the other hand, show no effects of mobilization on partisans but that Independents were
somewhat more likely to vote for Bush if contacted by both parties. In general then, these data,
in combination with Table 6 confirm the conventional wisdom that the effects of mobilization
efforts are mostly to bring citizens to the polls; there is little evidence of a consistent impact on
vote choice. On the other hand, neither are effects on vote choice completely absent.
While we see little evidence that mobilization from a single party persuades individuals
not otherwise disposed to vote for a candidate to vote for him Contact from the Republican party
only in the ANES data suggests that Independents and Democrats mobilized by the Republican
party were actually less likely to vote for George W. Bush. We cannot be certain of the causal
arrow, however; perhaps the Republican party focused some of its efforts on especially hard-to-
convince individuals, as opposed to actively turning their contacts off Bush. We see no such
relationship for the Democrats, although there is some hint in the Miami data that non-
Democrats who were contacted by Democrats only were more likely to vote for Kerry.
This begs the question of what predicts whether or not mobilization efforts will persuade
as well as bringing individuals to the polls. In the last of our analyses we examine the effects of
mode of contact. One way of analyzing the relationship between the mode of partisan contact
and vote choice is to replicate Table 8 with the addition of three-way interactions between party
identification, contact, and mode of contact. However, such models are unwieldy, difficult to
interpret, and as Table 7 shows suffer in this case from the additional problem of empty, or
nearly empty, cells (e.g., no Democrats who were contacted only by the Republican party voted
for George W. Bush). We therefore adopt a more straightforward approach in Table 9. We
include the same control variables as in Table 8 but we exclude non-partisan group contact,
which has no discernible impact on vote choice, and rather than examining the interactions
between party identification and contact we look at the interactive relationship between contact
19
Convention