Procedure to extract a weekly pattern of performance of human reaction time.

A study of reaction time in six tests (speed of simple identification, simultaneous and sequential letter identification, simultaneous and sequential stimulus word matching, and semantic verification) was carried out on 130 children aged 7 to 12 years. Subjects were tested Monday, Tuesday, Thursday, and Friday, using a cross-sectional design. The answers to each of the tests are binary in nature (Yes/No) measured by the recording of a click one of the two buttons of a microcomputer. Controlled variables were age, failure, sex, and day of test. The greatest percentage of the variance produced by the data was due to the age (typically 25%). The percentage attributable to the "day of test" (typically 5%) is hence masked by the variance due to the other factors. The method proposed herein makes it possible to isolate the percentage of the variance for the variable "day of test" alone and to provide evidence for a circaseptan effect on reaction time. It used a regressive correction on predominant variables. This effect shows performance is best on Thursday and weakest on Monday, the amplitude of the weekly rhythm being related to the task.