ERIC Number: ED351389
Record Type: RIE
Publication Date: 1992-Apr
Reference Count: N/A
A Binomial Test of Model Fit.
Fraas, John W.; Newman, Isadore
A new method for evaluating model fit that is easy to use and interpret is presented. The new method, which uses a binomial test of the number of hypotheses (paths) in a model that are supported by the data, has heuristic value when considering problems associated with other goodness-of-fit measures. An application of the binomial test as a goodness-of-fit measure is presented. The procedure is applied to a career development model studied by K. K. Sidhu (1988). A hypothesis (path) is judged to be supported by the data when the parameter estimate possesses the hypothesized sign and is statistically significant. For the model, the parameter estimates were expected to be positive and T values were expected to exceed two before the coefficient would be considered statistically significant. Since all seven of the parameter estimates had positive signs and T values in excess of two, all seven hypotheses are judged to be supported by the data. In the final step, the binomial probability value is calculated. Two tables and one figure illustrate the application. (SLD)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: N/A
Identifiers: Binomial Test; T Values
Note: Paper presented at the Annual Meeting of the American Educational Research Association (San Francisco, CA, April 20-24, 1992).