Ant Analysis

Discriminant Analysis (Dr See) •

Use to discriminate 2 mutually exclusive groups (e.g. Group 1: with M Ed and letters of recommendation and Group 2: with no M Ed but working experience) of candidates who can complete the Ph.D course. Can the regression equation created be able to analyze the likelihood of success for future applicants?

•

Similar to cluster analysis in dividing the sample or cases into 2 discrete groups and differs only in the procedure for creating the groups.

How to run a discriminant analysis? Example: There is a sample of 50 students selected for the Ph D course. The dependent variable (discrete variable) 1: candidate completed the PhD course , 2: candidate did not complete the PhD course, are used. The dependent variable is category (1 = finished the Ph.D., 2 = did not finish), and 17 independent or predictor variables are utilized to predict category membership in one of these two groups: 1. 2. 3. 4. 5. 6. 7. 8. 9.

gender: 1 = female, 2 = male age: age in years at time of applicatron marital: 1 = married, 2 = single gpa: overall undergraduate GPA areagpa: GPA in their area of specialty grearea: score on the major-area section of the GRE grequant: score on the quantitative section of the GRE greverbal: score on the verbal section of the GRE letter1: fir-stof the three recommendation letters (rated 1 = weak through 9 = strong) 10.letter2: second of the three recommendation letters (same scale) 11. letter3: third of the three recommendation letters 12.motive: applicant's level of motivation (1 = low through 9 = high) 13.stable: applicant's emotional stability (same scale for this and all that follow) 14.resource: financial resources and support system in place 15.interact: applicant's ability to interact comfortably with peers and superiors 16.hostile: applicant's level of inner hostility 17. impress: impression of selectors who conducted an interview

Step 1: Select the number of variables that should be small (eg 17 but not 100 which is not feasible) then compute a correlation matrix (pooled within-group correlation matrix ) of predictor variables.

2 1. 2.