Consider these self-descriptions the symbolic side of ideology. Respon-
dents, that is, in answering these sorts of questions are associating them-
selves with the symbols of whatever it is that “liberal” and “conservative”
mean. There is a problem with this symbolic ideology at the outset; about a
third of those asked the question do not answer. They tell us that they don’t
think in those terms, that they don’t know what the words mean, and they
are right.
15
So we are characterizing only two thirds of the American public,
generally those who are better educated and more involved in politics, when
we use these data.
So what do we learn simply by asking? The picture is quite different
than what we have seen from policy preferences. The most striking differ-
ence is that on average, the answer is opposite. While we have seen that
most Americans prefer a government that does more rather than less—they
are, that is, operationally liberal—asked to describe their political identity,
almost two of three choose the label “conservative” over the label “liberal.”
The joint conclusion, something we have long observed (Free and Cantril
1967), is that Americans on average are symbolically conservative and oper-
ationally liberal. This is a puzzle that needs a solution, a problem that will
gain my focus for the remainder of this paper.
Before observing the trends in symbolic ideology, it is worth noting
how much confusion this puzzle causes our politics. Because both sides
of the puzzle are reliably true, commentators on both side of American
politics can always make their case about the “real” America, even while
disagreeing fiercely with one another. Look at symbolic ideology and it is
true that conservatism dominates liberalism. Look at preferences for what
government does and it is true that preferences most of the time favor more
rather than less. It is tempting in this puzzling situation to resolve it by
asking which of these findings is real and which is illusion. But that won’t
get us far because both sides seem quite real. The contradiction is in the
views of the American people, not just in the analysis.
The basic facts of ideological self-identification are captured in Figure
9, a compilation of 19 different series of queries, 1162 individual readings in
all. The most striking fact of the responses is the dominance of conservative
over liberal identifications. Setting aside the large numbers of people who
15
On occasion prompts are added to force a response even after respondents have said
that they do not think in ideological terms, usually on the order of “Well, if you were
forced to choose one of those two, which one would it be?” What we learn from doing so
is that the responses to the follow-up seem quite meaningless, telling us that respondents
weren’t being reticent about their views, they truly didn’t know what the words meant.
27
Convention