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: Re: Flexible models for dce
From
Arne Risa Hole <[email protected]>
To
Carla Guerriero <[email protected]>, statalist <[email protected]>
Subject
st: Re: Flexible models for dce
Date
Tue, 28 Jan 2014 09:23:42 +0000
Carla
The -gmnl- command (SJ) can be used to estimate models in WTP space -
see the material available at
http://www.shef.ac.uk/economics/people/hole/stata for more
information.
Arne
On 27 January 2014 15:58, Carla Guerriero <[email protected]> wrote:
> Dear Arne and Stata users ,
> Is there a STATA command to estimate WTP space method ?
> I am using the command you created wtp to estimate wtp and 95% confidence
> interval from mixed logit model coefficient but reading your paper " Mixed
> Logit estimation of willingness to pay distributions: a comparison of model
> in preference and WTP space using data from health-related choice
> experiment" I would be very curious testing my results with the WTP space
> approach compared to the preference space
> Kind Regards
> Carla
>>>> Arne Risa Hole <[email protected]> 27/01/2014 09:36 >>>
> Carla,
>
> Your -mixlogit- (SJ/SSC) specification is correct if you want the mean
> *and* SD of the coefficient for risk reduction vary with the age
> group. If instead you wanted just the mean to vary, and the SD to be
> the same for each age group you could specify:
>
> mixlogit d p age2r age3r age4r, group(id) rand(r) nrep(500)
>
> I would not use the -iter()- option here. If the algorithm doesn't
> converge it is likely to indicate a problem with your model
> specification - in that case the best advice is to simplify the model.
>
> Note that if the respondents in your experiment make more than one
> choice you should specify the -id()- option. See
> http://www.stata-journal.com/article.html?article=st0133 for details.
>
> I don't understand your second question but, yes, if you are
> interested in estimating a latent-class conditional logit model then
> -lclogit- is an appropriate command to use.
>
> Arne
>
> On 25 January 2014 11:49, Carla Guerriero <[email protected]>
> wrote:
>> Dear Arne and Stata users,
>> My experiment present three unlabelled alternatives to test individual
>> willingness to pay for health risk reduction.
>> I started with the classic conditional logit but I want to implement more
>> flexible model..
>> I fund that age is a significant predictor for risk reduction marginal
>> utility .. my quesiotn is can I use the mixed logit command to estimate
>> random parameter coefficient of interaction variable (risksize * agegroup)
>> ?
>> es.
>> gen age1r = age1*r
>> gen age2r =age2 *r
>> gen age3r = age3*r
>> mixlogit d p , group(id) rand(age1r age2r age3r age4r) iter(15) nrep(500)
>> I used the iter(15) because it takes very long to estimate each iteration
>> but I am not sure this is correct ..
>>
>> My second question is more complex I have collected in questionnaire so
>> latent variables (binary measure of risk aversion) How can I include them
>> in
>> the latent class model ?
>> should I use the lclogit command ?
>> Thank you very very much
>> Kind Regards
>> Carla
>>
>>
>> Carla Guerriero
>>
>> Research Fellow in Health Economics
>> Health Policy Unit
>> London School of Hygiene and Tropical Medicine
>>
>> Tel:(+44)02079588282
>> Skype contact: carla.guerriero
*
* 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/