Need Help Finding Whats Stats To Perform

#1
Hi there,
I've been given a data set involving a food-training study.
In week 1, baseline weight and BMI were recorded for each participant.
In week 2, participants underwent four sessions of an online go/no-go response inhibition task. The task involved two treatments: food inhibition and control treatment.
In week 3, they did a final session of the go/no-go response inhibition task and participants were weighed again.

The proportion of errors (GoErrors, NoGoErrors) and reaction times for correct responses (GoRT) on the first and last sessions were recorded.

Weight was also recorded 1 month and 6 months after the study finished.

We've been asked:
Did performance on the go/no-go task improve between the first and last sessions?
and
Did baseline BMI predict the weight change from baseline to 6 months in the food inhibition group?

Could anyone help me figure out the different kinds of tests that can be done for this? I tried an independent sample T-test for the first one and assumed a regression is needed for the second. But I used a t-test for a different question and we've been asked to not repeat the statistical tests we use.

Thank you.
 

Karabiner

TS Contributor
#2
For the first question, you could use a oneway repeated measures analysis of variance, with
5 levels for the within-subjects factor. You can use a "reverse Helmert" contrast within this
procedure. That contrast compares the mean of the last level with the mean of the preceding
levels. Or, if the question has a slightly different meaning, you could use a "simple contrast"
with the last level as the reference category; this will compare each of the first 4 levels with
the last level.

Or, if only a comparison between the very first and the last session is required, simply use a
dependent (not independent) samples t-test.

The second is a repeated-measures analysis of variance (2 levels) with an additional predictor "BMI".
The interaction between BMI and the repeated-measures factor gives the answer. Or you
could do it more simple, and correlate weight difference with baseline BMI.

If all this is appropriate, partly depends on sample size, which preferably should be n > 30.

With kind regards

Karabiner
 
Last edited:
#3
I am new student of stats, I choose stats in my minor subject but in future I want it as my major course.
I want to ask one thing for now that why the answers the question can it cause hairloss? available on the back of book does not match the answer we solve in our note books?
Yes they are very few but why? I am very much tensed about it.
 
Last edited: