Burbach, “Rally ‘Round the Flag, or Run Away!”
Page 39
Table 4 presents results from regressions coded for the goal of each incident, and
success (to control against certain types of missions simply being less likely to succeed
than others. Results are presented for all events and major events only, as significant
results were found in both cases. The results strongly refute Hypothesis 4 and instead
support the “rational public” view that the public differentiates between interests at stake
in different interventions. Uses of force to defend allies or for foreign policy purposes
have a small but positive effect on approval. As expected, humanitarian missions are
much less supported; an approval loss is expected. Interventions advertised as protecting
Americans are distinctly popular, gaining 5% more than the “defense” mission for both
all and major uses of force. For the major events, success is significant and in the
expected direction – success leads to higher support. The estimated decay rate is also
quite a bit slower than found for undifferentiated rallies, stretching to a half-life of four
months. This could indicate that rally effects are longer-lasting and thus of more political
significance than usually thought, once their initial magnitudes are estimated more
accurately.
The difference in response to the different types of interventions is striking. In
the best case, a successful, major use of force for protection of Americans is expected to
produce a 10% gain in approval for a President. An unsuccessful, major humanitarian
incident would cost 17%!
77
That estimate hardly sounds reasonable, and a significant
driver of it is Clinton’s period of anomalously low polls (relative to the state of the
economy and historical precedent) from late 1993 through early 1995, which was also a
period of quite a few humanitarian missions. Several other factors may have been at
77
The only example of a major, unsuccessful humanitarian mission is Somalia, October 1993.
Convention