Statistical Analysis for Learning and Test-taking Behaviors in Basic Statistic Business and Application Course

Abstract

Learning behavior is key to academic success, particularly in hybrid learning systems that rely on technology for learning, research, and communication. These factors lead to varied learning behaviors among students. This study aims to: 1) investigate learning behaviors from subjective homework using Two-Step Cluster analysis, 2) analyze the differences in learning success across different learning behavior groups using the Kruskal-Wallis test, 3) analyze academic achievement from tests and final exam scores for each learning behavior group using the Mann-Whitney U Test, and 4) examine the correlation between multiple-choice and subjective final exam scores related to learning behaviors and test-taking using Pearson's Correlation Coefficient. The sample consisted of 129 students enrolled in the Basic Statistics for Business and Applications course at Prince of Songkla University in the first semester of the academic year 1/2022. The findings revealed that: 1) learning behaviors could be categorized into six groups: those who did not do homework, those who studied alone, in pairs, in small groups, in medium-sized groups, and in large groups; 2) the Kruskal-Wallis Test showed significant differences in learning success among the groups at the .05 significance level (χ2 = 40.23, df = 5, p < .05); the Mann-Whitney U test at a significance level of .05 revealed that students who did not do homework had significantly different success levels compared to other groups (4≤U≤13, p < .05), with the lowest success level while the large groups had the highest success level, which was not significantly different from the pair studying group (U = 226); 3) academic achievement from test and final exam scores showed that all groups had lower final exam scores compared to test scores. Meanwhile the paired t-test showed no significant difference between test and final exam scores for students who did not do homework, studied in pairs, and studied in small groups (t=1.62, 0.00, and 1.71 respectively) while other groups had significantly higher quiz scores than final exam scores at the .05 level; and 4) the Pearson's Correlation Coefficient showed no correlation between multiple-choice and subjective exam scores in all groups, with coefficients ranging from -0.30 to 0.40, indicating that good performance in multiple-choice exams does not necessarily correlate with good performance in subjective exams.