The Effect of Robotics Education on Gender Differences in STEM Attitudes among Dutch 7th and 8th Grade Students

: Because of its hands-on and integrative approach to STEM, educational robotics has become increasingly popular in recent years. Yet, a gender gap still exists in attitudes towards STEM studies and careers, especially among middle and high school students, potentially resulting in a lack of women in the STEM workforce. This study explores the effect of a robotics curriculum on Dutch 7th-and 8th-grade students’ attitude towards STEM subjects and careers, as assessed by the S-STEM survey. The results revealed no difference between the pre-test and post-test in attitudes toward STEM for both boys and girls. However, boys scored signiﬁcantly higher than girls on attitude towards technology, engineering and future STEM studies on the post-test. A post hoc analysis revealed a signiﬁcant difference between boys and girls on their attitude towards engineering and technology during the pre-test. These results demonstrate the difference between boys and girls in their attitudes towards STEM subjects and careers within the context of robotics education. Considering the lack of research on educational robotics among young teenagers, this ﬁeld needs to be further studied to assess its effect on gender differences within attitudes towards STEM.


Introduction
Robotics education has become increasingly popular in K12 education [1].It provides a hands-on and integrative approach to teaching STEM (science, technology, engineering and math) to primary and secondary school students [2].There are multiple robotic kits on the market: LEGO ® WeDo, LEGO ® Mindstorms, Robotis Dream and Arduino are some of them.Studies have shown that K12 students benefit from educational robotics in different ways, including STEM performance, 21st century skills, computational thinking, motivation and attitude towards STEM [3][4][5].The attitude towards STEM can be described as an interest towards STEM-related fields of study and professions.According to Unfried et al. (2015) [6], attitude includes self-efficacy, which is the belief in oneself regarding one's ability to complete a certain task [7].
Unfortunately, STEM-related fields remain understaffed [8], and are still mainly occupied by men.Despite the continued growth of the STEM workforce, women in technology fields are underrepresented [9].According to a study by Miller et al. (2015) [10], the Netherlands scored highest on gender-stereotypical images of science and engineering compared to 66 other countries.In addition, Dutch high school girls are much less inclined to study STEM subjects after high school than boys [11].Reasons for this include a lack of female role models in STEM and no conscious effort to motivate girls to choose STEM studies [11].However, previous studies have found a similar increase in positive attitude towards STEM after educational robotics for both boys and girls [12,13].To further stimulate the interest in STEM careers and the number of women in the STEM workforce, more research is needed.
Therefore, this study tries to answer the question of whether educational robotics can increase students', and especially girls', attitudes towards STEM.
Educational robotics is based on constructivism theory, which states that students learn best when they actively participate in the learning process by inventing and constructing products [14,15].Similarly, students construct and program the robots and see the reallife results of their programming work instantly.With the lack of female STEM teachers, there are fewer role models for girls in the field.However, due to educational robotics' constructivist approach, the teacher has less influence during the lessons.Furthermore, as robotics provides an integrative approach to teaching STEM-related subjects in one curriculum, it has a positive effect on STEM performance among students [16].An increase in STEM performance consequently creates an increase in self-efficacy, as becoming good at a task increases the belief that one is able to conduct the task successfully in the future [17].
A number of previous studies have analyzed the effect of educational robotics on attitudes towards STEM among K12 students, but the results have been inconsistent.Recently, Zhang et al. (2021) [5] published a systematic review on the effect of robotics education on students' attitudes towards STEM, analyzing the different effect sizes of seventeen studies published between 2010 and 2019.Meta-analysis results for learning effects showed that educational robotics had no significant impact on attitudes towards STEM.However, a few important studies published after 2019 were not taken into account [4,18,19].Overall, the most positive effects seem to have been found for attitudes towards engineering and programming, followed by sciences, math and the desire to study STEM subjects in the future [12,[19][20][21].Although older studies have not found that educational robotics greatly affects attitudes towards STEM, there appears to be a more positive trend in attitudes towards STEM in recent years.Yet, it remains unclear whether the overall effect of educational robotics on attitudes towards STEM has changed.
Research on robotics among middle and high school students is limited.Previous studies have shown the most positive changes in attitude towards STEM among kindergarteners [21] and elementary school children [3,4,16,22], while middle and high school students show less significant results [19,20,23,24].In addition, the literature on educational robotics has described the use of different methods of intervention, varying from after-school robotics programs [25], to robotics curricula during school hours [18] and robotics competitions [26].Due to the varying amount of time that students receive robotics education, it is unclear whether the number of hours of robotics training received has an effect on attitudes towards STEM.It is also very likely that students who have received robotics education during after-school programs and camps already have a high motivation towards STEM.According to Zhang et al. (2021) [5], experiments with a teaching period of four weeks, compared to longer teaching periods, had the largest effect on STEM attitudes, due to the newness of the lesson material.The question arises as to whether an even shorter teaching period would have a similar effect, and whether there is a difference between the effect on teaching periods of less than four weeks.
With respect to gender, recent studies found no gender gap after robotics lessons when students were asked about their desire to continue technical studies [12,13,18,27].Furthermore, Kaloti-Hallak et al. (2015) [28] found that, among middle school students, although motivation and attitudes towards STEM and robotics were already high from the start, it was mostly the girls whose motivation and attitudes towards STEM and robotics improved after a FIRST LEGO ® League competition, an international robotics competition for elementary and middle school students using LEGO ® Mindstorms.Although the gender gap seems to close after educational robotics, two recent studies did find a difference between boys and girls in their confidence levels after robotics classes [18,21].This difference in confidence level between boys and girls was also found in a study by Sullivan and Bers (2019) [26].Surveys after a robotics competition for middle and high school students showed that male students were more confident than female students in their general technical ability and their ability to put things together.There seems to exist some difference between boys and girls in their attitudes and confidence level after robotics education, but it is not clear whether this gap disappears after educational robotics.To investigate the gender gap in robotics, and thereby understand how to promote STEM among girls, it is important to further study this topic.
With an inconsistency in the results published on educational robotics and STEM attitudes so far, it remains unclear what the overall effect of robotics on STEM attitudes is.To promote STEM among children, and especially girls, who do not already have an interest towards STEM, it is important to investigate the effect of educational robotics on attitudes towards STEM during mandatory school hours.Based on previous research, the following hypotheses were made: -It was expected that there would be a more positive attitude towards STEM after the robotics curriculum than before, for boys and girls.-Girls' attitude towards STEM would improve more than boys' attitude towards STEM after the robotics curriculum.-There would be no difference between boys and girls in their attitude towards STEM after the robotics curriculum.-There would be a positive relation between the number of hours of robotics education that the intervention consisted of and attitudes towards STEM.

Materials and Methods
For this study, 90 students took part in the pre-test (35 girls and 55 boys), and 69 students took part in the post-test (25 girls and 44 boys).The students attended an urban high school in the Netherlands.They were recruited before they took part in the robotics curriculum, provided during school hours.The teachers of at least four different classes took part in the robotics curriculum and were asked to participate in this study with their students.The students received different hours of robotics education, ranging from 1 h, 2 to 4 h, 4 to 10 h and more than 10 h, within a timeframe of 1 to 4 weeks.The students' age ranged from 12 to 14 years.Each class received different hours of robotics education.Based on the previous literature, it was expected that there would be a 15% increase in the score of the S-STEM survey.Studies reported mean scores of 3.8 and 4.07 (SD = 0.74 and 0.68) on the pre-test [4,29].To achieve the expected increase in score and a power of 0.95, a sample size of at least 19 participants was needed.
To assess the students' attitude towards STEM, the Middle/High School (6-12th) S-STEM Survey [30] was used.This is a validated questionnaire designed for 6th through 12th-grade students to measure changes in students' attitude towards STEM subjects, studies and careers.[6,31].For this study, the S-STEM was translated from its original language in English to Dutch.Previously, the translated version of the S-STEM was used in a study by de Vink et al. ( 2022) [32].The S-STEM survey was translated by the researchers to be suitable for 10-12-year-old students.Most statements on the survey are answered on a five-point Likert scale, ranging from strongly disagree to strongly agree.Statements are divided into several subgroups: attitude towards math, science, engineering and technology and future STEM careers.Examples of statements related to STEM on the S-STEM survey are as follows: "Math has been my worst subject.","Knowing science will help me earn a living."and "I like to imagine creating new products".Statements on future STEM careers were answered on a four-point scale, ranging from not at all interesting to very interesting.Examples of STEM careers were as follows: medicine, earth science and computer science.The robotics curriculum consisted of multiple robotics lessons during school hours with the Leaphy Original and Leahy Flitz.These are two Arduino-based robotics sets using block-based programming as its software.The Leaphy original is slightly more advanced in its capabilities than the Leaphy Flitz.Students worked with a booklet that guided them through the construction process of the Leaphy robot and completed exercises using Easybloqs, the accompanying software program.
Because the students were under sixteen, they were first asked for informed consent from their parents.An informed consent form, with information about the study, was sent to the teachers of each class.Both the students and parents were required to sign the informed consent before participating in this study.Students were asked to fill in the S-STEM survey during class and before the start of the robotics curriculum.All students completed the survey on their own laptop in the classroom at the same time and were requested to do so in silence.Both the teacher and researcher guaranteed that the classroom was silent and that the students were not distracted.Students answered identification questions before the S-STEM survey to determine their age, gender and grade.To be able to pair the pre-and post-test results anonymously, students were also asked to fill in their house number and the last two numbers of their phone number.Both the pre-test and posttest were administered in this manner.The post-test was administered after the robotics curriculum.After the robotics curriculum, students were asked to fill in the S-STEM survey again.In the post-test version of the S-STEM survey, students were asked how many hours of robotics education they received.The pre-and post-test results on the S-STEM survey were compared and the variable gender was taken into account.

