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: ordered dependent variable


From   Nick Cox <[email protected]>
To   "[email protected]" <[email protected]>
Subject   Re: st: ordered dependent variable
Date   Thu, 15 Aug 2013 11:19:27 +0100

Sorry, no suggestions from me on how to do that. "conditional mean" of what?

For one thing, I don't recognise that reference.  As explained in the
FAQ, you are asked to provide complete references, not minimal name
and date references.

Nick
[email protected]


On 15 August 2013 10:39, Nicole Feliciani <[email protected]> wrote:

> Thank you very much for your suggestions Nick.
> As you guess my variable is coded from 1 to 6 and I was thinking about
> getting its conditional mean and the use it as the new dependent variable
> with the xtreg command.
> If  I understood rightly this is the tecnique suggested by
> Ferrer-i-Carbonell and van Praag 2004, but I do not know how to implement
> it in Stata.
> Any suggestion?

 Il giorno 15/ago/2013 10:13, "Nick Cox" <[email protected]> ha scritto:

>> I should add that nothing stops you calculating numerical scores or
>> even assigning them arbitrarily and there are methods that will do
>> that for you (e.g. correspondence analysis), but there is no single
>> agreed method to do that, and their applicability is contentious.
>>
>> The literature in your field will be a guide to what is considered as
>> widely acceptable. That doesn't stop you doing something different,
>> but be careful.

On 15 August 2013 08:50, Nick Cox <[email protected]> wrote:

>> > Explaining exactly what your dependent variable is would clarify your
>> > question, but I guess that you have ordered (ordinal, graded) scales,
>> > such as 5-point scales, indicating relative satisfaction. To apply
>> > ordinal probit, you would need to code those numerically, say as 1 to
>> > 5.
>> >
>> > Nothing then stops you using exactly the same values in a standard
>> > regression. No transformation is needed.
>> >
>> > But that's a big jump: you are now treating the values as if they were
>> > interval scale at least. (Being continuous or discrete is not quite
>> > the main issue.)
>> >
>> > So, now expect big arguments from any likely audience, many, perhaps
>> > most, of whom are likely to regard this as invalid or meaningless or
>> > at the very least problematic.
>> >
>> > "in Stata" doesn't seem important here. The issues are statistical and
>> > (social) scientific.

On 15 August 2013 05:58, Nicole Feliciani <[email protected]>

>> >> I am new to Stata and I am estimating the effect of holding a partime
>> >> job on job and life's satisfaction.
>> >> My dependent variable is an ordered one.  I know I can apply an
>> >> ordered probit, but I would like to perform also an OLS regression.
>> >> For this purpose I need to transform my ordered variable in a
>> >> continuos one.  Does anyone know how to do it in 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