October Coffee & Conversations: Polling and the New Hampshire Primary

October Coffee & Conversations: Polling and the New Hampshire Primary


Good morning everyone and welcome to our October Coffee & Conversations series presented
jointly by the Carsey School of Public Policy, the New Hampshire Bureau of
Education Training, and the New Hampshire Certified Public Managers Association.
We’re so thrilled Dr. Andy Smith, director of the UNH Survey Center, here.
I’m gonna just give you a brief background about Andy and then we’ll get
this thing started. So Dr. Andrew Smith, as I mentioned he’s
the director at the University of New Hampshire Survey Center. He has more than 20 years experience in academic survey research, currently as the director of
the survey center since 1999 following positions at the University of Wisconsin
Milwaukee and the University of Cincinnati. He received his Ph.D. from the
University of Cincinnati and dissertation is entitled “Survey
Questions and the American Public Opinion.” As director, Dr. Smith is responsible for
overall staff and financial management of the UNH Service Center. Dr. Smith is
also an associate professor of practice and political science in the UNH
Department of Political Science, where his teaching and research is focused on
survey methodology elections and public policy. Dr. Smith has been a member of
the American Association for Public Opinion Research since 1987 and is a
member of its New England chapter and is also a founding member of the
Association of Academic Survey Research Organization. His methodological research
has included work on question wording effects, question and response wording
effects, the issue of non attitudes and surveys, and methods of improving the
accuracy of pre-election surveys. He regularly presents at academic
conferences on public opinion and political science and we are thrilled to
have him here today to present on polling and the New Hampshire primary. Good morning, I see you came out bright and early to hear a nice boring talk about Survey Research method.
So if you remember back to your statistics classes in college, you’re
probably panicking about now. Well I’m not going to get into statistics because
I think most of what we have to deal with in Survey Research isn’t
statistically based it’s what I want to do today is talk a bit about going to
the New Hampshire primary kinda where we are a little bit of history go into
some of the specifics and really look at it with a jaundiced eye, because I think
too much especially this year we’re depending on the media, particularly
too much on polling, at the expense of other aspects of campaigns and I want to
emphasize some of the problems that we have a polling to maybe make you a
little bit more suspicious consumers of polling when when you’re looking at this
stuff. So first off everybody remembers back in 19… in 2016 polls in 2016
were wrong, right lacks of a word but it becomes part of the political narrative. Polling now is so much part of the narrative that there are so many polls
now that it’s difficult for the public to figure out what is a good poll and
what’s not what the bad poll are we like to believe the stuff that shows our guy
winning if it doesn’t show our guy winning, it must have some serious
methodological flaws of those polls that’s the environment we’re in now and like
cartoons, poll 4 in 5 don’t trust Washington, the survey says 9 in 10
don’t trust polls this begin is that now the belief in polling and how polls
relate to things are tied to they got tied to the perceptions of the media in
general and how they cover politics. Polls this is goes
back to the 2016 general election we’re talking about bias you can see this
polls are generally accurate if it’s Clinton and remember Clinton was leading to most
of the polls and the Trump people were all thinking now that pulled obviously
are wrong. This is and nobody thinks that they were biased in favor of Trump, but this
is the perception that’s it’s out there and it does roll into how we perceive
the media. So what I what I do talk about talk about first is just a very short
history of election polling, very short. So, first off they’re old almost as old
as the country. The very first poll occurred with all the strawpolls back
then in 1824. This is the Harrisburg Pennsylvania and it did a poll showing
that Andrew Jackson would win in the areas around Harrisburg Pennsylvania. Why did they have it in their newspaper back in 1824? It was an interest to their readers,
people like to read about this stuff. They wanted to have an idea
so it’s it’s an old thing and during the later part of the 1800s into the 1900s,
newspapers across the country conducted what they called straw polls in the 1896
election there were six different organizations that conducted national
straw polls and by 19 going through the 1920s and into the 1930s there were many
newspapers it didn’t certainly locally within their state or within their city
but nationally as well there was a magazine called a literary digest that
in that period became very well known for its final election poll and the
October issue before the election it was the literary digest that time was
