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: Mean test in a Likert Scale
From
Maarten Buis <[email protected]>
To
[email protected]
Subject
Re: st: RE: Mean test in a Likert Scale
Date
Thu, 6 Sep 2012 11:11:57 +0200
On Thu, Sep 6, 2012 at 10:55 AM, Yuval Arbel <[email protected]> wrote:
> Maarteen, I didn't quite follow the statement you made regarding free
> lunch and restricted models. I thought the less restrictive the model
> is the more powerful it is. It is well known, for example, that
> estimation of unrestricted models always yield higher R-squares and
> higher log-likelihood - so the LR statistics (or F-Statistics) for
> testing the validity of a restriction imposed on a model have to be
> positive
The term statistical power as I used it refers to the power of a
statistical test, i.e. the probability of rejecting an hypothesis when
you should, not the ability of a model to predict. The former
typically decreases as you relax constraints. The latter increases as
you make your model more flexible, at least within sample (at some
point you start to pick up random stuff specific to the sample, which
means you loose the ability to generalize to the population). Butthat
is another point. The point I made referred to the power of
statistical tests performed within a model, which tends to decrease as
you make the model more flexible.
Hope this clarifies it,
Maarten
---------------------------------
Maarten L. Buis
WZB
Reichpietschufer 50
10785 Berlin
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/