ABSTRACT
Twelve of the thirteen qualitative variables examined in the univariate analysis were used to develop a series of logit models, the exception being a change in a company's name. Logit models were then developed on the 111 matched pair company samples for all feasible combinations of the ratios, ensuring that one was selected from each category. The main requirement is that sufficient data are available relating to the births and deaths of companies. This is likely to be the case for a cohort of small firms with a common incorporation date; and for firms in an industry which have been subject to a common destabilising shock. Regrettably the data available for the present study are inappropriate for the application of survival analysis. Instead of having a time series of data running from a given 'birth date' the sample available only offers data running back from a given 'death date'.
