Revised version of physical education teachers ’ self-efficacy scale

The aim of the study is to revised the validity and reliability of physical education teachers’ selfefficacy scale. Sample group of the study was consisted of 567 physical education teachers who is working in different geographic regions of Turkiye. In the study it was performed that explanatory factor analysis and confirmatory factor analysis for determinin the subscales and construct validity. As the result of the analyses; 33 items whose factor loading values were over 0.57 were clustered under nine factors whose eigenvalues were greater than 1. It was found out that the percentage of variance explained by the nine factors was 78.362. When Cronbach Alpha internal consistency tests of the factors were analyzed, it was seen that α values of all factors were greater than 0.83. As a result; it is predicted that the scale can be used in a valid and reliable way in order to measure self-efficacy of the physical education teachers.


Introduction
Teachers, constituting the human capital of and serving in the educational system, are indispensible components of the system and they change and develop student behavior positively and educate human power to serve for the society.Today when more quality and competence is required; a modern teacher should not only be a person who teaches and assesses the students but also be a good leader, observer, manager and guide who organizes teaching-learning process.
Teachers' performing educational competences required by the teaching profession relates to getting a good education as well as to their beliefs in performing their tasks and responsibilities.
Hence the studies conducted indicate that people's beliefs affect their motivational levels, emotional status and behaviors.It is emphasized that one's belief in his competence affects and directs the way in which the behavior in question is done in terms of finalizing a behavior successfully (Bandura, 1977;1997;Enochs and Riggs, 1990;Gökçe, 2000).
Self-efficacy, defined as one's belief in his/her own ability to complete tasks and reach goals successfully by organizing the necessary activities in order to realize a performance, is based on our belief in our abilities and is necessary to organize certain behaviors in order to attain and fulfill certain goals (Schmitz and Schwarzer, 2000).Self-efficacy is not a function produced by the abilities of the individuals but all of the judgments about what can be done through their abilities (Gürcan, 2005;Kiremit, 2006).Self-efficacy beliefs affect one's thinking styles, problem-solving skills and emotional reactions.People with low self-efficacy think that things are more difficult than they are and their problem-solving skills are very limited.People with high self-efficacy are more confident and their problem-solving skills and determination are higher (Enochs and Riggs, 1990;Pajares, 1997).Bandura (1986;1997) argues that there are four starting points from which people develop their self-efficacy: mastery experience, vicarious experience, physiological and emotional stimulation and social influence.
Mastery experience includes experiences related to successes or failures; physiological and emotional stimulation include individual's psychologically or physically feeling good; vicarious experience includes one's increased beliefs in himself/herself by modeling others and observing their life styles and social influence includes being motivated by other people's encouraging words.On the other hand; self-efficacy requires four processes: cognitive process, motivational process, affective process and selection process.These processes play a key role in perceiving efficacy and they can prevent the desired behaviors from being realized or encourage them to be born (Bandura, 1986;1994;1997;Sckunk and Pajares, 2001;Bozgeyikli, 2005).
It is reported that self-efficacy beliefs can be used to explain individual differences in teaching activities and can make contributions to understand and to improve teacher behaviors (Enochs and Riggs, 1990;Pajares, 1997).Teachers' self-efficacy perceptions are their beliefs that they can successfully fulfill teaching function (Guskey and Passaro, 1994;Atıcı, 2000).
Teachers' self-efficacy perceptions are factors in their teaching activities and adoption to the education process (Pajares, 1997).When the self-efficacy concept is examined in terms of physical education teachers; we can describe the self-efficacy concept as teachers' beliefs that they can actively fulfill teaching function under all kind of negative and positive conditions and can make positive contributions to learning performance of the students (Yıldırım, 2012a).The studies conducted suggest that self-efficacy perception of a teacher affects teaching and learning; particularly, teachers' practices in the classroom.Depending on the self-efficacy beliefs; teachers say that their efforts spent for teaching, educational objectives and amount of teaching-desire change.It is argued that there is positive correlation between teachers' positive self-efficacy perceptions and student success, student motivation and student competence and it is believed that teachers' self-efficacy perceptions are one of the most important factors that predict teacher competence (Tschannen-Moran et al. 1998;Schmitz and Schwarzer 2000;Tschannen-Moran and Woolfolk Hoy, 2001).
When the studies on teachers' self-efficacy are examined; it is seen that the studies on teachers' self-efficacy -designed and conducted in foreign languages by Akkoyunlu et al. (2005), Ekici (2005) and Bıkmaz (2002)-are adaptation studies and scale development studies.On the other hand; although the studies conducted have focused on discovering teachers' general beliefs in self-efficacy, their beliefs in self-efficacy in relation with the academic branches have recently been investigated, too (Riggs and Enochs, 1990;Savran and Çakıroğlu, 2003;Gökyer, 2012).In this self-efficacy has been the subject of physical education teachers' has been the subject of the, some studies (Ünlü et al., 2008;Koparan et al. 2011, Yılmaz et al. 2010;Yıldırım 2012a, Yıldırım 2012b, Yıldırım 2013).

