trend analysis

I have a lot of ordinal independant variables in my dataset. When I
run logistic regression and adjust for other factors some (not a lot)
of the odds
ratios are positive. However, if you look closely there does seem to
be almost a dose dependant relationship for many of them regardless of
significance. The SPSS linear by linear association test is positive
for many of them. How useful is this test for trend and should I
consider it for my final analysis?

.

Quantcast