Print

Print
Thanks Richard for your input. 

From what little I have read, using the the change score has problems because of low reliability and the regression to the mean. Also now I understand that ANOVA on change scores and repeated measures ANOVA are basically the and gives the same F and p value. 

where are all the statisticians on this board :)


On Sun, May 4, 2014 at 6:40 PM, Richard Saitz <[log in to unmask]> wrote:
Just another thought here....the reason, in my opinion, that there isn't a "right" choice, that there is no consensus, is that the question really isn't one of statistics (at least not entirely--yes I know that adjusting for baseline will minimize variance etc). The answer really depends on what the question is.  For example, if we want to know if one group ends with a better value than the other, that would be a different analysis than if the question is what treatment leads to greater change.  You might imagine a clinical scenario in which getting below a threshold value  (i.e. below a diastolic blood pressure of 90) is more important than how much of a decrease there is, and vice versa.

Best
Rich
-- 
Richard Saitz
Editor Evidence-Based Medicine

BMJ, BMA House, Tavistock Square, London, WC1H 9JR
T: +1 617 414 7744
E: [log in to unmask]
W: ebm.bmj.com
Twitter: @EvidBaseMed_BMJ 
On 1 May 2014 13:29, Steve Simon, P.Mean Consulting <[log in to unmask]> wrote:
On 4/20/2014 5:29 PM, Anoop Balachandran wrote:

> I have two randomized groups (exercise) in a pre-post design.
>
> I wanted to find out what is the best statistical analysis for such a
> design. I just wanted to do a regular repeated measures ANOVa, but
> the more I read, the more I get the feeling that ANCOVA is the best
> option with pre-test as a covariate. My outcome measure at baseline
> is different for each group ( not significant) , so I assume ANCOVA
> is a better option.  I am guessing the reason for ANCOVA is better
> statistical power and for equalizing baseline scores.
>
> Does anyone have any information about it or any good books that
> might help? I tried reading some relevant papers but  just found it
> hard to wrap around the full concept.

Sorry for the late response. There is a third choice: change scores. All of your choices are controversial and there is no consensus in the research community on how to handle this. Using the baseline as a covariate is probably the least controversial approach. But there is an excellent book which outlines all of your choices and their implications.

Pepter L. Bonate (2000). Analysis of Pretest-Posttest Designs. Chapman& Hall/CRC. ISBN: 1-58488-173-9

Steve Simon, [log in to unmask], Standard Disclaimer.
Sign up for the Monthly Mean, the newsletter that
dares to call itself average at www.pmean.com/news





BMJ advances healthcare worldwide by sharing knowledge and expertise to improve experiences, outcomes and value. This email and any attachments are confidential. If you have received this email in error, please delete it and kindly notify us. If the email contains personal views then BMJ accepts no responsibility for these statements. The recipient should check this email and attachments for viruses because the BMJ accepts no liability for any damage caused by viruses. Emails sent or received by BMJ may be monitored for size, traffic, distribution and content. BMJ Publishing Group Limited trading as BMJ. A private limited company, registered in England and Wales under registration number 03102371. Registered office: BMA House, Tavistock Square, London WC1H 9JR, UK.