Purpose of the research
In the literature it was seen that some scale for measuring the self-efficacy of physical teachers, but in order to determine the self-efficacy of physical education teacher needs to valid and reliable scale, also needs more contemporary scales.According to this idea in this study it was aimed that the revised the physical education self-efficacy scale which was developed by Yıldırım (2012a).

Method Population and sample selection
In the sudy it was aimed that revised the physical education teachers' self efficacy scale.
According to the aim of this study it was consisted of the study group by simple random sampling.The population of the study was consisted of 576 physical education teachers (173 female physical education teachers and 394 male physical education teachers) who worked at the primary schools and high schools in different geographic regions of Turkiye .The average age of physical education teachers was 34.82±6.637(their ages varied between 22 and 59 years) in this research.

Data collection method
As the data collection method, the self-efficacy scale designed by Yıldırım (2012a;2012b;2013) with 36 items was used.Besides, a draft scale with 41 items, which was prepared by adding 5 extra items to the original scale after a literature review, was employed in addition to general teacher competence guidelines established by National Education Ministry (http://otmg.meb.gov.tr),academic competence guidelines for physical education teachers and self-efficacy scale for physical education teachers designed by Ünlü et al. (2008).The scale items had a 5-point Likert format with close-ended questions and the coding was on 9 point rating system as follows: 1. Never, 3. Slightly, 5: Somewhat, 7. Good and 9. Very good.

Analyzes of data
Adequacy of the data collected for the factor analysis was examined using Bartlett's Test of Sphericity while adequacy of the size of the study-sample was examined using Keiser-Meyer-Olkin coefficient.Subscales were established using explanatory factor analysis.In the explanatory factor analysis conducted to determine the scale items; it was paid attention to that eigenvalue of items should not be smaller than 1, factor loading values of items should be at least .50,items should be clustered under one factor and the difference in loading values of the items between two or more factors should not be smaller than 1 (Çokluk et al., 2010;Alpar, 2010).In the explanatory analysis; 25 degree varimax axis rotation test and principal components analysis were employed.After the explanatory factor analysis, confirmatory factor analysis was used through Lisrel package software.First, first-level and then second-level confirmatory analyses were conducted.Cronbach Alpha internal consistency tests was administered for the subscales found as the result of explanatory and confirmatory factor analyses and for the total scale in order to determine internal consistency of the scale.

