Implications of Realistic Utility Functions for Placement Using Aptitude-Treatment Interaction.
Abstract
When aptitude-treatment interactiorn (ATI) is used in placement decisions, it is generally assumed that a candidate should be placed in the treatment whose regression equation yields a higher predicted score. This can be justified using decision theory if utility functions are assumed linear and hence, unbounded. However, realistic utility functions ought to be bounded. Then the conventional placement rule is invalid. The optimum decisions depend o both the prediction equations and the utility functions. Therefore, the decision rule takes different forms in different situations. This is illustrated by assuming utilities to be proportional to normal distribution functions. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1980
- Accession Number
- ADA097459
Entities
People
- D. R. Divgi
Organizations
- University of Iowa