Learning How to Learn Scale: A Study of Validity and Reliability

The aim of this study is to develop a valid and reliable scale that can be used in investigating learning process of secondary school students. The study was carried out with students attending to different secondary schools in Mardin in 2018. Survey method was employed, and sampling was chosen with the help of simple random sampling method in this study. In order to create an item pool, 26 students having education in the 10 th grade were asked to write an essay about the concept and process of learning how to learn. Then the literature was searched, and a 5-point likert-type 66-item draft scale was formed out of the data obtained from this essay. The draft scale was revised in accordance with the expert views. The data achieved after the application of the scale were analysed using SPSS and AMOS programs. In the analysis, item total correlation, exploratory and confirmatory factor analysis and reliability analysis techniques were benefited. According to analyses performed, it was seen that the scale was of 21 items and had a structure of 5 factors, and the variant ratio that it explained was 52.630% and that also fit index values were acceptable, and they were at an excellent harmony level. It was observed that there was a positive direction and medium level relation Among all factors. Cronbach alpha internal consistency reliability coefficient of the scale changes between 0.655 and 0.864. As a result, it is comprehended that Learning How to Learn Scale prepared as 5 Likert type can be employed as a valid and reliable measuring tool for students attending secondary school according to findings obtained.


Introduction
Learning how to learn necessitates to know learning strategies an individual has and to be aware of strengths and weaknesses and to investigate appropriate education, guidance or opportunities (MEB, 2019). Competence of learning how to learn is the ability to continue learning and organize their own learning, including effective management of time and information, individually and in groups. These abilities consist of awareness of learning process and needs and determining available opportunities and overcoming the obstacles of a successful learning. In learning how to learn approach, students' previous learning and experiences are also important so that they will be able to achieve knowledge and skills (Education Council, 2006). Learning how to learn is expressed to be an experiential process based on interrogation (Priestly & Humes, 2010) and to be able to produce the necessary knowledge fpr new situations using available information (Özden, 2013: 77) and to provide the achievement of effective learning qualities (Özer, 2003). In addition, when learning how to learn approach is effective when it contains emotional and social aspects such as autonomy, selfconfidence, relations depending on corporation among the participants of process and participation of families and community (Moreno & Martin, 2007).
Learning how to learn is very important for motivation, confidence and competence of the individual (Education Council, 2006). The skill of learning how to learn takes place among the properties of student oriented education. Among the other characteristics of the students, the ability to have scientific thinking, produce knowledge and use this knowledge, have the ability to communicate and learn to learn (MEB, 2007). The skills that the students should possess the ability to think scientifically and can produce knowledge and can use this knowledge and the approach of learning how to learn are accepted among these characteristics (MEB, 2007). It is necessary that the student receive support to be equipped with the competence of learning how to learn. Attaining this competence is a case which student cannot overcome alone. The reason for this is that this competence necessitates inquiry of competence, examination, investigation, high motivation, patience and continual learning (Özdaş, 2018). Since learning how to learn is a mental activity, this skill cannot be assessed as any kind of knowledge or method. Learning how to learn is possible through the fact that the individual can use his/her intelligence and develop his/her thinking ability (Özden, 2013). Therefore, individual motivation plays a crucial role in continual learning how to learn and it's maintaining (MEB, 2019).
Learning how to learn is important as a requirement of living in an information society.
