Using the following three (3) cases as examples, answer the listed questions for each case, and differentiate between different types of missing data.

1. Suppose you conduct a survey of 850 randomly selected citizens and ask them about their policy preferences and political attitudes. Unfortunately, 20 of the surveys are lost.
   1. What kind of missing data problem is this?
   2. What should we do about it?
2. Suppose you are examining the predictors of civil war outbreaks. Each row in your dataset is a country-year. You collect data on variables such as GDP, regime type, state strength, religious heterogeneity, and ethnic heterogeneity. There are some missing data points in the ethnic heterogeneity variable, but you observe that (for the data you have) countries that have a high level of religious heterogeneity also have a high level of ethnic heterogeneity.
   1. What kind of missing data problem is this?
   2. What should we do about it?
3. Suppose you conduct a survey of political attitudes but find that many people don't answer some of the policy preference questions. Further, you are not convinced that the other variables in the survey are good predictors of these policy preference variables.
   1. What kind of missing data problem is this?
   2. What should we do about it?

Types of missing data: structurally missing, missing completely at random (MCAR), missing at random, or nonignorable (also known as missing not at random).