Self-regulated Learning Skills: Adaptation of Scale*

The aim of this study is to adapt and examine the psychometric properties of Achievement Goal Scale (AGS) originally constructed by Elliot and McGregor (2001) and the Motivated Strategies for Learning Scale (MSLQ) originally constructed by Pintrich, Smith, Garcia and McKeachie (1991) in order to measure the self-regulated learning skills of high school students' in a chemistry course. The study group was comprised of 862 high school students attending a chemistry course in different public schools. The construct validity of the sub-scales included in the scales were tested by confirmatory factor analysis. For the reliability studies, the internal consistency coefficient Cronbach's alpha (α) values as well as McDonald's ω (omega) coefficients were calculated. In addition, item-total correlations were calculated for the reliability of each item in the scales. When the confirmatory factor analysis results were examined, it was accepted that the fit indices met the goodness of fit criteria for both the Achievement Goal Scale and Motivated Strategies for Learning Scale. Factor loadings of the items in both scales were statistically significant. These results showed that the Turkish forms of both scales have enough psychometric properties in terms of validity and reliability for a chemistry course.


INTRODUCTION
Self-regulation is a cyclic process that individuals monitor their own behaviours; make a judgement by comparing based on their own criteria and regulate their behaviours. Self-regulated individuals affect, lead and control their own behaviours (Bandura; as cited in Senemoğlu, 2011). According to Zimmerman (2000) self-regulation is the thoughts, feelings and behaviours which individuals develop to achieve their goals and which emerge cyclically. Social cognitive theory contends that selfregulation develops in social environments and is internalised by individuals through time. According to the theory, self-regulation includes cognitive, metacognitive and motivational components in its structure (Zimmerman; as cited in Sakız & Yetkin Özdemir, 2014). Therefore, self-regulated students take on metacognitively, motivationally and behaviourally active roles in the process of learning, they set their own learning goals and they control this process (Zimmerman, 1989). Self-regulation is not defined as a mental ability or as an academic skill but rather as a self-directive process in which learners adapt their cognitive competencies in the form of academic abilities (Zimmerman, 2002).
Most of the learning models which have been developed by researchers in the field of education and which are based on self-regulation are based on Zimmerman's (1989) cyclical model. Pintrich's self- (McCoach & Siegle, 2003;Pintrich & DeGroot, 1990). According to Pintrich (2000a), self-regulation is a process in which learners set goals for themselves, follow them and try to organise their motivation, cognition and behaviours. This process is determined, organised and restricted by learners' goals and by the contextual properties of the environment they are in. Pintrich stresses that self-regulated learning is the learning actualised to develop self-efficacy and states that self-efficacy in addition to motivation is an important component of self-regulation (as cited in Sarı & Akınoğlu, 2009). Garcia and Pintrich also claim that motivation, an important component of self-regulation, is composed of individuals' beliefs about themselves such as personal goals, self-efficacy and value beliefs in addition to their perceptions about the classroom (as cited in Özturan, Sağırlı, Çiltaş, Azapağası & Zehir, 2010).
