I made an interventional study, with 2 groups control and intervention group, with measurement at baseline, after 3 months and 6 months, what is the best statistical test to evaluate if the intervention is statistically improving the studied parameter as compared to control?
1. calculate the difference between 3 months (and 6 months) to baseline in each group and then make comparison between these differences (for 3 months and for 6 months) between the 2 groups using independent T test or Man whitney U test
2. Compare results between the groups (at 3 months firstly the at 6 months) using 2 way ANOVA (2*2)
3. Can we get benefit from ANCOVA here? some studies use mixed model (ANOVA and ANCOVA)
So, please can someone explain that for me
Regards
Given what I understand to be your research design, I think that Option 1 is problematic (it doesn't take into account the "starting" / baseline positions of your participants). I confess I'm not sure how Option 2 would operate, but I'm pretty sure that Option 3 (ANCOVA) would be best.
Dear professor Robert, can you explain to me how to do ANCOVA by SPSS (which one is the dependent variable? I measure HbA1c in both groups so what will be the dependent variable?) and what is the aim of using ANCOVA if no statistical significance in baseline level between control and intervention group
Ehab Mudher Mikhael. There is a whole part of me that would like to help you, Ehab, but at the moment I am pretty busy working with two other people on their research, and that is taking up a lot of my time and energy.
Apart from that, ANCOVA is quite a complex statistical procedure, so it's not easy to explain how it works the way it's pretty easy to explain what happens with, and how to do, things such as correlations and t-tests.
In addition to that, a number of statistical assumptions underlie ANCOVA, and some of them are difficult to understand let along write about in a way that is highly likely to be meaningful.
Might I suggest, therefore, that you get some good stats books that explain things well, and also perhaps look at some presentations on YouTube that demonstrate how to do ANCOVA in SPSS. (Do you have access to YouTube?) Please be careful, however, with YouTube because not everything on there contains good advice.
There are also some posts on RG that I noticed when I was answering your question. They might be able to help you.
I confess that I don't remember whether lack of difference at baseline between an intervention or control group removes the need for ANCOVA (it is about 5 years since I used ANCOVA), but something in the back of my mind tells me the need for ANCOVA might still be there.
I'm sorry that I can't be more helpful just now. If you are having trouble identifying what the assumptions are concerning ANCOVA, please feel free to get back to me. Within the next day or so I could probably put something together for you.
1. see : ANCOVA sections in: https://www.amazon.com/Design-Analysis-Experiments-Douglas-Montgomery/dp/1118146921/ref=pd_sbs_14_2/130-9036087-8698950?_encoding=UTF8&pd_rd_i=1118146921&pd_rd_r=caba18d4-d8fb-45e2-9285-3b4be3d5f5fb&pd_rd_w=ElRlT&pd_rd_wg=dLWQJ&pf_rd_p=7cd8f929-4345-4bf2-a554-7d7588b3dd5f&pf_rd_r=KPPMACKE0E51RCW8B33A&psc=1&refRID=KPPMACKE0E51RCW8B33A
and
https://www.bing.com/search?q=repeated+measures+ancova&qs=SC&pq=repeated+measure+anco&sc=8-21&cvid=2A4A1EEF268144A99E0F614E173A4FD3&FORM=QBLH&sp=1
Here the repeated measure is Post-pre test score for each subject
2. The SPSS question is covered here:
https://www.bing.com/search?q=repeated+measures+ancova+in+SPSS&qs=n&form=QBRE&sp=-1&pq=repeated+measures+ancova+in+spss&sc=4-32&sk=&cvid=7F97BB3B6B7249AF85600CF3AC129130
Best wishes, David Booth
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA