Bookmark and Share

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]

Re: st: Re: Ordinal independent variables in probit regression


From   Nyasha Tirivayi <[email protected]>
To   [email protected]
Subject   Re: st: Re: Ordinal independent variables in probit regression
Date   Thu, 10 Apr 2014 11:21:37 +0200

Dear Joseph and Richard

Thank you for the suggestions! Very helpful!

Regards

N.Tirivayi
Netherlands

On Thu, Apr 10, 2014 at 6:09 AM, Richard Williams
<[email protected]> wrote:
> You could do something like this:
>
> webuse nhanes2f, clear
> probit diabetes i.health
> est store m1
> probit diabetes health
> est store m2
> lrtest m1 m2
>
> If the contrast is not significant you can treat the variable as continuous.
>
> Incidentally, it doesn't particularly matter that probit is being used; the
> same sorts of approaches could be used for regress.
>
> Joseph, I am not that good with the contrast command, so I wouldn't mind
> seeing examples of how it could be used instead.
>
>
> At 07:15 PM 4/9/2014, Joseph Coveney wrote:
>>
>> Nyasha Tirivayi wrote:
>>
>> How best can I use a likert scale/ordered predictor in a probit
>> regression? The variable has five response categories from Strongly
>> disagree to Agree (neutral is the third response option).
>>
>> Should I include the variable as it is, where one category becomes the
>> reference? Or should I consider the variable to be continuous? Or
>> should I instead use the tab command to create dummies for all five
>> response options, and include the ones I am interested in ( e.g.
>> strongly agree and agree responsea)?
>>
>>
>> --------------------------------------------------------------------------------
>>
>> There are numerous ways to include it as a predictor.  You could use
>> factor
>> variables and then use -contrast- after fitting the model.  You could put
>> the
>> scores in linearly (as a continuous predictor).  But it seems that you've
>> already hit upon the way that makes most sense from the standpoint of how
>> best
>> to address the question of scientific interest:  create three indicator
>> variables, one for strongly agree, one for agree, and one for all of the
>> other
>> scores--something like that below.  (You can accomplish the same thing
>> using a
>> factor variable and then constructing particular contrasts of interest
>> after
>> fitting the model.)
>>
>>     generate byte strongly_agree = likertlike_score == "Strongly Agree"
>>     generate byte agree = likertlike_score == "Agree"
>>     generate byte others = !strongly_agree & !agree
>>    probit response c.(strongly_agree agree others)
>>
>> Joseph Coveney
>>
>> *
>> *   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/
>
>
> -------------------------------------------
> Richard Williams, Notre Dame Dept of Sociology
> OFFICE: (574)631-6668, (574)631-6463
> HOME:   (574)289-5227
> EMAIL:  [email protected]
> WWW:    http://www.nd.edu/~rwilliam
>
>
> *
> *   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/
*
*   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/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index