Fit Log Linear by Linear Association models and Rasch family models by pseudolikelihood estimation