Hi R community,
I submitted this question to cross validate, but didn't get any
answers. Hoping someone on here can give me some clarification.
If I understand the phia documentation correctly, when using
testInteractions for factors, the value it returns is a contrast of
adjusted means, but if the "slope" argument is included, it is then
returning a contrast between two slopes, which I interpret to mean a
parameter coefficient. Is this correct?
So if I run:
testInteractions(model, pairwise="Factor")
and get:
P-value adjustment method: holm
Value Df Chisq Pr(>Chisq)
ConditionA-ConditionB 0.059987 1 1.453 0.2281
The correct way to report this would be: Condition A was not
significantly higher than condition B, m = .06, X^2(1) = 1.45, p .23.
But if I run:
testInteractions(model, slope="Covariate",
pairwise="Factor")
and get:
Adjusted slope for Covariate
Chisq Test:
P-value adjustment method: holm
Value Df Chisq Pr(>Chisq)
ConditionA-ConditionB -0.0094811 1 1.3427 0.2466
Then the correct way to report this would be: The slope for Covariate
was not significantly different for condition A and condition B, b -.009, X^2(1)
= 1.34, p = .25.
Is this correct? The fact that they both are simply labelled "Value"
makes me second guess this interpretation as it seems to me to be
implying that they represent the same type of statistic...
I am using the lmer function in lme4 to fit this model, if that makes
any difference.
Thanks!