QuBit The Christian Science Monitor Weekly April 23 2018 7 .pdf
Original filename: QuBit The_Christian_Science_Monitor_Weekly_-_April_23_2018_7.pdf
This PDF 1.7 document has been generated by PDF-XChange Editor 7.0.323.1 / PDF-XChange Core API SDK (7.0.323.1), and has been sent on pdf-archive.com on 19/04/2018 at 21:08, from IP address 122.162.x.x.
The current document download page has been viewed 128 times.
File size: 440 KB (1 page).
Privacy: public file
Download original PDF file
Social circle polling: What is it?
A recent paper looks at its role and effectiveness
Almost a year and a half since Donald Trump’s victory, political scientists continue to analyze the 2016 election. Some
researchers are exploring more accurate ways to conduct polls through innovative concepts such as social circle polling.
Q: What is social circle polling?
Social circle polling, in which potential voters are asked how their
friends plan to vote, can be more accurate than asking respondents
about their own intentions, claims a team of researchers in a recent
paper published in the journal Nature Human Behaviour.
Mirta Galesic, lead author of the paper and a researcher of human
social dynamics at the Santa Fe Institute in New Mexico, analyzed polls
from the 2016 presidential election in the United States and concluded
that social circle polling could have predicted Mr. Trump’s lead over
Hillary Clinton as early as August.
A respondent may find it less embarrassing to report a friend’s
support of a controversial candidate. And because relationships can
influence people’s political beliefs, social circle questions can predict
how respondents might change their minds before an election.
“What has not been fully appreciated is that these answers are not
just answers about oneself, but answers about their social circle,” Dr.
Galesic says. “Asking these questions, we learn something about the
general population beyond the respondent.”
Q: Could it have better predicted the
2016 presidential election?
Research has found that people know their
social circles well, and when it comes to judging characteristics of friends – such as potential
voting preferences – they are largely accurate.
Not only do social circle questions improve the
quality of polling, but they also offer a quantitative boost by increasing the study’s sample size.
Andreas Graefe, a forecast researcher at the
Macromedia University of Applied Sciences
in Munich, Germany, says social circle polling, or citizen forecasting, as he calls it, is the
undisputed accuracy winner – but is hardly
used. “People could say, ‘This sounds way too
easy to be true,’ ” he says. “People tend to think
that complex methods are necessary to solve
complex problems. But when we look at the
research, the simple often works.”
that we can say these will allow us to be hyper-accurate. But I do think
it is a very valuable tool, especially for socially desirable behaviors.”
Q: What about the coming US midterm elections?
As Democrats look to the 2018 midterms as a potential redemption
for 2016, with hundreds of state-level seats up for grabs and the possibility to control Congress, proponents say social circle polling could be
National polls were not that far off the mark in 2016: Mrs. Clinton
did, after all, win the popular vote by nearly 2.9 million votes. But
many pollsters, including Galesic, agree that polling was less accurate
at the state level. Trump won Wisconsin, Michigan, and Pennsylvania,
all states projected for Clinton, handing him the Electoral College and
thus the presidency.
“In 2018 it is going to be [all] about state-level predictions, and if
there is any polling that needs to improve in the US, it is state-level
polling,” Galesic says. “These social circle questions can help.”
Snapshots from 2016
Polling about people’s own voting intentions
Traditional polling techniques
saw Clinton ahead for the months
leading up to November.
Polling about the voting intentions of people’s social circles
Social circle polling
predicted a Trump
victory in August.
This polling technique provided a
more accurate picture of how
poll-takers eventually voted in the
Electoral College system.
SOURCE: “Asking about social circles improves election predictions,” Galesic et al.
Q: Has it worked before?
Along with analyzing 2016 US election data, Galesic and her
colleagues studied polls from the 2017 French presidential election.
Featuring a runoff between a liberal centrist (Emmanuel Macron) and
a far-right candidate (Marine Le Pen), that election was similar to the
US one. But polls predicted the French election with stunning accuracy.
Although it wasn’t explicitly social circle polling, mainstream French
polling largely avoided the so-called social desirability bias, or the embarrassment factor that some pollsters say caused a “shy Trump” effect.
Much of the French polling was conducted online, where voters might
feel freer to express their true political beliefs.
Chris Jackson, director of public polling at the global research firm
Ipsos, offers a caveat: “I don’t think these approaches are so proven yet
Q: Are other methods of polling under consideration for
With the use of landline phones decreasing and the potential for
hyper-focused, inexpensive web surveys increasing, political polling is
moving online more and more.
One of the biggest questions for polling’s future, says Mr. Jackson of
Ipsos, is the extent to which social desirability bias still exists. “Our ability to collect information from a whole variety of sources is unparalleled
in history,” he says. “Social polling will be a piece in that puzzle, but it
won’t be the biggest piece or only piece. We are in a world where if you
are not using a lot of different [polling] approaches, you aren’t doing it
– Story Hinckley / Staff writer
THE CHRISTIAN SCIENCE MONITOR WEEKLY
APRIL 23 & 30, 2018