Menu Adam R Brown

Notes navigation: Browse by titleBrowse by authorSubject index

Fiorina: Retrospective Voting in American Elections

Disclaimer. Don't rely on these old notes in lieu of reading the literature, but they can jog your memory. As a grad student long ago, my peers and I collaborated to write and exchange summaries of political science research. I posted them to a wiki-style website. "Wikisum" is now dead but archived here. I cannot vouch for these notes' accuracy, nor can I say who wrote them.

Fiorina. 1981. Retrospective Voting in American Elections. New Haven: Yale University Press.


In my view, his main point concerns the conceptualization of party ID. The social psychological school saw party ID as a simple, thoughtless attachment and challenged rational choice theorists to explain this attachment. Until Fiorina, they could not. Rational choice scholars could do no more than explain something _similar_ to party ID--for example, a "standing decision" of people to vote for the party that typically serves them best. Fiorina takes the social psychologists head on, and provides a rational choice theory that explains not only why party ID is so stable, but also why it changes. This is an advance from Downs, whose theory implies little or no independent role for partisanship. For Fiorina, Party ID is instrumental, therefore PARTIALLY ENDOGENOUS.


Key said that citizens are primarily concerned with real policy outcomes (purely retrospective). Downs said that citizens use the past only to evalute what a party will do in the future; retrospective voting is merely a cost-cutting variant of prospective voting. Now Fiorina says that retrospective voting is based on expectations about future welfare guided by evaluations of past policy end-states.

The Argument


Retrospective voting (as opposed to prospective voting) is important. It is based in reward-punishment theory.

  1. Reliance on retrospective voting vs prospective voting could lead to differing electoral outcomes
  2. Retrospective voting presumes that citizens are more concerned with policy outcomes than policy instruments
  3. Retrospective voting presumes that public policy formation is not constrained by the voters


If we were dropped into a new democracy with two parties, we might employ "simple issue voting"--that is, we might look at each party's issue platforms. But after some experience having one of the parties in government for four years, we might add to this calculus our estimation of how well the party in power has performed. We might further add an estimate of how well the other party would have performed if it had been in power, weighted according to our uncertainty about this estimate. Then, we can use all this to predict which party is more likely to serve our interests best in the future. Over time, these evaluations are reflected in our party identification, which is a "running tally" of how each party is treating me. Thus, my evaluations of the governing party's recent performance should elicit a change (one way or the other) in my overall evaluation of that party.

The model tested in this book focuses on changes in temporal party identification. He does not directly link party identification with vote choice. The model is based on a running tally of retrospective evaluations of party performances and promises. It looks at an individual's past experiences with political parties plus secondary factors such as parent's affiliation. His model allows party ID to change continuously.

Two Types of Retrospective Voting

  1. SIMPLE RETROSPECTIVE EVALUATIONS (SRE): Based mainly on personal finance, war, civil rights, and similar things. These are things that voters have direct experience with, so their retrospective evaluations are unmediated by the media or anything.
  2. MEDIATED RETROSPECTIVE EVALUATIONS (MRE) are those evaluations that depend on some intermediary; for example, I need the media to provide me with information about the macroeconomy, but I can also use my SREs to make mediated judgments (e.g. I can use my experience with personal finance to make judgments about the national economy). MREs are mediated by a citizen's choice of information sources and opinion leaders.


Uses two sets of panel studies:

  1. SRC (Michigan) panel study: 1956, 1958, 1960: Uses SRE as dummy variables. Finds that retrospective variables related significantly to Party ID.
    • For example, Republican identification increased significantly among those who believed that the domestic situation had improved and that foreign dealings had gone well.
    • Evaluations of financial situation and unemployment had the right sign, but their effects appeared to be smaller and statistically less precise.
    • As expected, the domestic affairs variable has a stronger effect in 1958 while foreign affairs is stronger in 1960 (this coincides with the conventional pictures of the campaigns)
  2. CPS panel study: 1972, 1974, 1976: Uses both SRE and MRE variables.
    • Evaluations of presidential performance and the Nixon pardon (by Ford) significantly contribute to the modification of a former Party ID
    • Crude but incontestable effects of personal economic experiences and party ID in 1976
    • MREs are, in part, a reflection of SREs

Research by the same authors

Research on similar subjects


Fiorina, Morris (author)American PoliticsVotingPartiesDownsian ModelQuasi-Experimental DesignPartisanship

Wikisum home: Index of all summaries by title, by author, or by subject.