topic Sample size calculation in New SAS User

Sample size calculation

emma19901 — Sun, 30 Dec 2018 03:47:42 GMT

Hi, I have a dataset I am working with., I have one variable of interest were the subjects answer, Yes or no " in this case the subject could only answer exposure to radon' based on their perception only once. I have the proportions of yes and the proportion of no's. I want to know how many no's can I include for further analysis. Should I use Proc Power one sample means or two sample means?

Thanks for your help in advance

Re: Sample size calculation

Reeza — Sun, 30 Dec 2018 05:32:48 GMT

@emma19901 wrote:

Hi, I have a dataset I am working with., I have one variable of interest were the subjects answer, Yes or no " in this case the subject could only answer exposure to radon' based on their perception only once. I have the proportions of yes and the proportion of no's. I want to know how many no's can I include for further analysis. Should I use Proc Power one sample means or two sample means?

Thanks for your help in advance

.That’s not how power analysis works. You do not calcualte power after an experiment, you use the data you have.

http://daniellakens.blogspot.com/2014/12/observed-power-and-what-to-do-if-your.html?m=1

If you’re doing this before your experiment, I would use ONESAMPLEFREQ.

Re: Sample size calculation

emma19901 — Sun, 30 Dec 2018 05:45:02 GMT

With their answers, I will create a new variable called the likelihood of true exposure using a Likert scale. I need a sample size for two major reasons a) I can not use only all the yes responses (I need to include also no) not doing so will be biased, B) is a large dataset and I want to compare later probably by using a Wilcoxon my interpretation of the likelihood with 3 more people that will do the same process. For time reasons they can only analyze a fraction of the dataset. 97/1,500 answered yes 1403/1500 said no

Re: Sample size calculation

VDD — Sun, 30 Dec 2018 13:50:56 GMT

the chance that anyone of the 3 being added to your sample with a reply of yes is 0.

1/15 would possibly answer yes.