Statistics Help @ Talk Stats Forum - Stata http://www.talkstats.com/ Stata usage and programming, Stata help en Wed, 19 Jun 2013 09:51:34 GMT vBulletin 60 http://www.talkstats.com/images/misc/rss.png Statistics Help @ Talk Stats Forum - Stata http://www.talkstats.com/ Egen rowtotal http://www.talkstats.com/showthread.php/44617-Egen-rowtotal?goto=newpost Tue, 18 Jun 2013 19:56:57 GMT I want to sum the values of three variables but I want to treat the missing values as missing values and not 0s. The egen rowtotal command treat... I want to sum the values of three variables but I want to treat the missing values as missing values and not 0s. The egen rowtotal command treat missing as 0s and for this analysis this is not correct. How can I treat the missing as missing and not 0s.

Thank you
Marvin ]]> Stata Marvin85 http://www.talkstats.com/showthread.php/44617-Egen-rowtotal <![CDATA[T test & Wilcoxon signed rank test in stata- for case control studies]]> http://www.talkstats.com/showthread.php/44611-T-test-amp-Wilcoxon-signed-rank-test-in-stata-for-case-control-studies?goto=newpost Tue, 18 Jun 2013 15:35:31 GMT Hi

I'm looking for some help with T-est and Wilcoxon signed rank test in Stata 12. I'm analysing data from a case control study where the cases and control are matched 1:1. I used an independant t test in stata without difficulty to compare the dependant variables when the outcome was a number (eg age of patient). I'm unsure of which test to use when the outcome of the variable is either yes or no, i.e. comparing the incidence of a variable between the group. I used the Wilcoxon signed rank test to compard two groups (cases and control) and the dependant variable (the same for each group). I'm not sure if this is the correct test to use and I'm also unsure of how to interpret the test in Stata.

I hope someone can help with this.

Thanks. ]]> Stata catherine_mcl http://www.talkstats.com/showthread.php/44611-T-test-amp-Wilcoxon-signed-rank-test-in-stata-for-case-control-studies Regression and ordinal data http://www.talkstats.com/showthread.php/44580-Regression-and-ordinal-data?goto=newpost Mon, 17 Jun 2013 11:35:05 GMT I'm using a 0-4 scale of severity which is generated from a larger questionnaire set to predict my outcome measure. In this instance, do you always have to use the 'i.' command? When I do this, the number in each cell is so low it makes the analysis somewhat meaningless. Would it be appropriate to take the groups out?

Thnx ]]> Stata loops22 http://www.talkstats.com/showthread.php/44580-Regression-and-ordinal-data