Bruce Hansen's book "Econometrics" defines a random sample as follows: "The observations $(y_i,x_i, z_i)$ are a random sample if they are mutually independent and identically distributed $(iid)$ across $i =1, ..., n.$"

I have two questions about this definition: $1)$ Can two observations be iid or does the concept of iid apply only to random variables? In other words, each observation must be associated with a random variable. And the different observations correspond to the result of the same experiment. $2)$ Since each observation is made up of three elements $(y_i, x_i, z_i)$, then should each observation be considered as the realization of a random vector? So the concept of iid refers here to the independence and identical distribution of random vectors and not random variables?

  • $\begingroup$ I would assume that each observation was a triplet, and so the realisation of a random variable (in $\mathbb R^3$ or whatever - if it helps to call this triplet a random vector, then you can do so but it is also a random variable). The sample is $n$ such realisations, mutually i.i.d. $\endgroup$ – Henry Jun 28 at 9:37

Both independence and having a distribution are properties that can be possessed only by random variables. In your example $y_i, x_i,z_i$ is a random vector, where for each $i$ they jointly have identical distribution and are independent of other such triplets. What we mean by such definition is that we treat the observations as realizations of the random variables.

| cite | improve this answer | |
  • $\begingroup$ Does the concept of independence apply between random vectors? $\endgroup$ – Juan Bromas Jun 28 at 18:26
  • 2
    $\begingroup$ @JuanBromas multivariate random variable is a random variable, so it has all the properties random variables have, and same laws of probability apply to it. $\endgroup$ – Tim Jun 28 at 19:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.