You might use my -powercal- package, downloadable from SSC, to calculate
power and sample size curves for your parameter of primary interest,
using your small dataset as a pilot study. Newson (2004) explains how
you can do this. The final example (Examople 3) gives a demonstration of
how this can be done even for rank statistics, and it should certainly
be feasible for -xtreg-type models.
I hope this helps.
Best wishes
Roger
References
Newson R. Generalized power calculations for generalized linear models
and more. The Stata Journal 2004; 4(4): 379-401. Also downloadable from
Roger Newson's website at
http://www.kcl-phs.org.uk/rogernewson/
Roger Newson
Lecturer in Medical Statistics
POSTAL ADDRESS:
Respiratory Epidemiology and Public Health Group
National Heart and Lung Institute at Imperial College London
St Mary's Campus
Norfolk Place
London W2 1PG
STREET ADDRESS:
Respiratory Epidemiology and Public Health Group
National Heart and Lung Institute at Imperial College London
47 Praed Street
Paddington
London W1 1NR
TELEPHONE: (+44) 020 7594 0939
FAX: (+44) 020 7594 0942
EMAIL: [email protected]
WEBSITE: http://www.kcl-phs.org.uk/rogernewson/
Opinions expressed are those of the author, not of the institution.
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Theo Serge
Sent: 09 February 2006 14:17
To: [email protected]
Subject: st: minimum sample size for xtreg
Dear Statalist,
I would be grateful if you could inform me about the
following problem:
I am analyzing an experimental panel dataset
comprising the measurements on 15 experimental units
in 7 time points (no missing values) and I am using
xtreg, random effects modeling (the latter being
appropriate according to the Breusch & Pagan Lagrange
multiplier test). My independent variables are two:
time (treated as continuous variable) and the main
factor (dichotomous variable, two subgroups of almost
equal size). Xtreg yields statistically significant
results, which are obvious even at the phase of
descriptive statistics. But the problem is: does my
number of observations (15 individuals * 7
observations per individual =105) yield valid results?
Is it acceptable to run xtreg in such a, comparatively
small, dataset?
Of course, 105 observations would be sufficient for
ordinary linear regression analysis, but given the
very limited number of individuals (15 experimental
units) in my dataset, I do not know whether this
represents a violation of the backgroung assumptions.
Is there a strict, or even empirical, formula giving
the minimum sample size (n units, p time points, so
n*p dataset) for bivariate, trivariate with / without
interaction, and multivariate xtreg analysis?
Please provide me with the appropriate references, if
they exist.
I thank you in advance.
PS. I have browsed the whole Statalist archive, and I
discovered only one relevant question (February 2005,
Mu Xu), but there was no response available online.
T.S.
University of Athens, Greece
*
* 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/
