Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
st: goodness-of-fit stats with MI data
From
Ryan Wells <[email protected]>
To
"[email protected]" <[email protected]>
Subject
st: goodness-of-fit stats with MI data
Date
Fri, 15 Mar 2013 06:09:00 -0700 (PDT)
I am trying to figure out how to conduct a relevant
goodness-of-fit test(s) in Stata after conducting a logistic regression
model for a multiply imputed dataset.
White, Royston and Wood (2011) say that goodness-of-fit test statistics
should not be combined using Rubin's rules. I know about the Wald-like
"mi test" Stata command, as well as the implementation of R-squared for regression analysis (following Harel, 2009) in the ssc-downloadable command mibeta. I am also aware that there are proposed alternatives (not using Rubin's standard pooling rules) for a pooled likelihood ratio test statistic and a pooled chi-square test
statistic (with associated p-values) that are applicable in a multiply
imputed framework, although to my knowledge these are not implemented in Stata. (If there is a user command for these somewhere, I'd like to
hear about it.) In my application (analyzing ELS education data from
NCES), I would have used a BIC or AIC in a non-imputed setting.
What is the recommended procedure to conduct and report a model
goodness-of-fit test for logistic regression analysis when using mi data in Stata? Thanks.
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/faqs/resources/statalist-faq/
* http://www.ats.ucla.edu/stat/stata/