Results
According to the results of KMO and Bartlett's Test of Sphericity seen in Table 1; it was accepted that it was adequate to administer factor analysis for the relevant data group.When slope graphs were analyzed in Figure 1, it was seen that there was a break after the 9 th point.First-level confirmatory analysis was conducted after explanatory factor analyses.There are various fit indexes to test the adequacy of the model constructed by the confirmatory factor analysis.The proportion of chi-square of the model to degrees of freedom was found out by RMSEA (Root-mean square error approximation), SRMR (Standardized Root Mean Square Residual), NFI (Normed Fit Index), NNFI (Nonnormed Fit Index) and CFI (Comparative Fit Index) (Schermelleh et al.,2003;Çelik, 2009;Yılmaz and Çelik, 2009).
When Table 4 was studied, it was seen that fit criteria were within acceptable fit limits and good fit limits (Schermelleh et al.,2003;Çelik, 2009;Yılmaz and Çelik, 2009).When Figure 2 and Table 4 was studied, it was seen as a result of first-level confirmatory analysis that X 2 /df value, RMSEA and SRMR values were within acceptable fit limits while NFI, NFFI and CFI values were within good fit limits.According to these values, it may be argued that the model constructed was a suitable model as far as the data used were concerned.Also, when Path Diagram of the first-level confirmatory factor analysis of the self-efficacy scale was examined, it was established that all of the subscales were in positive correlation with each other and each of the subscales and their explanatory items were in positive correlation (p<0.05).Therefore, the second-level confirmatory factor analysis was initiated.When Table 5 was examined, it was seen that fit criteria were within acceptable fit limits and good fit limits (Schermelleh et al.,2003;Çelik, 2009;Yılmaz and Çelik, 2009).When Figure 2 and Table 5 were studied, it was seen as the result of second-level CFA that X 2 /df values, RMSEA and SRMR values were within acceptable fit limits while NFI, NFFI and CFI values were within good fit limits.According to these values, it may be argued that the model constructed was a suitable model as far as the data used were concerned.
Also, when Path Diagram of the second-level confirmatory factor analysis of the self-efficacy scale was examined, it was established that all of the subscales were in positive correlation with each other and each of the subscales and their explanatory items were in positive correlation (p<0.05).When the correlation between self-efficacy scale and its subscales was analyzed, the highest correlation was found to be between "self-efficacy in classroom management and selfefficacy in planning" subscales and total of self-efficacy scale.

