4
$\begingroup$

What is an appropriate econometric model to describe count data with an upper bound?

$\endgroup$
  • 6
    $\begingroup$ Depends on why it has an upper bound: is it like a binomial?Describe your outcome! $\endgroup$ – Aniko May 14 '11 at 16:25
  • 2
    $\begingroup$ You can write your Poisson or neg binomial likelihood for the outcomes that are observed with a step of 1, and for the censored outcomes as a sum over the remaining probabilities from the upper bound to infinity. It's probably 15 lines or so in Stata. $\endgroup$ – StasK Sep 19 '12 at 3:07
  • 2
    $\begingroup$ There's not enough information to guess; it might be binomial, or you may need some modification of some other kind of count variate... you'd need tp describe the nature of the variable and how the upper bound operates. $\endgroup$ – Glen_b -Reinstate Monica Jul 6 '15 at 23:56
1
$\begingroup$

You may assume that data above some threshold are censored and not observed, and using the censored count model will be a way to go.

$\endgroup$

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