大規模学生調査から学習成果と学習時間の構造を掴む : 横断的・時系列的分析 <論考>
DaigakuRonshu_44_1.pdf 2.06 MB
The Structure of Learning Outcome and Learning Time in Japanese Students : Cross-sectional and Longitudinal Analyses <Article>
A growing body of literature suggests that given the trend toward universal access to higher education in Japan, learning outcomes, learning time and learner-centered assessment should be the focus of university reform. However, the context of educational services is still a black box and the relationship between student characteristics, college environments and college outcomes has not been systematically examined.
Using data from the National Students Survey in 2010 and National High School Students Survey in 2005-2010 (CRUMP) for 8,032 students enrolled in human and social science departments of 127 private institutions, we initially applied multiple correspondence analysis to visualize the IEO (input-environmentoutput) triangle. Secondly, the context of ratio of ‘excellence’ was examined using multilevel modeling in which students are “nested” each department. Thirdly, we applied growth curve modeling to longitudinal data to investigate level1 variability within individual growth trajectories of learning time.
Major findings are as follows: 1) The IE geometry consisted of two axes. Learning-related categories contributed to axis1 and an articulation between of high school and college, such as admission based on recommendation or examination characterized Axis 2. Organized along the two axes, we found that cognitive outcomes such as critical thinking and problem solving ability overlap axis1. 2) Ten percent of the variance in ratio of ‘excellence’ is explained by level 2 (between departments). Moreover, cross-level interaction of department ranking and recommendation admission shows a positive effect on the ratio of ‘excellence’, meaning that students who enrolled in high ranking private universities by recommendation subsequently earned good grades (vice versa). 3) Regarding growth rate of learning time at the individual level, we conclude that the average linear growth rate is 1 hour over three data points and varies among individuals. In conclusion, possibilities and implications of the national student survey for college impact researchers are discussed.
Copyright (c) 2013 Author