Discriminant Analysis

This analysis is used when you have one or more normally distributed interval independent variables and a categorical variable.

Linear discriminant performs a multivariate test of difference between groups. It is also useful in determining the minimum number of dimensions needed to describe these differences.

Procedure

1. From the menu, click Analyze- Classify- choose Discriminant
2. Select an integer- valued grouping variable(the variable that you think defines different groups).it will appear in the box with two question marks, you have to tell SPSS what the codes are for the two groups that you want to compare.
3. click Define Range to specify the categories of interest  g. “1” and “2” as the different values of the variable that you wish to compare
4. Select the independent (predictors) variables and move them to the Independent box
5. Click on the statistic button and check the “Means”

• This sub dialogue box lets you see descriptive statistics on each predictor variable for the different groups.

6. Click on continue to move back to the main dialogue box.
7. Click on the classify button. Check the box for “Summary table” and click continue
8. Click ok to generate the output.

The statistical significance of this model is explained by Wilk’s lambda value and the Eigen value

When Eigen value> 1, groups are distinct and hence the model has good discriminating power.

Wilk’s lambda is a measure of the extent of misfit of the Discriminant solution. The value of Lambda range from 0 to 1, a value closer to 0 indicates the groups are distinctly different and a value closest to 1 shows groups are overlapping.

When lambda is <0.5, solution is statistically significant and acceptable.