Horace Secrist’s (1933) Theory of Organizational Mediocrity: A Cautionary Tale
Bowling Green State University
In the midst of the Great Depression, Northwestern University statistics professor Horace Secrist made a great discovery. It was a discovery that had the potential to provide insight into the nation’s economic woes and perhaps even put America back on the road to prosperity. Secrist traced the fortunes of 49 department stores between 1920 and 1930, measuring their ratio of net profit or loss to net sales. He divided these stores into four groups—from lowest 1920 profits to highest 1920 profits. Secrist took the average performance of each of the four groups and traced it over the decade. Stores with higher than average profits performed steadily worse throughout the decade, whereas stores with lower than average performance performed steadily better. The overall trend was clear to Secrist: The performances of the businesses were converging on mediocrity.
Secrist did not rush his discovery into print. As a careful scientist, he examined other types of businesses, including hardware stores, railroads, and banks. All 73 of the different industries examined showed the same pattern. To ensure that he was not overlooking a simple alternative explanation, Secrist asked 38 economists and statisticians from America and Europe to provide comments and criticisms of his methodology. Confident that he had made a new and important discovery, the scientist published his 468-page book with 200 charts and tables entitled, Triumph of Mediocrity in Business (Secrist, 1933).
Initial reviews of Secrist’s book were favorable, noting that the research was meticulous and that the results were troubling. One reviewer, however, administered to the author what Stigler (1997) referred to as a “public flogging.” The well-known statistician Harold Hotelling pulled no punches. Observing that, despite even referring in his own book to Sir Francis Galton’s concept of statistical regression to the mean, Secrist committed the very same statistical fallacy. As Hotelling observed, the method of grouping used by Secrist ensured that he would observe the trend toward mediocrity (see Stigler, 1996, for a detailed discussion).
In business, as in sports, performance depends on chance as well as on skill. Thus, skill is not perfectly correlated with performance. Imagine you are at the driving range, and you just hit a perfect drive off of the tee. You can’t believe how far you hit it. Now, consider your next drive. Do you predict that it will go as far or farther? You conclude that it probably will not go near as far. Why? Because you have an intuitive sense of the statistical principle “regression to the mean.” When we choose to focus on those who exhibit extreme performance, we are focusing on people for whom error worked in a positive direction. Baseball hitters who have batting averages of .300 or higher in any season are 80% likely to have lower batting averages in the subsequent season (Schall & Smith, 2000). The second-year slippage of outstanding NFL rookies is often misinterpreted by the press as a “sophomore slump” (see Davis, 2009, for a recent example). Kahneman and Tversky (1973) observed that organizational leaders often misinterpret regression to the mean when they conclude that praise is often followed by poorer performance, and punishment is often followed by improved performance. Some managers draw the erroneous conclusion that performance feedback is only effective for poor performance.
It is easy to laugh in hindsight about Secrist’s folly, but regression to the mean is a complex and subtle phenomenon that has fooled the brightest of scholars. Blunders continued to occur in the literature, despite Secrist’s public embarrassment. R.L. Thorndike (1942) and Milton Friedman (1992), for example, exposed prominent researchers as merely repeating Secrist’s blunder. A recent example in our own field of neglecting regression to the mean is the debate over Collins’ (2001) book Good to Great. In a 2008 issue of The Academy of Management Perspectives, scholars observed that Collins’ “great” companies failed to outperform the S&P 500 in the decade following publication of his book. Although the commentators pointed out several flaws in Collins’ methodology, none pointed out the fact that focusing on a small subset of high-performing companies is a recipe for observing declining performance in subsequent years. This was analogous to Secrist focusing on only the top quartile of department stores.
The field of I-O psychology has become increasingly disenchanted with studies using cross-sectional designs, and there is a movement toward examining individual and team performance over time (Highhouse & Schmitt, in press). As scholars increasingly turn their attention toward performance trends, they would do well to guard against falling into the regression trap. In addition, a hot topic among I-O practitioners is to identify “high potentials,” or future organizational leaders—even suggesting that the identification of talent become a field of its own (e.g., Silzer & Church, 2009). Anytime people track the performance of a select group, however, they run the risk of drawing erroneous inferences based on a phenomenon observed by Galton over 100 years ago.
Horace Secrist died in 1943, at the age of 61. Although he authored 13 textbooks on economics and statistics, and was named director of Northwestern University’s bureau of business research, he will forever remain an unfortunate symbol of the regression fallacy.
Collins, J. (2001). Good to great. New York, NY: HarperCollins.
Davis, N. (2009, November 19). Second-year slippage afflicts Ryan, others. USA Today.
Friedman, M. (1992). Do old fallacies ever die? Journal of Economic Literature, 30, 2129–2132.
Highhouse, S., & Schmitt, N. (in press). A snapshot in time: Industrial-organizational psychology today. In N. Schmitt and S. Highhouse (Eds.), Handbook of psychology, volume 12: Industrial-organizational psychology). Wiley.
Kahneman, D., & Tversky, A. (1973). On the psychology of prediction. Psychological Review, 80, 237–251.
Schall, E. M., & Smith, G. (2000). Do baseball players regress toward the mean? The American Statistician, 54, 231–235.
Secrist, H. (1933). Triumph of mediocrity in business. Chicago, IL:â€ˆBureau of Business Research, Northwestern University.
Silzer, R. & Church, A.H. (2009). The pearls and perils of identifying potential. Industrial and Organizational Psychology, 2, 377–412.
Stigler, S. M. (1996). The history of statistics in 1933. Statistical Science, 11, 244–252.
Stigler, S. M. (1997). Regression towards the mean, historically considered. Statistical Methods in Medical Research, 6, 103–114.
Thorndike, R. L. (1942) Regression fallacies in the matched group experiment. Psychometrika, 7, 85–102.