I have attempted to use on line tools to work out a population sample size for an experiment I will be running. However I do not really understand the "Percentage" part

I have attempted to use this www.calculator.net Sample Size Calculator but I don't feel happy not understanding it even if most of the information I have read has said to leave it at 50 (the worst case scenario?). The description on the website does not seem to help at all

"Percentage: The percentage of a particular answer was chosen."

and some other resources seem only slightly better.

"Your accuracy also depends on the percentage of your sample that picks a particular answer..."

My issue is, what answer, and what question?, what if all my survey questions are Likert types scales and the user has to answer them? Everyone who undertakes this survey has to complete it or their answers are not recorded, so is my percentage 100%?

Any help is much appreciated

  • $\begingroup$ I don't really understand what question you are asking. Your original problem is to find out the required sample size needed to answer what question? And as a word of advice, don't use calculator sites that don't disclose the algorithm they are using. As a general rule, stay away from "calculator" sites altogether. If you can do basic algebra, you can figure out your sample size. $\endgroup$ – rocinante Mar 27 '14 at 0:38
  • $\begingroup$ @Glen_b Thanks I edited the question and added the direct link. $\endgroup$ – Deepend Mar 27 '14 at 0:47
  • $\begingroup$ @rocinante I am undertaking an experiment which requires users to rate various designs using metrics such as perceived trust, fairness etc. Population - Unlimited (The internet) Confidence Level 95% Confidence Interval 5%. Sorry for any confusion $\endgroup$ – Deepend Mar 27 '14 at 0:54
  • $\begingroup$ That still doesn't make sense. You want the sample size n at the 95% confidence interval for what? What is the question you are trying to answer? $\endgroup$ – rocinante Mar 27 '14 at 1:07
  • $\begingroup$ I am not sure you understand what it is that you are looking for here. So I think you need to understand what a confidence interval is. The Wikipedia article or the Khan Academy tutorial on the subject before going any further khanacademy.org/math/probability/statistics-inferential/… $\endgroup$ – rocinante Mar 27 '14 at 1:12

Different sample size calculators are used for different hypothesis tests.

It seems like you're using a calculator specifically designed to compute sample size for tests of/CIs for proportions (it's difficult to tell exactly what calculations are being done because they give absolutely no details, but one can surmise from the request for the proportion)

Yet it sounds like you're not doing a proportions test. So it's little surprise when it asks for information that makes no sense. It's like going to a general store to buy some nuts for dinner and instead looking in a section for hardware, and wondering why the assistant keeps asking you apparently pointless questions unrelated to food in an effort to give you the right kind of nut.

[Except in this case it's worse, because the site doesn't even tell you what situations it's giving the sample sizes for... so it's more like a store with no signs and nothing on display. No wonder there's room confusion; you have no easy way to tell you're not in the right section for food. Attempting to comprehend apparently randomly selected internet sites that don't clearly tell you what they're doing may not be a good way to go about your problem]

There are a number of questions here relating to sample size determination; perhaps start there, and then read more than one document that relates to finding sample sizes.

If you want clear advice here about what to do, ask a specific question that makes it clear what you want to do.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.