I am using lrm in Design package for a project using logit analysis. I think lrm is very useful for providing information about the number of missing values due to the inclusion of each variable. My first questions is: How to explore those observations that are automatically deleted from the lrm? For example, in a data I have more than 6,000 observations but after list-wise deletion I have only 2,411 observations, where there are 583 observations for the dependent variable of value 1 and 1,828 for the value 0. One independent variable of the model "cDisagLW" accounts for the deletion of 2,598 observations and I think this could cause problem in regression result. So, could you show me a way to find out "who" are those 2,598 observations? A further question is, how to make a subset of the original data that contains these 2,411 valid observations? With this dataset, I like to explore the underlying distribution of the regression data and contrast to the original data of 6,000 observations. Thank you in advance. Frank
Frank C.S. Liu Assistant Professor, Graduate Institute of Political Science E-mail: frankcsliu at gmail.com National Sun Yat-sen University (NYSYU) Office:+886.7.525.2000 #5555 Kaohsiung, Taiwan 804, R.O.C. FAX:+886.7.525.5540