Chapter Listing | Return to 1997 Topical Index


METHODS

The research design for this study was a cross sectional, mailed, self-administered questionnaire. Dillman's (1978) Total Design Method (TDM) was chosen as the foundation of the survey instrument design and mail implementation. There were two follow-up mailings.

Sample Selection

In this study, the small business owner/firm is acting as a surrogate for the entrepreneur/entrepreneurial firm. A small business owner was defined as "an individual who runs a business with 100 or fewer employees on a day to day basis and has an ownership interest in that business." The frame chosen for this final study was three large chambers of commerce in New Jersey that covered northern, middle and southern regions. These chambers had a combined membership of approximately 2500 members. From this frame a sub-sample of 1002 business owners meeting the above criteria was drawn, with approximately 1/3 from each region.

Survey Results

Of the sub-sample of 1002, 354 questionnaires were returned completed, 27 were returned blank and 54 returned ineligible or unreachable. Given the formula provided by Dillman (1978), the effective response rate was 37.34%. This rate compares favorably with that for mail surveys, which is rarely over 30% (Alreck & Settle, 1985). In an attempt to test the sample for response bias, non-response analysis was undertaken by telephone. The results warrant a reasonable level of confidence in the conclusion that the sample used in this study appropriately represents the population. The sample was 84.6% male with a mean age of 48.4 years old. The mean size of their firms based on employment in 1991 was 16.2 employees.

Description of Statistical Analysis

In order to test the direct causal relationships hypothesized by the models, linear structure equations were used. Moderated regression analysis was used to test the hypothesized contingency relationship (H2).

LISREL VIII (Joreskog & Sorbom, 1993), which estimates parameters of hypothesized causal models, was used to analyze the data and to test hypotheses one, three, four and five. Causal analysis is a method for studying patterns of causation among a set of variables (Blalock, 1971). Although causal analysis has been done in the past using ordinary least squared (OLS) regression, LISREL has been found to be a more robust and flexible analytical technique (Goldstein & Dillon, 1984).

Moderated regression analysis was used to test Hypothesis 2 and examine the hypothesized interaction effect of entrepreneurial posture and environmental turbulence. Although LISREL can be used to examine interaction and contingency relationships (Kenny & Judd, 1984), it was not used in this study for three main reasons. One, interaction effects are rarely tested using LISREL. Two, to test the interaction effect, LISREL mimics multiple regression techniques (Hayduk, 1987). Three, the main purpose of the study is to examine a system of relations between the perception of resource availability, entrepreneurial behavior and growth; so the interaction effect is not central to the study.

Top of page | Chapter Listing | Return to 1997 Topical Index


1997 Babson College All Rights Reserved
Last Updated 03/05/98