11.5 Truncated Models

Sample selection is determined by values of the \(Y\) variable. Do not observe x or y for truncated observations

  • Example: interested in effect of education on income, but only sample people below a certain income.
  • Example: studies of electoral success of newly formed political parties. Problem: likely only observe new parties whn their chance of success is likely high
  • Example: studies using newspaper reports on social movements to study predictors of violence in these movements. Problem: newspapers select which movements to report on- likely those with chance of violence.

Solution? Requires information about the mechanism that leads to the incomplete (or truncated) data set