The aims of this study were to identify cross-sectional dietary patterns (DPs) in a representative sample of 7–10-year-old schoolchildren, to examine how scores for these DPs tracked over a time period of five years (from age 7–10 years to 12–15 years), and to investigate longitudinal associations between changes in DPs scores and changes in BMI (Body Mass Index) z-scores. Children aged 7–10-years were examined in 2007 (n = 1158) and a subset of the sample participated in a follow-up in 2012 (n = 458). Factor analysis (FA) was applied to derive DPs at baseline. The change in DP from childhood to adolescence was analyzed by comparing factor scores using the complete cases, in which factor loadings were the ones evaluated at baseline. Associations of BMI change with DP change were assessed by multivariate linear regression. At baseline, four DP were identified that explained 47.9% of the food intake variance. On average, the factor scores of “DP II” (salty snacks, French fries, fast-food, sugary beverages) decreased in follow-up, while no changes were observed for “DP I” (rice, cooked beans, beef/poultry, leafy vegetables), “DP III” (fruits, cooked and leafy vegetables, fruit juices, pasta, milk, cheese), and “DP IV” (milk, coffee with milk, cheese, breads/biscuits). No significant linear association was shown between changes in BMI z-scores and changes in DP scores from childhood to adolescence. In conclusion, three out of four DP scores identified at baseline tracked slightly in adolescence.

A preocupação com medidas de traços psicológicos é antiga, sendo que muitos estudos e propostas de métodos foram desenvolvidos no sentido de alcançar este objetivo. Entre os trabalhos propostos, destaca-se a Teoria da Resposta ao Item (TRI) que, a princípio, veio completar limitações da Teoria Clássica de Medidas, empregada em larga escala até hoje na medida de traços psicológicos. O ponto principal da TRI é que ela leva em consideração o item particularmente, sem relevar os escores totais; portanto, as conclusões não dependem apenas do teste ou questionário, mas de cada item que o compõe. Este artigo propõe-se a apresentar esta Teoria que revolucionou a teoria de medidas.

Purpose: This study aims at developing an instrument to measure the latent trait propensity to drop out in face-to-face higher education.Design/methodology/approach: Based on Tinto's student integration theory, a 27-item scale was created to measure student propensity to drop out of undergraduate programs. Item response theory was used to evaluate the psychometric analysis of the items. Furthermore, different methodologies were used to evaluate and provide evidence for content validity, response process validity, internal structure validity and criterion-related validity.Findings: With the support of specialists in the construct, the interpretation of the scores for the use of the scale was defined in four levels of propensity: high, moderate, low and very low.Research limitations/implications: The latent trait propensity to drop out in face-to-face higher education allows the inclusion of new items and aspects in the instrument. Thus, it can be adapted to distance education.Practical implications: The students' propensity to drop out score can be useful for researchers and administration units in colleges and universities in the planning of permanent institutional actions and programs to take preventive measures.Social implications: Minimize dropout in order to raise the educational level of the population and make better use of the resources invested in education.Originality/value: This study points out when, why and how propensity to drop out can be measured and how scores can be interpreted in the context of the problem.