On the Relevance of Extremes vs. Means in Organization Science
published: Nov. 21, 2007, recorded: October 2007, views: 6832
Report a problem or upload filesIf you have found a problem with this lecture or would like to send us extra material, articles, exercises, etc., please use our ticket system to describe your request and upload the data.
Enter your e-mail into the 'Cc' field, and we will keep you updated with your request's status.
Scalability is a key element of complexity science. Many complex systems tend to be selfsimilar across levels—the same dynamics work at multiple levels. They are explained by scaling laws. Scalability results from what Mandelbrot calls fractal geometry. A cauliflower is an obvious example. Fractals often show Pareto distributions and are signified by power laws. Researchers find organization-related power laws in intrafirm decisions, consumer sales, salaries, size of firms, movie profits, director interlocks, biotech networks, and industrial districts, for example. Power laws signify Pareto distributions, which show “fat tails,” (nearly) infinite variance, unstable means, and unstable confidence intervals. Pareto distributions are alien to most quantitative organizational researchers, who are trained in Gaussian statistics and are trained to go to great lengths to configure their data to fit the requirements of linear regression, normal distributions, and related statistical methods. While normal distributions, and related current quantitative methods are still relevant for a significant portion of organizational research, power laws signify that Pareto distributions, fractals, and underlying scale-free theories are increasingly pervasive and valid characterizations of organizational dynamics. Where true, researchers ignoring power law effects risk drawing false conclusions and promulgating useless advice to practitioners. This because what is important to most managers are the extremes they face, not the averages. The implications for organization science, however, go beyond extreme events. Tools do not exist in a theoretical vacuum. The adoption of normal distribution statistics carries a heavy baggage of assumptions. Reliance on linearity, randomness, gradualism, and equilibrium influences how theories are built, how legitimacy is conferred, and how research questions are formulated. We begin with findings about 80 kinds of power laws. Then, we present sixteen scale-free theories that apply to organizations. Next, we discuss research implications. Then, we discuss implications in terms of the basic predictor function, y = f(x) + ε. How does basic thinking about prediction, data, statistics, and the error term have to change?
Link this pageWould you like to put a link to this lecture on your homepage?
Go ahead! Copy the HTML snippet !