Abstract:
This work discusses factors related to the age-grade gap, indicating the main determinants of it for Brazilian public-school students. Using data from PNAD and INEP from the year of 2013, it is evaluated how personal, family and school characteristics are related to the occurrence of age grade-distortion on public school students. More specifically, this work applies econometric models that deal exclusively with count data, such as Poisson and Negative Binomial, as well as models adjusted for excess zeros, such as Zero-Inflated Poisson and Zero-Inflated Negative Binomial. Among the results, it is highlighted a substitutability between the variables related to the educational and the individual-family context, with a reduction of the individual-family coefficients once educational variables are considered in the estimation. Furthermore, it is found that as students grow older, the absence of any previous age-grade gap is reflected in a smaller probability of it ever happening in the future.
Keywords:
age-grade distortion; count models, zero-inflated count models