I need an answer supported by an article or book... etc
Actually, am doing a research, the sample size is 120 subjects. I already assessed the reliability and validity on 20 subjects.
what I want now is a reference to show that my sample size for the reliability was enough.
Hello Abdullah,
There really is no fixed value, only guidelines. Twenty cases is likely too few to establish a stable estimate of the technical quality of scores. However, you may have legitimate conditions which limit your ability to collect additional data, and this/these may be your rationale for proceeding (as long as the initial, pilot analysis results were promising).
You can accurately answer the question, how much precision would one have in estimating a population correlation, assuming you have a random sample of X cases from that population, by showing either a confidence interval estimate or the estimated standard error for the statistic in question. I suggest that you give this approach a try, as you already have the data in hand to address precision with n = 20 cases.
Good luck with your work.
Dear Abdullah Alselwi ,
As David Morse said above, there is no fixed rule. However, it is a common practice in research to validate the outputs with 15-30 experts. I would start with 15 experts and stop when the saturation level is met (repetitive feedbacks and opinions from experts).
Best luck!
Abdullah Alselwi,
The cases you used (n = 20) indicate, in a rough estimate if considering your sample size (N = 120), a low confidence level (79%) and a big confidence interval (CI = 0.21). In the worst-case scenario, you should have selected 50% (n = 60) of your existing sample for piloting your instrument to achieve a general level of accuracy. Therefore, I don’t think there is a reference that could support your piloted sample size as enough for fulfilling a 95% confidence level. However, as David Morse pointed out, you may capitalize on the conditions of piloting your instrument if the initial results were encouraging.
For future reference, you may use the following online calculator to help you decide on a proper estimate of a sample size.
https://www.abs.gov.au/websitedbs/d3310114.nsf/home/sample+size+calculator
Good luck,
I suggest to use proc power of SAS. It is suggested to decide about mean and standard deviation of the trait. Please see:
https://stats.idre.ucla.edu/sas/seminars/proc-power/
- Badges
- Science topic
- Similar topics
- Books