Please let me know that can I select 400 sample size for a prevalence study where population size are more than 7 millions?
My objective is to determine the prevalence of psychiatric morbidity among the asthma patients. Where I will cover 7 divisions in our country and systematic sampling will be applied in divisional medical college and hospitals.
Any formula applied for 400 sample size? There is simple rule that level of homogeneity among the units under study. if this level increases sample size may also be increases. For this there is need to calculate the margin of error among units by a pilot study. Then at certain level of significance it may be calculated.
A much more important factor in the representativeness of the sample is the degree to which your sampling methodology approaches true randomness with replacement.
Hi zannat
The reference population is asthmatic pop. If this pop are 7 millions it seems, your sample size is very low.
You can use a multi stage sampling methods. The first stage you must assign a sample size to each decision proportional to size of its asthmatic patients.
Inthe second phase, you can use a cluster sampling or simple random if you have the list of patients.
A sample size of 400 will give you a confidence interval of ±5% around estimates of prevalence.
Bear in mind that you may be interested in subgroups. If a subgroup of interest has a prevalence of 25% then your sample of 400 will give you 100 people in that subgroup. A sample size of 100 has an associated confidence interval of ±10%
Is it worth taking a larger sample? Probably not. A sample of 1,000 reduces the confidence interval to ±3%, but at the cost of a huge extra effort.
Ahmad is wrong. The sample size does not depend on the size of the population uniless the population is < 5,000, and only depends significantly on the size of the population for populations < 1,000. Alex is right. The important thing about the sample, in your case, is that it is representative.
I believe a major consideration should be what the estimated parameters are going to be used for. As Ronan pointed out, this will determine the precision you need to aim for as well as whether that needs to be at subgroup level (divisions) or just for the whole population. Secondly if comparisons are going to be made between different groups (e.g. male vs. female subjects) then the sample size needs to make allowance for the requisite power to make such an analysis.
Dear friend
Your survey can be described as descriptive cross-sectional study . The size of the sample (n = 400) is predetermined and has not been calculated based on the expected prevalence (P) and of the desired precision (u ): n = 4P(1-P) / u² . This is legal but you must calculate the precision of your survey once the investigation is completed.
I understood that your divisions were all strata. If you know the distribution of your population in terms of the division, you can get the same distribution for your sample with the same sampling fraction. For all practical purposes , the attached part of my book " Statistique descriptive et sondages en sciences médicales " dedicated to the stratified sampling technique to estimate the proportion and precision of the survey. Best, to avoid long calculations , is to use software ( EpiInfo ... )
It might help you
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3775042/?report=printable
program OpenEpi can calculate Sample Size
http://www.openepi.com/SampleSize/SSCohort.htm
IF THE POPULATION WAS 80000 AND PREVALENCE IS 5 % THEN WHAT WILL BE THE SAMPLE SIZE IN CROSS SECTION SURVEY
I have created a simple excel dataset for determining the sample size for cross-sectional surveys. You just have to impute simple values and n (sample size will be as output). https://www.researchgate.net/publication/344363620_Sample_Size_Determination_for_Cross-sectional_Studies
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size