Knowledge bases of all disciplines and tools and methods used in these knowledge bases may change in a short time. In this process, the most important skill that the school will equip students with is to create desire in them to learn (Özden, 2013). The information society needs individuals who process and use the knowledge rather than the individuals who store the knowledge (Doğanay, 2012). It is Educational Policy Analysis andStrategic Research, V 15, N 3, 2020 © 2020 INASED impossible for the school to equip the students with enough information at a time when information changes rapidly. Instead of uploading knowledge to the students of the school, it is necessary to educate them in order to adapt to new situations and learn the information they need (Özden, 2013) because school is not just a place where knowledge is learned. The school is also the place where the individual is educated about the behaviours that he / she will gain to become a productive, happy and successful member of the society (MEB, 2007). To provide students with the necessary knowledge and skills to develop their interests and abilities and to prepare them for life in this way are among the main duties of the school (MEB, 1973). In this process, the school should encourage individuals to be more conscious and selective, create an environment for how they should be learned, enable them to explore in learning, and perhaps most importantly, teach systematic ways of accessing information (Balay, 2004). Learning how to learn consist of social, cultural and cognitive characteristics (Hoskins & Frederiksson, 2008). Within this context, learning how to learn is a very crucial tool for life-long learning. Therefore, there is a great need for providing learning environment to improve this qualification in education and training for all citizens including the ones who have less opportunity (Frederiksson & Hoskins, 2008).
Today, how to access information, structuring the acquired information and generating new information have gained significance; not what information is. Therefore, basic learning skills that should be prioritized in schools are active learning, problem solving, learning how to learn, cooperation and communication skills. These learning skills are important in terms of achieving and maintaining 21 st century skills (Louis, 2012;Cited in Bozkurt & Çakır, 2016). In this context, it is important how learners experience the learning-teaching process. In recent years, it has been observed that researchers have been carrying out studies to develop various data collection tools in order to reveal the learning-teaching process of individuals. Some of these studies can be given as examples as follows: Learning School Scale (Uğurlu, Doğan & Yiğit, 2014), M-Learning Attitude Scale (Çelik, 2013), Metacognitive Learning Strategies Determination Scale (Gündoğan-Çögenli & Güven, 2014), Learner Autonomy Support Scale (Oğuz, 2013), Teacher Candidates Competency Scale on Constructivist Learning (Yeşilyurt, 2012), Attitude Scale for Problem Based Learning (Turan & Demirel, 2010), Development of Metacognitive Learning Strategies Measurement Tool (Namlu, 2004), Attitude Scale for Learning (Kara, 2010), Lifelong Learning Competence Scale (Uzunboylu & Hürsen, 2011 (Büyüköztürk, Akgün, Özkahveci & Demirel (2004). It is observed that great majority of these studies were conducted in different populations with students attending to different departments of universities and university teachers.
As a result of these studies, it is seen that valid and reliable scales in terms of psychomotor properties which can be employed in the fields of strategies, styles, competency, metacognitive, selfregulation, lifelong learning, sensational characteristics (attitude, perception, motivation) during the process of learning and teaching that the individuals and students use in learning and teaching process have been achieved to literature. In this study, on the other hand, it was tried to develop a data collecting tool that can be used in investigation of learning processes for students attending secondary schools. Thus, this study was designed as a validity and reliability work of learning how to learn scale.

Method
Survey method was used in this study. Survey is a method which is not an experimental type, and which is carried out to collect data by employing questionnaires or interview protocols to establish the characteristics of a group (Christensen, Johnson and Turner, 2015;Büyüköztürk, Kılıç-Çakmak, Akgün, Karadeniz & Demirel, 2012). Survey method is an approach that aims at describing a case of past or present as it is (Karasar, 2005). Via this method, it is tried to explain the attitude of people combining their perceptions, views, attitudes and beliefs about an issue. Within this context, data are collected through survey/scale or interview (Toker-Gökçe, 2018). Therefore, a draft scale was prepared as a data collection tool for the development of learning how to learn scale.

Study Group
The study group was selected by simple random sampling method. In this type of sampling, each unit in the population is likely to be equal and independent in sampling (Balcı, 2018). In other words, all individuals have the same chance of being elected. In addition, the main feature of this method is that the sampling has a high power to represent the population (Büyüköztürk et al, 2012).
The study group chosen for the validity and reliability study of the learning how to learn scale consists of students (N=304) studying in different secondary schools in Mardin / Artuklu district in the 2018-2019 academic year. The students of the study group were also selected by using simple random sampling method (Table 1). When the characteristics of the study group were examined, it is seen that 41,1% of the participants were female, 58,9% were male, 23,4% were industrial vocational high schools, 31,9% were Anatolian high schools, 31,9% were Imam Hatip Anatolian High schools, and 12,8% of them are in vocational technical high school students.

