Some additional tests were conducted to see if the results could be improved by the addition of more accounting variables. Separately, and in combination, capital expenditures (quarterly average) from the cash flow statement and sales growth (quarterly geometric average) were tried and found to provide little or no improvement. The inclusion of other variables, in addition to or instead of these, may be argued for various reasons; however, there is no particular theoretical argument which would suggest a clearly superior choice of accounting variables, and ultimately this remains an empirical question. Use of the components of ROE and the inclusion of a variability measure were suggested by our examination of prior research and were robustly useful in the cluster analysis.