In this article, the author reports on the state job approval data set. Although the data set includes state-level job approval for governors and U. S. senators, it also contains similar data on the president. These data allow us to test propositions about public opinion toward the president at a level of aggregation lower than the national level but higher than the individual level. More than one thousand state-level presidential job approval readings exist in this data set. Such a large N allows us to test ideas and bypotheses that other limited N designs cannot accomplish. In the first part of this article, the author discusses the properties of these data. In the second part, he uses these data to test two competing theories of the impact of economic perceptions on presidential approval, the retrospection versus the prospection model, which provides one example of how these data can be put to use.