A scale development study on measuring science teachers ' autonomy on curriculum * 1

DOI: 10.31704/ijocis.2020.002 The purpose of this study is to develop a Teachers' Autonomy on Curriculum Scale. For this aim, an item pool consisted of 50-item was prepared for the study. These scale items were reduced to 29 items after expert review and pilot implementation. This preliminary form was applied to 178 science teachers working in secondary schools in Izmir, Turkey. Validity and reliability studies have been done and Cronbach's Alpha internal consistency coefficient was calculated as .82. The scale is four-dimensional and reveals 67.4% of the total variance. The scale has four sub-scales (Professional Autonomy, Process Autonomy, Assessment Autonomy, and Planning Autonomy). Confirmatory factor analysis results support that the scale consisted of four subscales (RMSEA= .05, CFI= .98, AGFI= .89, RMR= .05, GFI= .93, SRMR= .06). Article History: Received Revised Accepted Online 26 March 2020 04 May 2020 05 June 2020 08 June 2020


Introduction
The reform activities carried out in many areas of education have brought about significant changes in the way the teaching profession is defined and the areas of competence are specified. These changes have especially shaped teachers' in-class practices, and with the increasing popularity of studentcentred education, radical changes have occurred in the roles of teachers (Açıkgöz; 2016; Demirel, 2004;Sönmez, 2008). One of the first of these changes that comes to mind is the concept of "Autonomy".
There are numerous studies which deal with the concept of autonomy in the student, teacher and school dimensions (Bustingorry, 2008;Chan, 2001;Ingersoll, 2007;Little, 1995;Littlewood, 1996). Although all these dimensions include the basic components of autonomy (independence, freedom to make decisions, power of control, etc.), they have critical differences in structure. It is seen that these studies mainly concentrate on "school autonomy" and "learner autonomy" (Little, 1995;Öztürk, 2011), whereas in the last 20 years, the number of studies related to "teacher autonomy" has increased (Benson, 2010;Benson and Huang, 2008;Burkert and Schwienhorst, 2008;Çakır and Balçıkanlı 2012;Ding, 2009;Dymoke and Harrison, 2006;Hong and Youngs, 2016).
A more specific dimension of teacher autonomy is curriculum autonomy. Teachers' autonomy over the curriculum includes practices such as making decisions related to the curriculum, organising teaching plans, selecting teaching methods, and student evaluation (LaCoe, 2006;Pearson and Moomaw, 2005;Vangrieken et al., 2017). The fact that teachers have a large area of autonomy over the curriculum makes focusing on this aspect of teacher autonomy important.
In order to determine teachers' levels of curriculum autonomy, efforts to develop a scale with established validity and reliability are essential. When the literature is examined, it is seen that there are a limited number of scale development studies that focus on teachers' curriculum autonomy (Çolak and Altınkurt, 2017;Friedman, 1999;Ulaş and Aksu, 2015). In these studies, an attempt is made to reveal levels of curriculum autonomy among teachers from different branches and with different seniority levels. No study can be found that focuses on curriculum autonomy in science teachers.
The Science Curriculum was newly prepared and revised by the Ministry of National Education (MoNE) in 2005, based on a student-centred approach and constructivism, and was updated in 2013 and 2017 in the light of contemporary developments (MoNE, 2005(MoNE, , 2013(MoNE, , 2017. These programmes have components for supporting teacher autonomy in the areas of planning, organising the evaluation process, defining activities and implementing activities. In this case, it has become important to reveal the extent to which science teachers in particular perceive their autonomy in the curriculum implementation process. The aim of this study is to develop a Curriculum Autonomy Scale for determining science teachers' perceptions of autonomy over the curriculum. It is expected that the scale will be a resource that can be used as a tool to gather data for specialists in the field of "Curriculum and Instruction" in their studies on teacher autonomy.