With the emergence of self-regulation models based on social cognitive theory, the notion of the importance of context in self-regulation processes has emerged. Context can be defined as the circumstances creating an environment for a situation, an idea or an event (Context, 2018). With the emergence of the idea that context can influence the validity of findings, measurements were made sensitive to the context. Thus, measurements for different domains of learning gained more and more importance as cited in Özbay, 2008). Briefly, measurements sensitive to the context and directed to specific areas of learning and specific tasks instead of measurements based on generalisations became more important. Motivated Strategies for Learning Scale (MSLS) developed by Pintrich et. al (1991) is frequently used in the literature. Pintrich et al (1991) chose a course for university students as the unit of analysis in the scale (Özbay, 2008). MSLS was developed on the basis of the view that context had significant effects on the use of motivation and learning strategies and that different strategies should be used in different areas and tasks of learning (Özbay, 2008). MSLS contains five sub-dimensions as the indicators of students' cognitive regulation. They are labelled as rehearsal, elaboration, organisation, critical thinking and metacognitive self-regulation. There are some sub-dimensions on which cognition control activities and monitor measurements in the framework of self-regulated learning model suggested by Pintrich (2000a) and some performance control activities in the framework of the model suggested by Zimmerman (2000) are included. MSLS does not contain sub-dimensions for measuring motivational strategies related to organising motivation and feelings. Yet, there are sub-dimensions such as achievement goals containing performance and mastery, task value, self-efficacy for learning and performance and test anxiety at the forethought stage of Zimmerman's model. In relation to organising behaviours, MSLS includes three sub-dimensions. They are the sub-dimensions of effort regulation, time and study environment management and help seeking. Indeed, two self-regulation models which were developed by Zimmerman and Pintrich and which were based on social cognitive theory lay emphasis on such selfregulation strategies as performance control, time management, help seeking and environmental configuration. Lastly, MSLS contains two more sub-dimensions related to organising the context. They are called peer learning and time and study environment management. They are used to find how well students use their friends as sources of learning and how well they manage their study environment and time (Yumuşak, Sungur, & Çakıroğlu, 2007). Achievement goals included in MSLS influence learners' task determination and problem-solving efforts in addition to their study behaviours and recalling. According to Bandura, individuals' setting goals can cause increase in their motivation (as cited in Driscoll, 2005). When individuals set their goals, they evaluate their performance and their level intrinsically and they decide on the basis of extrinsic criteria. If they cannot attain such a standard, they will insist on their efforts. However, all these goals will not maintain this insistence. Goals set should have certain properties for this. Motivated Strategies for Learning Scale has two types of achievement goals labelled as mastery goals and performance goals. Yet, performance goals are divided into two as performance approach and performance avoidance in the literature (Elliot & Church, 1997;Skaalvik;as cited in Şenler, 2011). In later studies, however, mastery goals are divided into two as mastery approach and mastery avoidance in a similar vein (Elliot, 1999;Pintrich, 2000b). While performance approach goals involve such goals as doing better than others do and being the best, performance avoidance goals involve such goals as avoiding being ordinary. Mastery approach goals aim to learn and understand in depth whereas mastery avoidance goals emphasise not learning and misunderstanding (Elliot & Church, 1997;Elliot & McGregor, 2001;Elliot & Reis, 2003). Therefore, the need for using sub-dimensions for mastery 341 goals and performance goals available in MSLS arises. Achievement goal Scale (AGS) can be used in analysing mastery goals in MSLS as mastery approach goals and mastery avoidance goals and performance goals as performance approach goals and performance avoidance goals in four parts. Thus, Achievement Goal Scale has four components: mastery approach goals, performance approach goals, mastery avoidance goals and performance avoidance goals (Elliot & McGregor, 2001). The other items in the scale are not the items for goal orientation. However, the researchers developing the scale recommend that these items be included and implemented in the scale although they are not used.

