0

I was wondering how to set the standard deviations for average of an already averaged data! The following example shows 2 countries, each with 3 districts. From each district samples (n) were collected for analysis. The district means are shown in column 3 (counting from left) along with 2 times the standard deviation over square root of sample size (in column 4). If I want to average the district means to get country means, how would I apply the 2 standard deviation rule? Shall I divide 2 standard deviation with the square root of number of districts (n=3) as shown in column 6 or with the sum of total number of samples originally collected (column 7)? The differences in the two approaches is ~6 fold for country A and ~4 fold for country B! I assume dividing by n = 3 is more logical.

Example

ToNoY
  • 293
  • 1
  • 9
  • Derived in answer [here](http://stats.stackexchange.com/a/43183/4485). See also [wikipedia](http://en.wikipedia.org/wiki/Standard_deviation#Sample-based_statistics) – Affine Nov 16 '14 at 18:19
  • Also I want to point out this is applied to the calculation of `stdv`, and that when calculating your 2\**standard error* you should be using `sum(n)`, not `3` – Affine Nov 16 '14 at 18:27

0 Answers0