| Entrepreneurship comes in many shapes and forms, driven by a broad variety of motivations and in a diversity of contexts. While this heterogeneity contributes to making entrepreneurship fascinating it also makes it very challenging for entrepreneurship researchers to arrive at strong and credible conclusions regarding causal relationships. Building on experiences from within and outside of entrepreneurship research this article provides an integrated discussion of strategies for dealing with the problem of heterogeneity with particular application to the entrepreneurship domain. Specifically, we deal with three problems: 1) unobserved heterogeneity, i.e., that unmeasured or unavailable variables may bias estimated relationships; 2) causal heterogeneity, i.e., that the structure, strength, direction or form of relationships may vary among sub-groups of the studied population, and 3) instrument inequivalence, i.e., that the validity of chosen operationalizations also may vary by sub-group or context. We discuss how these problems can be reduced at different stages of the research process, i.e., through theory and theorizing; in choosing a basic design for the study (including sampling); at the operationalization stage, and through approaches chosen for analysis, respectively. We conclude each section with summarized advice that should help entrepreneurship researchers design more robust studies and arrive at more valid conclusions from extant data sets. Throughout, we illustrate with examples from entrepreneurship studies. |
| |
|
The Erasmus - EIM - Panteia Entrepreneurship Lectures Series is co-organized by Erasmus Research Institute of Management (ERIM.nl) and EIM Business & Policy Research (EIM/Panteia), an independent and international research and consultancy organisation, specialised in SMEs and Entrepreneurship. EIM is part of the Panteia group.
|
|
|
|
For more research resources visit our joint Entrepreneurship Research Portal
|
|
|
|
Contact information:
|
|
Andreas Rauch
|
Ingrid Verheul
|
|
Email
|
Email
|
|
|
