According to SDT, the reasons for engaging in physical activity can be classified along a self-determined continuum (Ryan & Deci, 2017) between intrinsic motivation, extrinsic motivation and amotivation. Intrinsic motivation implies engaging in an activity out of pleasure with a strong sense of volition. On the opposite end of the continuum, amotivation represents a lack of intention and therefore the absence of motivation. In between these two ends, extrinsic motivation refers to engaging in behaviours for external or instrumental outcomes rather than the participation itself. Further, with regard to extrinsic motivation SDT proposes four different types distinguishing the extent of self-determination (Ryan & Deci, 2002). The least self-determined type of extrinsic motivation is defined as externally regulated, which implies engaging in an activity for external means. In contrast, an introjected regulation refers to behaviours being relevant for self-esteem and driven by an inner demand; however, the reasons for engaging in the activity are only partially internalised. Identified regulated behaviours are more self-determined and judged as personally important and therefore relevant. The most self-determined extrinsic motivation involves an integrated regulation, which implies that reasons for engaging in an activity are considered part of the individual’s sense of self and are concordant with the individual’s value system (Deci & Ryan, 2000).

Ryan and Deci (2002) assume that activities that satisfy the three basic psychological needs for autonomy, competence and relatedness foster self-determined motivation. These basic psychological needs are defined as basic innate requirements regardless of an individual’s culture or age. Whereas the needs may vary in their expressions in different contexts, the core character remains unchanged. Accordingly, an educational environment created by the PE teacher that supports the needs of autonomy, competence and relatedness increases students’ need satisfaction (e.g. Ntoumanis & Standage, 2009). That is why SDT distinguishes between need support and need satisfaction. While need support refers to the social context, the conditions and the teacher’s strategies to support the basic psychological needs, need satisfaction refers to the individual’s feelings of being satisfied. An autonomy supportive teaching environment is characterized by providing choices for students, involving them in decisions in a meaningful way and responding to their lives and interests. Students’ competence can be supported by the teacher e.g. by providing positive feedback as well as challenging activities and directions to perform behaviours that fit the developmental level of the children. The teacher supports students’ feelings of relatedness by creating a context characterized by warmth, a sense of security and connection to others (Deci & Ryan, 2000). With regard to need satisfaction, the need of autonomy is satisfied if individuals perceive themselves as the origin of one’s own behaviour, i.e. performing according to one’s interests and innate values as a true expression of self. The need of competence is satisfied if individuals experience their own abilities and feel effective in interactions within the social environment. The need of relatedness is satisfied if individuals feel connected to others and have a secure sense of belongingness with them. Altogether, need support is expected to show positive relationships to self-determined types of motivation (intrinsic, integrated and identified) mediated by need satisfaction, and negative relationships to less self-determined motivations (introjected, external and amotivation; Deci & Ryan, 2000). The types of self-determined motivation will eventually lead to various positive cognitive, affective and behavioural outcomes (e.g. Ntoumanis & Standage, 2009).

While most findings with regard to the relationships between need support and types of motivation, mediated by the need satisfaction, refer to adolescents, only a few refer to younger children (for an overview see e.g. van den Berghe et al., 2014): A study amongst English 11-14-year-old students showed that the perceived PE teachers’ support for the students’ basic psychological needs (measured as one overall factor) predicted the students overall need satisfaction (Standage et al., 2005). The latter predicted positively intrinsic motivation and negatively external motivation and amotivation. In this study, the authors only examined relationships between the overall second-order factors need support and need satisfaction. They did not examine separate effects with regard to the three basic psychological needs. The relationship between an overall need support factor and students’ self-determined types of motivation was also confirmed with Estonian 12-16-year olds (Koka, 2013). The author further investigated the specific mediation effect of each of the three basic psychological needs. He found that only the satisfaction of competence (neither relatedness nor autonomy) was positively related to self-determined motivation. On the contrary, another study confirmed positive relationships between the satisfaction of all three basic psychological needs (three factors) and self-determined motivation reported by American upper-elementary students (8-11 years; Chen & Hypnar, 2015). The measurement of perceived need support was beyond the scope of this latter study. Contrarily, the importance of perceived need support was emphasized by a cross-cultural study which supported the positive relationship between perceived autonomy support and self-determined motivation for British, Greek and Singaporean samples (the only exception was a Polish sample; Hagger et al., 2005). Both Hagger et al. (2003) as well as Hagger et al. (2005), however, only concentrated on the support of autonomy and neither measured the support of competence nor relatedness.

Based on these empirical evidences it can be assumed that there is a positive relationship between the support of autonomy, competence and relatedness and self-determined types of motivation (intrinsic, identified) and a negative relationship between the three need support dimensions and more controlled types of motivation (introjected, external, amotivation). These relationships can be assumed to be mediated by the satisfaction of autonomy, competence and relatedness. However, as the prior studies were conducted primarily with older individuals, it is questionable if these relationships also apply for young children. Koka (2010) for example showed that for younger students (7^th grade) only the satisfaction of competence was significantly related to self-determined motivation, whereas for older students (12^th grade) competence, autonomy and relatedness were significantly related. Therefore, it is unsure if all theoretical relationships can be supported with a sample of young children. Beyond, the majority of the revised studies concentrate on single aspects in the assumed chain of constructs (e.g. support of autonomy), whereas only few studies examined all theoretical constructs addressed within SDT (three need support dimensions, three need satisfaction dimensions, types of motivation; e.g. Standage et al., 2005). In this regard, van den Berghe et al. (2014) provide a critical review of key findings on SDT in PE. Accordingly, educational research requires further investigations that distinguish between all of the three need support dimensions. Overall, it can therefore be stated that as a starting point suitable instruments that differentiate all aspects of SDT for younger students are necessary. To be able to do this, an instrument for younger students is needed that is able to differentiate all aspects of SDT – such an instrument, however, is not yet available, as will be shown in the following.

In the following, an overview of questionnaires (see Table 1) and their limitations with regard to the present research is presented. We first reflect on measures concentrating on need support, second on measures that concentrate on need satisfaction and we finish with measures concentrating on the types of motivation. We included German and English language instruments in the overview.

We found only a few instruments for children that cover at least one of the need support constructs. The English Sport Climate Questionnaire (SCQ) measures the support of autonomy (and nothing else) in sports training and has been used with children from the age of 13; it showed satisfactory reliabilities (α = .86; Amorose & Anderson-Butcher, 2007; Lim & Wang, 2009). However, it seems inadequate to use this scale as a measure of support of autonomy in addition to (other) scales measuring the support of competence and the support of relatedness, since the SCQ mixes aspects of other need supportive behaviours with the scale to measure support of autonomy. For example, the item “My coach conveys confidence in my ability to do well at athletics” also comprises aspects of the support of competence, as it refers to the students’ abilities. Moreover, the wording of the items might not be appropriate for 8-13 due to their language abilities, e.g. they might misunderstand words like “encourage” or “suggest”.

To measure the perceived support of competence and relatedness in PE, Standage et al. (2005) provided English scales for the use with children from the age of 11, which also showed good internal consistencies (both α > .80). It can also be questioned if the wording of these scales is adequate for younger children. More English scales including the support of autonomy, competence and relatedness are available for secondary school children (Sánchez-Oliva et al., 2018). However, statistical analyses showed high factor correlations and therefore insufficient differentiation of the three constructs. Another German questionnaire for children at the age of 9-12 measures two need support dimensions (support of autonomy, α = .63; support of competence and relatedness, α = .73) in the context of sports training (Kohake & Lehnert, 2018). It is also a weakness of this questionnaire that via exploratory factor analysis it was not possible to differentiate between a support of competence and a support of relatedness. Moreover, this questionnaire was developed for use in an extracurricular sport setting which implies the necessity of modifications and re-evaluation for the use in a PE context.

With regard to need satisfaction, some more scales are available. The Greek Basic Psychological Needs in Physical Education Scale (BPN-PE; Vlachopoulos, Katartzi & Kontou, 2011) covers all three subscales: the satisfaction of autonomy, competence and relatedness. Sánchez-Oliva et al. (2018) report acceptable reliabilities for the English translation in their study with secondary school children. However, the questionnaire has only successfully been used with children and youth not younger than 11 years (all subscales with α > .80; Vlachopoulos, Katartzi & Kontou, 2011). Moreover, scales previously used by Ntoumanis (2001) and Standage et al. (2005) with adolescents have been modified by Chen and Hypnar (2015) to suit the understanding of 8-11 year olds. They report reliabilities between .62 and .71. Additional scales exist in the context of sports training, e.g. German-language scales for children’s satisfaction of autonomy, competence and relatedness (Kohake & Lehnert, 2018) and English-language scales for youths’ and younger adults’ satisfaction of autonomy (Hollembeak & Amorose, 2005), satisfaction of competence (Amorose, 2003) and satisfaction of relatedness (Hollembeak & Amorose, 2005; all subscales with α > .70).