Purpose of the Study
The need for making measurements sensitive to the context emerges since context influences the validity of findings. For this reason, measurements directed to different areas of learning have been gaining more and more importance as cited in Özbay, 2008). Yet, it was found in studies that there were no reliable and valid scales for determining high school students' self-regulated learning skills in different courses. Therefore, scales are needed for primarily use in assessment so as to develop students' self-regulated learning skills in chemistry course. Besides, the fact that achievement goals available in MSLS are limited to two goals in the literature made it necessary to use MSLS along with AGS. Therefore, the two scales should be adapted and validity and reliability of the scales should be examined. In line with this need, this study adapts MSLS and AGS into chemistry course and analyses the psychometric properties to determine high school students' self-regulated learning skills.

Research model
This study employs survey model. Survey model is a research approach aiming to describe a situation existed in the past or existing at present as it is. When it is impossible to reach the population, study can be conducted with a small sample taken from the population in survey studies (Fraenkel & Wallen, 2000).

Participants
A total of 862 high school students who were the 9 th , 10 th , 11 th and 12 th graders in differing state schools in Ankara were included in the study. % 35.03 of the participants were female whereas 33.06% were male. In addition to that, 31.9% of the participants did not make any coding for gender. The participants' age ranged between 16 and 20.

Achievement Goal Scale (AGS)
Achievement Goal Scale (AGS) was developed by Elliot and McGregor (2001) and was adapted into Turkish by Şenler and Sungur (2007). The scale was adapted by Şenler and Sungur (2007) into science course and it was administered to primary school students. The 7-pointed Likert type scale was changed into 5-pointed Likert type. This study, on the other hand, adapts the scale into chemistry course for high school students using 7-pointed Likert type as in the original version by getting permission. The scale was administered to 862 students in total.
The scale has four sub-factors. The factor of mastery approach goals included items 1, 6 and 8; the factor of performance approach goals included items 4, 10 and 16; the factor of mastery avoidance goals included items 11, 14 and 17 and the factor of performance avoidance goals included items 2, 7, 13, 19, 20 and 21 (Elliot & McGregor, 2001). The other items included in the 21-item scale were not related to goal orientation. Yet, the researchers who had developed the scale recommended that these items be included in the scale and be implemented although they were not used. Thus, items 15 and 18 available in the scale were in the factor of competence expectancies (Elliot & Church, 1997) and items 3, 5, 9 and 12 were in the factor of challenge and threat appraisals (Elliot & Reis, 2003).

Motivated Strategies for Learning Scale (MSLS)
Motivated Strategies for Learning Scale (MSLS) was developed by Pintrich, Smith and McKeachie (1991) so as to be informed of university students' motivation in classes and of the learning strategies they used in those classes. The scale was adapted into Turkish by Büyüköztürk, Akgün, Özkahveci and Demirel (2004) and Sungur (2004). It is a 7-pointed Likert type scale. It has two main components called motivation and learning strategies. The motivation component is composed of six sub-factors. These are intrinsic goal orientation, extrinsic goal orientation, task value, control of learning beliefs, self-efficacy for learning and performance and text anxiety. The learning strategies part is related to different cognitive and metacognitive strategies students use and contains 31 items. In addition to the 31 items, there are also 19 items related to different resource management strategies. Learning strategies part includes nine sub-factors labelled as rehearsal, organization, elaboration, critical thinking, metacognitive self-regulation, time and study environment, effort regulation, peer learning and help seeking. High scores received from any factor in MSLS indicate that students have high levels of the property related to the factor (Pintrich et al., 1991;Büyüköztürk et al., 2004). Having received the necessary permission, the scale was adapted for use with chemistry course with high school students, and thus it was administered to 862 students.

Data Analysis
Prior to analyzing the data, the items which were stated negatively in the original version of the scale were coded inversely. First order confirmatory factor analysis was conducted for construct validity to see whether or not Achievement goal scale and Motivated Strategies for Learning Scale measured the intended structure. Because factor loadings were not equal, both Cronbach's alpha (α) and McDonald Omega (α) reliability coefficients were calculated so as to determine reliability in the sense of internal consistency. In this study, LISREL software was used for confirmatory factor analysis and SPSS and Excel software packages were used for reliability analyses.

Language Validity
Turkish adaptations of MSLS from English made earlier (Büyüköztürk et al. 2004;Sungur, 2004;Taştan, 2009;Yalçınkaya, 2010) and AQS study in Turkey (Şenler and Sungur) were examined in this study and expert opinion was consulted for the translated items which were determined. Efforts were made to see whether or not the translated items were equivalent to the original items and to see the degree to which the items in the Turkish version were compatible with Turkish grammar and were intelligible. After expert opinion was obtained, modifications were made, the resultant form was administered to a group of high school students having similarities with the students with whom the application would be done. The items of the revised version were checked in terms of content, and the language of the form was modified based on students' feedback.

