Revue de neuropsychologie


Three statistical tests for one-case pretest-posttest measurements and a fourth, omnibus test Volume 15, issue 3, Juillet-Août-Septembre 2023


  • Figure 1
Université Lumière Lyon 2, UR Étude des mécanismes cognitifs, Département de Sciences cognitives, Psychologie cognitive & Neuropsychologie, 5, avenue Pierre Mendès-France, 69676 Bron cedex, France
* Correspondance : G.A. Michael

The challenge for those interested in measuring the evolution of a patient’s performance over time is documenting that change. A common observation is that the use of statistics in the assessment of change is limited to comparing the patient’s score in phase 1 to a norm, then the score obtained in phase 2 to the corresponding norm. If in phase 1 the score is below the expected limits and the score in phase 2 is within the normal limits, then clinicians conclude that a change has occurred. However, showing a change in performance between two phases requires a statistical comparison between the phases. This article presents three different statistical tests allowing the comparison of a patient’s performance in two phases. Each one of these tests can be used in a specific situation: (i) a single norm is available, (ii) two norms are available from a single sample, (iii) two norms are available from two different samples. These tests require normative data in the form of “mean and standard deviation” from a fairly large normative sample. A fourth test is also proposed, with no requirement of normative sample size, and which can be used where the normative data is expressed in “mean and standard deviation” or in percentile form. A spreadsheet is provided allowing the use of all tests presented in this article.