Motivation Scale for STEM Fields

In this study, it is aimed to develop a measurement tool to be used to determine the motivation of high school students towards STEM fields. 29-item motivation scale trial form for STEM fields was applied to 462 high school students in their 2016-2017 academic years, studying in four schools of Kayseri. Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) was performed for the construct validity through the data set obtained from the study group. According to the results of Exploratory Factor Analysis (EFA), it was determined that the scale consisted of 22 items and 4 sub-dimensions. Scale subdimensions were named as: confidence, relevance, attention, satisfaction. It was determined that the four-factor structure related to the scale explained 78% of the total variance. The Cronbach alpha coefficient was 97 for the whole scale.


Introduction
Today, we are living the "industry 4.0" revolution, which was first mentioned in 2011, symbolizing digital industrial technology. "Industry 4.0" is a transformation process.
This transformation leads to the production of high-quality products at low cost, faster and more efficient production, thereby an increase in productivity, changes in the economy and labor profile, a reshape in the competitiveness of countries (BCG, 2019;Kuscu, 2018). Employment and Skills (UKCES), it can be seen that they include the occupations involving "industry 4.0" technologies such as programmers and software development experts, airplane pilots and flight engineers, web design and development experts, design and development engineers, information technology and telecommunications managers (UKCES, 2015). Therefore, countries should give importance to STEM fields and labor force in these fields in addition to "industry 4.0" technologies. Within this framework, many countries have started to carry out works on STEM education and STEM careers. The data contained in a report on STEM education including the information about 30 countries, supports this conclusion. The report states that approximately 80% of the countries currently define STEM education as a priority field at national level (Kearney, 2016). The target of most countries is to increase labor force in STEM fields. In order to increase the rate of employment in STEM fields, students should be interested in these fields, prefer these fields in their career path and first of all, they should have high motivation towards STEM fields. Because motivation is one of the variables that highly affect students' career choice and career interest (Wang, 2013;Christensen, Knezek & Tyler-Wood, 2015;Bahar &Adiguzel, 2016;Aeschlimann, Herzog & Makarova, 2016). In this context, the motivation of students towards STEM fields is gaining importance. However, the review of related literature revealed that there are limited number of motivation studies related to STEM fields. In addition, a scale related to STEM motivation was not found in the literature.
The articles related to STEM education between 1999 and 2013 were evaluated in a study reviewing the related literature and it was found that there are few studies related to motivation (Jayarajah, Saat & Rauf, 2014). In another study examining STEM education works performed between 2013 and 2015, no study related to motivation was observed (Mizell and Brown, 2016).
Regarding the researches related to STEM and motivation in the literature, it was concluded in a study that science motivation predicted STEM career interest of high school students (Robnett & Leaper, 2012). In another study, it was found that self-motivation is among the factors affecting STEM careers and STEM interest of high school students included in an academy (Christensen, Knezek & Tyler-Wood, 2015). A similar result was revealed in a study conducted by Bahar and Adiguzel (2016). In the study, it was found that the most important factor affecting American High School students' career in STEM related fields was self-motivation. When all these studies are examined, it is seen that there is no scale related to STEM motivation. Regarding the researches related to STEM motivation, it was found that motivation was generally addressed along with career and interest and not investigated as a separate variable. In this context, the research was conducted in accordance with the need in the relevant literature. Therefore, it is believed that the research will contribute to the literature.

Purpose of the Study
In this research, it was aimed to develop a scale that can be used to measure the motivation of high school students towards STEM fields.

Working Group
The scale was administered to 462 students from four high schools in Kayseri in Turkey, at the beginning of the second semester of 2016-2017 academic year. Information about the students is given in Table 1. Validity and reliability studies of the scale were carried out on the basis of student data given in Table 1.

Application
In this research, the Motivation Scale for STEM Fields was developed for high school students. The scale development principles specified by DeVellis (2014)

Clear Definition of the Construct to be Measured
A scale should be based on important theories about the fact to be measured.
Therefore, while developing the scale, the structural characteristics (subdimension/factor/component) of the fact to be analyzed are determined by reviewing the relevant field literature (DeVellis, 2014;Ozdamar, 2016). In this study, during the development of Motivation Scale for STEM Fields, similar scales and the theories and various studies on which these scales have been based were reviewed. As a result of the analyzes, it was decided to take ARCS Model as the base of the scale (Keller, 1984;cited in Keller, 2010). The reasons for this choice are; it is a higher-level model also explaining other theories, it is based on an extensive review of motivational literature, the categories explained by other theories are inadequate in practice, and other theories are inadequate outside of the field they explain (Keller, 2010).
The ARCS Model is described in four categories, which are attention, relevance, confidence and satisfaction (Keller, 2010). In the study, these four categories were determined as sub-dimensions while developing the Motivation Scale for STEM Fields.

Generating an Item Pool
When developing a scale, the item pool should be as large as possible and each item should be prepared to reflect the relevant construct (DeVellis, 2014). In this research, a pool of 33 items was prepared.

Specifying the Measurement Type
Likert type scaling was used in this research. Likert scales contain response options that indicate the level of agreement or approval to the sentence given as an item (DeVellis, 2014). In this research, the scale is of Likert type because the motivation scale attempt to determine students' level of agreement to the given items.

Review of Initial Item Pool by Experts
An expert evaluation form was prepared for the evaluation of the 33-item pool. Based on feedback from experts in STEM training, motivation and assessment, each item was reexamined and corrected and a scale containing 29 positive items was prepared. Examples of items in the item pool and resources used in writing items are given in Table 2. Some items  (2015), Keller (2010), Tahiroglu and Cakır (2014).

Review of the Items and Giving the Appropriate Form
While scoring the items in the scale, the respondents were asked to give a score between 1-5 to each item, by filling the blanks with science, technology, engineering and mathematics respectively. Then, the scores that students gave separately for science, technology, engineering and mathematics fields were summed and divided into four for each item. In this way, the item score of each item was formed. Therefore, item scores varied between 1 and 5. The items were analyzed after calculating students' scores for each item. The validity and reliability studies of the items and the final version of the scale are discussed in the findings section. While preparing the items, the word STEM was not used, and the scores were marked in separate columns for the disciplines (science, technology, engineering and mathematics). The reasons underlaying it; lack of definition of STEM department in higher education in Turkey, the word STEM is not meaningful for the students in Turkey and such a correction was proposed in expert opinions.

Data Analysis
Following the administration of the Motivation Scale for STEM Fields to the workgroup, exploratory factor analysis was performed, and the scale was finalized.
Confirmatory factor analysis, reliability analysis was performed on the final form of the scale; item variances, item mean scores, corrected item total correlations, difference between lower and upper groups of 27% were calculated. SPSS 22 and AMOS 24 software were used in the analysis of the data.

Exploratory Factor Analysis
KMO value of the scale was found to be 0.973 and Bartlett's test was found to be significant (p <.05). According to the results obtained from the KMO and Bartlett's tests of the scale's item pool, it was decided to continue the factor analysis.
In the study, the limit value for factor load was set as 0.30. There was no item with a factor load below 0.30. Article 28 was removed from the scale because of overlapping.
After removing the item, varimax, one of the vertical rotation techniques, was used.
Following the application of this technique, "rotated component matrix" table was checked, the overlapping items 7, 8 and 26 were removed from the scale and factor analysis was repeated using "varimax" technique. As a result of the analysis, it was found that all items Factor analysis was performed again on the 22-item scale after removing mentioned items, using "varimax" technique. The total variance table explained as a result of factor analysis is given in Table 3. According to Table 3, the scale consisted of four factors explaining 78% of the analyzed construct. "Scree Plot" graph of the scale is given in Figure 2.  Regarding Figure 2, the number of factors was found to be four.
In order to determine the sub-dimensions of the items, "rotated component matrix" The factors in the scale and the items are given below with their factor names.

Confirmatory Factor Analysis
In the confirmatory factor analysis, χ 2 / sd value was found to be 3.293, CFI value was calculated as 953. RMSEA value was found to be 071. Based on these results, it can be said that the scale construct that is obtained by exploratory factor analysis is validated and it is acceptable in terms of fit indices.
The standardized regression values, i.e. factor loads, obtained from the confirmatory factor analysis are given in Figure 3.

Figure 3. Motivation scale for STEM fields -Path Diagram
Regarding the factor loads of the scale, they ranged between 0.82-0.90 for confidence sub-dimension; 0.72-0.88 for attention sub-dimension; 0.81-0.91 for relevance sub-dimension; and 0.75-0.86 for satisfaction sub-dimension. Since p <.05 for these values, the items were appropriately assigned to the factors.

Reliability Analysis
The Cronbach Alpha coefficients of the whole scale and its sub-dimensions is given in Table 5. According to these results, both the whole scale and its dimensions have high reliability.

