How to calculate sample size for unknown population and as well as for known population in epidemiological studies (i.e. descriptive study, cross sectional study etc) ?
You need to read a basic text on sample size calculation. I've attached the RCSI guide to sample size calculation, which is an ongoing project. You need to read the introduction and then to skip to the section relevant to your project.
Here are a few things to consider:
1) What is the disease incidence in the population right now?
2) How fast is it changing?
3) How uniform is the distribution?
4) How expensive is a sample?
5) How time consuming is sample collection and processing?
6) How accurate do the results need to be? Think about sampling and sample size in this case. If you collect four samples, your choice is 0, 25, 50, 75, or 100% infection. Is this close enough? Also note that the 95% confidence interval will be very large and mostly meaningless.
7) How accurate are the test results: false positive, false negative?
8) You could consider pooling samples if the test is accurate, sensitive (detect one infected in a group of non-infected), and the prevalence is low.
9) Are animals (vertebrates) harmed in sample collection?
If you can, play with the bingroup package in R. If you do not pool samples, then pool size = 1. In pooling samples, the game is to use a pool size such that there are at least a few samples that test negative for the target.
https://cran.r-project.org/web/packages/binGroup/
Your final sample size will depend on the answer to the above question, the biological questions you are asking, and the ethical/social aspects of your project. Whether you read a book or consult a statistician, you will need the answers to these questions to make the best decision. I suspect that you will not be able to answer some of these questions.
After taking your best calculated guess, consider that the sample size estimate is a lower bound, and only as accurate as the guesses that you used to make the calculation. Also factor in that there might be missing values that will reduce your planned sample size. Thus, you might want to increase your planned sample size to try and correct for problems that might develop. Barring ethical considerations, it is better to end up with too many samples than it is to end up with too few.
Hi. It is usually done with a formula that considers the size of the population and the knowledge of the variance. But there is a CDC application that helps you. It is called Epi Info, you get it and there are several statistical tests, including the calculation of the sample size. You just have to put the size of the population and give the sample according to the degree of significance you want.
PS: Power and Sample Size Calculation is an useful online tool to estimate sample size. It provides you the option to calculate sample size for: Survival, t-test, regressions, dichotomous outcomes, Mantel-Haenszel, and log. It is free, here is the link:
http://biostat.mc.vanderbilt.edu/wiki/Main/PowerSampleSize
Please, let me know if you need help on how to use it.
Hello ...
This ppt presentation contains equations from chapter 5 (WHO_HRP_EPI_1994_(chp5) Sample Size) ... Also contains equations for increasing or decreasing sample size.
Also I recommended this website for OpenEpi
http://www.openepi.com/Menu/OE_Menu.htm
please check left column >>> sample size ( proportion ) is for cross sectional study that mainly at 95 % become 384 sample size.
As everybody has said, there are different formula to calculate the sample size for different types of studies. I have attached one book very useful for sample size calculation. Thanks and Regards
A big awareness question. I have collected a lot of information on this technical issue.
A big awareness question. I have collected a lot of information on this important topic.
A big awareness question. I have collected a lot of information on this important topic.
Check in the Epi info, it will give you alot of information in addition to formalae assigned to the specific kind of study considering the power you want to achieve
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size