multiple regression: multiple years and multiple variables
Mterkeurst
I am somewhat unexperienced in spss, so this question may seem simple
to most of you, but I still want to ask it since I could not find the
answer anywhere.
I have a dataset consisting of yearly data on labor productivity
growth for a 40 year period, for 16 countries in total. In the
regression I want to make, this is to be the dependent variable. As
independent variables I have several other economic data, also on all
16 countries and on the entire 40 year period, so my dataset is
complete. My problem is now that it is of course quite easy to do a
regression on each of these 16 countries. However, I want to do one
joint regression, that basically takes these 16 regressions and gives
me one coefficient per independent variable. I have already discovered
that some weighting might be necessary, so if this is needed I would
like to weigh countries by population. Could you please tell me how to
excecute this regression in spss? Thanks in advance!!
Greetings Maarten
Bruce Weaver
I'm no expert on analysis of longitudinal data, but one book I have on
my shelf (details below) suggests that generalized estimating
equations (GEE) and random coefficient models (multilevel models) are
two approaches that could be used. The latter would be done via the
MIXED procedure in SPSS. I don't know if there is any straightforward
way to do the former.
Details on the book I referred to:
Twisk, J.W.R. (2003). Applied longitudinal data analysis for
epidemiology: A practical guide. Cambridge University Press.
HTH.
--
Bruce Weaver
bwe...@lakeheadu.ca
www.angelfire.com/wv/bwhomedir
"When all else fails, RTFM."