Scale Development Process
The stages that should be followed in the scale development process were taken into consideration. The following stages were performed: 1 st defining the problem / setting goals, 2 nd forming scale items and seeking expert opinion, 4 th pre-application, 5 th validity and reliability (Erkuş, 2014;Tezbaşaran, 2008;Büyüköztürk, 2005).,

Problem Defining / Determining Goals
Knowledge, skills and attitudes related to learning how to learn take place among the qualifications in the curriculum developed by the Ministry of National Education. Students are expected to acquire these qualifications in primary and secondary education programs (MEB, 2018).
Therefore, a need arises to reveal knowledge, skills, attitudes and qualifications related to learning how to learn. In this study, it was targeted to develop a valid and reliable scale about learning how to learn.

Forming Scale Items and Expert View
In order to form the item pool of the draft scale for learning how to learn, the concept of learning how to learn was first explained to 26 students in the 10 th grade in secondary education and then asked to write essays on this subject. In the second stage, a literature review was conducted (MEB, 2018; European qualifications framework, 2018; Özdaş, 2018;Oral & Ayaz, 2016;Özden, 2013;Doğanay, 2012;Fredriksson & Hoskins, 2007;MEB, 2007;Moreno & Martín, 2007;Özer, 2003). As a result of the written opinions of the students and literature review, 86 items were formed.
Draft scale items were sent to four experts (1 Assoc. Prof. and 3 Dr. Lecturer) for their understanding of comprehensibility and clarity. After the feedback from the experts, the expressions of some items on the draft scale were changed and some items were removed from the draft scale. As a result, 66 items remained on the draft scale of learning how to learn. Each item in the learning how to learn scale form was converted into 5-point Likert-type expressions such as Never (1), Rarely (2), Occasionally Educational Policy Analysis and Strategic Research, V 15, N 3, 2020 © 2020 INASED (3), Mostly (4), Always (5).

Conducting the Application
Necessary permissions were obtained from Mardin Provincial Directorate of National Education for implementation. The schools were visited by the researchers in Spring semi-term of 2018-2019 education year and the implementation of draft scale. The students were given 30 minutes for implementation.

Data analysis
SPSS packet program and AMOS program were used for data analysis. The data were analysed by employing item total correlation, exploratory factor analysis and confirmatory factor analysis techniques.

Item Analysis
Item analysis was conducted to examine the relationship between the scores obtained from the items in the draft scale and the total score of the scale prepared to improve the learning to learn scale.
Item analysis was performed with corrected item-total score correlations from analysis techniques.
Büyüköztürk (2010)  and M66 were found to be greater than 0,30. However, these 10 items were excluded from the scale because they did not meet the relevant criteria (Table 2).

Construct Validity
In scale development and adaptation studies, factor analysis is the most commonly used method to obtain data on the construct validity of a scale (Seçer, 2017). Factor analysis aims to find a small number of conceptually significant new unrelated variables by bringing together p variables (Tekindal, 2009;Büyüköztürk, 2010). Factor analysis is conducted as exploratory and confirmatory factor analysis (Balcı, 2018;Seçer, 2017;Büyüköztürk, 2010). Exploratory and confirmatory factor analyses were conducted to determine the construct validity of the data obtained from the draft form of the learning how to learn scale.

