The Evaluation Center Construct Validity Of Simulation Research

Assessment center is a personnel evaluation method which assesses individual behavior performance from multiple perspectives. Many evaluation technologies are applied and many assessors estimate individual behaviors in special exercises. The assessors' evaluation opinions are collected through discussion and statistical method, then the overall assessment to individual are obtained. Usually, In-basket, Leadless Group Discussion, Role Playing and Management Game are used for the simulation technology of assessment center. The former researches indicate that, assessment center has good criterion validity and its content validity has been improving, but its construct validity has been less promising. The mainly influential factors to the construct validity of assessment center are dimension, exercise, assessor, assessee and rating method.This article applies assessment center, simulates the exercises of employing excellent mathematics teacher for some middle school, and confirms construct validity and the influences of different rating method to assessment center. Moreover, the article also discusses some questions about assessor rating consistency. In the study, 30 math department's senior students take role of assessees, they interview middle school's mathematics teacher. 32 graduates and experiential teachers take play of assessors, according to the behavior evaluation criterion; they evaluate the assessees' behavior performance in 10 minutes teaching, leadless group discussion and role playing.In study 1, by means of within-dimension rating method, Multitrait-Multimethod, Confirmation Factor Analysis and Generalizability Theory are used to testify the construct validity of assessment center. The results indicate that assessment center has well convergent validity, but its discriminant validity is low. Exercise effect is higher than dimension effect, and assessors estimate the assessees' ability by the person characteristic model.In study 2, by means of within-exercise rating method, Multitrait-Multimethod and Generalizability Theory are used to testify the construct validity of assessment center again. At the same time, influences brought by different rating methods are compared. The results indicate that assessment center has well convergent validity, but discriminant validity is low. Rating methods have little difference on construct validity.In study 3, assessor rating consistency is discussed in different exercises by Generalizability Theory. The results indicate that, assessor can understand well about the behavior evaluation criterion, but their rating consistency is low. Assessor rating consistency in 10 minutes teaching and Leadless Group Discussion is not good enough, but it can be accepted in Role Playing.According to the findings of the 3 studies, following conclusions were drawn:(1) Assessment center has well convergent validity, but discriminant validity is low.(2) Assessors rate assesses' behavior performance in person characteristic model.(3) Using a few dimensions and exercises, rating methods are not bringing great difference on the influence on construct validity.(4) Assessees have different behavior performance on the different dimensions. Assessors' rating can almost represent their real ability, but their rating consistency need be improved.(5) It is possible to meet the precision of measure that 3 or 4 assessors rate 3 dimension clusters.