Finally, there are several English studies with regard to the motivational constructs. Most instruments comprise four to six subscales along the motivational continuum. An intrinsic, identified, introjected and external motivation as well as amotivation can be assessed with the Exercise Self-Regulation Questionnaire for grade 3-6 students (.62 < α < .82; Ryan & Connell, 1989) as well as the Behavioral Regulation in Exercise Questionnaire (BREQ; Mullan et al., 1997) or BREQ 2 (Hein & Hagger, 2007). The scales showed good internal consistencies for the use with adults (.78 < α < .90; Mullan et al., 1997) and also for a modified version suiting the PE context and 11-15-year-old children (.75 < α < .84; Hein & Hagger, 2007). The Situational Motivation Scale developed by Guay et al. (2000) was also used with children from the age of 12 and contains all six subscales (Standage, Treasure et al., 2003; .83 < α < .90). All the previously mentioned questionnaires refer to the sports training. Also developed for sports training, a German questionnaire exists for the types of motivation for a similar age group (9-13 years) containing four subscales (intrinsic, identified, introjected and external; .66 < α < .83; Kohake & Lehnert, 2018). The only identified instrument measuring types of motivation in the PE context and a similar age group as targeted in this study is the English Children’s Perceived Locus of Causality Scale (Pannekoek et al., 2014; α > .60). Accordingly, it is possible to differentiate five motivational constructs for children at the age of 9-12 (intrinsic, identified, introjected, external and amotivation). However, the authors report strong correlations between the intrinsic and identified regulation scales also resulting in a lack of confident discriminant validity. These weaknesses did not occur when using a revised form of the PLOC with 10-11-year-olds (Vlachopoulos, Katartzi, Kontou et al., 2011).

All in all, there are many questionnaires to be used in adolescence or adulthood, but few were developed for younger students. In addition, it can be noted that most questionnaires were developed for sports training, but not PE. Finally, most questionnaires only cover certain parts of the SDT. However, in order to examine more differentiated relationships between the individual constructs, more differentiated instruments are also required.

Table 1
Overview of existing questionnaires for children in the context of SDT

Adhering to SDT, the present study aims to develop and to preliminary validate German measurements to investigate the perceived need support, need satisfaction and types of motivation of children aged 8 to 13 in PE. The SDT differentiates three basic psychological needs (Ryan & Deci, 2002). Therefore, we hypothesise a 3-factor structure with regard to the perceived need support (support of autonomy, support of competence and support of relatedness; Hypothesis 1). Comparably, we assume a 3-factor structure with regard to the need satisfaction (satisfaction of autonomy, satisfaction of competence and satisfaction of relatedness; Hypothesis 2). With regard to motivation, previous research indicates that young children are already capable of differentiating five types of motivation (e.g. Pannekoek et al., 2014). For this reason, we assume a 5-factor structure (intrinsic, identified, introjected, external and amotivation; Hypothesis 3). According to SDT, the types of motivation lie on a self-determined continuum (Ryan & Deci, 2002). Consequently, to test the construct validity of the types of regulations scale, we expect a simplex structure (Guay et al., 2000): i. e., we assume adjacent subscales (motivation types more proximally situated on the continuum of self-determination) to share more similarities and, consequently, these subscales should show higher correlations than subscales further apart (motivation types more distally located on the continuum; Hypothesis 4). Finally, to further test the predictive validity of the developed scales, we test whether the support of autonomy, competence and relatedness positively predict self-determined types of motivation (intrinsic and identified) and negatively predict more controlled types of motivation (introjected, extrinsic and amotivation; Deci & Ryan, 2000; Hypothesis 5.1). Also, with regard to previous research with older children and youth (e.g. Standage et al., 2005) we assume that the effect of a need support scale on the types of motivation is mediated by the corresponding need satisfaction scale (Hypothesis 5.2).

With regard to all scales for the development of the Students Motivation in Physical Education (SMoPE) instrument, we used 82 items in the preliminary analysis. The items were both derived from existing questionnaires (with a back-and-forth translation procedure) and newly developed. We used 30 items (10 items for need support, 9 items for need satisfaction, and 11 items for types of motivation) that were originally used in sports training (Kohake & Lehnert, 2018) and we modified them to suit the PE context. As some original items had limitations (e.g. a lack of differentiation between the three need support scales), we developed 47 additional items (20 items for need support, 13 items for need satisfaction, and 14 items for types of motivation) based on other already existing questionnaires (Hollembeak & Amorose, 2005; Standage et al., 2005; University of Rochester, 2017; Vlachopoulos, Katartzi, & Kontou, 2011) or following the work of Mageau and Vallerand (2003). Additionally, as sports training outside school is usually voluntary and, thus, amotivation does not play an important role in sports training, Kohake and Lehnert (2018) did not integrate amotivation in their study. However, amotivation might play an important role in PE classes. Therefore, we added five items of amotivation to the item pool (Pannekoek et al., 2014).

Five boys and three girls aged 8-11 (M = 9.6, SD = 1.3) who attended grades 3 to 6 in and around Hamburg, Germany, participated in a preliminary analysis. We performed cognitive pretesting procedure and conducted qualitative interviews with the students to test the first pool of items with regard to understanding and clarity (Bowen et al., 2004; Pannekoek et al., 2014). We explained the purpose of the study and both children and their parents agreed to audiotape the interviews. We ensured the students that the interviews would be kept in confidence and would not be available to their parents or teachers. The cognitive pretesting procedure contained 4 steps: At the beginning the child was asked to read the item aloud in order to identify possible reading difficulties with the particular wording. Secondly, the child chose the answer that best reflected his or her personal experience in PE. Afterwards, the child was invited to explain in his or her own words what the question means to him or her and why he or she chose the specific answer. Furthermore, children were encouraged to ask questions, articulate difficulties and express their thoughts throughout the whole interview(“think-aloud-method”). The interviews lasted between 30 and 60 minutes. Short breaks were taken regularly so that it could be ensured that the participants were sufficiently concentrated. After each interview, the item pool was reviewed and iteratively modified if applicable.

The cognitive pretesting showed that the students understood the majority of the items the way intended by the researchers. However, due to reading difficulties and misinterpretations, some items were deleted. In the context of extracurricular sports, items with negative phrasing were also shown to be especially problematic for children (Kohake & Lehnert, 2018) and were therefore excluded from further steps. We further reduced the item quantity to an age appropriate number to help elementary students stay focused while completing the questionnaire independently (Chen & Hypnar, 2015). Since the items of existing scales have already been validated, we excluded the newly formulated items where necessary.

The result of the preliminary study was a 33-item questionnaire, which was deemed to be understandable and clear for 8-13-year-old PE-students. The questionnaire contained 9 items measuring the perceived need support, 9 items measuring the perceived need satisfaction and 15 items measuring the types of motivation. Since four response options have been shown to be optimal for children (Borgers et al., 2004) and therefore avoid tendencies to choose the middle answer, response options ranged on a 4-point Likert scale from 1 (strongly disagree) to 4 (strongly agree). In addition to the wording and in order to facilitate the scaling for the children, answers were visualized by happy and sad smileys. Table 2 shows example items for all subscales.

Table 2
Sample items for the subscales of the main study

Five elementary schools and four secondary schools located in the North of Germany agreed to participate in the main study. In total, N = 1,011 3^rd to 6^th grade students from 66 classes completed the questionnaire (M = 15.3 students per class). They were between 8 to 13 years old (M = 10.54; SD = 1.14), 47.1% of the students were girls. Students were informed that participation was voluntary, that there were neither right nor wrong answers and that their answers would remain confidential. The faculty’s ethics committee and the education authorities of Hamburg and Lower Saxony granted permission to conduct this study.

