STEM in Transition from Primary School to Middle School: Primary School Students' Attitudes

This study aims to examine the STEM attitudes of primary school fourth-grade students who are in the transition period from primary school to middle school. More specifically, this study examines whether the students' attitudes towards STEM significantly differ according to gender, parents' educational background, science achievement, and mathematics achievement. The data obtained from 221 fourth-grade students through the STEM Attitude Scale were analyzed by independent sample t-test and one-way analysis of variance (ANOVA). The research results revealed that students move from primary school to middle school with a high STEM attitude. In this transition, girls' attitudes towards mathematics and boys’ towards engineering and technology were higher. The tendency in engineering and technology attitudes differs only in favor of male students but does not differ significantly in parents' educational background, science, and mathematics achievement. While there was no significant difference in general STEM attitudes of the students in terms of gender and mother’s education, there was a difference in father’s education, mathematics, and science achievement. Following, the research results were discussed, and recommendations for policymakers and practitioners were presented.


Introduction
I n today's 5.0 society, the rapid change in science and technology has enabled countries to change in many areas as health, industry, agriculture, technology (Murphy et al., 2018). In parallel with this change, professions in science, technology, engineering, and mathematics (STEM) fields, which are the sources of innovation of today's digital is gradually decreasing in developed and developing countries; and preference for these fields as a profession loses popularity (Brophy et al., 2008;Lavonen et al., 2008;National Research Council [NRC], 2011). According to Turkish Industry and Business Association (TUSIAD) estimation, requirement for the STEM-related employment for 2023 will be approximately 1 million in Turkey, and about 31% of this need will not be met (TUSIAD, 2017).
Concerning change in education systems has also become inevitable to meet the increasing professional needs of STEM fields and prepare students for an uncertain future. The uncertainty for today's students requires them to be prepared for and equipped with 21st-century skills such as communication, collaboration, critical thinking, innovation, creativity, and problem-solving. This requirement has led to the adaptation of new and different approaches into education systems. One of these approaches is the STEM education approach (Bybee, 2013). STEM education approach is an educational approach in which the fields of Science, Technology, Engineering, and Mathematics are integrated with daily-life skills (Stohlmann et al., 2012).
The STEM approach has attracted educators as it provides individuals with the skills and competencies necessary for the 21st-century business world (Walton & Johnson, 2015). Also, this approach is preferred to increase students' interest in STEM fields, teach the STEM-related concepts, and improve students' literacy and awareness towards these fields (Sarı et al., 2018). During the period from preschool to higher education, students can decide which fields they like, which fields they are good at, and which professions they will choose (Yıldırım, 2021). Consequently, it is recommended to integrate STEM education in classrooms starting from an early age (NRC, 2011;Yıldırım, 2021) to ensure that students have a positive attitude towards STEM fields (Gonzalez, & Freyer, 2014). STEM education approach is practiced in formal and informal learning environments in many countries such as South Korea, England, and the USA (Smith & White, 2019). There are many important reasons explaining why STEM education is practiced through formal and informal learning. To name a few, STEM education (1) enables school-industry connection, (2) improves 21 st century skills (Partnership for 21 st Century Skills [P21], (3) attaches importance to vocational high schools and (4) connects acquired knowledge with daily life (Guzey et al., 2020).
Previous researches have shown that there is a positive relationship between interest in STEM fields and STEM career intention (Christensen & Knezek, 2017). Also, there is a positive relationship between early career interest in STEM fields and career choice (Crisp et al., 2009). Besides, there is a positive relationship between attitude towards STEM fields and interest in STEM careers (Ciftci et al., 2020;Wiebe et al., 2018). Ciftci et al. (2020) indicated that approximately 43% of the total variance of interest towards STEM careers derived from attitudes towards STEM fields. They reported that attitude towards STEM was an important factor in determining interest in STEM careers. Therefore, it is important to examine students' attitudes towards STEM fields which have an influence on their choices for future careers.
The literature shows that students' attitudes towards STEM careers are not stable in primary and middle schools (Wiebe et al., 2018), and half of the students lose their high attitudes and interest in STEM fields when they reach the 8th grade (Allen, 2016). In this context, studies conducted in recent years (Akif Bircan & Köksal, 2020;Christensen & Knezek, 2017;Ciftci et al., 2020;Karakaya & Avgın, 2016;Koyunlu et al., 2020;Özcan & Koca, 2019;Yerdelen et al., 2016;Yıldırım & Selvi, 2017) have focused on determining the attitudes of middle school students towards STEM fields and their career interests in these fields. However, little is known about what the students' attitudes towards STEM are during the transition from primary to middle school. Particularly, determining the attitudes of primary school 4th-grade students towards STEM fields and examining these attitudes according to different variables can give important clues to policymakers, curriculum developers, teachers, and researchers on STEM education. Relatively, this study aimed to determine the primary school students' attitudes towards STEM, and the following questions are attempted to be answered: • What are the attitudes of primary school fourth-grade students towards STEM?
• Is there any significant difference in the attitudes of primary school fourth-grade students towards STEM in terms of gender, parents' educational background, mathematics achievement, and science achievement?

Research model
Aiming to determine the attitudes of fourth-grade students towards STEM, the survey model was used in this study. A survey model can be used to analyze the opinions of the participants about an issue or event or their interests, skills, abilities, attitudes, etc. It is used in cases where certain features are desired to be identified (Büyüköztürk et al., 2009).

Participants
Participants were selected according to the purposeful sampling method of non-probability sampling methods. In this sampling, the researcher decides on the sample to be conducted, and in this way, the most suitable sample is selected with the purpose of the study (Balcı, 2016). Choosing the most suitable sample in the research provides time and convenience for the people doing the research (Patton, 2002). While determining the participants, we focused on the students with family profiles from lower levels of socioeconomic status and educational background. The convenience sampling method was applied while reaching the participants. The total number of participants is 221 fourth-grade students from one of Turkey's western provinces. Table 1 shows the demographic information of fourth-grade students who voluntarily participated in the study.

Data Collection Tools
Demographic information form and "STEM Attitude Scale" were used as data collection tools in the study. The demographic information form developed by the researchers consisted of students' gender, parents' educational background, mathematics achievement, and science achievement.

Data Collection
The data collection was carried out in 4 stages. Firstly, possible primary schools to conduct the study were decided. Following, fourth-grade classroom teachers working in these schools were determined. Secondly, teachers were contacted by telephone and informed about the research. Eight teachers were willing to support the research. Thirdly, measurement tools were digitized with a Google Form, and the form's URL address was sent to volunteer teachers. Finally, the URL address of the form was shared with the students in virtual classrooms by the teachers. Since the data collection process was in the pandemic period, no face-to-face interviews were held with the teachers, and the data were not collected in the classroom.
With the help of teachers, fourth-grade students anonymously filled out the form within two weeks. When the given responses to the form stopped, the data collection was completed and the data analysis was started. Since the form was digital and each question had to be answered, the data were obtained without any errors and losses.

Data Analysis
To determine the statistical tests to be used in the analysis of the data, it was examined whether the data showed a normal distribution or not. Kurtosis and Skewness values of the scores obtained from the subscales and the scale were in the range of -1 +1 (see Table 2), indicating the scores do not deviate excessively from the normal distribution (Hair et al., 2019). With the assumption of normality and homogeneity of variances regarding group scores, data analysis was conducted through independent sample t-test and the one-way analysis of variance (ANOVA) test. In determining the source of the significant difference between the group averages as a result of ANOVA test, Scheffe test was used, since the groups did not have an equal number of participants.
Obtaining significant differences in the data analysis, Cohen's d was used to report the effect sizes of these differences. As Cohen (1992) guided, Cohen's d effect sizes were interpreted as small (0.20), medium (0.50), and large (0.80). Eta square (η 2 ) was used to report the effect sizes related to the significant differences found as a result of ANOVA test, and the values were interpreted as small (0.01), medium (0.06), and large (0.14) effect sizes (Büyüköztürk, 2015). Data analysis was proceeded via SPSS 23 statistical software, taking the significance level as .05.   Table 3 presents independent t-test results examined if there was any gender difference in the primary school 4 th -grade students' STEM attitudes.

STEM Attitude and Gender
As shown in Table 3, a statistically significant difference was identified in "Mathematics" (t = 2.02, p < .05) and "Engineering-technology" (t = 3.48, p < .01). While the female students' scores (M = 33.30, SD = 5.35) on the "Mathematics" were significantly higher than the male (M = 31.80, SD = 5.29), the male students' scores (M = 37.10, SD = 4.87) on the "Engineering-technology" were significantly higher than the female (M = 34.3, SD = 6.71). The Cohen's d effect sizes were, respectively, 0.27 and 0.47, suggesting a small and medium effect size. Except for the "Mathematics" and "Engineeringtechnology" subscales, no other subscales and total scale scores revealed significant differences between the male and female students. Table 4 shows the descriptive findings and one-way analysis of variance (ANOVA) results of the STEM  attitude scale and its subscales, according to the mother's educational background. There was no significant difference in scale and its subscales in terms of mother's educational background. This indicated that mother's educational background did not play a role in STEM attitudes of the students. Table 5 shows the descriptive findings and one-way analysis of variance (ANOVA) results of the STEM attitude scale and its subscales, according to the father's educational background. As seen in Table  5, significant differences were found in "Science" (F = 9.51, p < .01) and "Total" (F = 3.71, p < .05). The etasquare (η 2 ) effect sizes were, respectively, 0.08 and 0.03, suggesting a medium and small effect size. No significant father's educational background differences were found in other subscales. This indicated that father's educational background did play a role in STEM attitudes of the students.

Stem Attitude and Father's Educational Background
The Scheffe test results showed that the difference between the mean scores of the students in "Science" was due to the significant difference among students whose fathers were high school graduates and primary school graduates (t = 3.93, p <.01), and among students whose fathers were high school and middle school graduates (t = 3.66, p <.01). Students whose fathers were high school graduates had statistically significantly higher attitudes than students whose fathers were primary (M = 34.90, SD = 4.73) and middle (M = 35.20, SD = 5.53) school graduates. The Cohen's d effect sizes were, respectively, 0.66 and 0.60, suggesting a medium effect size.
In terms of the difference between the scores obtained from the scale total, the Scheffe test results showed that this difference was due to the difference between the students whose fathers were high school and middle school graduates (t = 2.48, p <.05). On the other hand, there was no significant difference between students whose fathers were high school graduates and primary school graduates (t = 2.25, p > .05). However, due to the absence of zero value in the 95% CI [-0.71, -0.04], regarding the effect size, it could be accepted that there was a significant difference among these groups in favor of the students whose fathers were high school graduates. Students (M = 156.00, SD = 17.60) whose fathers were high school

STEM Attitude and Science Achievement
This study also examined the role of science achievement in primary school 4 th -grade students' STEM attitudes dispositions. Students' science achievement was categorized into three groups; low is within less than 50 points, middle is 50 to 75 points, and high is more than 70 points. Table 6 shows the results of the one-way variance analyses (ANOVA). As shown in Table 6, except for "Engineering-technology" (F = 1.06, p > .05), statistically significant differences were identified in "Mathematics" (F = 13.00, p < .01), "Science" (F = 21.90, p < .01), and "21 st century skills" (F = 12.00, p< .01), and "Total" (F = 15.6, p< .01). The eta-square (η 2 ) effect sizes were, respectively, 0.11, 0.17, 0.10, and 0.13, suggesting a large effect size. That is, students' science achievement did play a role in STEM attitudes except for the "Engineering-technology" subscale. This also shows that the science and engineering-technology disciplines could not be adequately dealt with in the context of Turkey.
Scheffe test results showed students with high science achievement on the "Mathematics" was statistically higher than those of middle level (t = 3.96, p < .01) and low level (t = 3.93, p < .01). The Cohen's d effect sizes were, respectively, 0.92 and 0.87, suggesting a large effect size. Similarly, students with high science achievement on the "Science" was statistically higher than those of middle (t = 6.24, p < .01) and low (t = 3.40, p < .01). The Cohen's d effect sizes were, respectively, 0.92 and 0.87, suggesting a large effect size. In terms of "21 st century skills", attitudes of students with high science achievement were statistically higher than the middle (t = 4.66, p < .01), and attitudes of students with middle science achievement were statistically higher than low (t = 3.14, p < .01). The Cohen's d effect sizes were, respectively, 1.20 and 0.85, suggesting a large effect size. These findings mean that students with high science achievement have higher "Mathematics" and "Science" attitudes than those of middle and low, and as their success in science increases, their attitudes towards the "21 st century skills" also increase.
Attitudes of students with high science achievement on the total scale was statistically higher than those of middle (t = 4.53, p < .01) and low (t = 4.10, p < .01). The Cohen's d effect sizes were, respectively, 0.67 and 1.57, suggesting a medium and large effect size. This means that students with high science achievement have higher STEM attitudes than those of middle and low ones.

