Regression for Economics

Chapter 9

Pitfalls of Regression Analysis

Forming Incorrect Hypotheses

Using regression analysis without a theory to guide it is a waste of time, if not outright dangerous. Let us start with a simple hypothetical example. Afterward, real data will be used to demonstrate the problem.

Assume we have collected the following data, without any theory whatsoever.

Y	X1	X2
11	5	12
13	12	10
17	16	6
19	19	4
21	39	1

Enter this data into Excel. Perform regression on the following two models:

YA = β0 + β1X1 + ε (9.1)

YB = β0 + β2 X2 + ε (9.2)

The only reason for labeling Y with subscripts A and B is for reference. First regress Y on X1. The result is shown in Table 9.1.

Table 9.1. Regression of Y ...