The robust beauty of improper linear models in decision making.

  title={The robust beauty of improper linear models in decision making.},
  author={Robyn M. Dawes},
  journal={American Psychologist},
  • R. Dawes
  • Published 1 July 1979
  • Psychology
  • American Psychologist
Proper linear models are those in which predictor variables are given weights in such a way that the resulting linear composite optimally predicts some criterion of interest; examples of proper linear models are standard regression analysis, discriminant function analysis, and ridge regression analysis. Research summarized in Paul Meehl's book on clinical versus statistical prediction—and a plethora of research stimulated in part by that book—all indicates that when a numerical criterion… 

Tables from this paper

Dangerous predictions: evaluation methods for and consequences of predicting dangerous behavior
This thesis focuses on prediction in the social sciences. We begin by discussing the “clinical efficiency” of prediction methods as defined by Meehl and Rosen (1955), and present three equivalent
Interpretable classifiers using rules and Bayesian analysis: Building a better stroke prediction model
A generative model called Bayesian Rule Lists is introduced that yields a posterior distribution over possible decision lists that employs a novel prior structure to encourage sparsity and has predictive accuracy on par with the current top algorithms for prediction in machine learning.
Seeing the forest from the trees: When predicting the behavior or status of groups, correlate means.
When measures of individual differences are used to predict group performance, the reporting of correlations computed on samples of individuals invites misinterpretation and dismissal of the data. In
Using a Model of Analysts’ Judgments to Augment an Item Calibration Process
This work exploits a key finding from behavioral decision-making research by seeking a model to mimic how analysts integrate FT item level statistics and graphical performance plots to predict the analyst's assignment of the item’s status.
A Geometric Analysis of When Fixed Weighting Schemes Will Outperform Ordinary Least Squares
An analytic method to determine when a choice of fixed weights will incur less mean squared error than OLS as a function of sample size, error variance, and model predictability is presented.
Studies from psychology, education, personnel, marketing, and finance showed that bootstrapping forecasts were more accurate than forecasts made by experts using unaided judgment, and it can serve as a supplement for econometric models.


Estimating Coefficients in Linear Models: It Don't Make No Nevermind
It is proved that under very general circumstances coefficients in multiple regression models can be replaced with equal weights with almost no loss in accuracy on the original data sample. It is
The Relative Efficiency of Regression and Simple Unit Predictor Weights in Applied Differential Psychology
A very common problem in the behavioral and social sciences is the prediction of the standing of a person or thing on one variable, usually designated the criterion, from his or its standing on a
Man versus model of man: A rationale, plus some evidence, for a method of improving on clinical inferences.
Clinical psychologists, physicians, and other professionals are typically called upon to combine cues to arrive at some diagnostic or prognostic decision. Mathematical representations of such
A Comparison of Five Variable Weighting Procedures
an estimate of the multiple correlation coefficient in the population from which the sample was drawn, and the sample beta weights are taken as estimates of the population beta weights. It is these
Ridge Regression in Practice
Summary The use of biased estimation in data analysis and model building is discussed. A review of the theory of ridge regression and its relation to generalized inverse regression is presented along
Unit And Random Linear Models In Decision Making.
An experiment is reported based on the scheduling production problem in which the correlations coefficient is compared with actual costs and it is found that the unit and random rules yielded much higher costs than human judgment.
How to Use Multi-Attribute Utility Measurement for Social Decision Making
Abstract : The thrust of this paper is that a public value is a value assigned to an outcome by a public, usually by means of some public institution that does the evaluating. This amounts to
Three steps towards robust regression
The three most commonly used statistics, the arithmetic mean, variance, and the product-moment correlation, are most unfortunate choices when data are not strictly Gaussian. A new measure of