STEM Attitude and Mathematics Achievement
This study also examined the role of students' mathematics achievement in primary school 4 thgrade students' STEM attitudes dispositions. Students' mathematics achievement was categorized into three groups; low is within less than 50 points, middle is 50 to 75 points, and high is more than 70 points. Table  7 shows the results of the one-way variance analyses (ANOVA). As shown in Table 7, except for "Engineeringtechnology" (F = 0.07, p > .05), statistically significant differences were identified in "Mathematics" (F = 5.40, p < .01), "Science" (F = 4.24, p < .01), and "21 st century skills" (F = 6.61, p < .01), and "Total" (F = 15.9, p < .01). The eta-square (η 2 ) effect sizes were, respectively, 0.25, 0.04, 0.06, and 0.12, suggesting a large, small, medium, and large effect size. That is, students' mathematics achievement did play a role in STEM attitudes except for the "Engineering-technology" subscale. This also shows that the mathematics and engineeringtechnology disciplines could not be adequately dealt with in the context of Turkey.
Scheffe test results showed that attitudes of students with high mathematics achievement on the "Mathematics" were statistically higher than those of middle (t = 5.79, p < .01) and low (t = 6.92, p < .01).
There was also a significant difference in attitudes of students with middle mathematics achievement on the "Mathematics" compared to those of low (t = 4.22, p < .01). The Cohen's d effect sizes were, respectively, 0.84, 2.26, and 1.42, suggesting a large effect size. Similarly, attitudes of students with high mathematics achievement on the "21 st century skills" were statistically higher than those of middle (t = 2.72, p < .05) and low (t = 2.80, p < .05). The Cohen's d effect sizes were, respectively, 0.40 and 0.92, suggesting about medium and large effect sizes. On the other hand, there were no significant differences in students' "Science" attitudes (p > .05). However, due to the absence of zero value in the 95% CI [-1.34, -0.04] and [-0.62, -0.05], regarding the effect sizes, it could be accepted that attitudes of students with high science achievement on the "Science" subscale were statistically higher than those of middle and low achievement. The Cohen's d effect sizes were, respectively, 0.33 and 0.69, suggesting a small and medium effect size. This means that students with high mathematics have higher "21 st century skills" and "Science" attitudes than those of middle and low, and as their success in mathematics increases, their attitudes towards the "Mathematics" subscale also increase.
Attitudes of students with high mathematics achievement on the total scale were statistically higher than those of middle (t = 3.98, p < .01) and low (t = 4.41, p < .01). There was also a difference in attitudes of students with middle mathematics achievement compared to ones with low achievement (t = 2.57, p < .05). The Cohen's d effect sizes were, respectively, 0.59, 1.44, and 0.89, suggesting a medium, large and large effect size. These findings mean that as their mathematics success increases, their STEM attitude also increases.

