51 of 57 people found the following review helpful
Rich examples, interesting ideas, but inappropriate data,
This review is from: The Origin and Evolution of New Businesses (Hardcover)
Amar Bhide has written a richly illustrated book about new and growing firms, drawing eclectically on many social science disciplines. Although he makes frequent references to economics, he often invokes explanatory factors from cognitive psychology and organization theory. His resulting model thus has a strong ring of behavioral plausibility. Unfortunately, he draws on a database that is simply inappropriate for answering his central question: why do a small fraction of startups turn into promising firms, whereas the great majority do not?
Bhide committed a fundamental methodological error: he selected only successful firms and then tried to infer what differentiated them from the (non-selected) unsuccessful ones. He surveyed 100 founders of companies that appeared on the Inc. 500 list in 1989 and drew upon on several hundred case studies by his students at Harvard, plus cases of successful firms drawn from business periodicals and his own research. Although he shows some awareness of the methodological problem thus created, he nonetheless injudiciously draws strong inferences from empirical regularities among the successful firms.
Why is such selection bias problematic? Consider a hypothetical study, showing that 20 percent of the successful firms in the financial services industry were currently run by Harvard MBAs, compared to only 10 percent by Stanford MBAs. Would we be entitled to conclude Harvard MBAs were twice as successful as those from Stanford? What if we learned that Harvard MBAs started 80 percent of the firms in the financial services industry, compared to only 1 percent for Stanford? And, that most of the firms started by Harvard MBAs had failed? Now we see that Stanford MBAs are highly over-represented among the successful firms, compared to the initial population of startups, and that Harvard MBAs are substantially under-represented. Without information on the initial cohorts of firms starting out in an industry, we are in great danger of engaging in superstitious learning of the kind that Bhide actually reviews in his book (research conducted by Camerer, Kahneman, Tversky, and others).
By relying on information from firms that made it onto the Inc. 500 list or into the cases written up by his students, as well as on case histories of FedEx, Walmart, Microsoft, and other successful firms, Bhide cannot tell us why or how those firms got there. Only a research design that allowed him to follow startups and growing firms over time would give him the dynamic data he needs to answer the questions posed by his extremely interesting model.
Sort: Oldest first | Newest first
Showing 1-3 of 3 posts in this discussion
Initial post: Sep 13, 2009 3:36:27 AM PDT
John Persico says:
Dr. Howard Aldrich makes an excellent point here on ex post facto analysis of "success." Most books in the management literature purporting to show how success is achieved fall into this trap or is it just a "lazy" way of publishing research that will grab attention. Somewhat like the "seven secrets or seven habits or seven principles," books that dominate the management literature. Simple solutions to complex problems. But who would buy a complex solution?
Posted on Apr 8, 2010 1:22:29 PM PDT
Duncan Duke says:
As you mention, this book (like most non-academic business publications) commits the typical "sampling on the dependent variable" mistake. But this is only so if the question is "what makes some start-ups successful?". If the question is "what are some some the things successful start-ups do?", then the book is based on a decent sample.
In any case, many of the constraints to start-ups that Bhide describes surely apply to both successful and unsuccessful ventures (i.e. lack of credibility, lack of funding, etc.).
I don't think the book was meant to "advance theory" in an academic sense, most of what Bhide describes logically derives from extant theory. What he does do very nicely is show how some esoteric theoretical constructs (e.g. "cognitive legitimacy") are dealt with in the real world. As such, the book is a nice bridge between academic theory and practitioner reality.
In reply to an earlier post on Apr 11, 2010 6:54:03 PM PDT
Howard Aldrich says:
Duncan, I agree that even inadequate research designs can yield useful information, if one is willing to work hard to pull it out. I worry, however, when researchers do not discuss the obvious problems with their designs, but rather go right ahead & offer inferences from their 'findings.'
‹ Previous 1 Next ›