multinomial logistic regression.pdf

Singapore Med J 2005; 46(6) : 259 Biostatistics 305. Multinomial logistic regression Y H Chan Faculty of Medicine National University of Singapore Block MD11 Clinical Research Centre #02-02 10 Medical Drive Singapore 117597 Y H Chan, PhD Head Biostatistics Unit Correspondence to: Dr Y H Chan Tel: (65) 6874 3698 Fax: (65) 6778 5743 Email: medcyh@ .sg CME Article B a s i c S t a t i s t i c s F o r D o c t o r s Multinomial logistic regression is the extension for the (binary) logistic regression(1) when the categorical dependent outcome has more than two levels. For example, instead of predicting only dead or alive, we may have three groups, namely: dead, lost to follow-up, and alive. In the analysis to follow, a reference group has to be chosen for comparison, the appropriate group would be the alive, i.e. dead compared to alive and lost to follow-up compared to alive. The predictors used are two categorical (gender and race) and four quantitative variables (x1 – x4). In SPSS, go to Analyse, Regression, Multinomial Logistic to get Template I. Template I. Multinomial logistic regression. For the initial analysis, let us just use the two categorical independent variables (gender and race), put them in the Factor(s) option. Put the dependent variable Group (1 = alive, 2 = lost to follow-up, 3 = dead) into the Dependent box. The default Reference- Category is Last. Click on the Reference Category button to get Template II. Template II. Reference category definition. Change the Reference Category to “First category”. Leave the Category Order to “Ascending”, this means that the smallest value is the first category. The “Descending” option means that the highest category is the first category (a very misleading and redundant option – need to be cautious!). Click the Model folder in Template I to define the variables to be included in the model, see Template III. The Main effects option will include all the variables specified with no interaction terms whereas the Full f