the largest selling magazine in the entire country by far it was bigger than
sage Time Newsweek and US News working playing rolled together and it did this
this poll for years been particularly accurate but has been
and it was the only one that was really doing it at this sort of a scale of
waited for long so in 1936 they came up they did their poll and the literary
digest said that 43% of the public was gonna vote for Franklin Roosevelt 57%
throughout Clandon and the thing about this poll they sent out ballots this was
back before telephone visits they sent out ballots to over 8 million people got
about 2.1 million votes back County did not then what they got so if you were to
conduct calculated margin of sampling error on this which I should take you
can because of the requirements of the model but if you were the sampling error
would be about point zero zero zero one percent so you think this is really
accurate how did they get it so wrong well they got wrong because of their
sampling frame sampling frame is essentially the list where you draw your
potential respondents from they used motor vehicle registration lists and
telephone directors they see through people from those lists and set up
balance to them but what’s going on in 1936 depression
who’s most likely to have a car in or a telephone back that something with money
which at that time in nineteen nineteen thirty-seven once more likely to be a
Republican so they went out and interviewed a bunch of Republicans and a
those Republicans who they were gonna vote for the Republicans said Landon
much well at the same time George Gallup and a couple other researchers were
developing what they called scientific polling and Gallup was a bit of a sale
of an army there’s a lot of a salesman he went to newspapers and said two
things first off that he guaranteed that his polling would be more accurate than
the literary digest holding everybody back to one
the second thing he guaranteed men was that their subscription rates or the
number of copies they sold would go up when they publish the Gallup cases so
what they did with more accurate not that we would see considered
particularly reactor today five percentage points off in his estimate
Roosevelt but certainly more accurate than literary digest man of course
literary digest would bankrupt the next year not necessarily because of this and
wasn’t even present but they went out of business so these are the three guys
that founded what we think of as modern polling this is a art tsubame Crosley
elbow Roper and George Gallup these were the guys that came out of the market
research in psychology but the point that they did the sampling that they
used was what we call cluster sampling meaning that they would say we need a
certain percentage of men a certain percentage of women older people younger
people wealthier people lower-income people in the world they would fill
their quotas and try to make that up a more representative sample but it wasn’t
random based they would basically say I mean this many people that look like
this and they’re interviewers would go out and find them well you come up to
the 1948 election do you need Truman election this was the end of quota same
there was a big investigation afterwards by this time the survey research
business a fairly big business both and this was
a big block on the survey research we had to investigate what happened figure
out what was wrong well a couple things I found out first
was that the polls that were conducted were still done in person they were
largely completed by early October because it took a long time to get a
mail back and get them introduced or the rudimentary computers that the day
analyzed reported and put out so they were they were completed about a month
before the election and now with hindsight and Survey Research we forgot
there’s a good chunk of the electorate about happy electorate there really
isn’t paying much attention to the presidential race above out so conducted
polls close a second thing they did was they realized the quota seemed it wasn’t
good they were using quotas that were based on pre more estrogen
and so forth and they move to the idea of random sampling after this because of
you randomly sample people from country but then you can use the central
limit theorem which allows you to do those sorts of calculations departments
of sampling error and all that sort of stuff that we see today and it is a has
been proven to be much more accurate way of getting the representative sample then in the 1960s and 70s we move from
this fuel expert help today which of the following excuse would you get for not
voting a person polling to telephone Khloe I’m undecided but that doesn’t
mean I’m apathetic running for we moved to this why do you think we move from
in-person polling cheaper order of magnitude cheaper conducted back then to
conduct a national in-person survey would be even fifty to seventy-five
thousand dollars do a telephone poll via the five to seventy five hundred dollars that’s now it’s
and but orders of magnitude difference we’re in the midst of a similar sort of
thing right now we’re going to paradigm shift between other previous ways of
doing Survey Research and how do a more accurately the future and some of the
things that we’ve known to be true in the survey research world like started
with probability thanks Brandon base samples are going away and they’re going
away largely because of the cost so what we’re seeing now is a big increase of
web polls this blue line these are national these are national election
polls but you can see that we see this big spike in online polls we’ve seen a
decline in polls with live interviewers IVR stands for the interactive voice
response those are the Robo but you don’t have an interviewer you just have
a computer you press one for yes two for no orbitals
say who you’re gonna vote for well you’re seeing those things decline
because it’s much much more expensive as an example in
work that we do over the last five years it used to be that we could get about
one completed interview on a typical poll in an hour so the interviewer work
on the telephones for an hour they complete about one interview that’s down
to about twenty three five interviews per hour now these are costs and triples
in the last five years so now to conduct a statewide survey if you are a media
organization where you’re going to go don’t have the money to do the same so
what you do is you go to the less expensive one and what I was doing this
research did this first part title was Gresham’s law the iHub in the era of 538
538 refers to 538 website it does the whole aggregation how many know
aggression there you go the the van money drives
out the good right it comes from the idea of back in the days when you had
coins and that real value metal in it you’ve been king wanted to having a
little bit more money you would issue currency that had less gold in it or
less silver in it the public would figure that out so what they would do is
they would take their ninety five hundred percent gold coins is taken in
their pocket and spend the ones with less gold value in it so it drove the
good stuff out of the park I think that’s what’s going on in the polling
because we’re in this era of 538 which means if you look at that 538 website
they combined and aggregate all sorts of holes there regardless of methodology
and you’ve got cheaper ones they now make the bulk of what we see in these
polling aggregations it’s especially in state levels so in the nineteen or the
2016 election I was looking to the state of North Carolina which was a
battleground state this is one that everybody was really paying attention to
the four polls that made up the polling averages on these websites included two
ivr polls robo calls and two web-based polls one of the IBR poll was conducted
by a market research firm that was affiliated
little compartments there was nothing in there that any pulsed or anybody in
Survey Research wouldn’t consider a gold standard for the survey but that became
the average and that became what fed into those models and said that glint
had an 80% or 90% chance of winning but it doesn’t matter because the press do
with this and they have some reason for this busy table look for the little red
lines this is a certain research that we did going back to 2006 looking at every
statewide poll in battleground states where there was actually a competitive
election looking at how the final polls worked out president governor Senate
races live phone interviewers I’m ER that’s robo calls and then online polls
the red ones are the ones that were most accurate in that particular type of
election in that particular year do you see any pattern not really not really so
you could be running a crummy online poll say hey we’re just as accurate as
you guys are so why not use ups you can do it online poll 2500 5,000 bucks
through telephone poll an IVR to do an IVR
over it’s going to cost you $50,000 to do a
live interview before they’re all about the same in their accuracy problem is
with these online polls they don’t start with random samples there’s no such
thing as a random a way to randomly select email addresses so it’s a
convenience and it’s like bumping into people on the street and now what we’ve
got as an industry I say as an industry meeting both the academic and the Survey
Research but all the government business end of survey research is now we’ve got
the point with the big organizations and decided that probability based samples
it’d be nice if we could have but he’s not a probability sample there now
rationalizing what we should use nonprobability samples the main reason
is it’s cheaper but now we’ve got this rationalization problems process to
explain why we can throw out 50 to 60 years worth
traditional behavior and go back to essentially so it becomes a problem and
the problem is for consumers is who do you trust what I would say is there two
workmen one organization is called American Association of public opinion
research they’ve got a nice section on political polling worth reading to look
at the things that you should look at with who’s paying for it when it was
done however what the sample was constructed
how the data were collected what kind of waiting was done and so forth problem is
a lot of those organizations that are doing polls don’t make that information