Results
To conduct the main analysis, scores on the pre-and post-test were paired.The preand post-test scores consisted of 90 and 69 subjects.In the pre-test, there were 35 girls and 55 boys (mean age = 12.74, SD = 0.70).No outliers were removed from the pre-test scores.In the post-test, there were 25 girls and 44 boys (mean age = 13.13,SD = 0.51).One outlier was removed from the post-test scores as he reported being 98 years old and reported having not read the informed consent.Hypothesis 1. First, it was expected that there would be a significant difference in attitude towards STEM after the robotics curriculum compared to before, for both boys and girls.To test for this expectation, a Wilcoxon matched pairs test was conducted on the mean attitude scores.Because students were not always able to fill in their house number and the last two digits of their phone number, not all pre-test and post-test data could be paired.There were 30 remaining subjects with paired pre-and post-test scores, of which there were 11 girls and 19 boys (mean age = 12.93, SD = 0.52).No outliers were removed from the paired data, as no subjective scoring was unnatural to the population differences.
Table 1 shows the mean and standard deviations for each subgroup.A non-parametric test was used because the sampling distribution of the differences was not normally distributed for the subgroups: math girls, engineering and technology boys and girls, and future studies boys.A dependent t-test was conducted on the remaining subgroups.As can be seen from Tables 1 and A1, most pre-and post-test scores were similar to each other, except for math girls.There is a slight increase in the post-test scores (M = 3.42, SD = 0.95) compared to the pre-test (M = 3.30, SD = 1.19) scores for girls on the survey and their statements related to attitude towards math.Because multiple tests were conducted, the Bonferroni correction was used, which dropped the significance level to α = 0.006.The results from the Wilcoxon matched pairs test showed that this difference was not significant: T = 24.5, z = 0.92, p = 0.36, r = 0.28.There was also no significant difference between the pre-test and post-test for girls on science, t(10) = −0.03,p = 0.97, d = 0.98, engineering and technology, T = 28.0,z = −0.45,p = 0.66, r = −0.13 or future STEM studies, t(10) = 1.31, p = 0.22, d = 0.50.For boys, there was also no significant difference between the pre-test and post-test on math, t(18) = 1.43, p = 0.17, d = 0.34, science, t(18) = −0.30,p = 0.77, d = 0.95, engineering and technology, T = 59.5, z = −0.81,p = 0.42, r = 0.12 or future STEM studies, T = 77.0,z = −0.37,p = 0.71, r = −0.09.To test for this expectation, difference scores between the pre-and post-test were created by subtracting the post-test scores from the pre-test scores, and these were compared between the girls and boys.Tables 2 and A2 show the difference scores for the pre-and post-test.The variances were equal in each subgroup.Normality tests showed that the difference scores for girls in the subgroups of math, engineering for boys and girls and future studies for boys and girls were significant.A Mann-Whitney U test was conducted on the subgroups that violated the assumption of normality.An independent t-test was conducted on the remaining subgroup, namely science.To correct for multiple comparisons, the significance level dropped to α = 0.01.The Mann-Whitney U test showed that there was no significant difference between the boys and girls on the difference scores for math, U = 71.5, z = −1.44,p = 0.15, r = −0.26,engineering and technology, U = 111.5,z = 0.30, p = 0.76, r = 0.06 or future STEM studies, U = 126, z = 0.92, p = 0.35, r = 0.17.The independent t-test showed that there was also no significant difference between boys and girls on the difference scores for science, t(28) = −0.15,p = 0.88, d = 0.96.Hypothesis 3. The third expectation was that there would be no difference between boys and girls on their attitude towards STEM after the robotics curriculum.For this expectation, all subjects in the post-test were analyzed, consisting of 69 subjects (25 girls and 44 boys).During the post-test, the scores for math boys, science boys and girls, engineering boys, mean STEM boys and girls, and future studies boys were not all normally distributed.A Mann-Whitney U test was conducted to correct this violation.The variances were equal in each subgroup.Table 3 summarizes the results.
To correct the multiple comparisons, a Bonferroni correction was used, which set the significance level to α = 0.013.The Mann-Whitney U test showed that the difference between boys and girls on the post-test scores was not significant for math: U = 716.0,z = 1.89, p = 0.06, r = 0.23.It was also not significant for science: U = 480.0,z = −1.02,p = 0.31, r = −0.12.The difference between boys and girls was significant for engineering and technology, U = 768.0,z = 2.53, p < 0.01, r = 0.30.Boys scored higher (3.10,SD = 0.76) than girls (M = 2.76, SD = 0.81) on the post-test on their attitudes towards engineering and technology.The effect size for this difference was medium.The difference between boys and girls during the post-test for future STEM studies was also significant, U = 783.5,z = 2.72, p < 0.01, r = 0.32.Boys scored significantly higher (M = 2.27, SD = 0.60) than girls (M = 1.85,SD = 0.72) on their attitude towards future STEM studies on the post-test, with a medium effect size.Hypothesis 4. Finally, it was expected that there would be a positive correlation between the number of hours of robotics education received and attitudes towards STEM.In total, 14 students received 1-2 h of robotics education, 32 students received 2-4 h of robotics education, 22 students received 4-10 h and one student reported receiving more than 10 h of robotics education.To check for the expectation, a one-way ANOVA was conducted on the attitude scores during the post-test, and the number of hours of received robotics education.Table 4 summarizes the results.The variation within the different hours of robotics education was equal.Within the different hours of robotics education, there was a normal distribution for all the subgroups, except for the following subgroups: science 2-4 h and future studies 4-10 h.Normally, a one-way ANOVA would be robust enough to correct this violation, but because the group sizes were not equal across the number of hours of robotics education received, the Kruskal-Wallis test was conducted to test the expectation for these subgroups.The results of the one-way ANOVA showed that the attitude scores on math, F(3, 65) = 0.36, p = 0.78, and technology and engineering, F(3, 65) = 0.32, p = 0.99, were insignificant.There was no significant effect for the amount of lesson hours received on the attitude scores for math and technology and engineering.The Kruskal-Wallis test was used to test the expectation for science and future studies.There was no significant effect for the amount of lesson hours received on attitudes towards science, H(3) = 3.82, p = 0.94 or future STEM studies, H(3) = 5.47, p = 0.14.Post hoc tests were conducted on the pre-test to further analyze the differences between boys and girls.The pre-test consisted of 90 subjects (35 girls and 55 boys).On the pre-test, the scores for math boys, science boys, engineering boys and future studies boys were not normally distributed.A Mann-Whitney U test was conducted to correct this violation.Tables 5 and A3 summarize the results.A Bonferroni correction set the significance level to α = 0.013.The results of the Mann-Whitney U test showed that the difference between boys and girls on the pre-test was not significant for math, U = 1122.0,z = 1.32, p = 0.19, r = 0.14.It was also not significant for science, U = 833.5,z = −1.07,p = 0.28, r = −0.11,or for future studies, U = 1047.5,z = 0.71, p = 0.48, r = 0.01.There was a significant difference between boys and girls on the pre-test for engineering and technology, U = 1268.5,z = 2.54, p < 0.013, r = 0.27.Boys had a more positive attitude towards engineering and technology (M = 3.08, SD = 0.70) than girls (M = 2.76, SD = 0.70) during the pre-test, with a medium effect size.

