|
|
At 06:28 AM 8/31/2006, Paul Mcgeoghan wrote:
>I have a customer who is comparing data on 2 different financial
>websites but related to the same company. He is also comparing this
>over 2 years. He is interested in looking at the influence of a number
>of independents such as Company Size etc on a dependent variable which
>is the number of financial items reported on the website.
Let's see if I get this:
You have a number of companies, each appearing on the two Web sites in
each of two years. So you have four values of your dependent variable
for each company.
And you have a set of independent variables for each company. It sounds
like those do not vary from one year to the other for any company. They
can't very well vary from one Web site to the other.
>At the moment, he has 4 seperate regression models for the companies,
>looking at the same independents and dependent (one for each website,
>one for each year).
That is, running the regression model for dependent VS independents
separately for each of the four combinations of year and site.
>Should the customer look at just 1 regression model and if so, what
>type of model should be used as we do not have independent
>observations? Should we use a 2 factor repeated measures model (one
>factor is Time and the other factor being Site) with the independents
>treated as varying covariates?
Nobody else having responded, I'd say you have the classic situation
for exactly that model. There's a lot more to be said, and I'm not the
best one to say it, but I'd definitely proceed that way, and post again
with any questions about how to do it, and what the results mean.
|
Locked
