1. Introduction
Constructivist learning theories are used as a theoretical perspective supporting teaching, learning, and curriculum in physical education and sport [
1,
2,
3,
4,
5]. Constructivist principles provide conceptual foundations for redefining teachers’ roles and establishing new teaching practices. For example, it can be applied in contexts where the activity focuses on game understanding and the cognitive and social engagement of the learners in physical education and sports [
6,
7,
8]. Observations considering teacher education across theoretical perspectives have robustly noted that preservice teachers’ (PST) prior knowledge and experience influence their learning both positively and negatively [
4]. Constructivist research that examines the impact of prior knowledge is relevant in identifying pre-existing conceptions that are difficult to change, new concepts that are difficult to learn, the role of pre-existing knowledge in the learning process, and factors that facilitate or constrain learning [
9].
Recent observations in PSTs’ learning and the implementation of student-centered models in physical education noted the low impact of initial training on changing PSTs’ beliefs concerning constructivist-oriented practices [
10,
11]. Physical education PSTs start education programs with personal, preconceived knowledge regarding teaching and the nature of learning from their previous accumulated experience as students and athletes [
12,
13,
14,
15,
16,
17]. In addition, constructivist principles often conflict with naive teaching perspectives [
18,
19], given that PSTs can interpret learning as a transmissive process, during which PE teachers should rely on the reproduction of knowledge, and the teacher’s authority plays a determinant role in the learning process [
19]. Therefore, conceptual change is a difficult, long-term process [
9]. Physical Education Teacher Education (PETE) programs should support this conceptual change and help PSTs to reconceptualize teaching and learning in physical education [
16,
18,
19].
There are few instruments available with which to examine the perceptions of PSTs regarding constructivist teaching practices [
20,
21,
22] and behaviors [
23]. The Constructivist Teaching Practices Inventory in Elementary Physical Education (CTPI-EPE) was originally developed and validated in order to identify the constructivist-oriented teaching practices of teachers in the United States [
8]. The questionnaire has since been validated in other contexts, such as Turkey [
24], Greece [
25], and Brazil [
26]. However, in the Brazilian context, only content validity was performed. CTPI-EPE has been noted as a valid instrument by which to measure in-service teachers [
8,
24]. Data based on the questionnaire has provided specific direction for curricular and instructional approaches that successfully improve PSTs’ learning regarding constructivist-oriented teaching practices [
8]. Nevertheless, there is no information available concerning the cross-cultural validity of the questionnaire when applied to physical education PSTs’ constructivist-oriented teaching practices, or its use in exploring sources of variation across PSTs’ constructivist-oriented teaching practices, as was recommended during original conception [
8].
Lastly, the psychometric properties of questionnaires are often examined using frequentist methods, often leading to limited or inaccurate interpretations of the data [
27]. On the other hand, Bayesian methods consider the prior information available and the information contained in the data to update knowledge [
27]. In this study, we examined the validity of the adapted version of the CTPI-EPE to Portuguese in Brazilian physical education PSTs.
3. Results
The Portuguese version of the CTPI-EPE presented the content validity indexes of general Aiken
V (
V = 0.94), clarity of language (
V = 0.91), practical relevance (
V = 0.96), and theoretical dimension (
V = 0.96) (
Table 1).
BEFA showed that a 3-factor structure best fitted the questionnaire structure. All 25 items had posterior means higher than 3 (
Table 2). Factor 1 (Facilitating Active Construction of Knowledge in Games and Skills) is composed of five items (1, 4, 5, 7, and 8), Factor 2 (Facilitating Personal Relevance) is composed of seven items (9, 10, 11, 12, 14, 15, and 16), and Factor 3 (Facilitating Social Cooperation) is composed of 13 items (17, 18, 20, 21, 22, 23, 24, 25, 26, 28, 32, 33, and 36).
Based on the construct evidence from BEFA, we tested the factor structure in the BCFA. The first model (3-factor structure with 25 items) presented adequacy of the posterior values (>0.5;
Table 3), but inadequate fit indexes (
Table 4; BRMSEA = 0.08; BGammaHat = 0.88; adjBgammahat = 0.85; BMc = 0.44). Additionally, we tested items standardized residuals covariances (
Supplementary Table S1).
Based on this analysis, we found that some items presented high-standardized residuals covariances that could disturb model fit. Thus, items 10, 14, 17, 20, 24, and 33 were excluded. The second model (3-factor structure with 19 items) was run. Model 2 presented satisfactory posterior values (>0.5), better fit measures than Model 1 (meaning a better model than Model 1) and adequate fit indexes (BRMSEA = 0.06; BGammaHat = 0.94; adjBgammahat = 0.91; BMc = 0.74). However, items 5 and 23 presented high-standardized residuals covariances (>0.1). Although model fit presented satisfactory values, we decided to exclude item 23 to test a third model, and to avoid items from different factors with high-standardized residuals covariances.
The third model (3-factor structure with 18 items) presented satisfactory factor loadings, better fit measures, and similar fit indexes (BRMSEA = 0.06; BGammaHat = 0.94; adjBgammahat = 0.92; BMc = 0.77) than Model 2. Additionally, items did not present standardized residuals covariances closer to 0.1 or −0.1. Model fit measures are presented in
Table 5. Thus, Model 3 was considered the final model (
Figure 1). The final model showed good internal consistency across the three dimensions (FACK = 0.86; FPR = 0.90; FSC = 0.94). The structure is composed of Factor 1 (Facilitating Active Construction Knowledge in Games and Skills) with five items (1, 4, 5, 7, and 8), Factor 2 (Facilitating Personal Relevance) with five items (9, 11, 12, 15, and 16) and Factor 3 (Facilitating Social Cooperation) with eight items (18, 21, 22, 25, 26, 28, 32, and 36). The numbering of the 18 items was ordered for the final version (
Appendix A).
4. Discussion
This study examined the validity of the adapted version of the CTPI-EPE to Portuguese. The analytical framework was important to help sustain the structure of the questionnaire, and demonstrated adequate psychometric proprieties making it shorter and easier to apply. As a result, the Portuguese version of the CTPI-EPE presented good content validity (
V = 0.94), considered very high [
34] even compared to other studies in sports science [
37,
38,
42] and the initial content validity version in Brazil [
26].
Based on content validity, a three-factor structure was deemed the best fit, and all 25 items presented posterior means higher than our cut-off point [
8]. Furthermore, the three-factor structure (Facilitating Active Construction of Knowledge; Facilitating Personal Relevance; Facilitating Social Cooperation) is theoretically supported by the three fundamental constructivist principles [
4,
8], consistent with the Turkish version [
24]. Additionally, our results are consistent with observations made using an observational instrument that examines constructivist teaching practices in physical education research [
23].
Regarding the distribution of items in the factors, three items (5, 7, and 8) originally from the dimension “Facilitating Personal Relevance” migrated to the dimension “Facilitating Active Construction of Knowledge”. The changes in the items emphasize the importance of learners’ active and constructive involvement in their learning.
These observations highlight the importance of providing the opportunity for students to develop responsibility for their learning, actively involving them in discovery, self-regulation, problem-solving, and encouraging them to be autonomous thinkers and evaluate their learning [
4,
7,
8]. In the dimension “Facilitating Personal Relevance” (Factor 2), only item 12 from Factor 1 was included in the model. It refers to teaching situations that establish a relationship between the learner and the content, considers their previous knowledge of organizing learning activities [
4,
8], and achieves personal relevance and the attribution of meaning to the game in their lives [
43]. Although there are differences between the dimensions, both dimensions are related to psychological constructs and are explained by the cognitive constructivist perspective of learning. For example, the Facilitating Active Knowledge dimension develops group strategies that encourage students to take responsibility for their learning, and the Facilitating Personal Relevance dimension addresses strategies to integrate students’ new understanding with their previous knowledge [
4,
8]. This is an important aspect to consider when interpreting the differences between these dimensions.
In the “Facilitating Social Cooperation” dimension (Factor 3), all items (20, 21, 22, 23, 24, and 25) from the original version remained. In addition, six items (17, 18, 26, 28, 32, and 36) from Factor 1 and one item (33) from Factor 2 were included in this dimension. Our observations contrast with both the original and the Turkish version. It is likely that Brazilian physical education PSTs perceive that providing opportunities for student involvement in games are strategies that encourage active “participation”. These strategies involve learning and social interaction with others (peers and teachers), related to the sociocultural constructivism perspective [
3,
5,
8]. Through interaction, students share ideas, expose common limitations, establish comparisons, negotiate meanings, ask questions that contribute to individual understanding, make decisions together, listen, and carefully observe each other for mutual group adjustment [
2,
44]. Thus, there is strong empirical support for the social domain as an essential pillar of learning, as long as all the students demonstrate competence and active learning throughout the process [
45].
BCFA confirmed the three factors had good fit indices [
40,
41]. However, seven items (10, 14, 17, 20, 23, 24, and 33) were excluded. Items 14, 20, and 33 displayed cross-loading in the original version, subjectively included in the original questionnaire in the respective dimensions [
8]. Additionally, the subjective inclusion of these items in the original version might have supported the maintenance of many controversial items. Thus, excluding them did not have any effect on the final model and its fit indexes. The exclusion of items 17, 23, and 24 can be justified by interpreting the Brazilian PSTs’ active involvement as similar to social cooperation concepts previously discussed. Additionally, these items were excluded in the confirmatory factor analysis of the Turkish version [
24]. Brazilian PSTs showed similar perceptions regarding items 9 and 10, as both items referred to students’ understanding and previous experiences of games and skills, and their relationship with learning new content [
4,
8], resulting in the exclusion of item 10.
Finally, the Portuguese version of the CTPI-EPE obtained a final structure with 18 items distributed across three factors. The distribution and number of items diverged from the Turkish version, with 16 items [
24]. It is important to highlight that in addition to cultural differences, the present Brazilian version is validated for preservice teachers. In contrast, the original version [
8], Turkish version [
24], and versions used in Greece [
25] and Brazil [
26], apply to in-service teachers. Limitations of the study were related to the majority of the sample consisting of physical education preservice teachers from public universities in southern Brazil. A more balanced distribution of the sample in different regions of Brazil, and validation in other cultures of Portuguese speakers (e.g., Brazil and Portugal), is recommended for future studies.