Abstract:

Background:
STEM Fluency (SF) is an online training system developed to enhance student mastery of basic STEM skills through explicit practice (Mikula & Heckler, 2017). The STEM Fluency system has been implemented in various STEM courses and has enhanced student performance on essential skills (Andrew & Nieberding, 2023; Heckler & Mikula, 2016). Despite the overall positive effects, it is not clear whether students benefited equally from the online training. To address the gap, the study examined the differences in students’ behaviors and progress using latent profile analysis. Latent profile analysis (LPA) refers to methods that uncover latent groups in data by determining the likelihood of individuals belonging to various groups (Masyn, 2013). In education, LPA has been used to identify subgroups of students who may have different individual characteristics, motivational profiles, learning behaviors, and learning needs (Hong, Bernacki, & Perera, 2020; Miller, Perera, & Maghsoudlou, 2021).

The Study:
The goal of the study to examine the differences in students’ behaviors and progress when students participated in SF online training. Participants were 104 undergraduate students in an algebra-based physics course. Students participated in five weeks of SF online training on a set of essential skills. Pre- and post- tests were administered to test students’ mastery of the essential skills.

Learning Gains:
A paired sample t-test was conducted to understand the effectiveness of the online training on improving student performance. The results suggest that students’ post-test scores (M = 10.79, SD = 3.50) are significantly higher than their pre-test scores (M = 8.23, SD = 2.76), t = 9.73, p < .001.

Latent Profile Analysis. LPA was conducted to identify student profiles based on their learning behaviors and their pre- and post- test results using the ‘mclust’ package in R. A number of commonly used model fit measures were calculated and compared against each other to determine the best model. Due to the space limit, only the results of the analysis are provided in Table 1.

Table 1. Mean and (standard deviation) of the four profiles.
Group 1 (N=57) Group 2 (N=25) Group 3(N=10) Group 4 (N=12)
Pre-test Score 9.31 (2.80) 7.92 (2.15) 6.02 (1.63) 6.20 (1.98)
Post-test Score 12.18 (2.94) 11.37 (2.83) 7.31 (2.25) 6.24 (1.95)
Total Correct 71.91 (23.50) 88.53 (25.63) 141.29 (24.33) 25.67 (15.11)
Total Attempted 119.45 (55.00) 154.93 (63.87) 322.82 (71.04) 56.89 (34.80)
Total Time (s) 4582.61 (1814.03) 11698.47 (2426.02) 5354.00 (1504.36) 2233.05 (1718.43)

Significance:
The latent profile analysis reveals that, although there is an overall improvement on student performance on the learning topics, not all students behaved similarly, nor did they all benefit from the learning activity at the same level. Further discussion will be provided in the presentation. The study is significant for several reasons. First, the study identifies how subgroups of students performed differently in the online training. It acknowledges the importance of understanding individual differences in online learning environments. Second, by identifying student subgroups, it helps researchers and educators consider ways to improve the design of the system and enhance the overall effectiveness of the system. Finally, it urges future research examining how to provide more individualized support to meet diverse needs in online learning environments.

Keywords:

Latent profile analysis, online learning, individual differences.