Abstract
Although the method of constant stimuli is commonly used in psychophysical investigations, special statistical procedures for testing its reliability are rarely employed. Measurement errors are usually ignored, and their potential effect on the overall efficiency of the measurement procedure is not taken into consideration. In reality, the decision-making strategy of the subject may change during the course of the experiment, and responses may be correlated.
We suggest a completely nonparametric method for testing hypotheses about a subject's decision-making strategy. It is based on the analysis of contingency tables and includes the method of primary-data grouping. This technique was applied to address two questions: (1) whether the results of a set of experiments, conducted independently of each other, may be considered as statistically equivalent; (2) whether the subject's response depends statistically on his preceding decisions.
Specially designed experiments on discrimination of the length of lines were performed with a large number of untrained subjects. Exploratory analysis of primary data typical for experiments of this kind has revealed some features that are usually not detected by ordinary statistical data processing, but may affect the final results. It appeared that untrained subjects often used a decision-making strategy biased towards the preceding response. This does not seem to be dependent on the personal characteristics of the subject. Some psychophysical procedures in fact encourage subjects to use this strategy.
Get full access to this article
View all access options for this article.