We conducted Confirmatory Factor Analyses (CFA) with MPlus (Muthén & Muthén, 1997-2017) to assess the construct validity of the need support, need satisfaction and motivation types constructs (Hypotheses 1, 2 and 3). We took the multilevel-structure of the data into account by using the complex-option and including learning group as the cluster variable. We used Maximum Likelihood with Robust Standard Errors Estimation, which is relatively robust for violations of normal distribution and data independence (Muthén & Muthén, 1997-2017). By considering the Full Information Maximum Likelihood (FIML) estimation procedure, the small amount of missing values (< 0,01% per item) could be estimated based on the model. According to recommendations by Hu and Bentler (1999) we evaluated the overall model fit by means of multiple goodness-of-fit indices, including the χ² test. The p-value associated with the χ² test is supposed to be nonsignificant. However, as the χ² test is highly oversensitive to sample size (Schermelleh-Engel et al., 2003), we present alternative fit indices, including the Comparative-Fit-Index (CFI), the Tucker-Lewis-Index (TLI), the Standardized-Root-Mean-Square-Residual (SRMR), and the Root-Mean-Square-Error-of-Approximation (RMSEA). A model that fits the data well is indicated when values for the CFI as well as the TLI are greater than .95 (good fit) or .97 (excellent fit) and for the RMSEA as well as the SRMR are less than .10 (good fit) or .05 (excellent fit; Marsh, 2007; Schermelleh-Engel et al., 2003).

For the need support construct we compared a G-factor (single-factor) model (Model 1) with a 3-factor model (Model 2) in order to examine if children are already capable of differentiating the three basic needs or rather perceive an overall need support. Comparably, with regard to need satisfaction we also compared a G-factor (single-factor) model (Model 3) to a 3-factor model (Model 4). Regarding the types of motivation scale, we compared the 3-factor model (Model 5) to a 5-factor model (Model 6). Thereby we tested whether it was better only to distinguish between rather self-determined motivation (intrinsic and identified), non-self-determined motivation (introjected and external) and amotivation or whether the more differentiated 5-factor model was appropriate. A 3-factor model might be appropriate as a self-determined vs. a controlled motivation mark the endpoints of the motivational continuum, while amotivation, on the contrary, marks the absence of motivation. Other studies have also shown the necessity of a reduction to three factors (e.g. Standage, Duda, et al., 2003).

For hypothesis 4, we analysed correlations between the motivational subscales. According to the continuum assumption, we assumed correlations between adjacent subscales (e.g. intrinsic and identified regulations) to show higher correlations than subscales further apart (e.g. intrinsic and extrinsic regulations).

To test if the perceived support of autonomy, competence and relatedness predict types of motivation, we conducted multiple multivariate regression analysis (Hypothesis 5.1). We entered the predictors (support for autonomy, support for competence and support for relatedness) simultaneously (single step) since no variable is given greater priority than another (Model 7). If the assumed regressions were shown to be significant, we conducted further latent path analyses: For significant regressions we tested in separate models for each need if the relationship between the support of the need and the specific type of motivation was mediated by the satisfaction of this same need (Hypothesis 5.2; Model 8 and 9).

With regard to the factor structure of need support, the G-factor model (Model 1) showed a poor model fit whereas the 3-factor-model showed a good model fit (see Model 2 in Table 3). All standardized item-factor loadings were significant (p < .01) and above .50. The final Model 2 contained three items on each of the three factors support for autonomy, support for competence and support for relatedness. Cronbachs’s Alpha for all subscales is provided in Table 5. Moderate latent correlations between the three subscales indicate that the constructs can be sufficiently distinguished (r_{Autonomy,Competence} = .69; r_{Autonomy,Relatedness} = .57; r_{Competence,Relatedness} = .80).

For the need satisfaction scale, the fit of the G-factor was not acceptable (see Model 3 in Table 3). The model fit remarkedly improved when three factors were specified and showed good fit to the data (Model 4). All standardized factor loadings were significant (p < .01) and at least moderate in size with one exception (.29), ranging from .46 to .81. Sufficient differentiation of the three constructs is supported by moderate latent correlations between the subscales (r_{Autonomy,Competence} = .28; r_{Autonomy,Relatedness} = .46; r_{Competence,Relatedness} = .47).

Table 3
Model fit indices of the proposed models

For the analysis of the types of motivation, a CFA of the 3-factor model showed an unsatisfactory model fit (see Model 5 in Table 3). Yet, the 5-factor model showed a better fit to the data (see Model 6). In this model, all items loaded significantly (p < .01). With one exception (.38) standardized factor loadings were greater than .50, ranging from .55 to .88.

Complementing the analyses to test hypotheses 1 to 3, we conducted a CFA in which all 11 subscales were modelled simultaneously. The model fit is satisfactory (see Model 7 in Table 3), indicating that the hypothesized constructs are adequately represented side by side.

