Measuring latent variables: social, cultural and economic capital in PISA data from Peru

Abstract

The purpose of this paper is to study the scales used in educational sociology as important factors in the explanation of school performance. Initially the non-parametric item response theory (IRT) was used to study the properties of the scales of the latent variables social capital, cultural capital and economic capital of Peruvian students, based on the contextual questionnaire of the PISA 2000 international exam. Then the confirmatory factor analysis (CFA) is used as an alternative to analyze the model fits and to verify the dimensionality of the three scales constructed. The results of the TRIN showed great compatibility between the observed results and the theoretical conception of the latent constructs. These revealed that the scale obtained for the economic capital construct presents a high scalability coefficient (Loevinger's H), generating a strong scale, with which the students can be ordered and discriminated with certainty. The social capital construct presented intermediate scalability and the cultural capital construct showed weak scalability. For its part, the CFA corroborated the results found with the previous methodology, where the evaluation of the model indicated a good fit. Although the results were satisfactory, there is a need to improve the measurement of social capital and, even more so, of cultural capital.