Conceptual Framework
A number of different definitions of the concept of autonomy can be found in the literature. Oshana (2003) defines autonomy as an individual's having a say in actions and choices that direct his/her own life. According to a similar definition made by Oğuzkan (1974), autonomy is an individual's possession of a certain amount of independence in directing his/her own behaviours. Littlewood (1996) defines autonomy as an individual's ability to make choices in directing his/her own behaviours and having an independent capacity to carry out these behaviours. By drawing attention mostly to the conscientious significance of autonomy, Piaget (1932) explains autonomy as the directing of an individual's behaviours by the self, without any external or internal pressure (Piaget, 1932, cited by Moomaw, 2005. Pitt (2010) defines autonomy as modern humans' quality of being unaffected by external influences and of using their free will.
The common aspect of these definitions of autonomy consists of individuals having a say in actions and choices that direct their own lives. An individual who can direct his/her own behaviours works independently, initiates new activities, and in order to adapt these to changing conditions, makes changes to existing situations. In this study, autonomy is defined as a person's ability to freely make decisions related to his/her own behaviours, and, while making these decisions, to remain independent of external influences as much as possible.
Studies can be found which reveal that autonomy has an effect on students' academic performance, and on adults' job satisfaction and professional performance (Garcia and Pintrich, 1996;Hmel and Pincus, 2002). By its nature, teaching is one of the professions in which autonomy is strong. The existing structure of schools induces teachers to work independently of external control in self-contained classes (Anderson, 1987). Therefore, focusing on teacher autonomy has become important.
There are various definitions of teacher autonomy in the related literature. Pearson and Moomaw (2005) explain teacher autonomy as teachers' capacity to control themselves and the learning environment. According to Little (1995), autonomy can be expressed as teachers' capacities for independent actions, reflective thinking and objectivity. Ingersoll (1997) defines teacher autonomy as teachers' ability to make joint decisions in planning the learning process and in instructional matters. Benson (2010) regards autonomy as teachers' ability to make free decisions independently of external control and pressure. Edgar and Warren (1969), however, express teachers' control over a specific area of duty and ability to make their own decisions as active autonomy.
In its broadest sense, teacher autonomy is defined as levels of freedom possessed by teachers in determining curriculum outcomes and contents, choosing course books, selecting teaching methods, and deciding on assessment activities (Eurydice, 2008). Teacher autonomy is regarded as one of the professional qualities (Eraut, 1994). A teacher's right to have control over his/her own practices constitutes an important dimension of autonomy (Sachs, 2000).
According to Friedman (1999), besides being active in educational activities, teachers should also be active in planning, developing and directing all the instructional processes. Teachers' natural leadership and autonomy over teaching processes will become more meaningful with an increase in their autonomy in the other areas.
Teacher autonomy can be examined in five different levels (Freidman, 1999): No autonomy: Teachers' views are not consulted in relation to planning, implementing and evaluating instruction or to participation in school processes, nor are they allowed to display autonomous behaviours.

Scant autonomy:
Teachers are allowed limited authority within the boundaries of teaching programmes defined by school administrators, and are granted a weak area of choice.
Moderate autonomy: Teachers are permitted to make different plans, generate new ideas and develop programmes, but to implement these, a stringent procedure is applied and they have to obtain the necessary permission.
High autonomy: Teachers are granted considerable authority to develop and implement new teaching programmes, plans and methods, within the boundaries of general regulations and principles.
Complete autonomy: Teachers are granted complete freedom to develop and implement new teaching programmes, plans and methods within the framework of generally accepted moral and legal principles.
The rigid, centralised structure of education systems, inadequacy of school-based curriculum development, existence of a centralised exam system, and lack of authority in determining outcomes and contents, all prevent teachers from having a high degree of autonomy. Despite all these obstacles to autonomy, teachers have much greater authority in the curriculum implementation process. When teachers are in charge of instruction, they develop analytical and reflective strategies for the learning process, they do not remain limited to the framework drawn up by the curriculum, and they implement different instructional activities effectively. Teachers' regarding themselves as a competent authority and their direction of the learning process with their own decisions and creation of personalised rules in class are an indicator of their autonomy (Franklin, 1988). Teachers' determination of work methods and resources and their consideration of in-class practices and lesson planning are associated with their autonomy (Burkert and Schwienhorst, 2008).

