0

# Help deciding on number of pairwise comparisons

Thu 20 Jun, 2019 05:13 am
0

I have to complete my dissertation in a couple of days. I am proposing a RCT with two independent variables: treatment( group 1 : avatar cognitive therapy , group 2: CT) and time(pre-test, post-test and follow-up). I also have 2 dependent variables(severity and power beliefs) I have no actual data since this is a proposal, but I need help proposing the statistical analysis. So I will perform two mixed anova(2X3) tests to see the effects of my intervention on each outcome.

How many post hoc pairwise comparisons do I need to do ? my tutor said something about t test between subject at post-test and follow up(so these are 2 tests?) and t tests within subject to see how do the effects evolve in time( so 3 tests? 1 testing pre-post , one post-follow up and one pre-follow up?
If so, how do i calculate sample size a-priori with bonferroni correction? (estimated effect size, based on previous research would be 0.8) Thank you !!!
• Topic Stats
• Top Replies
Type: Question • Score: 0 • Views: 820 • Replies: 1
No top replies

engineer

2
Thu 20 Jun, 2019 06:43 am
@ioanac96,
Do you have any idea of what the data has looked like in the past? Sample size depends on the size of the signal change you want to see and the expected variance in the data. It takes a lot more data to see a tenth of a sigma shift in distribution than one sigma. If you had historical data, you could use that to find out what the expected variance is which would give you part of your answer. The other part would be part of your proposal. "I will analyze for a 10% shift in treatment effectiveness."
0 Replies

### Related Topics

1 in 5 women get raped? - Discussion by maxdancona
How will you die? - Discussion by Robert Gentel
Statistics - probability - Question by jrl06
Sampling formula - Question by Moses mbeya

1. Forums
2. » Help deciding on number of pairwise comparisons