Statalist The Stata Listserver


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: multilevel or hierarchical regression modeling question


From   "Austin Nichols" <austinnichols@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: multilevel or hierarchical regression modeling question
Date   Tue, 12 Sep 2006 11:56:38 -0400

Hillel Alpert--
Others may disagree, but I don't think your data apply to your
research question. I don't know this literature, but I think: Whether
average gas mileage (which should be measured as energy usage per unit
of output, or gallons per mile) per car type changes over time depends
on the supply and demand for car types in the car market, and (at
least in the US) on "Corporate Average Fuel Economy" standards or
"fleet requirements," c.f.
http://thomas.loc.gov/cgi-bin/cpquery/T?&report=hr475&dbname=109&;

I don't even see number sold in your data; this would be the very
first thing to get, but in any case, you would probably want an
-xtivreg- type framework, which would require policy variables or more
plausible instruments...

If you don't believe my concerns about simultaneous equations for
supply and demand and possibly other stuff, you could try something
like this, but I for one will never believe your results (and
multilevel modelling will not help your research design):

sysuse auto, clear
gen fakemodel=word(make,1)
bys fake (make): gen year=1978+_n
gen gpm=1/mpg
egen type=group(fake)
xtreg gpm year, i(type) fe

The short answer is likely to be: fuel used per mile per pound of car
moved is decreasing (mostly, except in those cars moving from 4-cyl to
V6 to V8 etc.) over time, but pounds per car is increasing, so gallons
used per mile is increasing over time.

On 9/11/06, Hillel Alpert <halpert@hsph.harvard.edu> wrote:
Would anyone be able to advise on the appropriate model and Stata
command syntax for the following analysis?

The dependent variable is gas mileage (mpg), and I wish to determine if
average gas mileage per car type has changed (increased or decreased)
over an eight-year period.

Each observation is a make and model car in a particular year. A given
make and model has from one to eight observations, depending on the
number of years that make and model was produced. Other car features may
be included as potential covariates. The manufacturer couild also be
included in the model.

Perhaps this calls for a multilevel, hierarchical, and/or repeated
measures model. Specific advice regarding how to model the analysis
including command and the specific parameters would be much appreciated.

Variables:

mpg
manufacturer
make
model
year
covariate1
covariate2
covariate3

Many thanks in advance.

Hilllel Alpert
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index