Abstract
One of the problems in using spreadsheet packages for regression analysis in the business environment is binary dependent variables, where the variable takes on a value of zero or one. An alternative to standard ordinary least squares (OLS) regression is provided that more accurately specifies that relationship between dependent and independent variables when the dependent variable is binary. The proposed Lotus 1-2-3 regression template ensures that the estimated probabilitiy falls within the range of 0%-100%. Using a sample of 18 failed companies matched with 18 nonfailed firms based on asset size and sales for the fiscal year before bankruptcy, results from the logit template are compared to those of the OLS routine. In terms of goodness of fit, the logit template outperforms the OLS function. It provides a higher percentage of firms correctly categorized for each cutoff value, and ex post, the mean square error is smaller for the logit estimated equation than for the OLS estimated equation.