Hello, I am doing a research where I have to send surveys to customers with obesity of a specific company. Based on national numbers, 14,2 percent of the adult customers have obesity which would be 511 200. From those numbers there are two main subgroup, people with obesity and whom are aware of their condition ( 460 080) and those who have obesity and are not aware ( 51 120). When it comes to the sample size that is needed for the survey, do I calculate the sample size of all people with obesity (511200) or do I calculate the sample size of each subgroup and add them together?
I'm not exactly sure what you're asking, so please forgive me if my answer is off.
Assuming you will run an independent samples t test to compare the group means, power analysis/sample size calculation programs such as G*Power will give you the optimal sample size for each group. G*Power also provides the total sample size (both groups combined). You can even get this computed for unequal group sizes (by choosing the relevant "Allocation ratio" in G*Power).
Given that your two populations vary greatly in size, you will have to take this difference into account when collecting your data so as not to "undersample" the smaller population--unless there is a way to know beforehand who belongs to which group/population.
Carlijn Alles - I took this as meaning you have a question about strata versus subpopulations. The optimal stratification sample size is for an overall mean or total, etc. But you need a larger total sample size if you are reporting for each group/subpopulation, and have to 'calculate' them separately. Also, not being sure what you are doing, I want to warn you about online sample size calculators which often are only for yes/no (proportion) data, simple random sampling without a finite population correction factor, and for the worst case scenario regarding standard deviation, which is meaningless for continuous data. There are many considerations regarding methodology, including model-based approaches. Also, if you have qualitative data, there may be other considerations of which I'm unaware, but I think in all cases you need to find what sample size gives you the needed accuracy based on inherent variance and methodology used. - Cheers - Jim
PS to my response above -
Bias has to be under control so that standard error is the only accuracy concern. Then all sample size needs considerations boil down to what sample size is needed, with what inherent variability (standard deviation), under the relevant/used methodology, to obtain the desired accuracy for your application.
In this case I assumed the question was strata for an overall result versus each of more than one subpopulation. But, Carlijn, you can tell us more after you think about it. The above "formula" fits a variety of cases, including prediction (i.e., regression), comparing groups, and finite populations.
I think you have one population, number of obesity which is = 511 200.
determine the sample size of this from the Cohins sample size determination table table or Taro Yamene sample size determination formula.
from the sample sample size gotten, use the proportional allocation formula to allocate to the two groups (those who are aware that they have obesity and those who are not aware)
I agree that estimating the sample sizes for your groups relies on your study purpose and objectives. That being said, you could use optimum allocation of sample sizes to different strata, which can be decided by minimizing the survey cost, keeping the variance of the estimator fixed, or by maximizing the survey efficiency (minimizing variance), keeping the survey cost fixed. For relevant procedures, you could refer to Chapter 7, “Stratified Sampling,” of the book by Arnab (2017), referenced below.
Arnab, R. (2017). Survey sampling theory and applications. Elsevier/AP, Academic Press, an imprint of Elsevier. https://www.sciencedirect.com/book/9780128118481/survey-sampling-theory-and-applications
Good luck,
Hi Dr Carlijn Alles . See the following link: https://www.geopoll.com/blog/sample-size-research/
See the following link: https://www.google.com/amp/s/www.questionpro.com/blog/determining-sample-size/
Also see the following link: https://www.qualtrics.com/experience-management/research/determine-sample-size/
Kindly see also the following link: https://uncw.edu/irp/ie/resources/documents/qualtrics/determining-sample-size-2.pdf
Depends on your aims and objectives. Follow this article, it may help. Thanks
Preprint Food consumption behavior of Pakistani students living in Ch...
Also check https://www.questionpro.com/blog/determining-sample-size/
Check on this paper.
https://www.statisticshowto.com/probability-and-statistics/find-sample-size/
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size