Discussion
In the study which was conducted to develop a scale of self-efficacy for physical education teachers, construct validity of the scale was established using explanatory factor analysis and confirmatory factor analysis.
For social sciences; factor analyses are made in order to establish construct validity in scale development studies and scale adaptation studies.Factor analysis is a statistical method with multi-variables used to describe variability among observed, correlated variables in terms of a potentially lower number of unobserved variables called factors and to investigate whether a number of variables of interest are related to a smaller number of unobservable factors (Çokluk et al., 2010;Büyüköztürk, 2011).There are two factor analyses: Explanatory Factor Analysis and Confirmatory Factor Analysis.Explanatory Factor Analysis (EFA) does not only reduce the number of the variables and identifies the underlying relationships among the variables but also serves to identify latent constructs underlying a behavior.In other words, it is a kind of explanatory analysis employed to question whether or not certain items clustered under a factor are indicators of the theoretical construct.
Confirmatory Factor Analysis (CFA) is used to test whether the observed data of a model are consistent and fit with the measurement model identifying the underlying variables and presents detailed statistics regarding to what extent these data fit with the model.In the construct validity tests, it is a common and accepted method to employ confirmatory factor analysis after explanatory factor analysis (Çokluk et al., 2010;Worthington and Whittaker, 2006).
Before explanatory factor analysis, Kaiser-Meyer-Olkin (KMO) test was administered in order to test whether or not sample-size was suitable for factorialization.When the literature is examined, KMO values between 0.50 and 0.60 are considered bad, between 0.60 and 0.70 weak, between 0.70 and 0.80 moderate, between 0.80 0.90 good and over 0.90 excellent.It is emphasized that factor analyses cannot be performed if KMO is smaller than 0.50.It is accepted that the nearer KMO value gets to 1, the more suitable it is to administer the factor analyses.However, if the KMO value is smaller than 0.5 factor analyses cannot be proceeded because the correlation between variable-pairs cannot be explained by other variables (Bayram, 2004;Alpar, 2010;Çokluk et al., 2010).In this study, it was found out that KMO value of the self-efficacy scale for the physical education teachers was 0.923 (Table 1); by which it was concluded that sample size was "excellent" to perform factor analyses.Besides, when Bartlett's Test of Sphericity was studied, the Chi-square value obtained was seen to be significant (x 2 (528):14144.995;p<0.001).
Therefore, it was accepted that the data followed a normal distribution with multi-variables.As the result of the findings of KMO and Bartlett's Test of Sphericity, the relevant data group was considered suitable for factor analysis and explanatory factor analysis was initiated.
Varimax orthogonal axis rotation test and principal components analysis were employed as factorialization method in order to explore the factor construct of the self-efficacy scale for the physical education teachers.The first phase of the explanatory analysis was made with 41 items.
As the result of the first-level explanatory analysis; 6 th , 10 th , 11 th , 12 th , 37 th and 41 st items were removed from the scale because their factor loadings were smaller than 0.50 and they appeared in more than one subscales and there was adjacency of these items.The second phase of explanatory analysis was made with 33 items.As the result of the repeated EFA Varimax orthogonal axis rotation test, it was found out that the smallest factor loading value was 0.577 (Table 2).Factor values are standardized regression coefficients and indicate the correlation between variable and factor.Smaller factor loading demonstrate that the item is not strongly correlated with that factor.Although it is stated in literature that minimum factor loading value should be 0.30, there are also other theoreticians that argue that minimum factor loading value should be ≥ 0.40.In general, loading values between 0.30 and 0.59 are identified as moderate while those ≥ 60 as high (Çokluk et al., 2010).In line with that information in literature, it was established that factor loading values of the items that made up self-efficacy scale for the physical education teachers were high and very good.
As the result of the EFA conducted with 33 items; when slope graphs were analyzed in Figure 1, it was seen that there was a break after the 9 th point and there were 9 subscales whose eigenvalues were greater than 1 (Figure 1 and Table 3).The eigenvalue of an item reflects the power of the correlation between the factor and original variables.Eigenvalues are used to estimate variances identified by factors and to decide the number of the factors.In the factor analyses, those factors whose eigenvalues are ≥ 1 are accepted as stable (Alpar, 2010;Çokluk et al., 2010).As the result of the factor analysis made, it was decided that self-efficacy scale for the physical education teachers would have 9 factors because there were 9 components whose eigenvalues were ≥ 1 and there was a break after the 9 th point in the slope graphs.Therefore; the subscales were termed as follows: Self-efficacy in the usage of technology, Self-Efficacy in classroom management, Selfefficacy in special field knowledge, Self-Efficacy in finding resource and providing support, Self-Efficacy in verbal and non-verbal communication, Self-Efficacy in planning, Self-efficacy in special education, Self-Efficacy in assessment and evaluation and Self-Efficacy in improving physical performance.As the result of the analysis, it was discovered that factor loadings of the items which were clustered under 9 subscales were over 0.57 and the percentage of variance explained by the nine factors was 78.362.The size of the variance explained indicates the strength of the factor construct of the scale.In the practice; with multifactorial models, it is enough for a variance to be between 40% and 60% in scale developing in social sciences and there are some examples in the studies (Alpar, 2010;Çokluk et al., 2010;Ünlü, 2012).In this sense, it was concluded that the percentage of variance found for the self-efficacy scale for the physical education teachers was considerably high and acceptable.
After conducting explanatory factor analysis, we proceeded to confirmatory factor analysis.It is kind of analysis through which a pre-identified and predetermined construct is tested to decide whether it is confirmed as a suitable model or not.Besides, this analysis is used to confirm "theoretical structure" or "theoretical model", too.Whether or not self-efficacy scale for the physical education teachers could be confirmed was tested using first-level confirmatory factor analysis and second-level confirmatory factor analysis.Fit indexes are used to determine the adequacy of the model constructed in confirmatory factor analysis (Table 4).These indexes are were significant differences between the two scales.In the first study of Yıldırım (2012a), construct validity of the self-efficacy scale was performed using explanatory factory analysis and it was seen as the result of the analysis that the percentage of variance explained for the scale with 36 items and eight factors whose eigenvalues were bigger than 1 was 75.091 for total variance and internal consistency test presented Cronbach's Alpha value as 0.950.In another study conducted by Yıldırım (2012bYıldırım ( , 2013) ) to determine construct validity using explanatory and first-level confirmatory factor analyses; it was found out that the scale was consisted of 32 items and eight factors whose Eigen values were bigger than 1.
The percentage of variance explained was 74.747 for total variance.In internal consistency test, Cronbach's Alpha value of the scale was found to be 0.950.As for our study; the percentage of variance explained for the scale with 33 items and nine factors whose Eigen values were bigger than 1 was 78.362 for total variance.In internal consistency test, Cronbach's Alpha value of the scale was found to be 0.942.In the current self-efficacy scale, there were three new items and these items were clustered under one factor which was termed as "Self-Efficacy in improving physical performance".Besides; it was noted that percentage of variance explained for the scale was higher than other scales.Confirming the self-efficacy scale for the physical education teachers through first-level confirmatory factor analysis and second-level factor analysis after the explanatory factor analysis, exploring a high percentage for the total variance and high Cronbach's Alpha values will increase the importance of the scale and will indicate that the scale has a "valid, reliable and strong theoretical basis".

