I have two dependent samples $\mathbf{X}$ and $\mathbf{Y}$ of $s$ and $t$ observations, respectively, that is $$ \mathbf{X} = \{X_1, \dots, X_s\} \;, \\ \mathbf{Y} = \{Y_1, \dots, Y_t\} \;. $$ For example, you could imagine that there is a random variable Z and functions $f, g$ such that $\mathbf{X} = f(Z), \mathbf{Y} = g(Z)$.

My quantities of interest are the sample means $\bar{X}$ and $\bar{Y}$ of $\mathbf{X}$ and $\mathbf{Y}$, respectively.

Is there a standard estimate for the covariance $\mathrm{Cov}[\bar{X}, \bar{Y}]$ of the sample means?

  • $\begingroup$ I am happy for suggestions for a better title and tags! $\endgroup$ Dec 5, 2021 at 19:07
  • $\begingroup$ The answer can be found here $\endgroup$ Dec 5, 2021 at 22:41
  • $\begingroup$ @AlejandroCelis That's just the simple case of paired observations. $\endgroup$ Dec 6, 2021 at 10:34


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.