Can We Still Trust Polls?


– Can we still trust polls? After the 2016 U.S. presidential election, a common reaction from
across the political spectrum was that polling must be broken. Now that researchers
have had time to dig into what happened in 2016 and
in polling more generally, we can bring a lot
of data to bear on this. Here are the basics, Some people who were surprised by the outcome of the 2016 election still can’t get past the feeling that they were led astray by the polls. But the fact of the matter is that national polls in
2016 were very accurate. On average the national polls
indicated that Hillary Clinton would win the popular vote by about three percentage points and she won by two points. – When a commission got down to the data and looked at what happened in the 2016 election, they found that most of the problems were in the state-level,
where there are folks who are trying to do this kind of polling with not the resources you might
have at the national level. What happened with some
of those polls was that they were out of the field
and missed a late swing among those undecided
voters toward Donald Trump. Another issue they
had was under-representing people who didn’t have college educations. And all those let to big surprises in the Midwest on election night. – Another common criticism
is that pollsters favor a particular political party. But a look at the direction
of polling misses in the U.S. over time shows that whether
the polls tend to miss in the Republican direction
or the Democratic direction is pretty much just a coin flip. Some of the best evidence
about the performance of election polling comes
from rigorous analyses of the accuracy of polls over time. If anything, national polling
in presidential elections is more accurate now than
it was a few decades ago. Since 2000 the average absolute error in national presidential polls has been around two or
three percentage points, which historically-speaking,
is really quite good. – And an even broader look
at polling in 45 countries around the world over the past 75 years, done by Will Jennings
and Christopher Wlezien, found a similar result. Election polls worldwide have
a good record of accuracy, and despite some forecasting
misses of the past few years, the average level of
accuracy has not declined. – While those overall
trends are reassuring, unexpected election outcomes still rattled the public’s confidence in polls. A good example of this is Brexit. Believe it or not, Brexit was
not a massive polling failure. While betting markets in
Britain did strongly suggest that Britons would vote to
remain in the European Union, the polls were actually
showing a close race, and for most of the final month, the polling showed “leave” ahead. So while Brexit was certainly not a shining moment in
the history of polling, it also wasn’t an instance
where the polls were wildly off. – The fixation on election
polls is totally understandable and probably unavoidable, but they’re just not the best barometer for how good a job polling is doing. And that’s because election
polls have to do something that other polls measuring policies or attitudes don’t have to do. They have to figure out, of the people they’re talking to on the phone, who is actually going to go out and vote in some day in the future and
then how are they going to vote? For those of you who have
studied human behavior, even if it’s just in
your friends and family, You know that those kind of
predictions are really hard. – While polls may not always
be able to call the winner in very close elections, they
remain very well equipped for reliably tracking the
contours of social attitudes, measuring peoples values,
their preferences, priorities and concerns. Take the trend over time in attitudes toward same-sex marriage. This is an example of
polling, not from an election, where real world events arguably validated the ability of polls to track real movement in public opinion. – You had different polling organizations using different questions
to measure support for a legal right for gays
and lesbians to marry, but all of them told the
same story over time: majority opposition in
the 1990s and early 2000s, followed by gradual gains in support through the last decade and
the emergence of majority support in recent years. And these trends were
corroborated to some extent by the fact that supporters
of same-sex marriage began winning statewide
referenda on the issue about the time majority
support emerged in the polling. – Another way to evaluate
the performance of polls is called benchmarking. We take some questions asked
in large federally-funded studies with high response rates and ask those same questions
in our regular polls. The difference between our
poll numbers and the numbers from the benchmark surveys, gives us some sense as to our accuracy. We recently benchmarked
our telephone surveys against some high profile federal surveys and found our polls get very
similar numbers on issues like “What share of Americans identify as Republicans versus Democrats?” – And the same goes for “What
share of Americans identify as Protestant, Catholic, another religion, or no religious affiliation?” In these analyses, the fact
that the trends from the polls line up very closely with the trends
from the benchmark surveys is evidence that the polls
remain quite accurate. During the timeline shown, the
response rate was declining, but despite what some people claim, the accuracy of these polls
has basically held steady. – Pew Research Center takes
the accuracy of polling very seriously because we that believe polls have a very important role
to play in a democracy. They’re designed to give every citizen, whether rich or poor, well-educated or less educated, an equal voice. They also serve as a check on politicians on both sides of the aisle
who could otherwise just claim to know what the public thinks or why they voted the way they did. – Now, that doesn’t mean
that people in government should be slaves to poll numbers or that all polls are trustworthy. But polls do perform a valuable
function as a reality check on how the public is
thinking about major issues and what’s going well or going
poorly in people’s lives. – And even though some recent
elections have put a spotlight on the limits to just how
accurate polls can be, there is a lot of research
showing that well-designed polls still work and still
produce useful information. So, bottom line, can polls be trusted? As long as we’re realistic
about the precision that polls can actually provide, yes. (light soft sounds)

