When we explained Presidential Approval in Chapter 2, we did so in the way that others have done. That is, we found an election-honeymoon sequence and powerful effects of the performance of the domestic economy. In that analysis we did not predict who would be president or account for the performance of the domestic economy. We just took those as historical facts and exploited them. And the same with Macropartisanship: We saw that it moved in response to incumbent approval and performance and we took the party of the incumbent as historically given. John Kennedy did defeat Richard Nixon in 1960, for example, and so we assigned the White House to the Democrats for 1961 through 1964. We did explain elections (in Chapter 7), taking as given factors such as Mood, Macropartisanship, Economic Performance, and the like. And so on with Policy Preferences, Policy Activity, and, most recently, Policy. Each explanatory focus takes as given something else, which is itself to be explained in other chapters.

One might quickly imagine a hopeless quandary to such explanation, a grand circularity in which everything explains and is explained by everything else. Luckily, there is no such quandary; this dog chases no tails because everything in this explanatory system works on future values of the effects it produces. Thus performance produces future approval, which influences future partisanship, which influences future election outcomes. But, to stay with our example, John Kennedy's Approval rating does not influence the 1960 outcome.