Statalist


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

Re: st: Conditional logit - how to account for within-subject correlation


From   Shehzad Ali <drshehzad_ali@yahoo.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Conditional logit - how to account for within-subject correlation
Date   Mon, 3 Aug 2009 03:25:20 -0700 (PDT)

Sorry to nudge again on this; since the question was sent over the weekend, I thought some potential respondents may have missed it.
Any help on this would be appreciated.
Thank you,
Shehzad



----- Original Message ----
> From: Shehzad Ali <drshehzad_ali@yahoo.com>
> To: Stata List <statalist@hsphsun2.harvard.edu>
> Sent: Saturday, August 1, 2009 1:10:19 AM
> Subject: st: Conditional logit - how to account for within-subject correlation
> 
> Hi Listers,
> 
> I am estimating a conditional logit model for a choice experiment where each 
> individual is asked to select their preferred treatment between two 
> alternatives. Each individual is presented with 10 choice sets (with two 
> alternatives in each set). Alternatives are labelled A and B and have meaning 
> because of the associated attributes.
> 
> The alternative specific variables are: adverse_events, disability_dummy, 
> frequency_2, frequency_3, cost 
>  
> I also have individual specific variables: age, sex, education, ethnicity, 
> employment_status income
>  
> Previously I was advised that, to estimate the effect of case-specific variables 
> on choice made, I should use interaction terms. However, interacting all the 
> case-specific variables with all alternative-specific variables would generate 
> too many variables. 
>  
> So I have three questions:
>  
> a) Is there any way I could model the effect of case-specific variables without 
> having to use a too many interaction terms?
>  
> b) If I want to account for the fact that each set of 10 choicesets was 
> responded by one individual with the same case-specific variables, whats the 
> best way to do so? My -clogit- command below takes account care of the fact that 
> the alternatives belong to the same choice set but not that each set of 10 
> choice sets are responded by the same individual.

> clogit choice adverse_events disability_dummy frequency_2 frequency_3 cost, 
> group(choiceset)
> 
> c) If I use the above model, are case-specific variables assumed to be random or 
> are they part of the intercept?
> 
> I would appreciate any help with this.
> Thank you,
> Shehzad
> 
> 
>       
> 
> *
> *  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/



      

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index