hw4inst - alphabetically sorted as well before you run...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
A few caveats for running log-linear and logit models: ______________________________________________________ 1) The key thing is to type in the data contained in a table. Remember to declare the frequency variable as a numeric variable. Follow the data format in the handout given in class for the defendant's race, victim's race and death penalty verdict example. 2) Before you start Analyzing go to Data-weight cases and weight cases by the frequency variable. 3) When you run a log-linear model always make sure you click on the model option and specify the model you need. The default choice saturated model may give you 0 values for all statistic. You may have to choose only main effects in your customized model (problem 7 and 10 only ask for main effects). 4) For dealing with ordinal data, you have to have the categories
Background image of page 1
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: alphabetically sorted as well before you run ordinal regression. For Regression Diagnostics __________________________ 1) For running stepwise, backward or forward models, input all your X's in the space for independent variables (predictors), SPSS will automatically select from your list. You do not have to choose them a'priori. 2) For regression diagnostics, you will go to analyze-regression-linear and then input your Y and X's then click on the small box saying "save" at the bottom of the screen. Once you click on that box, a screen will appear with tons of choices, tick on the diagnostics you want. The output will be stored NOT in the output file but in the DATA FILE, to help you sort through the list....
View Full Document

This note was uploaded on 07/08/2011 for the course STA 6127 taught by Professor Mukherjee during the Fall '08 term at University of Florida.

Ask a homework question - tutors are online