Contributing Factors to Success Versus Failures of the Startup Business: Startup Business Success Versus Failure Prediction Models

Date of Submission


Document Type


Degree Name

Doctor of Science in Management Systems (Sc.D.)




David Morris

Committee Member

Robert Baeder

Committee Member

Louis Mottola

LC Subject Headings

New business enterprises--New England, Business failures--New England--Econometric models

Call No. at the Univ. of New Haven Library

AS 36 .N290 Mgmt. Syst. 1992 no.4


Business failure is both frequent and potentially damaging yet central to the efficient operation of the market economy. A better understanding of business failure can help lead to fewer business failures in the future, resulting in better utilization of our limited resources.

The purpose of this study was to develop and test a generic startup business success vs. failure prediction model through causal survey research in New England. Four generic models were developed and tested, all significant at the .05 level, using discriminant analysis and factor analysis. The four models will reliably outperform random classification of success vs. failure 95 percent of the time. The accuracy rate of the four models’ ability to predict a specific business as being successful or failed varies from model to model between 62 to 65 percent. There is no significant difference between the four generic models. The eight variables distinguishing between success and failure Generic Model 1 are: use of professional advisors, planning, staffing, education, parents owning a business, minority, industry experience, and record keeping and financial control.

The sample of 216 businesses (108 failed businesses and their 108 successful company matches based on industry, size, location, and age of the business) was subdivided by industry. Five industry models, all significant at the .05 level, were developed. All five industry models are more accurate at predicting businesses as successful and failed than the four generic models. The five industry models, with their predictive ability are: manufacturing- 95%, finance- 86%, construction- 81%, service- 80%, and retailing- 80%. The sample of 216 was again subdivided by size of business. Two size models, significant at the .05 level, were developed for firms employing 0-10 and 11-25 people, with accuracy rates of 75 and 81 percent respectively. For a comparison of generic, industry and size models, see Figure 17 and Table 64 on pages 182-184. Specific models can be used by entrepreneurs, investors, lenders, suppliers, educators, consultants, and public policy makers to aid in decision making. It is recommended that further study focus on models developed for specific industry groups based on company size.

In addition to developing a startup business success vs. failure prediction model, the study answered two questions (through testing 30 hypotheses): 1. Do successful and failed business CEOs agree on the contributing factors to startup business failure? 2. Do successful and failed businesses start with equal resources? The CEOs disagreed, at the .05 level, on the importance of three of the fifteen variables: being a minority, product/service timing, and parents owning a business. The businesses did not start with equal resources in three areas: failed CEOs stated that their business had less staffing difficulties, made greater use of professional advisors, and they have a higher level of education than successful CEOs.