Research Model
In this study, validity and reliability analyses are made for a measurement tool developed with a survey model in order to determine perceived autonomy over the curriculum among science teachers employed in secondary schools located in the province of Izmir, Turkey.
Factor analysis was used to determine the structural validity of the scale. Factor analysis examines whether scores obtained from a scale measure the characteristic that the test wishes to measure (Büyüköztürk, Kılıç-Çakmak, Akgün, Karadeniz, and Demirel, 2011). Moreover, factor analysis also enables a large number of items included in the scale to be expressed with a smaller number of new data structures (Karasar, 2003;Özdamar, 1999).
In this study, since the factors of the scale used were not known prior to analysis, structural validity analyses were begun with Exploratory Factor Analysis (EFA) with the aim of revealing the factor structure of the scale. In the EFA performed in this study, principal components analysis, which is frequently consulted as a factor extraction technique, was used (Büyüköztürk, 2014). In the process of forming the factors in the factor analysis, the following criteria were taken into consideration: 1. The factor loading values of the items should be .40 or greater (Büyüköztürk, 2014;Şencan, 2005), 2. There should be a difference of .10 or more between the loading value of a factor and the loading values of other factors found in the items (Büyüköztürk, 2014;Çeçen, 2006), 3. Items grouped under each factor separately should be consistent with each other in terms of meaning and content (Çeçen, 2006), 4. The eigenvalue of each factor should be at least 1 or higher (Büyüköztürk, 2014;Çeçen, 2006;Şencan, 2005), 5. There should be at least 3 items in each factor (Şencan, 2005).
In order to confirm the factors that emerged as a result of the EFA, Confirmatory Factor Analysis (CFA) was performed on the same sample group. Performing EFA and CFA on the same sample does not create a problem (Jöreskog and Sörbom, 1993;Thompson, 2005).

Study Group
The study group of the research consisted of 178 science teachers who were employed in secondary schools located in the province of Izmir, Turkey, during the 2018-2019 academic year. Information related to the study group of the research is included in Table 1.

Development Process for Curriculum Autonomy Scale
The Curriculum Autonomy Scale was prepared with the aim of determining science teachers' perceived autonomy over the curriculum. Following a review of the literature, an item pool of 50 items considered to be related to teacher autonomy was created. The scale was prepared as a 5-point Likerttype, and is scored as "Never" (1), "Rarely" (2), Sometimes (3), Frequently (4) and "Always" (5).
During the development of the scale, an item pool of 50 items aimed at determining teachers' perceptions of curriculum autonomy was created. Of these items, which were examined by 3 academicians who are experts in the field of scale development, 12 were removed from the form, as they were considered not to be related to autonomy. The form consisting of the remaining 38 items was administered to 8 science teachers as a pilot study. Following the pilot study, it was decided to remove 9 more items from the scale. Thus, a scale form consisting of 29 items was obtained. This form was administered to 185 teachers employed in 52 different schools in 8 districts of Izmir province during the 2018-2019 academic year. It was determined that 7 teachers gave the same answer to all items, and since reliability could not be ensured, these forms were not included in the research. Consequently, analysis was performed on the remaining 178 teacher forms. In the literature, sample sizes between 100 and 200 are stated to be adequate, especially when the factors are strong and distinct (Büyüköztürk, 2002).

Data Analysis
Validity and reliability analyses of the scale were carried out in line with the data gathered from the 178 science teachers who participated in the research.

Reliability Analysis
During the reliability analysis of the scale, item-total test score correlations and Cronbach's alpha internal consistency coefficients were examined with the SPSS 23.0 software program. The Cronbach alpha internal consistency coefficient provides an insight into correlation between items forming a scale (Tan, 2016). Alpha coefficient values of .70 and over show that the reliability level is adequate. The item-total test score correlation explains the relationship between item scores and scale scores. A high, positive item-total test score correlation is interpreted as a good level of internal consistency (Büyüköztürk, 2014).

