Timeline for How to Estimate Population Variance from Multiple Samples
Current License: CC BY-SA 3.0
5 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 17, 2021 at 13:26 | comment | added | whuber♦ | @Anomaly I wouldn't go quite so far as to say "not valid," but you're certainly right to call into question what such a statistic might mean and to wonder what it could be estimating. Such calculations come into their own in situations where it's possible the subsamples do come from a common population and your objective is to test that hypothesis. My estimator is not the same as the sample variance of the pooled observations, though, as you can see by considering a situation with two subgroups of two observations each and each subgroup has a tiny variance but their means differ greatly. | |
| Mar 17, 2021 at 12:44 | comment | added | The_Anomaly | Of course, if each sample is drawn from a population with a different mean (but same variance), then the sample variance of the pooled observations would not be valid. But if the means are identical, then I think your estimator should be equivalent to the sample variance of the pooled observations? | |
| Mar 17, 2021 at 12:20 | comment | added | The_Anomaly | +1 Very clear explanation. How does this estimate compare to pooling observations from all the samples and using the sample variance? That is, to consider all the observations as one large sample and calculate the sample variance. Is it equivalent? | |
| Nov 22, 2016 at 3:09 | vote | accept | Delyle | ||
| Nov 17, 2016 at 16:15 | history | answered | whuber♦ | CC BY-SA 3.0 |