Premium
Testing case‐descriptive models
Author(s) -
Starbuck William H.
Publication year - 1961
Publication title -
behavioral science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.371
H-Index - 45
eISSN - 1099-1743
pISSN - 0005-7940
DOI - 10.1002/bs.3830060304
Subject(s) - descriptive statistics , normative , descriptive research , frame (networking) , referent , norm (philosophy) , descriptive geometry , test (biology) , computer science , mathematics , statistics , mathematics education , linguistics , epistemology , telecommunications , paleontology , philosophy , biology
Crudely, there are three reasons models are built. First, models are either normative or descriptive. Within the class of descriptive models, there are norm‐descriptive and case‐descriptive ones. Case‐descriptive models, unlike the other kinds, are not generalizations. They represent a set of referent causal processes by detailed analysis of one, or a few, specific cases. Then such questions arise as: Is the model sufficient? Does it represent the real situation? How can one test its value? For case‐descriptive models, the usual likelihood tests are not appropriate because they are designed to evaluate generalizations. Outlined here, in addition to frame of reference for the building of case‐descriptive models, are suggestions for testing them.