SPSS: Mixed design ANOVA or paired sample t-test for both groups separately

#1
------Dears TalkStats, Your help is highly appreciated! :) -------


I want to test wether the decrease in sleep quality of my sample is different for the two types of conditions. I did a paired sample t test while using split file (so seperately for condition 1 and condition 2) and i did a mixed within between subjects anova where i tested the interaction time(sleep before, sleep after)*condition

I expected to have the same outcome for both test but this is not the case. The interaction in the ANOVA is not signficant, while the ttest showes a significant decrease in 1 condition and not in the other.

How do i interpret this?

Thanks in advance!:tup:
 

Karabiner

TS Contributor
#2
The difference between significant and not significant is itself not significant.
http://www.stat.columbia.edu/~gelman/research/published/signif4.pdf

I'd assume that the second group also showed a decrease, but it just was not
large enough to reach "signficance" in the t-test for that separate group. And
I'd assume that the decrease was somewhat different between groups, but the
difference of decrease just wasn't large enough to be distiguished from chance.

With kind regards

K.
 
#3
Dear Karabiner,

I really appreciate your answer, thank you! Your assumption about the second group was right indeed.
I am now thinking about how to report this in my article. Would you suggest reporting the ANOVA result and not the t-test? (i will also report the mean scores before and after for both groups).

Thanks in advance,

Kind regards,

Evalien
 

Karabiner

TS Contributor
#4
I do not know the context, or study details, or your research questions.
But a mixed ANOVA maybe is the appropriate analysis, providing you
with the answers you seek. In contrast, the difference between the t-test
results seems hardly interpretable.

With kind regards

K.