关键词: 个性化学习, 结构方程模型, 预测, 本科生科研

Abstract: Objective To establish a method to predict style, experience and gains of research participation by undergraduates' personal traits, thus to provide decision-making reference for individualized learning. Methods Gather data through the relevant questionnaire, and analyze reliability and validity of the questionnaire; then extract 4 latent variables (personal traits, research participation style, research participation experience and research participation gains) from observed variables in the questionnaire, and build structural equation modeling (SEM) consisting of the above 4 latent variables; at last establish predicting equations of personal traits and style/experience/gains of research participation respectively based on SEM analysis results. Results With 94.99% (417/439) efficiency, the alpha reliability coefficient of the questionnaire is 0.774, KMO of the validity analysis is 0.804, Bartlett's test is significant and extracted common factors can explain 53.94% of the total variance; three SEMs do not show negative variance, and the three SEMs' Chi-square P＞0.05, Normed Chi-square are all greater than 1 less than 3, AGFI and GFI are all greater than 0.9, RMSEA are all less than 0.05, and NC are all greater than 200; the three predicting equations are all linear functions (with the independent variable: personal traits and the dependent variable: observed variables of research participation style/experience/gains respectively). Conclusions This questionnaire is highly efficient with good reliability and validity. The constructed SEMs have no unreasonable parameters and fit the data well. Based on SME, the method to predict style, experience and gains of research participation through undergraduates' personal traits is objectively and feasible.

Key words: individualized learning, structural equation modeling, prediction, undergraduate research