# POL222H1 Lecture Notes - Lecture 7: Confounding, List Of Zeppelins

## Document Summary

Many important and interesting political science questions cannot be examined through experiments, i. e. we cannot control and randomly assign x. The feasible second best solution is to use the actual observation of x, y and z in the real world. Divide observations according to values of x. Compare the average value of y across observations with different values of x. Relationship between x & y, controlling for z. Divide observations according to values of x and z. Compare the average value of y across different values of x among observations with the same value of z. Variety of observational research: individual survey data, we study individuals randomly sampled from a population. Economic performance election outcomes: hypothesis = real gdp growth governments party vote share, we can also test the same theory using individual survey data (survey of individual voters, feeling thermometers are used to gage ratings.