THE BLOG
03/14/2007 02:19 pm ET | Updated May 25, 2011

Primary Polling Primer II: More on Turnout

Continuing with my Presidential Primary Polling Primer, I
know I promised in the first
installment
to talk next about issues of timing, but before moving on, a
few more points about national primary turnout are in order. I want to consider
below both total turnout and the somewhat unique approach of the CBS/New York Times surveys.

My 2004 turnout summary was based on participation in the
Democratic primaries only because George W. Bush was essentially unopposed that
year. According to data collected by Rhodes Cook (shared via email), Bush had
the primary ballot to himself in eleven states and faced token opposition in
sixteen more. Because there was no contest, the number of states holding Republican
presidential primaries fell from 43 to 27.

So a better way to consider the turnout picture is too take
the longer view. Cook's text, The
Presidential Nominating Process
,
includes a table (p. 49) that provides
total national turnout in primary elections going back to 1912. The table below
shows turnout in the "primary era" (since 1972), and includes numbers for 2004
that he kindly provided via email.**

03-14%20primary%20turnout.png

The total Turnout in 2004 as a percentage of the general election
vote for president declined from 29.6% to 19.7% for two reasons. First, again, Republicans
lacked a contest (note the similar pattern in 1984). Second, the general
election turnout increased. According to Michael McDonald's data archive, presidential
turnout as a percentage of eligible voters increased from 54.2% in 2000 to 60.3% in 2004.

But taking the longer view, it is clear that the total
primary state turnout is likely to be, at best, 35% to 40% of those who will
ultimately cast ballots in the November 2008, or at best, roughly 20% to 25% of
those who will be eligible to vote in 2008.

Now consider again the way national media pollsters
typically report presidential primary vote preference. Many are simply ask the
primary vote questions of those who identify with or lean to the Democratic or
Republican parties, a "screen" that captures roughly 90% of all adults. Some also
screen out the roughly 20% who will tell pollsters they are not registered to vote. So the results
for Democratic and Republican primary preference are typically based on roughly
70% to 90% of all adults, when only about 20% to 25% of adults are likely to participate
in presidential primaries or caucuses.***

The CBS/New York Times
poll does it a bit differently, and their approach is worth focusing on. Unlike
the other national media pollsters, they first ask a question about past
primary voting:

Next
year, are you more likely to vote in a Democratic presidential primary or
caucus, or a Republican primary or caucus or aren't you likely to vote in a
primary or caucus at all?

They then ask the primary vote preference questions among
registered voters who say they are likely to vote in the appropriate party primary
or caucus. The weighted interview counts provided in the latest CBS release
indicate that 63% of their adult sample qualified as registered voters likely
to participate in either the Democratic (38%) or Republican (26%) primaries. That
still amounts a significant over-reporting of likely turnout (a common
challenge
that pollsters face), but it gets us a lot closer to the
population of interest: the voters who will actually chose the delegates that
select each party's nominees. The tabulation of the voting question by party in
the most recent survey (see q46, p. 20)
shows that it excludes roughly half of independents and 20% to 25% of
partisans. That's a start, at least.

So one bit of intrigue for poll consumers will be to watch
whether the CBS/NYT results start to diverge from other national polls. If so,
I will put more faith in the CBS/NYT results than those who simply report on
the 90% of adults who lean to one party or another.

Ok, so next, on to primary timing and how that effects the
polling puzzle...

**The primary vote totals do not include caucus state
participation nor the votes cast in two party-run primaries in Michigan and New
Mexico.

***For the Democrats, the total participation in the 14
caucus states in 2004 amounted to just 4% of the total primary and caucus turnout.
Including the caucus participants would boost turnout as a percentage point of
the general election vote by less than a single percentage point.

Series continues here