Section 10 Count data

In this section, we will talk about models that are appropriate for estimating outcomes that are structured as “count data”– non-negative integer data.

The key characteristics of count data are:

  • \(Y\) is not continuous and \(Y_i \geq 0\)
  • Typically the count can be small for many observations (not always)
  • Describes number of occurrences in a specified time period (the observation period) or space (e.g., per state) (e.g., vetoes during a term in office)

The following supplemental resources may be useful:

  • King, Gary. 1998. Unifying political methodology: The likelihood theory of statistical inference. University of Michigan Press. Chapters: 5.5-5.10
  • Gelman and Hill Chapter 6. (on Canvas)
  • Resources for R related to count data
    • R help for negative binomial from UCLA
    • R help for poisson