Discussion
The goal of this study was to investigate the influence of robotics education on attitudes towards STEM and to look at the gender differences within this effect.The results show that there was no difference in attitudes towards STEM after the robotics curriculum compared to before the curriculum, for both boys and girls.In addition, the girls' attitude towards STEM did not improve more than the boys' attitude towards STEM after the robotics curriculum.In addition, after the robotics curriculum, boys scored significantly higher than girls in their attitudes towards technology and engineering, and future STEM studies.Fourth, there was no positive relation between the number of hours of robotics education that the students received and their attitudes towards STEM.Finally, post hoc tests revealed that, during the pre-test, there was a significant difference between boys and girls on their attitudes towards engineering and technology.Boys scored significantly higher than girls on statements relating to these subjects.
It was expected that there would be a more positive attitude towards STEM after the robotics curriculum than before, for boys and girls, but this study found no significant difference between the pre-test and post-test for both genders.Because previous studies have reported this effect [4,18,20,23], the question arises as to what the difference was between our study and previous research.First of all, it is notable that, compared to the long-term studies by Sisman et al. (2021) [4], with an intervention of 4 h x 31 weeks, and Chang and Chen (2020) [20], with an intervention of 3 h x 8 weeks, this study was relatively short.The students in this study only received 1 h lessons within a timeframe of 1-4 weeks, which might have been too short to influence STEM attitudes.In addition, the results of our study revealed that there was no difference in attitudes towards STEM for the different hours of robotics education that the students received.Surprisingly, Zhang et al. (2021) [5] found that a 4-week robotics curriculum was the best amount of time to improve students' attitudes towards STEM.Though this finding is in contrast with the study by Sisman et al. (2020) [4], which found that educational robotics affected attitudes towards STEM after 31 weeks, the general consensus of previous findings is that the longer the robotics program, the more students lose interest after the initial freshness of the subject.Yet, the number of hours of robotics education in the current study and the differences between the hours of robotics education might still have been too small to see any effect.It is recommended that future studies increase the timeframe of the intervention to find an effect.
Second of all, previous studies with positive findings regarding the effects of educational robotics on attitude towards STEM mainly focused on younger populations [4,18].It might be that younger students are more flexible to attitude changes.Indeed, Lamptey et al. (2021) [23] found that the younger group, aged 6-9, had a positive attitude change towards STEM careers, compared to the older group, aged 10-14, who did not.What is also interesting is that, compared to Sisman et al. (2021) [4] and a pilot study by Leonard et al. (2016) [33], the students in this study scored much lower on average on the S-STEM survey.Because both of the aforementioned studies assessed elementary school students and this study assessed middle school students, it is possible that students' attitude towards STEM is, on average, less positive among older students than younger students.This is in line with previous research findings stating that interest in STEM decreases with age [34].
The second expectation of this study was that the girls' attitude towards STEM would improve more than the boys' attitude towards STEM after the robotics curriculum.This result was not found.Girls did improve more than boys on the subject of math, but this difference was not significant.Thirdly, it was expected that there would be no difference between boys and girls in their attitude towards STEM after the robotics curriculum.The opposite was found.During the post-test, boys scored higher than girls on two subscales, engineering and technology, and future studies.Although this result is not in line with previous studies [12,13,18,27], which found that there was no gender gap after robotics education, this result is consistent with studies reporting a higher confidence level by boys compared to girls on STEM subjects [18,26].In addition, in line with previous research, post hoc results showed a gender gap between boys and girls during the pre-test.Considering previous findings by Miller et al. (2015) [10] and Walma van der Molen (2020) [11], the question arises as to whether the cultural context has an influence on changes in attitudes towards STEM after a robotics curriculum.Considering how high the Netherlands scored on gender-stereotypical images of science and engineering, it may be possible that it is harder for a Dutch student population to experience a change in attitude towards STEM than for students from other countries.
There are a couple of limitations to this study.First of all, because there was no control group, other factors might have contributed to the results of the intervention.In addition, the student identification system appeared to be less effective than expected.Because some students did not know their house number or the last two digits of their phone number, not all the pre-test data could be paired to the post-test data, and, as a result, the groups were smaller than expected in the paired samples test.Third of all, as students were not individually observed by a researcher, but observed as a group in the classroom by the teacher, it is unclear whether they filled in their answers in a hurry.Because the test takes 10-15 min to complete, it is possible that students were not motivated or got bored quickly.Furthermore, students in the age group of twelve to fourteen might have never filled in a 5-point Likert scale before.Because students might have been unfamiliar with this way of answering questions, it is possible that their answers did not reflect their true beliefs.In a future study, a control group should be made to analyze the differences between groups that received educational robotics and that did not receive educational robotics.Furthermore, a different system should be used to pair the pre-test and post-test results.Finally, to correct the possibility of the hasty completion of the surveys, researchers could individually monitor the students.

