Compte renduducolloque: Methodes Quantitatives en Sciences Humaines (MQSH) 2024

On Friday, June 7, 2024, the 13th annual Quantitative Methods in the Human Sciences (MQSH) conference was held at TÉLUQ University in Montreal. Five speakers presented their work in the field. Eric Frenette discussed the evolution of the Bifactor model, from its proposal by Holzinger, following Spearman's work in 1927, to its contemporary version, a version worthy of Frankenstein's monster, which it became by transfiguring its underlying hypotheses and patching together parameters that increase its explanatory potential. Sophie Vanbelle, guest speaker for the event, presented the main statistical measures and their interpretation to determine reliability and agreement in the case of a binary scale. She discussed the importance of reporting different indices of agreement, calculating the sample size needed to have reliable measures of inter-rater (and intra-rater) agreement, and the need for guidelines for researchers. André Achim proposed an innovative approach to exploratory factor analysis based not on matrix decomposition, but rather on signal cancellation, which allows for the extraction of all relevant information, the number of factors, and the factor structure, while reducing the number of decision options. Ibtissem Ben Alaya presented the seven-step validation process of a questionnaire on teachers' socio-emotional skills while discussing the challenges encountered and the psychometric choices made. Louis Laurencelle, in his presentation, criticized the reasoning and criteria for altering the type I error rate (alpha threshold), such as the Bonferroni and Tukey corrections, which torpedo the statistical power of research in the humanities. He examined and argued the harmfulness of such an approach and its absurdity. In conclusion, this 13th edition of the MQSH conference was worthy of its banner: competence, criticism and creativity.