Jmp Categorical

Stat 621, Categorical variable, workout

November 5, 2007

1

Objective

• Know how to take the JMP output from a regression including categorical X-variables, and identify the individual regression line for each group. • Illustrate with case book examples

2

Example 1. Two level categorical, no interaction: manager.jmp.

• Output on page 169. • Goal: pull out the External’s regression line. • Method: from the output pull all coefficients that are to do with externals – there are 3 relevant terms. • Common intercept term: -1.843 • External’s difference from the common intercept: -0.257 • Common slope (on Salary) 0.107 • Put them together: F it = (−1.843 − 0.257) + 0.107Salary.

3

Example 2. Two level categorical, with interaction: manager.jmp.

• Output on page 173.

1

• Note that the interaction terms have been centered about the mean of the continuous variable (in this case Salary). • Consequence: part of the intercept term will be found by expanding the interaction. • Goal: pull out the External’s regression line. • Method: from the output pull all coefficients that are to do with Externals – there are 5 relevant terms. • Term 1: the common intercept -1.815. • Term 2: External’s difference from the common intercept -0.254. • Term 3: the common Salary coefficient 0.107. • Terms 4 and 5: these are both found by expanding the interaction term for the Externals. (Salary - 71.63)*Origin[External] ... -0.00185. • Because we are looking at the Externals you can view the above equation as simply substituting in External’s = 1. • Consequently on expansion of this interaction term we obtain two parts: – Contributing to the Salary slope: Salary * -0.00185 – Contributing to the intercept: -71.63 * -0.00185 = 0.133 • The first of these terms is a part of the Salary slope for the External’s, the second is a part of the intercept (as it does not depend on Salary). • Collect terms: – Intercept terms: -1.815 -0.254 + 0.133 = -1.936 – Slope terms (0.107 - 0.002) Salary...