Psychometric properties of a dropout prediction tool for students in Andalusia

Abstract

This article analyzes the psychometric properties of the scale used in the research corresponding to the topic “Stories of Abandonment. A Biographical-Narrative Approach to the Academic Dropout in Andalusian Universities. Multicausal Analysis and Proposals for Prevention.” The scale is composed of 71 items, with five Likert-type response alternatives. The participants forming the convenience sample were 970 from six Spanish universities. The study dimensions are: A.-Motivation, B.-Commitment, C.-Attitude and behavior, D.-Socioeconomic conditions, and finally E.-Permanence. The study begins by calculating the statistical power and the effect size, thus determining the type I and type II error. Similarly, reliability has been calculated based on the intercorrelation of items through Cronbach’s alpha (0.906). The construct validity was carried out through exploratory factor analysis, for which the correlation matrix was studied using the Kaiser-Meyer-Olkin measure of sampling adequacy (KMO coefficient), in this case the value is 0.886, the result of Bartlett’s test of sphericity is 0.000 and the Determinant 4.009E−12, so we continue with the analysis of the correlational structure, extracting the factors through the principal components method and determining in this way the communalities with the highest and lowest values. Next, we calculated the total variance explained, obtaining 16 factors and an accumulated variance of 57.315%. Finally, the model was determined by distributing the items according to the highest level of saturation by factors, obtaining a 30-item scale with a Cronbach’s alpha of 0.893, which is reliable. In conclusion, the questionnaire used complies with the psychometric aspects necessary to be an optimal scale.