Cem July 2012

cem@lists.gking.harvard.edu

1 participants
1 discussions

Trouble integrating imputed data into imb and cem using miname and misets options

by Julie Rajaratnam

Hello cem-list, I am a new user of cem and for some reason, the google search function of the archives returns zero results when I attempt to search for previous posts on this topic. I have browsed the past 5-6 months of postings and have not found a post on this particular error, so I will go ahead and post my question. I have created 5 MI-imputed datasets in Stata, and placed them in my current working directory. Per the cem manual, I have the first one open, and I attempt to run imb using the miname and misets options. I get the following error: variable __000000 not found st_data(): 3500 invalid Stata variable name imbalance(): - function returned error <istmt>: - function returned error r(3500); When I run imb or cem on just the first dataset in memory (i.e. the command line is exactly the same, just without the miname and misets options), it works just fine. So something appears to be going wrong when I specify the mi options. I did verify that the file names have the appropriate stem ("imputed") and they reside in the working directory, and there are 5 of them (misets = 5). The set trace command in Stata shows the error occurring at this step: - mata: imbalance("`varlist'","`breaks'","`treatment'",`uw',"`miname'",`misets') = mata: imbalance("age_begin ea_urban hf_type hf_level hf_cqi hf_baylor hh_wealthq_dhs m_educx","scott","hf_arm",0,"imputed",5) Also, I am using Stata 12 on a Windows 7 machine. Many thanks for any tips on this! Julie Rajaratnam

11 years, 9 months

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

Cem July 2012