Mr Tyman Stanford
Doctor of Philosophy student
Office: 666 | Telephone: +61 8 8313 1605
Statistical methods for the analysis of protein mass spectra data
Using linear models to estimate home ground advantage of Australian Football League teams
Home ground advantage is a common concept in sport. It is often talked about but not actually quantified. Byron J. Ga jewski  analysed the home field ad- vantages of college football sides in the United States of America using linear models. He successfully showed there was an advantage in playing at home grounds. An Australian equivalent of this is to analyse the home ground advantage of AFL teams. This paper will fit statistical linear models to estimate the home ground advantage of Australian Rules Football (AFL) teams. The structures explained by Gajewski and Harville  are extended to account for the complexities of the AFL. Some of these considerations relate to how home ground advantage is defined. Such modifications are required because not every AFL club has one home ground, some AFL clubs share home grounds and some AFL clubs play 'home' games at grounds other than their normal home ground. Issues of identifiability in the modelling of parameters arise as a result of limitations in the data and are discussed and extended to more general situations. Six models are proposed to estimate home ground advantage. Each model is able to estimate significant home ground advantages. By statistical tests we are able to show some models are more preferable than others, and eventually a single model that best suits our purposes. An additional discussion of whether the model can incorporate not only home ground advantage parameters, but also parameters to factor for the existence of a disadvantage incurred by playing away from home. We conclude statistically significant home ground advantage does exist in AFL competition. The estimated home ground advantages without doubt favour non-Victorian sides.