- 基礎心理学研究 (ISSN:02877651)
- vol.36, no.2, pp.236-242, 2018-03-31 (Released:2018-06-16)
When researchers analyze data from an experiment with multiple experimental stimuli, they tend to aggregate responses to the experimental stimuli before performing a statistical test (e.g., t-test, analysis of variance). This common practice, however, ignores sampling errors of experimental stimuli, resulting in a substantial increase in Type-1 error rate. This article reviews the relevant literature and provides conceptual explanations about the mechanisms underlying the inflation of Type-1 error rate. The article also illustrates how linear mixed-effects model with random-stimulus effects can address the issue, with the emphasis on the correct model specification when using linear mixed-effects model.