Validity Analysis
Factor analysis was used in the structural validity analysis of the scale. Firstly, the Kaiser-Meyer-Olkin (KMO) test of sampling adequacy and Bartlett's test of sphericity were used with the SPSS 23.0 software program to determine whether or not the scale was suitable for factor analysis. Once it was ascertained that the scale was suitable for factor analysis, Exploratory Factor Analysis (EFA) was performed and Varimax orthogonal rotation was applied to ensure that factor variances with fewer variables were maximised (Tavşancıl, 2014). The sub-dimensions of the scale were defined as a result of the EFA. Then, Confirmatory Factor Analysis (CFA) was performed with LISREL 8.0 software and the validity of the relevant dimensions was confirmed.

Findings
First of all, prior to the exploratory factor analysis, the standard deviation value, which gives an idea about the reliability of items, and the anti-image matrix, which shows partial correlation among items, were examined (Hair et al., 2010). Standard deviation values close to 1.00 were obtained for the items. Since no coefficients below .50 were found for the anti-image correlations, all items were included in the analysis.
Prior to factor analysis, the Kaiser-Meyer-Olkin (KMO) test for adequacy of sample size, and Bartlett's test of sphericity for determining whether data come from multivariate normal distribution, were used (Arıkan;2012;Seçer, 2015). In the KMO, a value between 0 and 1 is taken, and a value close to 1 gives the idea that data are adequate. It is recommended that the KMO value should be at least .70 and above, and that for the most suitable data set, a value of .80 is taken as the basis (Arıkan, 2012;Büyüköztürk, 2014;Seçer, 2015;Şencan, 2005). In Bartlett's sphericity test, the degree of significance is examined, and a value below .05 is interpreted as suitable for sample size factor analysis. According to the values obtained (Kaiser-Meyer-Olkin Test = .806, Bartlett's Sphericity Test = 786.703, sd= 78, p= .000), the data were determined to be suitable for factor analysis, and EFA was begun.

Findings Related to Exploratory Factor Analysis
The specific number of factors obtained with EFA were subjected to axis rotation. Axis rotation reveals which items have stronger relationships with the determined factors. In this study, Varimax, which is one of the frequently used orthogonal rotation techniques, was utilised (Büyüköztürk, 2014;Özdamar, 1999).
EFA was performed on the scale consisting of 29 items, and items were loaded onto 8 factors. By paying attention to cases of items loaded on more than one factor and of items loaded on two different factors with a difference of less than .10, the EFA was repeated several times by means of item removal. As a result of the analysis, the items numbered 4, 5,7,8,9,10,17,18,19,20,21,22,23,25,27 and 28 were removed from the scale because, despite having high loading values in different factors, they overlapped. After these items had been removed from the scale, a KMO value of .81 and a Bartlett's sphericity test value of 786.70 (p=.000) were calculated for the remaining 13 items. Finally, a structure made up of 13 items and 4 factors appeared. In addition, the scree plot was examined with the aim of confirming the number of factors in the scale (Figure 1). As can be seen in Figure 1, the slope of the line levels off after the fifth point. By counting the intervals between the points up to the fifth point, the scale was interpreted as having a four-factor structure.
The factor loadings, anti-image correlation coefficients, mean variances, means, standard deviations and item subscale total correlations that were formed following the EFA are shown in Table  2.  The dimensions that include the scale items formed as a result of the exploratory factor analysis were named. At this stage, known as "labelling", naming by using terms that are familiar and include meaningful expressions is taken as the basis. The dimensions that include the items of the scale and the names given to these dimensions are shown in Table 3.

