The estimation of a logit model can be problematic when there are a few observations from one outcome (i.e. qualified statements) relative to the other one (i.e. unqualified statements). The reason is that the “information content” of such a sample for model estimation is quite small, leading to relatively imprecise parameter estimates (Palepu, 1986). One approach to tackle this problem is to use a choice-based sampling approach (i.e. equally matched sample of observations from the two groups) to increase the sampling rate of qualified financial statements. An alternative procedure that is being used in the present study is to weight the data and compensate for differences in the sample[10]