The bivariate correlation matrix with regard to the types of motivation constructs supported a simplex-pattern (see Table 4): Types of regulations that are theoretically closer (e.g. intrinsic and identified regulation) showed higher positive manifest correlations (r = .66**) than more distal regulations (e.g. intrinsic and introjected: r = .05; intrinsic and external: r < .001; intrinsic and amotivation: r = -.37**). Likewise, the manifest correlation between the introjected and external regulations were higher (r = .60**) than those between the external and intrinsic (r < .001), the external and identified (r = .03) and the external and amotivation (r = .24**). This picture is also evident with regard to the latent correlations, which are naturally somewhat higher. Overall, a simplex pattern was largely supported by the data.

Table 4
Manifest (above the main diagonal) and latent (below the main diagonal) correlations between types of motivation

Table 5
Descriptive statistics, reliabilities and factor loadings on subscale level

The proposed multivariate multiple regression model fitted the data well (see Model 8 in Table 3). In line with our assumptions, the analysis showed significant positive effects of support of competence on both intrinsic motivation (β = .57*) and identified motivation (β = .57*). Additionally, the analysis showed a significant negative effect of support of autonomy on amotivation (β = -.18*). In contrast to the assumptions, no further paths were significant (see Table 6).

Table 6
Results of multivariate multiple regression analysis

Since only three paths in the above-mentioned regression analysis turned out to be significant, only these relationships were of interest for further investigation with regard to the mediating effect of the need satisfaction subscales.

To test whether the relation between support of competence and intrinsic motivation was mediated by the satisfaction of competence, the latter was included in the model (Model 9). The model-fit indices were acceptable. The standardized regression coefficients shown in Figure 2 indicate significant direct latent correlations between the support of competence and the satisfaction of competence (β = .35**) as well as the satisfaction of competence and intrinsic motivation (β = .20**). Additionally, the mediation model still shows a significant direct effect between the support of competence and intrinsic motivation (β = .51**). However, we also found a significant indirect effect (β = .07**) of support of competence on intrinsic motivation through the satisfaction of competence. Consequently, the correlation between the support of competence and intrinsic motivation is partially mediated by the satisfaction of competence.

We found the same pattern for the relation between the support of competence and identified motivation (see Figure 3: The Model 10 with the satisfaction of competence included shows acceptable model fit. The standardized regression coefficients show significant direct latent correlations between the support and satisfaction of competence (β = .36**) as well as the satisfaction of competence and identified motivation (β = .35**). Moreover, there is a significant direct effect between the support of competence and identified motivation (β = .53**). Nevertheless, we found a significant indirect effect (β = .13**) of support of competence on identified motivation through the satisfaction of competence. Comparable to the previous analysis, we conclude that the relation between the support of competence and identified motivation is also partially mediated by the satisfaction of competence.

All analyses of this study relied on one set of data. Even though the initial evaluation has produced acceptable psychometric characteristics, cross validations with additional samples could further strengthen the present results. Moreover, the data were cross-sectional and therefore not adequate to examine change over time, which could be interesting for future research.

Considering that assessments in classes always build on data with a multi-level structure, future research should also investigate to what extent the individual assessments of SDT constructs differ between the children in a class and between the classes (e.g. Lüdtke et al., 2006).

The measurements presented in this study can be used for further research that aims at investigating relationships between children’s perceptions of their PE teacher’s behaviour, their types of motivation and the resulting outcomes, e.g. extracurricular physical activity engagement. The results underline the importance of examining the impact of each need support separately in order to be able to predict the impact on specific types of motivation.

Building on the Work of Koka (2010), future research should further investigate possible age effects by comparing children of different age groups. Future research might also consider adding further subscales that focus on, e.g. controlling teacher behaviours (in contrast to need supporting behaviours) and need thwarting (in contrast to need satisfaction). Instruments examining need thwarting in the sports training context have already successfully been validated for youths and adults (Psychological Need Thwarting Scale; Bartholomew, Ntoumanis, Ryan & Thøgersen-Ntoumani, 2011). Depending on the research question, controlling behaviours might explain additional variance: while need support and satisfaction in the sports context have shown to be highly related to positive outcomes, controlling behaviours and need thwarting were better predictors of negative outcomes, e.g. emotional and physical exhaustion (Bartholomew, Ntoumanis, Ryan, Bosch et al., 2011). However, the scales’ invariance across participants’ age as well as the suitability for the PE context is still to be tested.