cslogistic (version 0.1-2)

cslogistic: Perform an Analysis of a conditionally specified logistic regression model

Description

This package contains functions for likelihood and posterior analysis of conditionally specified logistic regression models.

Details

Assume that for each of $n$ experimental units the values of $m$ binary variables $$Y_{i1}, \ldots, Y_{im}$$ are recorded. The 'MleCslogistic' and 'BayesCslogistic' functions fit a conditional specified logistic regression model, such that for $i = 1, \ldots, n$ and $j = 1, \ldots, m$,

$$\mbox{logit} P(Y_{ij}=1 | Y_{ik}=y_{k}, k \neq j) = X_{ij}\beta_j + \sum_{k=1, k \neq j}^m \alpha_{jk} y_k$$

where, the parameters $\alpha_{jk}$ have interpretation as conditional log-odds ratios and the parameters $\beta_j$ correspond to the regression coefficients associated to the vector of covariates $X_{ij}$. For compatibility of conditional distributions it is assumed that $\alpha_{jk} = \alpha_{kj}$, $j \neq k$.

References

Garcia-Zattera, M. J., Jara, A., Lesaffre, E. and Declerck, D. (2005). On conditional independence for multivariate binary data in caries research. In preparation.

Joe, H. and Liu, Y. (1996). A model for multivariate response with covariates based on compatible conditionally specified logistic regressions. Satistics & Probability Letters 31: 113-120.

See Also

MleCslogistic, BayesCslogistic.