Help deciding on a statistical test for my RCT

Ask here about academic and research issues, like designing studies, recruiting participants, choosing statistics, submitting for publication, etc.
Post Reply
MegAtron
Posts: 3
Joined: Sat Sep 24, 2016 12:48 pm

Help deciding on a statistical test for my RCT

Post by MegAtron » Sun Dec 10, 2017 12:27 am

Hello!

I was wondering if someone can help me choose an appropriate test for an RCT. The study is investigating the effects of therapy on cannabis use and PANSS score. PANSS score and cannabis use were measured pre and post therapy.

The design is mixed:
1 IV (Group) is between-subjects with 3 levels (Long therapy/brief therapy/standard care)
1 IV (Time) is within-subjects with 2 levels (Pre and post therapy)
2 DVs - PANSS score and cannabis use measured pre and post therapy

I also have data on how long the participants have been diagnosed with psychosis, how long they've been a cannabis user, dosage of antipsychotic medication they take, their age and gender.
Which statistical test would be most appropriate for this study? My thoughts were either a Mixed-ANOVA or an ANCOVA but I'm unsure.

Thanks! :)

Advertisement
Pearson Clinical Assessment publishes a wide range of assessments to support psychology professionals including the Gold Standard Wechsler range. To view our range please visit: pearsonclinical.co.uk/cpf
User avatar
hs577
Posts: 85
Joined: Mon Apr 12, 2010 11:02 am

Re: Help deciding on a statistical test for my RCT

Post by hs577 » Sun Dec 10, 2017 6:19 pm

I’d go with a MANOVA due to your two dependent variables of if you wish to control concomitant independent variables a MANCOVA. Good luck :)

MegAtron
Posts: 3
Joined: Sat Sep 24, 2016 12:48 pm

Re: Help deciding on a statistical test for my RCT

Post by MegAtron » Mon Dec 11, 2017 11:34 am

hs577 wrote:I’d go with a MANOVA due to your two dependent variables of if you wish to control concomitant independent variables a MANCOVA. Good luck :)
Thank you! I decided to run 2 ANCOVAS (for baseline-followup PANSS score and Cannabis use) as i didn't feel that confident running a MANCOVA :')

I originally ran a Mixed-ANOVA for both but unfortunately the Levene's statistic was significant for the baseline indicating that the variance wasn't homogenous, hence why I wanted to control for this imbalance. The ANCOVA worked great for the PANSS scores but for the cannabis use it's identified that the baseline value has failed the test of normality (one of the 3 shapiro wilk values was significant and looking at histograms i can see theres a large positive skew) would you recommend i transform the data? I really don't want to for the PANSS scores only the cannabis use as that's what's failed the assumption.

Sorry for all the questions, nothings ever simple with stats I find :')

User avatar
hs577
Posts: 85
Joined: Mon Apr 12, 2010 11:02 am

Re: Help deciding on a statistical test for my RCT

Post by hs577 » Tue Dec 12, 2017 9:46 am

I'd check with your tutor now as you say as it may get to the point where you look into non-parametric alternatives. (There aren't any for a MANOVA but there are for ANOVA). Pallant (2010) has a chapter on "choosing the right statistic" that summarises this well from memory.

Post Reply

Who is online

Users browsing this forum: No registered users and 1 guest