Open Access Article SciPap-963
Logistic Regression Model for Longitudinal Data
by Viera Labudová 1,* and Martina Lakatová 2

1 Fakulta hospodárskej informatiky, Katedra štatistiky, Ekonomická univerzita v Bratislave, Dolnozemská 1, Bratislava 852 35, Slovakia

2 Fakulta hospodárskej informatiky, Katedra štatistiky, Ekonomická univerzita v Bratislave, Dolnozemská 1, Bratislava 852 35, Slovakia

* Authors to whom correspondence should be addressed.

Abstract: The objective of this paper is to describe particularity of longitudinal data and methods which can be used to analyse them. The assumption of usual tools used for analysis is the independence of observations. In order to analyse of longitudinal data, we have to make provisions for their particularity, which is the dependence of observations. Therefore, while we analyse them, we must employ methods that are adjusted to that dependence. Several approaches have been proposed to model binary outcomes that arise from longitudinal studies. Most of the approaches can be grouped into two classes: the population-averaged and subject-specific approaches. The generalized estimating equations (GEE) method is used to estimate population averaged effects. In this paper, we investigate the Generalized Estimating Equation (GEE) capabilities of PROC GENMOD for correlated outcome data to fit models using unspecified (unstructured) correlation structure. The data from EU SILC was used to find out how material deprivation of households in the Slovak Republic (material deprivation: yes (1), no (0)) is linked to their available characteristics.

Keywords: Material Deprivation, Longitudinal Data Analysis, Generalized Estimating Equation Model, EU SILC

JEL classification:   C10 - General,   M31 - Marketing,   O10 - General

SciPap 2018, 26(3), 963

Received: 1 September 2017 / Accepted: 27 June 2018 / Published: 23 November 2018