|
|
Hi all, I'm in a bit of a dilemma,
I'm explaining variance in English scores, using Hierarchical
multiple regression.
I have three intelligence variables and I am adding Interests variables
to see if interests are able to explain additional variance after
accounting for intelligence.
Artistic interests are able to explain an additional 6.5% of variance .
One of the three intelligence variables (Raven's matrices) does not add
significant variance, nor does it appear to be acting as a suppressor
variable on the other variables in the regression.
If I remove the intelligence variable not contributing to the explained
variance, the sample size changes from N = 98 to N = 109, as not all of
the participants have completed the Raven's matrices.
On the removal of Raven's, Artistic interests explain only 2.3% of
additional variance which is not significant. HOWEVER, If I keep the
sample size constant at N = 98 by adding Raven's in the last step of the
regression , Artistic interests explain an additional 6.7% of
variance .
My dilemma , is that I've only just discovered this difference and have
already completed my analysis (I was editing and re-checking!!) . So do I
re-do my analyses, so that every time Raven's is removed" (and this
is often!) it is "left in" by adding it in the final step
of the analysis.
I think I already know what I will be doing. ....re-running the
analyses!!
I was hoping someone could convince me that I should just leave it as is,
provided I record the different sample sizes.
Karen
PS: Aren't Statistics fun!!
|
Locked
