SAGT: NEW SOFTWARE FOR GENERALIZABILITY ANALYSIS

RESUMO

Antonio Hern�?¡ndez-Mendo1, �?ngel Blanco-Villase�?±or2, Jos�?© Luis Pastrana1, Ver�?³nica Morales-S�?¡nchez1, Francisco Jes�?ºs Ramos-P�?©rez1

This paper describes a new software application for generalizability analysis. The characteristics of this software overcome the limitations of existing programs (GENOVA, GT and EduG) at both the numerical and graphical level, as well as in relation to data import/export. Its environment is also more user-friendly. The effectiveness of the new software is illustrated by analysing data and evaluating results from an intervention programme whose aim was to reduce gender discrimination in physical education classes. This part of the study used an observational design and a partially-nested orthogonal generalizability design with six facets. The intervention programme lasted six months and involved 100 children aged 10-12 years (51 girls and 49 boys) who were distributed across four classes in primary year five. The results of the generalizability analysis indicated that the observation instrument was a reliable, valid and precise tool, and also that observers were highly reliable (k = 0.73-0.81, G = 1). The ANOVA carried out indicated that the intervention programme had been effective. The SAGT software was likewise shown to be effective and user-friendly.