Comparing Academic Performance of Elementary Education Majors in General Education Science Courses

PLOS ONE | https://doi.org/10.1371/journal.pone.0320137 March 19, 2025 7 / 15 PLOS ONE Elementary Education Majors in General Education Science Courses TYPE) as a possible fixed effect. Default settings for the MIXED command in SPSS were used unless otherwise noted. We followed the steps outlined by Theobald [74] for each course as follows. First, we ran an empty model that included random intercepts for each section (i.e., GRADE ~ (1 | SECTION)) in order to calculate the intraclass correlation coefficient for the random effect. For all courses, the ICC for SECTION was less than 0.1, and it was often less than 0.05 (Biology: 0.074; Physical Science: 0.051; Statistics: 0.038; Humanities: 0.051; Human Development: 0.037; History: 0.015). Based on this step alone, we were not confident the ran- dom effect was needed in our models. To settle the question, we selected random effects by using the Akaike information crite- rion (AIC) to compare full models that did and did not include SECTION as a random effect (i.e., GRADE ~ MAJOR TYPE + (1 | SECTION) compared to GRADE ~ MAJOR TYPE). Restricted maximum likelihood was used when fitting these models. For all courses, including SECTION improved the model (decreased the AIC), so we proceeded with including random intercepts by SECTION when modeling GRADEs in each course. Next, we selected fixed effects by using the AIC to compare models with and without grouping by MAJOR TYPE (i.e., GRADE ~ MAJOR TYPE + (1 | SECTION) compared to GRADE ~ (1 | SECTION)). Maximum likelihood was used when fitting these models. For all

courses, including MAJOR TYPE as a fixed effect improved the model (decreased the AIC), so we proceeded with including this in our final model for each course. Finally, we reran the final model for each course (i.e., GRADE ~ MAJOR TYPE + (1 | SEC- TION)) using restricted maximum likelihood to get accurate parameter estimates. Because MAJOR TYPE had multiple categories (see above), the 95% confidence intervals of parameter estimates were used to determine which broad fields were statistically distinguishable from each other with ELED as the reference MAJOR TYPE. We also calculated estimated marginal means of the grades for each MAJOR TYPE to be used in data visualizations. The results of this final step are included in the Results section below. Sample To provide a clear understanding of our sample and facilitate comparisons with other popu- lations, we also gathered ACT score and high school GPA data for the students in our data- set. We used one-way ANOVAs to compare by MAJOR TYPE. For ACT score, there was a significant effect for MAJOR TYPE, F(5, 1183215) = 730.20, p < 0.001), and post-hoc Tukey comparisons between ELED and the other majors showed that the mean score for ELED majors (M = 26.56, SD = 3.26, N = 7296) was significantly lower than the mean score for Business (M = 27.47, SD = 3.49, N = 26631), Humanities (M = 27.63, SD = 3.48, N = 16669), and STEM majors (M = 27.66, SD = 3.71, N = 69238). For high school GPA, there was also a significant effect for MAJOR TYPE, F(5, 170121) = 340.01, p < 0.001, and post-hoc Tukey

comparisons between ELED and the other majors showed that the mean GPA for ELED majors (M = 3.81, SD = 0.21, N = 6903) was significantly higher than all other major types (Arts and Communications: M = 3.72, SD = 0.28, N = 19297; Business: M = 3.77, SD = 0.27, N = 25286; Humanities: M = 3.72, SD = 0.28, N = 15508; Social Sciences: M = 3.72, SD = 0.28, N = 38978; STEM: M = 3.77, SD = 0.26, N = 64155). Because ELED majors differed from their peers in significant ways (based on ACT and high school GPA), we could have used these variables as covariates to account for the impacts of these differences on final grades before comparing by major type. However, we were not interested in whether general education courses served ELED majors after accounting for incoming differences. Rather, we wanted to investigate how ELED majors compared to other students regardless of background because we are interested in what knowledge ELED majors