19 Comments

  • Polling must be imperfect

  • We never did.

  • Fucking idiots.

    "Late swing" my arse.

    Trump won hands down ALL THE WAY as you were predicting his demise.

    He defied your lying shit and this whiney excuse ridden effort is just a continuation of your lies.

    Fuck off you dumb liars. Nobody believes you.

  • β€œLate swing”. Yeah, I call Comey sabotaging HRC a lot of things too.

  • Excuse me.

    Brexit was a colossal polling FAILURE.

    "Believe it or not"…… Lol

    That's what you should add as a rider to you fucking polls, you stupid lying cunts.

    Guess what? We don't fucking believe you.

    Even the fucking WEATHER forecast here in the UK was more reliable.

    πŸ˜‚πŸ˜‚πŸ˜‚πŸ˜‚πŸ˜‚πŸ˜‚πŸ˜‚

    You're a fucking JOKE

    πŸ˜‚πŸ˜‚πŸ˜‚πŸ˜ƒ

  • How about loaded multiple choice terms?

  • BREAKING NEWS! COMPANY WHOSE BUSINESS MODEL RELIES ON PEOPLE BELIEVING POLLS SAY POLLS ARE STILL ACCURATE!

  • PEW is a propoganda machine, always was, always will be, especially when it comes to the constitution. They hate it, always did. "Trust the polls", even though they were 100% wrong just a few years ago . And who was the big poll taker that the msm used? Pew maybe? Pew stinks to high heaven, always did!

  • a research company posting a video championing the efficacy of polls. LOL

  • Polling isn't broken it's rigged by the remocrat conspiracy's psychic terrorism, the 'suck', to determine dempublican apathy, complacency and overconfidence. All republicans are told to 'just not bother to answer polling surveys, etc.', for three months before the election, skewing the polls extremely in dems favor. Then the rems are told to answer polls 3 weeks before the election and on, which realizes a probable lessening in dems GOTV advocacy, because of the multi-media conspiracy projections 'dems are shoving rems face in dems extreme victories to come, running up the score, etc.', which aren't true; then the fake rem building moment in the last 3 weeks seems unstoppable and the swing vote, some of whom just vote for who they think will win, swings to the rems, apathy, complacency and overconfidence, as well as cynicism take hold of some dems and they give up or do less, etc.. Of course this is multiplied by the rems conspiracy having fake person online 'bots make believe they're hundreds of people posting rems supposed lies about dems, and positve stories about rems; combined with global hackers doing the same and worse equals invisible coups now visible stealing elections- and they supposedly win when they actually didn't, duh. 35 years agao I said dems had to win by 1 % to win, because of rem voter crime, then 30 years ago 3 %, then 20 years ago 4 %; and we saw in Hillary's CASE SHE WOULD HAVE HAD TO WIN BY 5 % IN ORDER TO WIN- HER WINNING BY 4 % DIDN'T GET IT DONE. When's Mueller going to be done, before or after Trumpler's visible coup steals the midterms? If you didn't vote for Hillary, you voted for Prump/Tutin. "…We(e),…" must protect the vote, vote early, GOTV, and protect the results more than ever, before the country gets used to being drunk on democracy's backslider's wine.