Conclusions
This study found that the educational robotics curriculum had no effect on attitudes towards STEM within the timeframe of the intervention, which was 1-4 weeks.There was a difference between boys and girls on the pre-test and on the post-test.Boys scored higher than girls on their attitude towards STEM-related subjects and careers.This is in line with previous research, stating that boys have a more positive attitude towards STEM subjects than girls.A short robotics education intervention, with different hours of intervention, could not change this gender gap.Considering the other benefits that educational robotics has, such as the integration of different STEM subjects and its effect on STEM performance, educational robotics is still a beneficial part of STEM education and is an interesting topic for further study.In addition, as STEM professions remain understaffed and mainly occupied by men [8,9], it is important to keep researching ways in which to promote STEM from a young age.

Table 1 .
Mean paired scores and standard deviations (between brackets) on the S-STEM pre-and post-test for boys and girls.The second expectation of this study was that the girls' attitude towards STEM would improve more than the boys' attitude towards STEM after the robotics curriculum.Similar to the first hypothesis, this test consisted of 30 subjects (11 girls and 19 boys), as this was the paired dataset.

Table 2 .
Difference scores of the pre-test and post-test and standard deviations (between brackets) for boys and girls.

Table 3 .
Mean scores and standard deviations (between brackets) on the S-STEM post-test for both boys and girls.

Table 4 .
S-STEM scores on the post-test per number of hours of robotics education received.

Table 5 .
Mean scores and standard deviations (between brackets) on the S-STEM pre-test for both boys and girls (post hoc).

Table A2 .
Difference scores of the pre-test and post-test and standard deviations (between brackets) for boys and girls.

Table A3 .
Mean scores and standard deviations (between brackets) on the S-STEM post-test for both boys and girls.

Table A4 .
Mean scores and standard deviations (between brackets) on the S-STEM pre-test for both boys and girls.