Conclusions and recommendations
The self-efficacy scale for the physical teachers is a measurement tool with 33 items which was clustered into 9 subscales and developed to measure professional self-efficacy of teachers.The subscale of Self-efficacy in the usage of technology is consisted of 5 items and measures teachers' belief that they can utilize technology in education by following technological advancements.The subscale of Self-efficacy in special field knowledge is consisted of 4 items and measures teachers' belief about professional knowledge and skills about physical education and sports.The subscale of Self-efficacy in finding resource and providing support is consisted of 4 items and measures teachers' belief that they can find resources and provide support by using private sector, public sector and other external factors for the needs such as fields, sportive materials and etc.The subscale of Self-efficacy in planning is consisted of 4 items and measures teachers' belief that they can rationally design teaching activities in order to attain and to realize program objectives in teaching programs.The subscale of Self-efficacy in special education is consisted of 3 items and measures teachers' belief that they can meet the needs of those students whose educational abilities and individual characteristics are different from the level that should be expected.The subscale of Self-efficacy in verbal and non-verbal communication is consisted of 3 items and measures teachers' belief that they can develop effective communication through (effective) speaking, rhetoric, tone, looks, gestures, body movements, body signs and mimics.The subscale of Self-efficacy in classroom management is consisted of 4 items and measures teachers' belief that they can prepare and maintain the best classroom environment and conditions for an effective learning and teaching.The subscale of Self-efficacy in assessment and evaluation is consisted of 3 items and measures teachers' belief that they can attain a standard of judgment about the subject measured by observing students' different qualifications, expressing numerical data and symbols and comparing the final results to criteria.The subscale of the development of physical performance is consisted of 3 items and measures teachers' belief that they can improve the students through sportive activities in line with their interests and abilities.
As a result; it is predicted that the scale can be used in a valid and reliable way in order to measure self-efficacy of the physical education teachers.

Figure 2 .
Figure2.Path Diagram of first-level confirmatory factor analysis of self-efficacy scale (Technolo: Self-efficacy in the usage of technology; Assesme: Self-efficacy in assessment and evaluation; Class: Self-efficacy in classroom management; Communic: Self-efficacy in verbal and non-verbal communication; Field kn: self-efficacy in special field knowledge; Resource: Self-efficacy in finding resource and providing support; Planning: Self-efficacy in planning; Special: Self-efficacy in special education; Physical: Self-efficacy in improving physical performance).

Figure 3 .
Figure 3. Path Diagram of second-level confirmatory factor analysis of self-efficacy scale

Table 2 .
Rotated Component Matrix: Physical education teachers self-efficacy scale It was seen inTable 2 that 33 items whose factor loading values were over 0.577 were clustered under nine factors.

Table 3 .
Total Variance Explained and Cronbach's Alpfa (α) When Table3was examined, it was seen that accountability of the self-efficacy scale clustered into 9 subscales was 78.362.When internal consistency test of the self-efficacy scale was examined, general Cronbach's Alpha value was found to be 0.944.

Table 4 .
Fit criteria and the results of First-level confirmatory analysis of the self-efficacy scale