This study was concerned with the use of a graphic technique in predicting business bankruptcy. The study looked for possible improvements in the construction of schematic faces which would make them more informative. In the search for improvement, the study examined three factors: (1) The group of financial ratios to be used in constructing the face, (2) the distribution assumption of these ratios, and (3) the association between financial ratios and facial features.

To accomplish the goal of this study, an experiment was designed in which Introductory Accounting Students used the graphs to discriminate between bankrupt and nonbankrupt firms. Two groups of ratios were used. One group consisted of Dun and Bradstreet's key financial ratios and the other group represented ratios found by other researchers to be useful predictors of business bankruptcy. For the second factor, the ratios were scaled based on the assumption that they are normally distributed versus the actual distribution for the industry. For the association between ratios and features, the study examined two mappings under each type of distribution. The study found that: (1) For the non-bankrupt firms the Dun and Bradstreet ratios and the normal distribution provided graphs which resulted in more accurate predictions. The mapping had little effect. (2) For the bankrupt firms the combination of variables proved important with the best results obtained from the combination of the normal distribution, prior ratios, and mapping two. It is concluded that further research is necessary to gain an understanding of the factors which will improve the predictions of a bankrupt firm.