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.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: QMLE
Nick Cox <email@example.com>
Re: st: QMLE
Tue, 19 Mar 2013 19:35:52 +0000
Please give full references, not just name (date) references. This
request is often repeated in Statalist.
In principle, you can write your own program to fit such a
distribution by maximum likelihood estimation, or any other method
that takes your fancy.
In practice, you could search for any distribution fitting program
written by Stephen Jenkins and posted on SSC; and then then clone it.
If you've not done any or much Stata programming that would be harder
than if you have.
On Tue, Mar 19, 2013 at 6:56 PM, Lane, Scott D <Scott.D.Lane@uth.tmc.edu> wrote:
> In 2003, Brown & Heathcoat published an QMLE method applicable to "ex-Gaussian" distributional analyses, which produces three parameters (mu, sigma, and tau, representing then mean of Gaussian, the sd of the Gaussian, and the mean and sd of the exponential, respectively).
> In cognitive psychology, particularly in the domains of attention, reaction time, and eye tracking, this model has been very influential in analyzing the full distribution of reaction times, as the Tau parameter is often more informative and malleable to experimental manipulation than mu or sigma. However, the original QMLE coding was written in Fortran, some updates have been homespun in Matlab, and there is a DOS-based .exe file floating around, but most of these are outdated. Rather than recode in Matlab, Python, etc., including convergence and fit statistics, I am hoping this distributional analysis exists in some form (perhaps as an ado) in Stata — perhaps under a different name. We have a large dataset of eye-tracking based reaction times to which we would like to apply this analysis, and we otherwise use Stata for most data analyses.
* For searches and help try: