Averaging P values and beta coefficients + exporting outputs to excel

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

Averaging P values and beta coefficients + exporting outputs to excel

Jose von Roth
Hi Everybody,

I was wondering if somebody can give me some advice. It's about
logistic regressions with a stepwise backward selection (Wald).

I am creating logistic models on 100 data samples (bootstrapped with
replacement), and since I split the data comparing by group (using
the sample variable as differentiation index) I am getting 100
output tables, actually they all come in the same table, and
seperated by sample. So I want to average the coefficients of the
last stepwise selection of all the 100 samples. Does anybody know a
way of how to code this in SPSS Syntax (one of several problems I am
encountering is that the number of steps used to get to the final
model are not always the same, saying that in one sample it may
have taking 6 steps to get to the final model, and in the next
sample it took 10, and so on)?

Same story for the P values, I want to get the average P values  of
the betas of the final 100 models computed kind of automatically.
Any suggestion?



The other option would be to export the whole output into Excel, and
sorting in there in a manual fashion, but this is a pain, and I am
encountering the problem that when I export it and open it in excel
I am just able to see the hearders and a huge blank space where the
results should be of the table I am exporting (might be due the
selection criteria, because when I am not using it, i have no
problem with the exporting). Any feedback about this?


Thanks,

Jose