Kleinbaum and Kupper (Applied Regression Analysis and ...



Canonical Correlation Exercise: Predicting Medical Knowledge

Marascuilo and Levin (Multivariate Statistics in the Social Sciences, 1983, Brooks/Cole) provided the data for this exercise. The data file, MEDICAL.dat, is available on my Data Files page. The following SAS code will read the data file for you:

options pageno=min nodate formdlim='-';

TITLE 'Marascuillo and Levin, 1983, page 214';

DATA Medical; INFILE 'C:\StatData\Medical.dat';

INPUT Year ID SES MAT Order Nurturance Medicine Psychiatry Ob_Gyn Pediatrics

Pub_Hlth Surgery Specialty @@;

PROC

The research participants were medical students. MAT is score on the Miller Analogies Test. ORDER is score on a psychological test of need for order. NURTURANCE is score on a psychological test of need for nurturance. ORDER, NURTURANCE, MEDICINE, PSYCHIATRY, OB_GYN, PEDIATRICS, PUB_HLTH, and SURGERY are scores on measure of knowledge in those domains.

Name your first set of variables KNOWLEDGE, and include in that set all of the variables which measure medical knowledge (medicine through surgery). Name your second set of variables PREDICTORS, and include in that set MAT, order, and nurturance. Use the ALL keyword (rather than the REDUNDANCY keyword) to get all optional statistics from PROC CANCORR.

Extract three pairs of canonical variates, test their significance, and comment on the magnitude of their correlation. Use the canonical coefficients and the loadings to label the six canonical variates obtained. Conduct a redundancy analysis. Summarize and interpret the results of your analysis.

Cut and paste your SAS program and SAS output into a Word document. Add to the top of that document your name and your summary statement and attach the document to email sent to me.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download