Timeline for Are the differences between sampling clusters and sampling strata, conceptual, methodological, neither or both?
Current License: CC BY-SA 4.0
9 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 16, 2020 at 20:30 | comment | added | Huy Pham | Ok I've given the best answer i can to the question. I leave it to others to discuss the linked paper. Best wishes in your research. | |
| Jul 16, 2020 at 20:11 | comment | added | Alexis | Your recent comment muddles the methodological distinction between clustering and stratification. Hagopian, et al. were pretty explicit about using a two-stage cluster sample, and detailed their whole approach. | |
| Jul 16, 2020 at 20:08 | comment | added | Huy Pham | But if the clusters were different, then I guess that's where stratified sampling comes in. Could it be the case that that paper just used the phrase "cluster sampling" in its own idiosyncratic way? But actually meaning stratified sampling? | |
| Jul 16, 2020 at 20:01 | comment | added | Alexis | Ok. I wonder if the motivation you describe is aiming at within cluster estimates, whereas my suggestion that between cluster variability may be acceptable aims at estimates for a whole population (as in the Hagopian paper case)? | |
| Jul 16, 2020 at 19:47 | comment | added | Huy Pham | I do think so, yes. What I've described in my answer is two stage cluster sampling. In one stage cluster sampling, it's even simpler: you just simple random sample the clusters and then take ALL elements within that/those clusters. This could only work if the difference in means between clusters is small, and the variance within one is large. If there is a wide variation in one cluster(s) and all the other clusters are the same as it, then it could capture the variation within the population. I don't know of many examples of where this would be useful, but it is what it is. | |
| Jul 16, 2020 at 18:48 | comment | added | Alexis | I am not sure I buy the proffessed "differences between the means of the clusters is small, and the variance within a cluster is large" desideratum, having seen cluster sampling used specifically as a method of estimating a heterogeneous population (e.g., the Hagopian, et al. paper on post-second Iraq invasion excess mortality paper). If the differences in means and variances were small, then a single school would be representative of the whole population of schools, right? | |
| Jul 16, 2020 at 18:44 | history | edited | Alexis | CC BY-SA 4.0 |
deleted 2 characters in body
|
| Jul 16, 2020 at 18:38 | comment | added | Huy Pham | I don't think that this answer has any particularly interesting insights for OP's needs such that it is worthy of the bounty. But I put it here just to help others who might wander into the question. | |
| Jul 16, 2020 at 18:35 | history | answered | Huy Pham | CC BY-SA 4.0 |