RESULTS
This section presents the findings obtained from the analyses done for validity and reliability of both scales. The results for confirmatory factor analysis conducted for structure validity of the scales are shown in Table 1. When the fit indices of the Achievement Goal Scale were examined in Table 1 and Figure 1, it was concluded that the values apart from Chi square/df (5.26)-which were fit indices-met the criterion for good fit (χ²/df < 3.0; RMSEA<.08; CFI>.95; IFI >.90; GFI>.90; NFI >.90; AGFI >.85; NNFI >.95; SRMR <.1) (Çelik & Yılmaz, 2013;Schermelleh-Engel, Moosbrugger, & Müller, 2003).  Figure 1 and Table 2 show the variance values described with t values which were found to be significant for each item, the factor loadings (λx) and error variances (δ). Accordingly, it was found that factor loadings in the sub-factors were found to range between .22 and .87. to perform the reliability analysis for the scale, McDonald's coefficient (omega)which is recommended when the factors loading in each factor were not equal-in addition to Cronbach's alpha values was also found (Zinbarg, Revelle, Yovel & Li, 2005). Cronbach's alpha reliability coefficients were found to range between .67 and .85. In addition to reliability analyses, total item correlation suggesting the consistency of each item with the whole factor in which the item belonged was also analysed. It was found in consequence that only the total correlation for item 20 was smaller than .30 yet, some studies in the literature (Briggs & Cheek, 1986, for instance) point out that item correlation coefficient in .15-.50 interval would be sufficient for scales measuring more comprehensive properties. Clark and Watson (1995), on the other hand, state that the values between .15 and .20 would be adequate for total item correlation in scales measuring more comprehensive properties. Since AGS measured the different properties of both mastery and performance, decision was made to include this item in the study.  Following the confirmatory factor analysis conducted for the motivation section of Motivated Strategies for Learning Scale, it was concluded that the values apart from Chi square/df (5.22)-which were fit indices-met the criterion for good fit (See Table 3 and Figure 2 Figure 2 and Table 4 show the variance values described with t values which were found to be significant for each item, the factor loadings (λx) and error variances (δ). Accordingly, it was found that factor loadings in the sub-factors were found to range between .37 and .80. Cronbach's alpha reliability coefficients were found to range between .61 and .87. Following the confirmatory factor analysis conducted for the learning strategies section of Motivated Strategies for Learning Scale, it was concluded that the values apart from χ²/df (3.99), GFI (.83) and AGFI (.81) -which were fit indices-met the criterion for good fit. (See Table 5 and Figure 3). χ²/df (3.99), GFI (.83) and AGFI (.81) (  Table 6 show the variance values described with t values which were found to be significant for each item, the factor loadings (λx) and error variances (δ) for MSLS learning strategies section. Accordingly, it was found that factor loadings in the sub-factors were found to range between .22 and .74. Cronbach's alpha reliability coefficients were found to range between .59 and .84.

