Abstract

LAROS, Jacob Arie  and  MARCIANO, João Luiz Pereira. Análise multinível aplicada aos dados do NELS:88. Est. Aval. Educ. [online]. 2008, vol.19, n.40, pp.263-278. ISSN 0103-6831.

In Psychology and related fields there is growing awareness that the independent observations derived traditional regression analysis are not always reflected in reality. The objective of this article is to illustrate multilevel analysis, an alternative to traditional regression analysis, using a dataset of public access, NELS:88. (National Educational Longitudinal Study). NELS:88 is conducted by the Ministry of Education of the United States of America and was designed to identify the effect of socio-economic and school variables on educational performance. The dataset contains data of 21,580 eighth grade students of elementary school distributed amongst 1,003 schools. In the NELS:88 study a mathematics test was administered together with a questionnaire about socio-economic and cultural characteristics of the students. It can be concluded that the NELS:88 dataset serves very well to illustrate in a practical way how a multilevel analysis is carried out.

Keywords : Multilevel Analysis; Hierarchical Linear Model; Intraclass Correlation; NELS:88.

        · abstract in Portuguese | Spanish     · text in Portuguese     · Portuguese ( pdf )