range between 0.80 to 0.87. Finally, Gonçalves and Alchieri (2010) performed the cross-cultural adaptation and validation of the scale for Brazilian Portuguese version. The results showed that the 26 items were not acceptable using Confirmatory Factor Analysis (CFA) standards [χ²₍₂₈₉₎ = 757.75; GFI = 0.83; AGFI = 0.80, and RMSE = 0.07], although the authors considered that the values were acceptable. In addition, Gonçalves and Alchieri (2010) results showed acceptable values of the Cronbach’s alpha (between 0.75 to 0.88). Therefore, results around MPAM-R Brazilian version suggest the importance of psychometric review.

Moreover, in all MPAM-R versions, the factor scores are calculated by the mean of scores by items of the factor (for example, mean of the scores of the items 6,15,21,28, and 30 for social factor in MPAM-R). Although, this method is largely used in Sports Psychology Scales (Lonsdale et al., 2008; Pelletier et al., 2013) there are several other methods (for more details, see DiStefano et al., 2009), that can be interesting. One option would be the weighted method (DiStefano et al., 2009), that used the factor loading extracted by the EFA or CFA. In the common method (mean of scores by items), all items on a factor are given equal weight, regardless of the factor loading value extracted in EFA or CFA. In summary, the common method does not involve item loading values, thereby disregarding the strength (or weight) for each item. Therefore, items with relatively low loading values are given the same weight in the factor score as items with higher loading values. On the another hand, in the weighted method, the factor score is based on factor loading extracted by EFA or CFA, to each item. Therefore, one advantage this method is that items with higher factor loadings have greater weight in the factor score, as items with lower factor loadings have less weight have less weight in the factor score (DiStefano et al., 2009).

Thus, the present study aims to perform a new translation and cultural adaptation of the MPAM-R for the Portuguese languages and analyze its psychometric properties. In addition, verify the agreement between common (mean by items) and weighted methods to calculate factor scores.

The aim of this study was to perform a cultural adaptation of the MPAM-R for the Portuguese languages, analyze its psychometric properties and compare the factor loadings calculated by mean and by weighted method. In summary, although the Portuguese version of the MPAM-R has fewer items when compared with the original version (Ryan et al., 1997), our version had acceptable psychometric proprieties and internal consistency. In addition, our results showed that the factor score calculated by mean and by the weighted method are different in three of the five factors.

In general, most of the sports psychology scales were developed in English-speaking countries, so cross-cultural and international collaborative studies, as well as the possibility of testing theories are needed. Thus, researchers need reliable and valid instruments in other languages. Nowadays, there are well-established methodological approaches for translating, adapting and validating instruments (Beaton et al., 2000; Sperber, 2004). Although, there was no clear consensus on how each method should be used. To the best of our knowledge, the MPAM-R was adapted to Spanish (Celis–Merchán, 2006; Murcia et al., 2007b) and Portuguese (Gonçalves and Alchieri, 2010). In general, the translation process used by those of studies was quite mixed, being more (Gonçalves and Alchieri, 2010) or less rigorous (Celis–Merchán, 2006). Thus, ours and Gonçalves and Alchieri (2010) studies have used the most rigorous approaches for translating the MPAM-R.

EFA and CFA are widely used multivariate statistical procedures that serve as tool to scale validation. EFA was used in Spanish versions (Celis–Merchán, 2006; Murcia et al., 2007b). However, EFA in Celis–Merchán (2006) study showed many problems. Instead, EFA showed acceptable results in the final version of the scale purpose by the Murcia et al. (2007b). Although the use of EFA is an interesting and useful method in scales validation, it is important to note that none of the Spanish studies performed an assessment of model fit. Model fit is obtained through the several statistical tests used to determine how well the model fits to the data (Brown, 2006; Kline, 2011) and it is considered a robust statistical technique to validate scales. Unlike the Spanish versions (Celis–Merchán, 2006; Murcia et al., 2007b), Gonçalves and Alchieri (2010) performed a CFA and reported model fit. However, using the recommendations indicated by some authors (Brown, 2006; Kline, 2011), the results showed that model fit was not acceptable [χ²₍₂₈₉₎ = 757.75; GFI = 0.83; AGFI = 0.80, and RMSE = 0.07], although the authors considered acceptable. A similar model fit as also found in another study (Wilson et al., 2002) that aimed to verify the psychometric properties of the MPAM-R. Meanwhile the authors considered that the values were not acceptable. Thus, our study used rigorous statistical approaches and the final version of the MPAM-R indicated that all parameters recommended for an excellent model fit were found [X²(205) = 414.672; p < 0.001; CFI = 0.98; TLI = 0.97; RMSEA = 0.058; SRMR = 0.03].

Reducing the number of items is a common practice among translation and cross-cultural adaptation of scales. Especially in MPAM-R, Murcia et al.’s (2007b) and Gonçalves and Alchieri’s (2010) studies had final versions with 28 and 26 items, respectively. In the present study, two items were excluded due to factor loading higher than 0.30 in two dimensions (8: “to improve existing skills” and 25: “to get better at activity”); and other two items were excluded because showed factor loading lower than 0.30 in dimension that were conceptually contemplated by the original model (1: “want to be physically fit” and 12: “to keep up current skill level”). Although the P-MPAM-R has a smaller number of items than the original scale, the psychometric properties showed good values. Furthermore, the smaller the scale is, the shorter will be the time for your application, which ease data collection process. Moreover, as in other studies (Celis–Merchán, 2006; Murcia et al., 2007b; Gonçalves and Alchieri, 2010), the final version of the scale presented the same structure of five factors (enjoying, competence, appearance, fitness, and social) found in the original study (Ryan et al., 1997).

Regarding reliability, Cronbach’s alpha is a generalization of the internal consistency reliability coefficient (Field, 2009). As can be seen in Table 6 all versions (Ryan et al., 1997; Murcia et al., 2007b; Gonçalves and Alchieri, 2010), including the present study, showed higher values than recommended (>0.70).

In the Brazilian context, other scales can be used to investigate the motives to physical activity, since that the most used is IMPRAFE (Motivation to Regular Physical Activity Inventory (in Portuguese “Inventário de Motivação para a Prática de Atividade Física”) (Barbosa, 2006). The scales were composed by some factors named: stress control, social, competence, appearance, enjoyment and fitness. Barbosa (2006) conducted the validation process of the IMPRAFE with 126 items and its short version with 48 items. As in Gonçalves and Alchieri (2010) study the results showed that in 120 items version of the IMPRAFE, the model fit was not acceptable [GFI = 0.859, AGFI = 0.854 and RMS = 0.065]. However, the short version with 48 items showed satisfactory [GFI = 0.952, AGFI = 0.948 and RMS = 0.058]. In both versions, the IMPRAFE showed Cronbach’s alpha higher than 0.70. In summary, the IMPRAFE-short version (Barbosa, 2006) can be an excellent scale to investigate Motivation for Physical Activity in a Brazilian context. However, the fact that was not translated or adapted to other contexts (for example, English) does not allow cross-cultural and international collaborative studies. In addition, the fewer of items of MPAM-R reduces application time to IMPRAFE, which facilitate data collection process.

Overall, the present study compared two ways to calculate the factor score. In general, factor scores are calculated by the mean of scores by items of the factor. Therefore, when performing this procedure it is suggested that all items have the same weight for factor calculation. However, in EFA or CFA, the factor loadings show that some items load more than others in the same latent variable. For this reason, we believe that the current way factor scores are calculated needs to be reviewed, though being largely used. In the present study, we used an adaptation of the weighted method (DiStefano et al., 2009), that used the factor loading extracted by the EFA to create weights for each item of the factor. Our results showed that when compared two methods to calculate factors scores, only “Enjoyment” and “Appearance” factors showed agreement between methods to calculate factor scores. On the other hand, Paired t-test found that all others comparisons were significant (p < 0.001). Hence, values calculated by the two methods exhibit statistical significant differences. In addition, disagreements (see Figure 3) of this factors (“Competence”; “Fitness”; and “Social”) do not present a pattern (e.g., overestimation or underestimation the values). Thus, it is possible to consider that the theoretical changes in the way of calculating the factor score modified the results. Basically, we considered that the weighted method proposed in the present study presents a more appropriate theoretical background. For this reason, we believe that the weighted method can be a more appropriate alternative to calculating factor scores. Although even more refined methods exist (for more details, see DiStefano et al., 2009) we chose to adapt the weighted method because it enables the comparison between the factors since the units of measure are the same as the original scale (variation between 1 and 7). In conclusion, we believed that our not much common weighted method could be an alternative to calculating factor scores, once it takes into account the factor loading of the item extracted by the Factorial Analysis.

Our study suffers from some limitations. Firstly, reliability, in a practical definition, is the ability of an instrument to measure consistently (Hopkins, 2000), and it is closely associated with its validity (Tavakol and Dennick, 2011). In the present study, we used the Cronbach’s alpha as the reliability measure, however, this measure has received a lot of criticism (Sijtsma, 2009). Although understanding the limitations to the Cronbach’s alpha (Sijtsma, 2009), and that other methods could have been used (Hopkins, 2000), we chose it because it is probably the most used reliability measure in scales validation, as well as, all other studies (Celis–Merchán, 2006; Murcia et al., 2007b; Gonçalves and Alchieri, 2010), including the original scale (Ryan et al., 1997), also used this reliability measure. In the end, it was not possible to test our final model proposed by the EFA with a new CFA in a different and large sample, however, this may be a proposal for further study.

The present study provides evidence of validity and excellent psychometrics properties of the Portuguese version of the MPAM-R in which 7-items measure enjoyment, 4-items measure competence, 6-items measure appearance, 4-items measure fitness, and 5-items measure social. Therefore, the Portuguese version of the MPAM-R may be used appropriately and successfully to measure the motives for physical activity. In addition, we showed another way to calculate factor score named weighted method that is different from the common method (mean of the score), but can be a more appropriate alternative to calculate factor scores.

MA, ML, JdP, and VdC designed the study and analyzed data. LF collected the data. All authors participated in the interpretation of the data. MA drafted the manuscript. All authors helped with the writing of the manuscript. All authors approved the final version of the manuscript submitted for publication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.