Abstract

TAVARES, Heliton Ribeiro; CARVALHO, Adriana Moraes de  and  LEITE, Walter Lana. Growth curve for Mathematics via Item Response Theory. Est. Aval. Educ. [online]. 2016, vol.27, n.64, pp.242-266. ISSN 1984-932X.  https://doi.org/10.18222/eae.v27i64.3790.

This study proposes models to follow the evolution of average educational performance of a group of individuals in Mathematics. They were evaluated over time using Item Response Theory, enabling estimation of average skills in periods not evaluated and improving the interpretation of scales. The linear, quadratic, logistic and Gompertz growth curves were evaluated. The longitudinal feature leads to interdependence among skills at the various evaluation times, with the proposal of some covariance structures to model this dependence. The parameters of the items were known from a prior calibration. Real data from the Plano de Desenvolvimento da Escola (PDE-Escola/MEC) [School Development Plan] of the Ministry of Education were used. It was found that the logistic growth curve, associated with the correlation structure AR(1), fit better to data, with a very high correlation between the estimated skills and that the greatest gain in skill occurs by the end of the 6th school year.

Keywords : Item Response Theory; Growth Curves; Mathematics; Evaluation of Education.

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