Setup
        . webuse sysdsn3

    Fit multinomial logistic regression model
        . mlogit insure age male nonwhite site2 site3

    Same as above, but use 2 as the base outcome
        . mlogit insure age male nonwhite site2 site3, base(2)

    Replay, reporting relative-risk ratios
        . mlogit, rrr

命令的查stata自带的帮助

stata10的一些选项改变了

但是基本的是不变的，所以你的查帮助

For answering your question, I found the help resources of Stata.com in http://www.stata.com/help.cgi?mlogit.

In that webpage, you can find the correct command of basecategory such as "baseoutcome(#), base(#) or b(#)", not "basecategory(#)".

However you didn't make a categical variable before using mlogit, I think.

Because you said "... 但是无法指定对照组."

In the mgolit command of Stata,the default is to choose the most frequent outcome.

For example, if "q" of your dependent variable is composed of 5 high income groups(=1), 4 middle income group(=2) and 10 low income group(=3),the base outcome is automatically 3, that is low income group, and also you just do mlogit without base(#).

When you want to change the base outcome or group for additional analysis, you must use the command of bsae(#).

So you must check whether your dependent variable "q" is correct and then do mlogit.

I hope you enjoy Stata's wonderful world.

[此贴子已经被作者于2008-9-1 22:13:41编辑过]

如果选择缺省的话，stata会自动默认对照组为1，你的”0“似乎不能作为对照组。仅供参考。