Behavioral Criteria for Grounding Entrepreneurship Education and Training Programs: A Validation Study^*

^*This paper was funded in part through an SSHRC grant to the first author and by Memorial University's Hubert W. Kelly Chair in Youth‐Focused Technological Entrepreneurship. We thank the registries of Alberta, British Columbia, Nova Scotia, New Brunswick, Newfoundland, the Northwest Territories, Ontario, Prince Edward Island, Quebec, Saskatchewan, and the Yukon for their assistance with the sample.

Abstract

The entrepreneurship literature lacks a systematically developed and validated framework to ground educational programs. We previously developed behavioral observation scales (BOS) consisting of 9 dimensions and 47 behaviors. In this study, we validated the BOS using 12 performance measures and a national survey of 149 entrepreneurs. The BOS were found to be valid. All 9 BOS dimensions, as well as the total score on the BOS, correlated significantly with many of the 12 nonbehavioral performance measures. These BOS provide entrepreneurship education and educators with a validated and systematically developed instrument that can be used to appropriately ground education programs.

Notes

^* This paper was funded in part through an SSHRC grant to the first author and by Memorial University's Hubert W. Kelly Chair in Youth‐Focused Technological Entrepreneurship. We thank the registries of Alberta, British Columbia, Nova Scotia, New Brunswick, Newfoundland, the Northwest Territories, Ontario, Prince Edward Island, Quebec, Saskatchewan, and the Yukon for their assistance with the sample.

1. Results of the frequency analysis reported later in the paper confirm that respondents were able to differentiate between behavioral frequencies. A subsequent study (n = 38) requesting small business owners to assist with fine-tuning the properties of the BOS instrument included two questions explicitly asking for the identification of any items “confusing or difficult to interpret.” Only one item was cited by more than one respondent. Results indicated the negative wording associated with the financial item “Does not spend excessive amounts on luxury or personal items” may be potentially confusing, as two of three respondents citing this item noted “double negative” even though the wording was technically correct. Overall, there was strong support for the meaningfulness of all items.

2. We provided an “n/a” category to avoid having to use negative labels like “loss” or “decline.” Our coding rule here was to treat an “n/a” response as missing data unless we were able to verify that the firm also experienced zero or negative employee growth; in the latter instances, we coded the response as “0.” All instances of “n/a” turned out to warrant a coding of 0. We were unable to identify any plausible alternate interpretations of an “n/a” response.

3. We also examined the surveys returned by Canada Post. In total, 1,134 surveys were returned to us. Of that number, we were able to find contact information for 169 (14.9 percent) in order to resend the survey. This suggests that as few as 14.9 percent of the companies for whom surveys were returned were still in business.

4. When applying an eligibility requirement (i.e., “surviving” firms), one first must estimate the number of eligibles among the nonrespondents (Wiseman and Billington 1984).

5. A factor analysis can also be used to assess the fit of BOS items to BOS dimensions when there are “three to five times as many individuals to be rated as there are behavioral items” (Latham and Wexley 1994, p. 90). Our data barely met this ratio. As argued by the creators of the BOS method (Latham and Wexley 1994), the main advantage of factor analysis is that “it saves time” (p. 91). Moreover, they assert that factor analysis seeks to ensure that behavioral criteria (or dimensions) are independent of each other, which is unlikely to take place for workplace behaviors given that the criteria are frequently logically related. Thus, many works report Cronbach's alpha coefficients and do not conduct factor analysis on the items contained in BOS (e.g., Brown and Latham 2000, 2006; Morin and Latham 2000; Sue-Chan and Latham 2001). Hence, we only focus on the “human judgment” model in this paper. However, we did conduct a factor analysis. In brief, each of the nine factors had an Eigen value greater than 1.3 and the nine-factor model explained 57.61 percent of the variance in the sample. For full results of the factor analysis, please contact the authors.

Additional information

Notes on contributors

Travor C. Brown

Travor C. Brown is professor and director of Master of Employment Relations program in Faculty of Business at Memorial University.

Dennis Hanlon

Dennis Hanlon is associate professor of entrepreneurship in Faculty of Business at Memorial University.