3.3.4 Statistical analysis plan

Two important considerations in jurimetrics are sample size and the potential problem of over-determination. The sample size in this research cannot be said to be very large, especially given that it is used for the purpose of regression analysis.1 I acknowledge that many rules of thumb exist with regard to sample size when undertaking multiple regression analysis. In the context of jurimetrics, it has been argued that the number of cases should be more than 2^(u-1), where u is the number of variables.2 My data set complies with this criterion.3 More importantly, several methods have been developed in legal scholarship to overcome the potential problem of over-determination. For instance, Nagel focused on a limited set of predictor variables for conducting a multiple regression analysis.4 Haar et al. selected only highly significant variables.5 Accordingly, I chose to apply logistic regression analysis on a limited set of factors – behavioural and contextual legal case factors – which I had derived from existing case law and subsequently only made use of the highly significant variables.6