DISCUSSION and CONCLUSION
Students' individual differences are the properties that should be taken into consideration in teachinglearning process. This is because the teaching-learning approaches students choose and their responses to teaching change according to the difference in their individual properties. Their individual properties can be divided into cognitive, affective, social and physiological categories. Several factors which can be described as individual differences such as having different levels of motivation, difference in perceptual preferences, intelligence level and psychological factors are influential in individuals' teaching-learning processes (Kuzgun-Deryakulu, 2004). One of those individual differences is students' self-regulated learning skills. Therefore, the validity and reliability of the scales were deemed adequate to reveal students' self-regulated learning skills in teaching-learning environments. The emergence of the view that the importance of contexts in self-regulation processes could not be ignored with the arise of self-regulation models based on social cognitive theory made us feel the necessity for scales which could be used with differing courses. For this reason, this study adapted Achievement Goal Scale and Motivated strategies for Learning Scale for chemistry course and analysed the psychometric properties so as to measure high school students' self-regulated learning skills. The sub-factors in the scales were analysed by means of confirmatory factor analysis. In addition to Cronbach's alpha-which was an internal consistency coefficient-McDonald's Omega coefficient was also calculated. Moreover, total item correlations were also analysed for the reliability of each item in the scales.
On examining the results for confirmatory factor analysis performed for Achievement goal Scale, the fit indices for the scale were found as RMSEA= .07; GFI=.94; NFI= .97; AGFI=.91, NNFI=.97; CFI= .98 and SRMR=.042. An examination of fit indices makes it clear that only chi-square/df ratio is below 3. Yet, on considering the other fit indices, it can be concluded that there is good fit. Garver and Mentzer (1999) state that NNFI, CFI and RMSEA can be used in determining model-data fit.
Considering the acceptability of RMSEA below 0.8 and having RMSEA of 0.7 in this study along with the other fit indices, it was regarded that the model had good fit (Schermelleh-Engel, Moosbrugger, & Müller, 2003). Besides, due to the fact that NNFI and CFI (>.90) had acceptable values in this study, the scale was assumed to have construct validity. It can be said that the fit indices obtained in this study yields results similar to the ones in the original scale and the ones in other adaptations. On examining the results of confirmatory factor analysis performed for Achievement Goal Scale developed by Elliot and McGregor (2001), it was found that Chi-square (48, N=148) = 60.49, p=.11; RMSEA= .042, Tucker-Lewis Index (TLI) = .99 and CFI= .99. Another adaptation made by Pamuk (2014) found, on examining the results of confirmatory factor analysis performed for each subfactor, that fit indices were perfect for three sub-factors apart from the sub-factor of performance avoidance. The fit indices for performance avoidance was reported as Chi-square/df=22.55, NFI=.97, CFI=.97, SRMR=.04 and GFI=.98.
The results of reliability analyses done for Achievement Goal Scale indicated that the Cronbach's Alpha (α) found for mastery approach was .85, it was .79 for mastery avoidance, .77 for performance approach and .67 for performance avoidance. Nunnally (1978) suggested that reliability coefficient be .70 as a general rule. But O'Rourke, Hatcher and Stepanski (2005) pointed out that values below .70 were also adequate and that social scientists even reported values below 60 occasionally (for example Dekovic, Janssens & Gerris, 1991;Holden, Fekken & Cotton, 1991). Therefore, when considered along with all other results for the scale, it was concluded that the factors of the scale satisfied the reliability criteria. Additionally, it was found that the other adaptations of this scale made in Turkey had also calculated similar reliability indices. Cronbach's Alpha-which was the internal consistency coefficient-calculated for Achievement Goal Scale developed by McGregor (2001) ranged between .83 and.87. Şenler and, on the other hand, found that Cronbach's Alpha took on values between .64 and .84. Cronbach's Alpha coefficients ranged between .65 and .76 in Pamuk (2014). Examining the results for the scale and the adaptations made in Turkey, it can be said that achievement objectives, a component of self-regulated learning skills, can be measured more comprehensively with this scale (Şen, 2015). The results of confirmatory factor analysis conducted for the learning strategies part of the scale developed by Pintrich et al (1991) were as in the following: Chi-square/sd = 2.26; RMR=.08; GFI=.78. Sungur (2004) found the values in an adaptation for biology course as: Chi-square/sd = 4.5, GFI = .71, and RMR = .08. Büyüköztürk et al (2004) found the fit indices for the learning strategies part as: Chisquare/sd =4.73, GFI=.80, AGFI=.77, CFI=.70 NNFI=.67 RMR=.22, SRMR=.06 and RMSEA=.07.It was found that the fit indices in the adaptation made in Turkey did not meet the model-data fit values as in the fit indices for the learning strategies part of the model in the scale prepared by Pintrich et al (1991). Considering the original scale and the fit indices of the adaptations in Turkey, it was regarded that the fit indices for the scale met the indices for good fit. Besides, Pintrich et al (1991) state that students' use of strategies differs according to students' individual differences, teachers' properties and the structure of courses; and that therefore researchers consider the values they find as acceptable. For this reason, considering the adaptations made for MSLS (Büyüköztürk et al., 2004;Pintrich et al., 1991, Sungur, 2004 it may be said that the reliability indices found are acceptable.

Journal of Measurement and Evaluation in Education and
Cronbach's Alpha found for the factor of rehearsal of MSLS was .76, it was .68 for the factor of organisation, .78 for the factor of elaboration, .76 for the factor of critical thinking, .84 for the factor of metacognitive self-regulation, .75 for the factor of time and study environment, .69 for the factor of effort regulation, .71 for the factor of peer learning and .59 for the factor of help seeking (Şen, 2015). On examining the adaptations in the literature and the original version, it can be stated that the reliability indices found in this study are higher. Cronbach's Alpha was found as .52-.80 by Pintrich et al (1991), as .57-.81 by Sungur (2004)