Exploratory Factor Analysis (EFA)
If the number of structures or factors thought to be measured by the measurement tool is not known, within this context, in cases where it is not possible to test a certain hypothesis, exploratory factor analysis is performed in order to obtain information about the nature of the factors measured by the measurement tool (Balcı, 2018;279). In this analysis, there is a process to find and define factors based on the relationships between variables (Tekindal, 2009;Büyüköztürk, 2010). In other words, exploratory factor analysis is done with the aim of determining how many subheadings of the items in a measurement tool can be collected and what kind of relationship there is between them (Seçer, 2017).
The construct validity of the learning how to learn scale was carried out in four stages such as the examination of the suitability of the data for factor analysis, obtaining factors, rotating the factors and naming the factors (Kalaycı, 2005).
While factor analysis is carried out in a scale development study, initially, scale data are controlled whether they are suitable for factor analysis. With this aim, the adequacy of the sample size to which the scale is applied should be considered. For the adequacy of the samplings used in the investigations, Kaiser-Mayer-Olkin (KMO) coefficient, namely sampling adequacy, and Barlett test are used (Büyüköztürk, 2005;Metin, 2015). That KMO coefficient value is greater than 0,60 and Barlett test is meaningful with the data is an indicator that they are suitable for factor analysis (Büyüköztürk, 2010). In addition, in factor analysis, reaching between 300-500 individuals for sample size and reaching 5 or 10 times more individuals can be used for qualification (Seçer, 2017). The data obtained to improve the learning how to learn scale was found to be suitable for factor analysis. In addition, it is observed that the number of students (N=304) and sampling size were in the recommended range.
Educational Policy Analysis and Strategic Research, V 15, N 3, 2020 © 2020 INASED Principal Component Analysis was used in exploratory factor analysis to create factors.
According to Büyüköztürk (2010), in the exploratory factor analysis, the following criteria are generally taken into consideration for the removal of items that do not measure the same structure: Load values at the factor in which the substances are located are high: Factor load is the relationship between the factor and the factor that is tried to be measured by means of an item. It was stated that factor load value of 0, 45 and higher would be a good measure for selection (Büyüköztürk, 2010;Can, 2014).

It should be taken into consideration that the load values of the items in the factors should
have high load values in one factor and low load values in the others.
Here, it has been suggested that the difference between a factor in which an item gives a high load value and a load value in a second factor should be at least 0, 10. Similarly, it was stated that items with high load values in more than one factor should be removed from the scale (Çokluk, Şekercioğlu & Büyüköztürk, 2014;Büyüköztürk, 2010).
Factors with eigenvalues greater than or equal to 1 were also considered as important factors for the number of factors (Büyüköztürk, 2010).
When exploratory factor analysis was performed, it was seen that the items were collected under In order to decrease the factors, the process repeated 6 times. As a result of this process, in each factor, Varimax Vertical Rotating (Büyüköztürk, 2010) analysis technique was employed so that the items exhibiting high relations could be brought together. It was found that the 21 items remaining in learning how to learn by Varimax rotation were distributed under 5 factors (Table 3). According to the result obtained from rotated component matrix of exploratory factor analysis, it is seen that in factor 1, there took place 4 items (factor load value is between 0,767-0,623), and in factor 2, there took place 5 items (factor load value is between 0,662-0,470), in factor 3, there took place 4 items (factor load value is between 0,736-0,607), in factor 4, there took place 4 items (factor load value is between 0,776-0,586), in factor 5, there took place 4 items (factor load value is between 0,721-0,577) (Table 4).

Table 4. Variance explanation percentages of factors
In factor analysis, the concept of eigenvalue is a condition that shows the variance explained by a factor alone. In factor analysis, the eigenvalue of a sub-dimension is expected to be at least 1 (Seçer, 2015;Büyüköztürk, 2010) and each sub-factor is expected to explain at least 5% of the total variance in the scale (Seçer, 2015).
As a result of varimax rotation applied to the learning to learn scale, it was explained that in eigenvalue greater than 1 and consisting of a total of 21 items, 5-factor structure, factor1 total variance was 27,339%, factor2 was 7,359%, factor3 was 6,765%, factor4 was 5,695% and factor5 5,473%. There are various views about total variance value which should be explained. However, as general tendency, it can be said that the variance rate explained in a measurement tool should definitely be higher than that of unexplained variance rate (Seçer, 2015). As a result, according to this statement, it is seen that the explained variance rate of learning how to learn scale is at the expected level.