Item Variances
It is important that the items of a scale have high variance. The variances of the items of the Motivation Scale for STEM Fields are given in Table 6. According to Table 6, the variance statistics for all scale items are higher than 0.703.

Item Mean Scores
Since item scores vary from 1 to 5 in the study, another feature desired in the scale is that the mean scores of the items being close to 3. In the study, the mean scores of the scale items ranged from 3.09 to 3.48.

Item Total Correlations
Items on a scale should have a strong relationship with the item group outside of them. In the study, corrected item total correlation for all items of the scale was greater than 0.640.

Difference between the Lower and Upper Groups of 27%
T-test results of the scale between the lower and upper groups are given in Table 7. As p <.05, there is a significant difference between the lower and upper groups in terms of overall scale scores. The mean score of the upper group was 90.62, whereas the mean score of the lower group was 52.94. Therefore, it can be said that the scale can measure the distinction between students with low and high scores.
As a result of the factor analyzes, item analyzes and reliability analyzes described above, a valid and reliable 22-item and four-factor Motivation scale for STEM fields was developed (Appendix 1).

Discussion and Conclusions
In the research, Motivation scale for STEM fields was developed for high school students. When developing the scale, the scale development steps suggested by DeVellis (2014) were followed, which have been also followed in many scales developed in the field of education (Baypinar and Tarim, 2019;Ilhan and Cetin, 2013;Seker and Kaya, 2018).
Many analyzes were performed to check the validity and reliability of the item pool.
First, exploratory factor analysis was applied to the scale. As a result of the analysis, it was found that the scale explained 78% of the construct and consisted of four factors. Regarding the "Scree Plot" graph of the scale, it was seen that the factors contributed to the variance after the fourth factor were very close to each other. In addition, the slope is flattened after the fourth factor. Therefore, the number of factors was set as four (Cokluk et al., 2010). After the number of factors was set, "rotated component matrix" table was examined to determine the sub-dimensions of the items (Secer, 2013 In the confirmatory factor analysis, χ 2 / sd value was found to be 3.293, A χ 2 / sd value greater than 5 indicates a discordance (Ozdamar, 2016). CFI value was calculated as 953. A CFI value greater than 0.90 is acceptable (Ozdamar, 2016). RMSEA value was found to be 071. This value being lower than 0.10 is an acceptable result (Ozdamar, 2016, p. 185). Based on these results, it can be said that the scale construct that is obtained by exploratory factor analysis is validated and acceptable in terms of fit indices.
In the confirmatory factor analysis, the regression values indicate the power of the observed variables to predict the sub-dimensions (Karagoz, 2016). Regarding the factor loads of the scale, they ranged between 0.82-0.90 for confidence sub-dimension; 0.72-0.88 for attention sub-dimension; 0.81-0.91 for relevance sub-dimension; and 0.75-0.86 for satisfaction sub-dimension. Since p <.05 for these values, the load of items on factors is significant. In other words, it can be said that the items were appropriately assigned to the factors (Karagoz, 2016).
In addition to these results, covariance, correlation and variance values among subdimensions are also important. Since p <0.05 for the covariance, correlation and variance values of paired sub-dimensions, it can be said that these results are statistically significant (Karagoz, 2016). According to the results, there is a strong relationship between the subdimensions. The high variance of the scale items will indicate that the respondents did not responded to the item in the same way (DeVellis, 2014). The variance of the items of the motivation scale for STEM fields was found to be greater than 0.703. The mean of a scale being close to the center of the score range is a desirable feature (DeVellis, 2014). Mean items of the scale ranged from 3.09 to 3.48. Therefore, this means close to 3 satisfy this condition.
Items on a scale should have a strong relationship with the item group outside of them (DeVellis, 2014). This relationship is expressed by the correlation coefficient and this value should be greater than 0.20 (Tavsancil, 2010). In the study, corrected item total correlation for all items of the scale was greater than 0.640.
It was found that the scale developed in the research can measure the distinction between the upper and lower groups. As a result of the factor analyzes, item analyzes and reliability analyzes described above, it was concluded that a valid and reliable 22-item and four-factor Motivation scale for STEM fields was developed (Appendix 1).

Suggestions
Motivation scale for STEM fields developed for high school students can be used in experimental studies or descriptive studies. Validity and reliability studies can be performed by applying the scale to different education levels and it can be used in the researches.