Suppose I have a set of 5,000 proposals submitted to a competition for funds. After review 10% were Rated Highly, 15% Average, 75% Poor.

I want to take a sample from the group of proposals and test them for plagiarism.

How many should I select from the 5000? How many from each of the 3 groups (high, average, poor)? Couldn't I just do a simple power calculation and draw that many from the pool at random, or select the required sample size such that 10% are rated high, 15% average, and 75% poor? Suppose 95% confidence interval with 5% margin of error.

Does it matter if plagiarism is abundant 25%, or scarce 2%, in my population?


  • $\begingroup$ Unless there's some chance the average and poor proposals would be funded, then why would you want to consider them any further? Regardless, your question is unanswerable because it does not specify any quantifiable objective. $\endgroup$ – whuber Oct 6 '15 at 0:34
  • $\begingroup$ Thanks WHuber. I think I answered my own question. Upon a bit more research I suspect I need a stratafied sampling approach. A power calculation for population 5000 with 95% ci and 5 % marg. Of error is a sample of 357 (resulrs from an online sample calc.) A stratafied approach would be 75% of 357 = 268 from the poor group, 15% of 357 = 54 from average, and 10% of 357 = 36 from the high group. Thats 358 total with the rounding. I suppose one could argue the poor proposals are likely to have more plagiarism but I dont know how to nip that bit of bias out. --Tim $\endgroup$ – TimK Oct 6 '15 at 1:03

Your Answer

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

Browse other questions tagged or ask your own question.