0

Hi I am really struggling here any help would be muchly appreciated, please use laymen terms where possible.

research question: a comparison of quit rates between 3 smoking cessation treatments. Each participant group will be randomly assigned a treatment. At the end of 12 weeks quit rates (quit vs not quit) will be measured to ascertain if 1 novel treatment is comparable to the other two.

This is for a research protocol essay, I have to show manual workings out of a power calculation to show how i got to my chosen sample size. I have spent hours reading cohen, posts on here etc and I am just getting more and more confused :(

I could also use some help on what analysis post test should be used - the furthest I have got with this is that it should be an ANOVA.

Thankyou in advance.

becky
  • 1
  • 1
  • Welcome to CV! Take a look at the answer to this [question](http://stats.stackexchange.com/questions/63391/free-internet-or-downloadable-resources-for-sample-size-calculations/63392#63392), in particular the very first link and the link to Quick-R. The Quick-R link gives a nicely approachable introduction to power. – Sean Easter Nov 08 '14 at 15:20
  • Thanks SeanEaster im having a quick look at Friendlys: calculation for ANOVA Number of levels (a) of effect 'A'- would be 3? Total number of levels (b) of all other factors crossed with effect 'A'.- would be 1? Error level (alpha) for which you want power or sample size calculated - would be 0.05? – becky Nov 08 '14 at 16:09
  • me again! I am also struggling with the aforementioned links as they do not show manual workings which I need for the assignment. – becky Nov 08 '14 at 16:26
  • Hmm, are there no examples of manual power calculations given in Cohen? To be clear, you're expected to show analytic calculations of sample size given a certain power level. Is that right? – Sean Easter Nov 09 '14 at 15:47
  • I would see what you can make of formulas in the book and compare them to the results of the `pwr.anova.test` function in this [R package](http://cran.r-project.org/web/packages/pwr/pwr.pdf). That would at least give some confidence in your calculations. – Sean Easter Nov 09 '14 at 15:48
  • hi SeanEaster thanks for your comment. I am trying to estimate the sample size needed to get an 80% power - i have to show manual workings for how I got there. The following study protocol is very similar to my own, but I do not understand from their sample size description how they got there http://www.biomedcentral.com/content/pdf/1471-2458-13-210.pdf to put it in context I have never completed a sample size calculation before and have not been instructed how to do so during my course. Cheers – becky Nov 22 '14 at 14:17
  • Ah, that's helpful. The paper mentions, "and the treatment groups will be compared using chi-squared tests with multiple logistic regression analysis adjusting for other variables as appropriate." I don't have a copy of Cohen, but previews say chi-squared tests are covered in chapter 7. Perhaps his illustrative examples might have something resembling the smoking study, which will hopefully guide your calculations. (Wish I could be more help, sorry!) – Sean Easter Nov 22 '14 at 22:45

0 Answers0