Analysing count data related to football match scoring outcomes using multilevel modelling

Abstract

This dissertation aims to present an in depth study of the theory of two and three level models assuming that the dependent variable, which comprises counts, follows a Poisson distribution. These multilevel models are implemented using the facilities of gllamm command which runs in the STATA® (StataCorp, 2003) software package. The fitted multilevel models are applied to real football data set, which was accessed from an online free football betting portal. The dependent variable is the total number of goals scored during a match. The aim of this modelling technique is to examine the relationship between the total number of goals scored during a football match and a number of game-related predictors. Moreover, the models will determine whether the total number of goals scored during a match has changed over the course of the last 10 years, and whether there is a notable difference in the number of goals scored per match between the Bundesliga and Serie A football leagues. Predictors that explain a significant portion of the total variance in the response variable will be identified; thus outlining the main football events during the match that are more likely to lead to a goal.