Abstract:

The validity and reliability must be guaranteed to any psychological experiment. The size of experiment material is a critical factor that will significantly affect the results of experiment. This paper aims to study the selection of optimal size of the psychological experiment material under the frame of multivariate generalizability theory. The results show that, with the IAT experiment, 50 will be the optimal size of each combined tasks, and the reliability will be 0.905; with the Picture-word Interference Paradigm, when the material size is about 48, the reliability will be 0.954; with the Cue Paradigm, when the number of valid cue is 50%, the optimum size of valid cue and invalid cue is about 35, and the reliability will be 0.973. This study shows that MGT can help us to select an optimal size of the psychological experiment material.

Key words: multivariate generalizability theory, the reliability of experiment, size of experiment material