MODELING TRIVARIATE CORRELATED BINARY DATA | ||||
| Al-Azhar Bulletin of Science | ||||
| Article 3, Volume 26, Issue 2-B, December 2015, Page 1-13 PDF (383.9 K) | ||||
| Document Type: Original Article | ||||
| DOI: 10.21608/absb.2015.22635 | ||||
 View on SCiNiTO | ||||
| Author | ||||
| AHMED MOHAMED MOHAMED EL-SAYED* | ||||
| Department of High Institute for Specific Studies Management Information Systems Nazlet Al-Batran, Giza, Egypt. | ||||
| Abstract | ||||
| This paper provides the estimation and the test procedures for the measures of association in the correlated binary data associated with covariates. The generalized linear model (GLM) using the serial dependence and the developed alternative quadratic exponential form (AQEF) procedures are employed for the trivariate binary correlated outcome variables. The log-odds ratios, as measures of association, are estimated and the appropriate tests are suggested. For comparison between the two procedures, we used the simulation study and an application to an ecology problem which involves the estimation of the measures of association and their tests. The over-dispersion criteria is investigated for these procedures. Finally, the deviance and scaled deviance are used as tests for the goodness of fit of the model to determine the best procedure. | ||||
| Keywords | ||||
| Trivariate Bernoulli Distribution; Markov model; Generalized linear model; Deviance; Likelihood ratio test; Maximum likelihood estimators; Alternative quadratic exponential form | ||||
|
Statistics Article View: 100 PDF Download: 280 |
||||
