Notice: On March 31, it was **announced** that Statalist is moving from an email list to a **forum**. The old list will shut down on April 23, and its replacement, **statalist.org** is already up and running.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

From |
"Ploutz-Snyder, Robert (JSC-SK)[USRA]" <robert.ploutz-snyder-1@nasa.gov> |

To |
"statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu> |

Subject |
st: Canonical Correlation with -mim- |

Date |
Mon, 24 May 2010 13:13:32 -0500 |

I have a set of X predictors and a set of Y outcomes, and I want to perform a Canonical Correlation to better understand potential relationships between/among linear combinations of X's and linear combinations of Y's. -> [FYI, there's a nice walk-through of Canonical Corr with Stata at UCLA's Academic -> Technology Services website at http://www.ats.ucla.edu/stat/stata/dae/canonical.htm] However there are a lot of X's and a lot of Y's, and missing data problems plague my listwise n. Thus, by Bayesian multiple imputation, I've compiled a "mim-compatible" dataset and wish to use the -mim- prefix to canonical correlation. The rub: The -mim- prefix command uses Rubin's rules for combining estimates on multiple imputed dataset and supports a number of Stata estimation commands, however the -canon- command is not currently among the list of supported routines. My read of the -mim- help files suggests that I should be able to specify the mim_options category(mim_type), which I assume would be cat(combine), but I'm stumped when trying to further define this option by the estimates(??) that is required when using the cat() option. There is a lot of estimation involved with canonical correlation, so I don't know what to use for the est() option?? Example: mim, cat(combine) est() : canon sy1 sy2 sy3 sy4 = sx1x2 sx5 sx6 sx7x8 sx13 sx14 Question #1: HAS ANYONE TRIED using -mim- with -canon-, and if so, can you provide guidance here? Question #2: IS IT POSSIBLE/LEGITIMATE to use some other -mim- supported command (ex. reg) to merely generate a var/covar matrix, and then somehow feed that matrix into the -canon- routine? Seems like that could be an option for getting the analysis I want, utilizing the multiply imputed data set that I have, but I'm not sure if it is possible or legitimate. Any guidance is much appreciated. Rob * * For searches and help try: * http://www.stata.com/help.cgi?search * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/

**References**:**st: Panel Multinomial Logistic Model***From:*"Lee, Albert" <Albert.Lee@hud.gov>

**Re: st: Panel Multinomial Logistic Model***From:*"Michael N. Mitchell" <Michael.Norman.Mitchell@gmail.com>

- Prev by Date:
**Re: st: Panel Multinomial Logistic Model** - Next by Date:
**Re: st: can the total sampling weights be applied to subsample analysis** - Previous by thread:
**Re: st: Panel Multinomial Logistic Model** - Next by thread:
**Re: st: Panel Multinomial Logistic Model** - Index(es):