Discriminant Analysis (DISCRIM)

1. Open the data file Discrim 8 (attached) . These are data from n = 465 women, and include measures on the following variables:

• locus of control (CONTROL), with larger values indicating an external locus of control;
• satisfaction with current marital status (ATTMAR), with larger values indicating less favorable attitudes;
• attitude toward women's role (ATTROLE), with larger values indicating more conservative attitude;
• attitude toward homemaking (ATTHOUSE), with larger values indicating less favorable attitudes.

A grouping variable, role satisfaction, is also included. The grouping variable has three levels: (1) working women, (2) role satisfied housewives, and (3) role-dissatisfied housewives.

PERFORM THE FOLLOWING:

1. Screen the data for outliers, are there any? Yes or no? If yes what are they?
2. Are any variables severely skewed? Yes or No? If yes which one?
3. Are sample sizes approximately equal? Yes or No?
4. Should you use equal or weighted prior probabilities?

e . Is equality of covariance matrices satisfied?

f . Create a "training sample" in the following manner:

Using the "select cases" procedure, select a random sample of 50% of the cases. This will create a binary indicator variable. Now use "select cases" again and re-select all cases (note that the binary indicator variable you created remains).

1. Carry out DISCRIM using the training sample (i.e., using the indicator variable as a selection variable in the DISCRIM dialog box). Interpret the results.
2. C reate a new variable that indicates the predicted group status for all cases (you needn't show any output for this new variable).
   1. Interpret t he discriminant function plot.

Price: $14.61

Solution: The downloadable solution consists of 10 pages, 461 words and 13 charts.
Deliverable: Word Document