Autocorrelation Screening: A Potentially Efficient Method for Detecting Repetitive Response Patterns in Questionnaire Data




Valid data are essential for making correct theoretical and practical implications. Hence, efficient methods for detecting and excluding data with dubious validity are highly valuable in any field of science. This paper introduces the idea of applying autocorrelation analysis on self-report questionnaires with single-choice numbered, preferably Likert-type, scales in order to screen out potentially invalid data, specifically repetitive response patterns. We explain mathematical principles of autocorrelation in a simple manner and illustrate how to efficiently perform detection of invalid data and how to correctly interpret the results. We conclude that autocorrelation screening could be a valuable screening tool for assessing the quality of self-report questionnaire data. We present a summary of the method’s biggest strengths and weaknesses, together with functional tools to allow for an easy execution of autocorrelation screening by researchers, and even practitioners or the broad public. Our conclusions are limited by the current absence of empirical evidence about the practical usefulness of this method.