Friday, July 08, 2005

Good to Great, by Collins

Good to Great
by Jim Collins

Because Jim Collins writes such a great deal about the research supporting this book, this review will focus on the validity of, and possible flaws in, that research.

Although apparently thorough, Collins’ research process is questionable at a number of points. He limited his pool of companies to “Fortune 500” listees (p 220), he excluded companies without an “obvious shift to breakthrough performance” (p 222), he eliminated Coca-Cola and Pepsico for not substantially exceeding their industry (p 227) rather than considering that these two dominating companies may have driven the strong industry performance, he improperly applies the scientific method, and he appears to have ranked acquisitions subjectively (p 259-260). These combine to weaken his otherwise strong argument about the nature of “good to great” companies.

Limiting the pool to only Fortune-listed companies is justified by the statement that the Fortune list only includes large companies, and that most Fortune firms are publicly traded (this was an implicit screen). However, many publicly traded companies of a reasonable size have not been listed in Fortune, or may have come on and then fallen off the list in years he did not survey (Collins only looked at the 1965, 1975, 1985, and 1995 lists). By limiting his “company universe” in this way, Collins created the possibility of a selection bias.

It is entirely possible that a company’s stock value would show a gradual shift from “good” to “great”, and in fact 19 companies from the limited universe did. If Collins had included the 18 of these companies with no other elimination criterion, it is possible that his finding may have been different. For example, he compared the company performance to the apparent breakthrough in the hatching of a chicken from an egg. In these additional companies no breakthrough was present, but a gradual shift did occur. Perhaps he might have asked why some companies appeared to make the gradual shift, while for others a breakthrough appeared.

Collins’ exclusion of Pepsico and Coca-Cola appears to reflect a misconception of industry v. company performance. The truth is that an industry is simply a grouping of like companies, and if the larger industry players are successful the industry will appear to be successful. As dominant companies, it should be expected that Coca-Cola and Pepsico would not substantially exceed their industry performance; rather that they would pull the industry average up with them.

Finally, Collins improperly applies the scientific method. He emphasizes repeatedly that his recommendations are not based on any preconcieved notion, but solely upon research. This actually undercuts, rather than strengthens his case. Proper application of the scientific method would call for his hypotheses to be tested against data, but he has turned this backwards and created hypotheses to fit his data. One way around this might have been to split the study group in two, using one group to establish the hypotheses, and then "testing" the hypotheses against the other group. Since it is obviously too late for that, another good possibility would be to use the additional years of data that have accrued since he completed this study to identify new "good to great" companies, and test the hypotheses against them.

Despite these shortcomings in methodology, Good to Great is well formatted and provides highly useable suggestions for how to make the transition in any type of organization. Possibly the most helpful sections were those in which Collins gave non-business “good to great” examples, such as his wife in the Ironman triathlon and the high school track team, because they demonstrated the broader applicability of his findings.

Tags: , ,

No comments: