In 2016, there was a national political poll analysis “bubble” that led
almost everyone, including the pundits, to misread the presidential
election. Most polls undermeasured conservatives and Trump voters,
but their numbers were not s far off as were the interpretation of
them, particularly to discern the difference between those polls which
measured all of the voters and the voters state-by-state. Hillary did
win the popular vote as the polls predicted, but a presidential election
is a state-by-state electoral college contest. Thus, huge Clinton
margins in California, New York and Illinois were not anticipated to
be offset by smaller margins in more states (such as “rust belt”/
midwestern Pennsylvania, Ohio, Indiana, Michigan, Wisconsin and
Iowa) --- states that gave Donald Trump his upset win.
I pointed this out in this column in October, 2016, and I note that
Democratic pollster Mark Penn has reiterated it in a recent column
in The Hill. Mr. Penn is a savvy observer, and he notes that pollsters
don’t seem to have learned the lessons of 2016, and might well be
missing how voters really feel abut the upcoming midterms.
I will go further, and state that the pollsters are creating another
polling bubble in 2018 --- this one even more undermeasuring
conservative, Republican and independent voters in many races.
These polls are the primary basis for a number of pundits to
predict an imminent “blue wave” in November. The latest observer
to do this, Larry Sabato, sees the Democratic takeover of the house
(based on polling). Mr. Sabato was convinced in 2016 that Hillary
Clinton would win, and said so as late as election eve.
Just as there were innumerable signs in autumn, 2016 that the polls
were wrong about the outcome, there are many signs that 2018 is
not yet likely to be so “blue.” Yes, the Republican U.S. house majority
is mathematically vulnerable, as incumbent senate Democrats are
also mathematically vulnerable, but the most valuable clues come
from each party’s turnout in the midterm primaries.
The evidence so far is that both Democratic and Republican voters
are highly motivated, mostly around their attitude about President
Trump. Too often, published polls (already hampered statistically
by the difficulty in getting participation from those they are trying
to poll) are not successfully identifying truly “likely” voters. Polls
which do not measure “likely voters” are virtually meaningless,
especially at this date so close to the election.
Remember the notorious national exit polls on election night, 2004?
Across the nation, those exit polls (of those who had already voted)
showed John Kerry winning almost everywhere, usually with
startling margins). It turned out that exit polltakers were primarily
polling liberal voters. George W. Bush won that election.
When we get to election day a little more than three months from
now, a highly volatile electorate could indeed vote in a blue wave, or
a lesser blue surge, The Democrats could even win back the U.S.
house without a full wave. But there could also be a red surge, and
Republicans could add to their now small U.S. senate majority
while keeping house control. The two party bases ARE energized,
and the economy is booming. Trade issues are unsettled, and
foreign policy controversies are unresolved. Immigration is still a
hot issue. The pot is still boiling. Dinner is not yet served.
I’m convinced that current public polling reflects more about what
some pollsters wish for than what the voters will likely actually do
in November. As always, of course, the polling just before the
election will be more useful.
Wait and see. Wait and see.
Copyright (c) 2018 by Barry Casselman. All rights reserved.