In 1990, Budge and Hofferbert (B&H) claimed that they had found solid evidence
that party platforms cause U.S. budgetary priorities, and thus concluded that mandate
theory applies in the United States as strongly as it does elsewhere. The implications
of this stunning conclusion would mean that virtually every observer of the American
party system in this century has been wrong.
King and Laver (1993) reanalyzed B&H’s data and demonstrated in two ways
that there exists no evidence for a causal relationship. First, accepting their entire
statistical model, and correcting only an algebraic error (a mistake in how they computed
their standard errors), we showed that their hypothesized relationship holds
up in fewer than half the tests they reported. Second, we showed that their statistical
model includes a slightly hidden but politically implausible assumption that a new
party achieves every budgetary desire immediately upon taking office. We then specified a model without this unrealistic assumption and we found that the assumption
was not supported, and that all evidence in the data for platforms causing government
budgets evaporated. In their published response to our article, B&H withdrew their
key claim and said they were now (in 1993) merely interested in an association and
not causation. That is how it was left in 1993—a perfectly amicable resolution as
far as we were concerned—since we have no objection to the claim that there is a
non-causal or chance association between any two variables. Of course, we see little
reason to be interested in non-causal associations in this area any more than in the
chance correlation that exists between the winner of the baseball World Series and
the party winning the U.S. presidency. Since party mandate theory only makes sense
as a causal theory, the conventional wisdom about America’s porous, non-mandate
party system stands.
Also see related research.