An Empirical Study On Leaderless Group Discussion Test Of Undergraduates

Leaderless Group Discussion is a situational evaluation method commonly used in assessment center at the rate of 59%, along with wide application in personnel assessment and group training development. Researches abroad on the reliability of LGD have mainly focused on reliability and the criterion-related validity. Overall results have illustrated most studies have proved that LGD has high reliability, and the criterion-related validity also has reached the medium level. However, comparatively few studies have probed into construct validity, cross validity and incremental validity. Domestic studies of LGD have started quite late and now are still in the exploratory stage. With increasing application of LGD in domestic personnel assessment, empirical researches will be further substantiated in breadth and depth .Based on previous studies, with university undergraduates as subjects, this thesis has investigated the reliability, construct validity, cross validity, criterion-related validity and incremental validity of LGD assessment. The results show as follows.(1)With random samples of university graduates as subjects, this thesis carried out a LGD Test in five dimensions including Language Skills, Analyzing and Sorting Capabilities, Organizing and Coordinating Ability, Adaptability and Pressure Sustainability, Behavior and Manners. The results reveal that reliability coefficient of LGD was .96, reaching the desired level with good distinction. Correlation of LGD Evaluation between any two dimensions ranges from .85 to .93,correlation and some distinction exist between any two dimensions. The correlation between the score of each dimension and the total score varies from .93 to .97,each score has significant contribution to the total score and there is no multicollinearity between them. The thesis inspected the construct validity of LGD Test by confirmatory SEM, fit index of the model reveals that each absolute fit index and goodness of fit index have had best fitting standard with good construct validity. Simultaneous analysis of multiple groups shows the factor structure tested in LGD contains invariance in different gender groups. With extra-curricular activities as the criterion, the results show good criterion-related validity.(2) Studies of training effects on LGD evaluation show that training effects have significantly influenced the raters. Systematic training can promote more objective and impartial evaluation with increasing rating accuracy. The results also reveal that subjects in training groups have better performance in Analyzing and Sorting Capabilities, Organizing and Coordinating Ability, Adaptability and Pressure Sustainability, Behavior and Manners .Furthermore, the total results conspicuously surpassed the scores of the control group. Systematic study of personnel assessment courses and special training of LGD will contribute to subjects'better performance in some aspects and enable them skillfully respond to LGD evaluation.(3)Investigation into the relationship between FFM and LGD test. The results show that, except neuroticism has negative correlation with LGD evaluation results, other FFI personality dimensions have different levels of positive correlation with LGD test. By hierarchical regression analysis, the result showed LGD had good incremental validity over FFM test. The incremental reached .30.