Findings Related to Confirmatory Factor Analysis
Confirmatory factor analysis (CFA) of the Curriculum Autonomy Scale was performed on the same sample for which the EFA was carried out (n=178). The values obtained from the confirmatory factor analysis and the intervals for the acceptability limits of these values are shown in Table 4.  (2003) When Table 4 is examined, c 2 /sd, CFI and RMR are within the limits of good fit indices, while GFI, AGFI, RMSEA and SRMR values are within the limits of acceptable fit indices. In this case, it is seen that the four-factor structure of the scale is confirmed. The path diagrams revealed by the confirmatory factor analysis are shown in Figure 2 and Figure 3. In the path diagram displayed in Figure 2, the error variance and factor loading values of the 13 items in the scale are shown. Accordingly, it is seen that the standardised factor loading values of the scale items range between .57 and .84. Factor loadings of all items are greater than .40 and show a good fit with the four-factor structure. The factor loadings of all items are statistically significant (p <.01). Moreover, it is seen that the highest correlation is between Factor 1 (autonomy in professional development) and Factor 2 (procedural autonomy), while the lowest correlation is between Factor 1 (autonomy in professional development) and Factor 3 (evaluation autonomy). In the path diagram displayed in Figure 3, it is seen that no red arrow is found related to t values, and that therefore, all items are significant at a level of .05 (Jöreskog and Sörbom, 1996). It was determined that the t values in the latent variables of the factors range between 7.31-11.25 and that since they are greater than 2.76, they are significant at a level of .01 (Schumacker and Lomax, 2010).

Discussion and Conclusion
The main aim of this study was to determine the validity and reliability of the Curriculum Development Scale developed to reveal science teachers' perceptions of autonomy. With this aim, the research findings that were obtained are discussed.
Teacher autonomy is teachers' ability to make decisions in planning the learning process and in instructional matters (Ingersoll, 1997). Considering the role of teacher autonomy in developing educational environments, it is very important for teachers to display autonomous behaviours (Freidman, 1999;Türk Eğitim Derneği [TED], 2015). Blumber, Wayson and Weber (1969) state that teachers who have a say in important matters and participate in management work more productively. Moreover, teachers' autonomous decisions also increase their job satisfaction (Öztürk, 2011). When all these characteristics that define autonomy are considered, revealing teachers' perceptions of autonomy becomes important.
When the studies conducted on the subject of autonomy are examined (Çolak and Altınkurt, 2017;Friedman, 1999;Ulaş and Aksu, 2004), it is seen that autonomy is discussed in several dimensions. However, in these studies, curriculum autonomy is dealt with as a general phenomenon, and an approach to autonomy that also covers specific areas of the science curriculum (experimentation, observation, projects, etc.) is not reflected in the research. Therefore, in this study, it was considered important to develop a scale aimed at revealing science teachers' levels of perceived autonomy.
A 50-item pool was created in the scale development process, and following expert examination and a pilot study, the number of items was reduced to 29, while after the necessary analyses had been made, the overlapping items were removed and a 13-item scale was obtained. In this study, a Cronbach's alpha internal consistency coefficient of .82 was calculated for the scale. Exploratory Factor analysis was applied to the scale, and the suitability of the sample size was approved with KMO and Bartlett statistics (KMO= .81, x2 = 786.70, p= .000). As a result of the exploratory factor analysis, 4 dimensions were obtained for the scale, namely Autonomy in Professional Development, Procedure, Planning, and Evaluation. Autonomy in professional development was found to be consistent with Ulaş and Aksu (2015) and Friedman (1999). Procedural and planning autonomy were found to be consistent with Ulaş and Aksu (2015), Friedman (1999) and Pearson and Hall (1993). However, no study revealing the dimension of evaluation autonomy could be found. When the sub-dimensions and items of the scale were examined, it was seen that the items including areas specific to the subject of science (experimentation, observation, projects, etc.) were eliminated. This leads to the conclusion that the basic dimensions of teacher autonomy do not differ depending on different branches (Friedman, 1999;Pearson and Hall, 1993;Ulaş and Aksu, 2015). In order to test the suitability of the model related to Exploratory Factor Analysis, Confirmatory Factor Analysis (CFA) was applied. When the chi-square, GFI, AGFI, CFI, RMR, SRMR, and RMSEA fit indices were examined, it was seen that the CFA results supported the four-factor model.
In conclusion, it is thought that the Curriculum Autonomy Scale that has been developed within the scope of this study, and whose validity and reliability have been established, can be used to determine perceptions of autonomy not only of science teachers, but also of teachers in many other branches.