14.03.2014 Views

Modeling and Multivariate Methods - SAS

Modeling and Multivariate Methods - SAS

Modeling and Multivariate Methods - SAS

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

144 Fitting Stepwise Regression Models Chapter 4<br />

Models with Crossed, Interaction, or Polynomial Terms<br />

Table 4.5 Description of the Current Estimates Report (Continued)<br />

SS<br />

Reduction in the error (residual) sum of squares (SS) if the term is entered<br />

into the model or the increase in the error SS if the term is removed from the<br />

model. If a term is restricted in some fashion, it could have a reported SS of<br />

zero.<br />

“F Ratio” Traditional test statistic to test that the term effect is zero. It is the square of a<br />

t-ratio. It is in quotation marks because it does not have an F-distribution for<br />

testing the term because the model was selected as it was fit.<br />

“Prob>F”<br />

Significance level associated with the F statistic. Like the “F Ratio,” it is in<br />

quotation marks because it is not to be trusted as a real significance<br />

probability.<br />

R Note: Appears only if you right-click in the report <strong>and</strong> select Columns > R.<br />

Multiple correlation with the other effects in the model.<br />

Step History Report<br />

As each step is taken, the Step History report records the effect of adding a term to the model. For example,<br />

the Step History report for the Fitness.jmp example shows the order in which the terms entered the model<br />

<strong>and</strong> shows the statistics for each model. See “Example Using Stepwise Regression” on page 135.<br />

Use the radio buttons on the right to choose a model.<br />

Figure 4.8 Step History Report<br />

Models with Crossed, Interaction, or Polynomial Terms<br />

Often with models from experimental designs, you have cross-product or interaction terms. For continuous<br />

factors, these are simple multiplications. For nominal <strong>and</strong> ordinal factors, the interactions are outer products<br />

of many columns. When there are crossed terms, you usually want to impose rules on the model selection<br />

process so that a crossed term cannot be entered unless all its subterms (terms that contain it) are in the<br />

model.<br />

Example Using Interaction Terms<br />

1. Open the Reactor.jmp sample data table (Box, Hunter, <strong>and</strong> Hunter 1978).

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!