Reliability Study
Cronbach's alpha reliability coefficient is a statistical technique used to determine the internal consistency of a test (Demircioğlu, 2012 learning scale was found to be 0,758 for factor1, 0,710 for factor2, 0,655 for factor3, 0,6766 for factor4, 0,650 for factor5 and 0,864 for the whole scale. It is accepted that Cronbach Alpha coefficient is within the range of 0,70 and above (Christensen, Johnson & Turner, 2015;Büyüköztürk, 2011;Tekindal, 2009;Şencan, 2005). However, in some sub-dimensions of the scale, it is seen that Cronbach Alpha coefficient is slightly below this value range. Although the internal consistency coefficients of the factors are not very high, they can be considered sufficient for such scales that try to measure affective sensational characteristics (Özbek & Bindak, 2018). When these data are examined, it is seen that the Cronbach Alpha coefficient of the developed learning how to learn scale is in a reliable range for all factors (Table 5). The correlation coefficient is used to find and interpret the amount of the relationship between the two variables (Büyüköztürk, 2010). Pearson correlation coefficient was used to correlate the factors of the learning to learn scale. In the research, when the correlation coefficient was found to be between 1, 00-0, 70, it was interpreted as high, and when it was between 0, 69-0, 30, it was medium; and when it was 0, 29-0, 00, was interpreted as a low level relationship (Büyüköztürk, 2010).
According to this, it is seen that there is a weak positive relationship between F4 and F1 (r = 0,370, p <0, 05) and F4 and F3 (r = 0,355, p <0,05). Again, it is seen that there is a positive relationship between all other factors except these factors. This result is that scale factors are in a positive direction and correlated with each other (Table 6). Educational Policy Analysis and Strategic Research, V 15, N 3, 2020 © 2020 INASED

Naming of Factors
As a result of exploratory factor analysis and reliability analysis of the learning how to learn scale, 21 items were named collected in 5 factors. According to this, Factor1 was named as Learning/Thinking Style (Ö/DS), Factor2 was named as Theoretical and Practical Unity in Learning (ÖTPB), Factor3 was named as Individual Awareness in Learning (ÖKF), Factor4 was named as Emotionality in Learning (ÖD).

First Level Confirmatory Factor Analysis
In the scale development and adaptation process, confirmatory factor analysis is based on an advanced technique depending on testing the theories of latent variables of a model or structure previously determined by exploratory factor analysis (Seçer, 2017;Çokluk, Şekercioğlu & Büyüköztürk, 2014). In this analysis, the researcher demonstrates whether the data in his hand conforms to the original structure. In other words, he wants to show whether the data he has in his hand is compatible with the factor structure that was previously constructed. The aim here is to test factor structure of the variable (Meyan & Şeşen, 2015).

