On Tue, 18 Feb 2020 09:53:23 -0800 (PST),
lukas.kineziku@gmail.com
wrote:
Let me get into things briefly. I've got two independent groups
(samples): A) one as experimental and one as B) control group. Each
sample has variables X and Y. If more precisely, X is some
particular posture which dependent variable Y depends on. I am going
to apply only a confouding variable (intervention) to treat X on
EXPERIMENTAL and NO VARIABLE (e.g PLACEBO) on control group and
check if any changes being made upon Y (pain). I am not going to
talk nor about distribution nor about linear regression model so to
check any significant differences available in between independent
groups.
The question is only in the Subject line:
Subject: What SPSS procedure to follow in order to measure each
independent group's dependent variable's mean?
That is modified by "to check any significant differences".
presumably, especially, in Y. We are also told that Y depends
on X. And that you don't want to use a regression model.
Okay. You have requirements that conflict. Make up your
mind.
You can check means and the difference in X with a t-test.
If the groups differ by more than the least amount in X, then
X "confounds" whatever you see for a difference in Y -- A
simple t-test would be a poor test for "significance."
If X and Y are both continuous, the obvious comparison
would be the covariate/regression approach. An ANOVA
with a covariate gives means and adjusted means. But
there is an assumption that the regression lines are parallel.
Look at a scattergram of X and Y with the two groups
marked by different symbols to see what is going on --
whether lines are parallel and which means may differ.
--
Rich Ulrich
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)