topic Re: really low sample size, Can we still do something???? in SAS Procedures

really low sample size, Can we still do something????

Ruhi — Wed, 17 Dec 2014 17:45:07 GMT

Hi all

I have a got a data from a study, in which some measures are computed on children at the beginning of study and later after intervention. But unfortunately many parents didn't wanted to participate after intervention . So we have outcome measure for only 10 subjects. We spent time and money in this project and now we don't want to throw it in trash just because we have observations for only 10 subjects. Can we still do something????

Thanks in advance for your help.

Re: really low sample size, Can we still do something????

PGStats — Wed, 17 Dec 2014 18:09:37 GMT

I wouldn't trust any conclusion from such a study. Not because of the low sample size, that can be dealt with statistics, but because I would expect the choice to continue participation after intervention depended on the outcome and was not random.

I can only hope that some of the wasted money did at least benefit some children.

Re: really low sample size, Can we still do something????

Reeza — Wed, 17 Dec 2014 18:46:32 GMT

Why didn't parents want to participate after intervention?

Your learnings may not be quantifiable.

Re: really low sample size, Can we still do something????

SteveDenham — Wed, 17 Dec 2014 19:15:25 GMT

I would report the median and the range of the differences, and specify that the results are per protocol, rather than intention to treat.

Steve Denham