39
war’, so that hard-liners on Vietnam leaned Democratic, soft-liners Republican, while
hard- and soft-liners on national security voted strongly in the opposite direction.
Finally, by 1972, Vietnam had obliterated all else in the foreign-policy domain.
With Vietnam items in the analysis, national security had disappeared, and Vietnam itself
jumped up to rival social welfare as the only statistically significant—and major—
relationships to the vote. This was truly ‘the Vietnam election’. This time, however, the
partisan situation had reversed. Vietnam was now “Nixon’s war”, so that hard-liners
were voting Republican, soft-liners Democratic. In response, the coefficients for national
security and foreign engagement moved in opposite directions as well. Now, with
Vietnam in the analysis, national security fell from a positive to an irrelevant (and
marginally negative) relationship to the vote, while foreign engagement moved from
negative to positive, with isolationists tending Democratic, as they would forever after.
The domain of international relations is also especially good at introducing a
distinctive measurement problem, perhaps the worst single case in our effort to have five
consistent sub-dimensions across fifty years, while demonstrating both the way a
comprehensive model of an issue context can address this problem methodologically and
the robustness of the substantive results derived from the model under altered conditions.
For the year 1980 was the only one (after 1948) when we had no item that could stand in
for one of our five sub-dimensions of policy preference. In that year, we simply could
not generate a measure of foreign engagement. The marker items, isolationism and
foreign aid, were both missing, while the three available items were all proven aspects of
national security.
Convention