Construção e evidências de validade de uma bateria de testes de habilidades metacognitivas baseada em desempenho

Abstract

Among the variables that involve the active subject-object knowledge interaction and that play an important predictive role in educational outcomes, metacognition has a prominent place, as it demands intense self-regulatory activity from the subject. The field of metacognition measurement is dominated by the use of self-report instruments and judge assessment protocols (think aloud), however, these instruments produce considerable respondent and confirmatory bias, respectively. Performance-based tests offer na alternative to deal with the aforementioned biases and allow identifying a strong prediction of metacognition about academic performance. This thesis aims to develop and present a set of initial evidence of validity of a battery of performance-based tests that aims to assess specific metacognitive abilities in higher education students. This thesis is composed of three studies. The first study focuses on presenting the development of the Meta-Performance battery, in terms of its conceptual basis, construction strategies and structure. Evidence of its content validity is also presented. The results of the judges' evaluations attested the validity of the battery's content, as well as the target audience declared the test of adequate understanding and execution. The second study analyzes the first evidence on the structural validity of one of the battery tests, the Meta-Text, through the analysis of the test's dimensionality and reliability in a sample of 655 Honduran university students. The bifactorial model containing a general cognition regulation fator and a specific judgment factor was rated as the best model (CFI = 0.992; RMSEA = 0.021). The results of this study generate information not previously identified by the area, suggesting that judgment may actually be a component of the metacognitive knowledge dimension, having little participation in the regulation of cognition. The third study aimed to analyze the predictive validity of performance-based testing in contrast to self-report-based testing in the field of metacognition. This study was conducted on a sample of 505 university students. Two measures of academic performance were used as outcome variables. The results indicated that when metacognition is measured by a performance-based measure, its predictive strength is statistically significant and much greater (β = 0.81 and β = 0.38; p < 0.001) than when it is assessed by self-report (p >0.05). Finally, based on the findings presented in the three studies, we discussed the contributions of the thesis to the area of metacognition measurement, the conceptual implications for the theorizing of metacognitive components and the applications for the practice of diagnosis and psychopedagogical intervention in higher education.