作者: Anton K. Formann
DOI: 10.1080/01621459.1992.10475229
关键词: Statistics 、 Goodness of fit 、 Multinomial logistic regression 、 Identifiability 、 Expectation–maximization algorithm 、 Latent variable model 、 Generalised logistic function 、 Mathematics 、 Latent class model 、 Logistic regression
摘要: Abstract For latent class analysis, a widely known statistical method for the unmixing of an observed frequency table into several unobservable ones, flexible model is presented in order to restrain unknown sizes (mixing weights) and response probabilities. Two systems basic equations are stated such that they simultaneously allow parameter fixations, equality certain parameters as well linear logistic constraints each original parameters. The maximum likelihood this “linear analysis” given, their estimation by means EM algorithm described. Further, criteria local identifiability tests (Pearson- likelihood-ratio-χ 2) goodness fit outlined. practical applicability analysis demonstrated three examples: mixed regression, Bradley-Terry paired comparisons with ties, lo...