Using Quasi-variance to Communicate Sociological Results from Statistical Models

In three recent papers British statistician David Firth has advanced a method to assist in the presentation and interpretation of statistical models with categorical explanatory variables (Firth 2000; 2003; Firth and Menezes 2004). At the current time, to the best of our knowledge, these papers are not widely known within the British sociological research community and we suspect that this is partly due to their mathematical nature. This paper describes Firth's idea of 'quasi-variance' as a solution to the 'reference category problem' in presenting outputs from statistical models. We provide a number of examples to illustrate the flexibility of the quasi-variance approach, and discuss the circumstances when it is most relevant to sociological research. In addition we have mounted public access data, and a number of example files which use the popular packages SPSS and Stata, to help sociological researchers practice and apply this technique. We conclude that quasi-variances should be routinely presented as output when the results of statistical models with categorical explanatory variables are reported within sociological analyses.