Regression Analysis of Subjects' Self-Reported Life Changes in an Attempt to Improve Illness Prediction,
Abstract
A regression model was applied to the prediction of illness from recent life changes information. Subjects in the validation sample were 2,378 enlisted men aboard three Navy cruisers. The cross-validation sample sonsisted of 1,085 enlisted men from a battleship. The efficacy of multiple regression equations using combinations of single items for prediction of illness based upon the total validation sample and upon specific subgroups differing in age and marital status was compared. Multiple correlations based upon equations for specific subgroups were superior to those for the total sample, but cross-validations of regression equations for specific subsamples were not statistically significant. Life change event specificities for different subgroups of men, as well as insufficiently high illness rates for these subjects, appeared to partially explain the insignificant cross-validation results. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1972
- Accession Number
- AD0749351
Entities
People
- E. K. Eric Gunderson
- Phyllis D. Jensen
- Richard H. Rahe