Discussion, Conclusion and Recommendations
This study was carried out in order to develop a valid and reliable scale on learning how to learn. For the purpose of the study, in order to form the item pool of the scale for learning to learn, 26 students in the 10 th grade of secondary education were first explained the concept of learning how to learn, and then they were asked to write essays on this subject. In the second stage, literature review was done. As a result of the opinions obtained, a draft scale of 86 items was formed. The draft scale was shared with the instructors who are experts in the related field and their opinions were obtained.
As a result of expert opinions, 22 items from the draft learning scale were deleted and 66 items remained. Each item in the learning how to learn scale form was transformed into 5-point Likert-type expressions such as Never (1), Rarely (2), Occasionally (3), Mostly (4), Always (5). After the permissions obtained, the scale was applied.
Theoretically proposed analysis techniques were considered in order to develop the learning how to learn scale. Learning to learn items were analysed by item total correlation. The total correlation criterion of at least 0.3 items proposed for each item was considered and as a result, 10 items were eliminated.
The learning how to learn scale was analysed by Kaiser-Mayer-Olkin (KMO) coefficient, that is to say, sample adequacy and Barlett test. KMO coefficient of the scale was obtained as 0,870 and it was seen that the Chi-square value of the Barlett test was also statistically significant (χ 2 =1602, 424; p <0, 01). These results can be interpreted as the KMO coefficient is excellent. Principal Component Analysis technique was used for factor extraction for the scale. In the analysis, it is considered that the load values of the items in the factor in which they are located is 0, 45 and higher will be a good measure for selection. In addition, it was taken into consideration that the difference between a factor that gives a high load value of a substance and a load value given in a second factor should be at least 0,10. In order to determine the number of factors, the factors with eigenvalues greater than or equal to 1 were considered as important factors.
According to the results of the first exploratory factor analysis, the items were collected in 15 factors and the last 2 factors included at most 1 item. 35 items, which do not meet the criteria load factor of 0,45 and are also considered to be overlapping items, were eliminated by factor reduction because they were included in more than one factor with a difference of less than 0.10. This process was applied six times in total to reduce the number of items. However, 10 items were eliminated from the scale with total item correlation. As a result, 45 items were eliminated from the learning how to learn scale.
As a result, it was seen that 21 items remained in the developed learning how to learn scale and these items were collected under 5 factors. Varimax Vertical Rotation technique analysis was used to bring together the items that are highly correlated with each factor.
Educational Policy Analysis andStrategic Research, V 15, N 3, 2020 © 2020 INASED When the data obtained from the rotated component matrix were examined, it was seen that 4 items were collected in the Ö/DS factor, 5 in the ÖTPB factor, 4 in the ÖKF factor, 4 in the Öİ factor and 4 in the ÖD factor. In addition, it was observed that the item load values were between at least 0,470 and 0,767 for all factors.
Factor1 was named as Learning/Thinking Style (Ö/DS), and Factor2 was named as Theoretical and Practical Unity in Learning (ÖTPB), and Factor3 was named as Individual Awareness in Learning (ÖKF), and Factor4 was named as Willingness in Learning (Öİ), and Factor5 was named as Emotionality in Learning.
As a result of varimax rotation applied to the learning to learn scale, the Ö/DS factor with an eigenvalue greater than 1 in 5 factored structures explains 27,339% of the total variance, 7,359% of the ÖTPB factor, 6,765% of the ÖKF factor, 5,695% of the Öİ factor and 5,473% of ÖD factor. 52,630% of the total variance was explained with the help of 5 factors.
The developed learning how to learn scale was between 0.655 and 0.864 for the Cronbach Alpha coefficient. It is understood that this value is a highly reliable scale by having a value in the value range of Cronbach Alpha coefficient which is theoretically suggested.
When Pearson correlation coefficient for learning how to learn scale is examined, it is seen that there is a positive relationship between all factors. However, it can be said that this relationship is weak and moderate.
For the confirmatory factor analysis, first level confirmatory factor analysis was performed for the model which consisted of 5 latent variables (Ö/DS, ÖTPB, ÖKF, Öİ, ÖD) and 21 observed variables which were originally determined in the original scale in the study. When the fit indexes of first level confirmatory factor analysis were examined, it was found that the obtained conformity index values were [χ2(sd,N)=257,95;179, 304), χ2/sd =1,441, RMSEA=0,0038, RMR = 0,046, IFI = 0,96 NNFI = 0,98, NFI = 0,97, CFI = 0,97, GFI = 0,87, AGFI = 0.86. According to this result, if the ratio of the chi-square value to the degree of freedom is χ2 / df ≤2 =, it can be seen that it is perfectly compatible (Tabachnick and Fidell, 2001;cited, Çokluk, Şekercioğlu and Büyüköztürk, 2014). In addition, it was found that RMSE, RMR, IFI, NNFI, NFI, CFI values exhibited an excellent conformity and GFI, AGFI values displayed an acceptable conformity. As a result, it is seen that the learning how to learn scale obtained with the applications suggested for scale development studies is a valid and reliable scale. This scale is recommended to be used by researchers.