    reality

  • Corrected comment: Polling isn't broken it's rigged by the remocrat conspiracy's psychic terrorism, the 'suck', to determine dempublican apathy, complacency and overconfidence. All republicans are told to 'just not bother to answer polling surveys, etc.', for three months before the election, skewing the polls extremely in dems favor. Then the rems are told to answer polls 3 weeks before the election and on, which realizes a probable lessening in dems GOTV advocacy, because of the multi-media conspiracy projections 'dems are shoving rems face in dems extreme victories to come, running up the score, etc.', which aren't true; then the fake rem building momentum in the last 3 weeks seems unstoppable and the swing vote, some of whom just vote for who they think will win, swings to the rems. Apathy, complacency and overconfidence, as well as cynicism take hold of some dems and they give up or do less, etc.. Of course this is multiplied by the rems conspiracy having fake person online 'bots make believe they're hundreds of people posting rems supposed lies about dems, and positive stories about rems; combined with global hackers doing the same and worse equals invisible coups, now visible, stealing elections- and they supposedly win when they actually didn't, duh. 35 years ago I said dems had to win by 1 % to win, because of rem voter crime, then 30 years ago 3 %, then 20 years ago 4 %; and we saw in Hillary's CASE SHE WOULD HAVE HAD TO WIN BY 5 % IN ORDER TO WIN- HER WINNING BY 4 % DIDN'T GET IT DONE. When's Mueller going to be done, before or after Trumpler's visible coup steals the midterms? If you didn't vote for Hillary, you voted for Prump/Tutin. "…We(e),…" must protect the vote, vote early, GOTV, and protect the results more than ever, before the country gets used to being drunk on democracy's backslider's wine.

    reality

  • Batman

  • your just another branch of the fraud,.N.F.L

  • Wow – when are you people going to face reality. Robert Mercer and Cambridge Analytical ran their algorthyms programs – only in Michigan and Wisconsin – legal – but surreptitiously which threw over the entire table with results known just over three hours before the election ended. The polls – as polls are taken were correct. They just have no way of polling this new technology. The test was Iowa. Mercer gave Cruz $10 million – Cruz was polled in last place – they ran their program and Cruz won. He said to Trump – it works – (Mercer's choice was not Trump – instead it was the front runner he would use his program on.) As Trump emerged Mercer said "your gonna win" so say what you like, insult who you like, name call who ever you want to, you've already won! Polls didn't fail. They simply have not kept up.

  • I'm more concerned about polls other than elections, trying to determine what people are for or agains. Very important to know how the questions were and how they were asked. See my post just now on https://Facebook.com/John.S.Wren

  • Not yours lol

  • πŸ€«πŸ€«πŸ™„πŸ™„

  • A polling company, saying polls are good. What a shocker! Next there will be a tobacco company saying cigarettes are cool.

  • There is no god and there is no jesus and Christianity is harmful and evil. The united states was built on freedom from religion if you have been mentally abused by a church please contact proper authorities that will handle the case and sue them. They are not god or jesus and they can't judge anyone. LGBTQ need to start taking a stand. Sue the churches for mental and physical abuse. Make them accountable. God and Jesus are not real Christianity is harmful and causes pure evil. Stop going to church stop cultural Christianity it's based on slavery and they are trying to enslave everyone. Stand against the evil churches. Hold them accountable contact your proper law enforcement agent who is an atheist and or nonbeliever and contact aclu. Christianity needs to end now. Christians cause the most suffering in the world and war and want to take away medicaid and medicare from people that need. Conservative Republican Fundamentalist are harmful and evil need to be put in their place. USA is not a Christian nation and it's time for us Non Christians to take it back.

Leave a Comment

Your email address will not be published. Required fields are marked *