Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.

# Re: st: MIXLOGIT: marginal effects

 From Arne Risa Hole To statalist@hsphsun2.harvard.edu Subject Re: st: MIXLOGIT: marginal effects Date Mon, 6 Feb 2012 17:25:24 +0000

```Thanks Maarten, I take your point, but as Richard says there is
nothing stopping you from calculating marginal effects at different
values of the explanatory variables (although admittedly it's rarely
done in practice). Also the LPM is fine as an alternative to binary
logit/probit but what about multinomial models?

Arne

On 6 February 2012 16:56, Maarten Buis <maartenlbuis@gmail.com> wrote:
> On Mon, Feb 6, 2012 at 3:03 PM, Arne Risa Hole wrote:
>> I disagree when it comes to marginal effects: I personally find them
>> much easier to interpret than odds-ratios. In the end the choice will
>> depend on your discipline and personal preference.
>
> My point is that it is fine if you prefer to think in terms of
> differences in probabilities, but in that case just go for a linear
> probability model. If you are only going to report marginal effects
> than you will summarize the effect size with one additive coefficient,
> which is just equivalent to a linear effect. By going through the
> "non-linear model-marginal effects" route you are doing indirectly
> what you can do directly with a linear probability model. Direct
> arguments are more clearer than indirect arguments, so they should be
> preferred.
>
> Even if you are uncomfortable with a linear probability model,  the
> "non-linear model-marginal effects" route is still not going to help.
> The non-linear model will circumvent the linearity which is in such
> cases a problem, but than you are undoing the very reason for choosing
> a non-linear model by reporting only marginal effects.
>
> In short, there are very few cases where I can think of a useful
> application of marginal effects: either you should have estimated a
> linear model in the first place rather than post-hoc "fixing" a
> non-linear one or you are undoing the very non-linearity that was the
> reason for estimating the non-linear model in the first place.
>
> Hope this clarifies my point,
> Maarten
>
> --------------------------
> Maarten L. Buis
> Institut fuer Soziologie
> Universitaet Tuebingen
> Wilhelmstrasse 36
> 72074 Tuebingen
> Germany
>
>
> http://www.maartenbuis.nl
> --------------------------
> *
> *   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/
```