Discussion
This study aimed to examine the STEM attitudes of primary school 4 th -grade students. Particularly, this descriptive study aimed to reveal whether students' attitudes towards STEM significantly differ according to gender, parents' educational background, science, and mathematics achievements.
In terms of general STEM attitudes, results show that students have high STEM attitudes. This finding is in parallel with many studies in the literature (Karakaya & Avgın, 2016;Özcan & Koca, 2019;Yıldırım & Selvi, 2017;Yerdelen et al., 2016). Students' STEM attitudes and interests are high at an early age, but they may decrease over time (Allen, 2016). Since students' attitudes towards STEM affect their early career choices in STEM fields (Brophy et al., 2008), determining it from an early age is necessary.
The results show no significant difference in STEM attitudes of the students in terms of gender. This finding is parallel to many studies (Aydın et al., 2017;Balçın et al., 2018;Brown et al., 2016;Ciftci et al., Özyurt et al., 2018). Özyurt et al. (2018) found no significant difference between primary school student's attitudes towards STEM and gender. On the other hand, in this study, it was found that girls' attitudes towards mathematics were higher than boys', while boys' attitudes were higher than girls' in engineering and technology. This finding coincides with the results of the research conducted by Ciftci et al. (2020) andKoyunlu Ünlü &Döküm (2020). Similarly, there are studies showing boys are more interested in engineering than girls and the perception that engineering is a more suitable profession for men (Capobianco et al.2011;Ketenci et al., 2020;Knight & Cunningham, 2004;Yıldırım & Türk, 2018). There are also studies indicating that student's attitudes towards STEM change depending on gender (Christensen & Knezek 2017;Karakaya et al., 2018). These studies show that boys' attitudes towards STEM are higher than girls' (Reinking & Martin, 2018;Legewie & DiPrete, 2014). Gender pressure can be the major rationale for this conclusion (National Science Foundation [NSF], 2003).
When STEM attitudes of the students are examined in terms of parents' educational background, the findings show that students' STEM attitudes do not differ according to their mothers' educational background, but they do in terms of their fathers' educational background. Students whose fathers are high school graduates have higher Science and STEM attitudes than students whose fathers are middle and primary school graduates. Sivrikaya (2019) also has reached a similar finding. Similarly, Fan (2003) has indicated that students' STEM attitudes differ according to parents' educational background. As opposed to these findings, Aydın et al. (2017) have stated no difference in the students' attitudes towards STEM regarding parents' educational background. Families, mothers in particular, with low socio-economic and education level generally work in agriculture and service sectors, so it may have effect on these results.
In terms of students' science and mathematics achievement, the results show that students' mathematics, science, 21 st century skills, and STEM attitudes significantly differ, except for engineering and technology. Also, the findings show that as students' science achievement increases, their 21 st century skills attitudes increase; as students' mathematics achievement increases, mathematics and STEM attitudes increase significantly. Previous studies show that students' academic achievements positively affect their attitudes towards STEM (Karakaya et al., 2018;Olivarez 2012). There is a positive relationship between students' science and mathematics achievements and their attitudes towards these courses (Liu et al., 2011). Previous research results are consistent with the results of the study.

Conclusion
The research reveals that students move from primary school to middle school with a high STEM attitude. In this transition, girls' attitudes towards mathematics and boys' towards engineering and technology are seemingly higher. The tendency in engineering and technology attitudes differs only in favor of male students but does not differ significantly in parents' educational background, science and mathematics achievement. While there is no significant difference in STEM attitudes of the students according to gender and mothers' educational background, there is a difference in fathers' educational background, mathematics and science achievement.

Implications for Policy and Practice
If the high level of attitude in the transition to middle school can maintain during the secondary school period, students' career preferences towards STEM fields may increase. For this reason, it is recommended to make changes in middle school curricula. Instructional interventions depending on the curriculum changes may include STEM-based design-oriented activities in which knowledge turns into products.
It is striking that although the girls have high attitudes towards mathematics; their attitudes towards engineering and technology are low. If there is no intervention in the primary school period, it may continue during the middle and high school periods (Ketenci et al., 2020). Therefore, instructional activities can be organized to increase girls' engineering and technology attitudes. With after-school and summer camp activities for girls, their design and production skills can be improved. Raising girls' engineering and technology attitudes at this age can positively affect career choices in forward-looking engineering and computer science fields.
Another important result of the study is that there is no significant difference in engineering and technology attitudes of students with high mathematics and science achievements from other levels of education.
It may indicate that STEM applications are not sufficiently implemented in primary school. Updating primary school curricula to include STEM activities can both increase students' existing STEM attitudes and make it easier to increase girls' positive attitudes towards engineering and technology.

Limitations
This study has some limitations. The first limitation of the study is that the number of fourth-grade primary school students is limited; therefore, the generalizability is low. This means that findings cannot be interpreted to reflect all fourth-grade primary school students in Turkey. However, research findings will make a sense in understanding fourth-grade primary school students' attitudes towards STEM. The second limitation of the study is that when examining the students' attitudes towards STEM, the variables were gender, parents' educational background, science achievements, and mathematics achievements. Different variables can be used in future studies. The third limitation of the study is that it is a quantitative study, and STEM attitudes of students can be investigated in more depth with qualitative studies.