Linear Model Building

13 Pages Posted: 5 Apr 2010

See all articles by Phillip E. Pfeifer

Phillip E. Pfeifer

University of Virginia - Darden School of Business

Abstract

This note provides guidelines for the effective construction and application of linear models. We begin with a discussion and categorization of the various uses for a linear model. Whether a given model is appropriate for a particular use depends on the way the independent and dependent variables relate to each other. Next, we describe the kinds of possible relationships between Y and X variables, and then we emphasize the kind of relationship between Y and X necessary for each potential application of a linear model.

Excerpt

UVA-QA-0293

LINEAR MODEL-BUILDING

Linear (regression) models are used in a variety of business situations for a variety of purposes. One reason for their popularity is that they can be easily understood and implemented. All that is needed is data on two or more variables and a computer or calculator. With a minimal amount of effort, a manager obtains (1) an equation for the relationship between the dependent variable (Y) and one or more independent variables (the Xs), (2) a forecast of Y for any given set of X-values, and (3) a host of statistics to complement the analysis. Because so much is obtained so easily, the potential for misuse is great if the model builder is not careful about how the model is built and applied.

This note provides guidelines for the effective construction and application of linear models. We begin with a discussion and categorization of the various uses for a linear model. Whether a given model is appropriate for a particular use depends on the way the independent and dependent variables relate to each other. Next, we describe the kinds of possible relationships between Y and X variables, and then we emphasize the kind of relationship between Y and X necessary for each potential application of a linear model.

Even when a manager understands what kind of relationship is necessary for the particular way the model will be used, there still remains the question of how best to combine this manager's knowledge of the situation with available data to build an effective model. In the fourth section, we propose a procedure for building a linear model when there is only limited prior knowledge about the underlying relationships among the variables. The procedure suggested attempts to maximize the use of the managers' prior knowledge and judgments, yet allows for the constructive and intelligent use of the available data. The objective is to guard against overreacting to peculiarities of the data and to build a sound and effective model that is consistent with both the modeler's prior knowledge and the information found in the data. In contrast to this advice on how to go about building a good model, we present a list of things not to do, explaining some common mistakes associated with the building and using of linear models.

Uses of the Linear Model

. . .

Keywords: model

Suggested Citation

Pfeifer, Phillip E., Linear Model Building. Darden Case No. UVA-QA-0293, Available at SSRN: https://ssrn.com/abstract=1584517

Phillip E. Pfeifer (Contact Author)

University of Virginia - Darden School of Business ( email )

P.O. Box 6550
Charlottesville, VA 22906-6550
United States
434-924-4803 (Phone)

HOME PAGE: http://www.darden.virginia.edu/faculty/Pfeifer.htm

Here is the Coronavirus
related research on SSRN

Paper statistics

Downloads
23
Abstract Views
332
PlumX Metrics