readily available the transparency initiative which paid for is went
through it is not like a a seal of approval for the poll but it means it’s
a seal of approval that you’re willing to put out all the methods and how you
conduct them the look for organizations that won’t do the aiport transparency
initiative because at least they’re willing to say this is this is all the
stuff that we’ve done all right so that’s another one histories of polling
why we do election polling the first place well you’re in a campaign you’re
judging the effectiveness of your message resonating with people
we’re working that way second is academic to do this because people pay
attention when they get engaged in elections what are the issues that seem
to be moving people it’s certainly content for TV shows and it’s content
because sales advertise same thing that George Gallup said in 1936 is absolutely
true today when television organization went into the polling business in the
1970s they did it because more people watch the TV shows and do shows when
they had old on them and otherwise and now the day where we’ve got the Internet
everybody’s got a website guess what people click on so this is why so many
poll eivol plus clicks equals money if you’re in an or have the media here
today the pull of the organization’s on your website if you have a poll listed
people click on it and if they click on it you make more money that’s the
revenue stream so that’s the big reason you’re seeing there’s a proliferation of
holes but polls because it doesn’t really matter what poll you put up there
people will click on and the headlines would get rid of because of that are
really kind of scary because I was looking today Elizabeth Warren the
new front brother and I looked at the analysis and this guy was saying well
Elizabeth Elizabeth Warren is at 26.5% and Joe Biden is at 26% time Elizabeth
Warren really hasn’t been in a national poll or a staple with maybe one an
outlier anywhere in the country over the last couple months even though now
Elizabeth Warren isn’t the front right she’s up by two percentage points here
three percentage points there it’s all in the margin is being a sampling error
it’s no ways and it’s noise at the time when nobody’s really paying that much
attention to the election so the reason that that is listed that way though is
people who click on a headline that says there’s a new front brother he’s in the
lead same reason go to the baseball team you
wonder the health of the score is if only people really interested in the
difference between some guy slider and his swerves curveball now let’s get into the process it’s
important to remember about the nomination process kind of where we are
when you think about Polhemus primaries firstly the invisible primary there’s
only four ways of measuring how good a candidate is what we mean by the
invisible primary is everything that happens before somebody caps above
everything that happens before the I’ll call and when you mention this is
endorsements especially national figures or governor’s senators and so forth
money how much money you’re able to raise polling is the third way of doing
this and you can create either a vicious virtuous circle where you’re begging to
you you raise a lot of money so the press will covers that which means your
polling numbers are likely to go up because people read these good stories
about you which means some of the candidates or some of the politicos out
there looking for the bandwagon a hop-on do they say kind of like this time if
you find em they endorse that becomes another story and this certainly this
goes on and on and on you’re seeing that right now Elizabeth Warren is giving the
additional endorsement she’s raised a ton of money over the last quarter and
she’s rising in the polls because she gets cover for the candidates who don’t
get covered don’t make money don’t get endorsements and bingo in kind
of a vicious circle that dairy market breakthrough and I think that’s even
more true today because we have debates that are entry the debates so we have
this invisible primary then we have the early contests and we tend to think here
in New Hampshire the girl independent of everything our voters are smart gauge
voters not really much more than anybody else we have a higher turnout but we
also have a billion dollars in advertising but what happened in Iowa
impacts what happens in New Hampshire there’ll be a lot of candidates that are
relative al at night where they won’t be on the ballot in two p.m. sure what they
want incredible is after New Hampshire them go down even further so the people
in South Carolina Nevada will have even a more limited pool of people and
they’ll know who’s the winner and who the loser at that point so polling is
very difficult route to look at especially national polling because what
it’s going to happen in these states especially states after these for these
days is completely dependent on what happens in this sequential process so
looking at national polls absolutely nonsensical because most of
the people who vote won’t be able to see most of the candidates who are currently
running when they go to the ballot box it is not going to be so and then
typically by market we’re done so primaries are different low turnout
generally low to moderate interest even presidential level and that sequential
process I talked to so it makes polling very very difficult when you look at the
election polls as I mentioned the hardest things that we haven’t
electricals is not asking people who they’re going to vote for it who’s going
to show up on Election Day the hardest thing even here in New Hampshire where
we have high turnout the primary it might be 50 percent this year how do you
figure out when to those did he pretend oh the people are going to show up when
there’s this we call the social desirability bias to say
the election I’m gonna be there to be late changes in preference and
especially in primary polling there’s no real hope that people have I think
people are anchored to the candidates that are out there if you’re an activist
maybe but then regular voters what’s the difference between Elizabeth Warren’s
health care plan or Bernie Sanders is health care plan or Joe Biden’s health
care plan your real walk you might know the general public they don’t care
they’re all pretty much Democrats plans there’s no difference between these
candidates other than some personality issues or some funny thing they hear and
say in a debate or some screw up they make it a debate so there’s not a lot of
anger to any of the vote they move around all the time and then finally is
what do you do those people that are undecided visit any holding from an
issue we’re going to have five to ten percent of the electorate who have not
made up their minds on Election Day in particular that’s true
so as I mentioned it’s no since it’s the national primary electorate so if you
look at these national polls they’re just they’re worthless what they’re
doing is basically measuring name recognition and the answer and I want a
little bit better the polls here are better because they’re actual campaigns
going on but I’d argue even in New Hampshire at this point in time polls
are not predictive of what’s going on at all both candidates are not going to be
on the polls and it’s really important to remember that the psychology of
voters in a primary process they’re interested in electing somebody who can
win in the general election that’s whole point right picking a nominee for your
party how do you know you’ve got a candidate that’s gonna win they way to
the primaries lose so if you win in the early primaries that we see tell me and
Jimmy Carter is a great example wins the Iowa caucuses in 1976 actually
I’m committed one money came in second he got media attention from that he came
to New Hampshire he was able to campaign for both the 6-8 weeks back then he was
able to win the New Hampshire primary he became a winner and all those other guys
that didn’t understand this sequential process of the importance of momentum
there was a little I’ll worry about that Massachusetts they were losers Rudy
Giuliani in 2008 says I don’t have to concentrate on the 30 states I’m going
to focus on Florida by the time that got around in Florida he had lost everything
early States he was a loser in no team so you gotta win winners and losers and
voters make up their mind when they need to at the end of the campaign so other
things they need to pay attention to you’re interested likely voters New
Hampshire we have a real problem trying to figure out who’s a likely voter we
can’t go back with lists and say well these people voted in the last couple people who show up on Election Day
we’re registered what are you doing about that funding this we have higher
turnout in New Hampshire than those places so we can’t just go back and look
at those activist who vote really routinely in a state like Iowa turnouts
10% 15% I think it’d be percent of the maximum they’ve ever had in their caucus
is there we’ve got the most in the country between two thousand four
different people and move into the stage 22 percent of
whom do take that period the other ones weren’t old enough to vote in 2000 but
now they’re old enough to vote in 2008 so well that’s a while now we did it
between 2008 2016 it was 30 percent of the electorate we’re different people
now not all of those are going to vote but if you’re a pollster and you’re
trying to figure out all of these people can vote what do you do you can’t go
back and use your old lists because you’re going to systematically exclude a
large chunk of the electorate makes it very difficult to hold the primary so
they mentioned at state and national polls as a whole can’t tell you’re gonna
win the nominees they’re not going to tell you who in turn everything around
so this is a decision 2008 I like 2008 this repose on averages in New Hampshire
Rudy’s the purple line there you can see by early 2008 he was already second to
third place John McCain dipped down that he came back up
Romney’s leading here and the other guys are down in the noise area big Tom
campaign disappeared but you can see that in New Hampshire its McCain’s
identified as going up pretty early there Giuliani’s an afterthought by
October November of 2008 the leader broody significant lead over everybody
else second places McCain he’s down at 8% Romney’s at 10%
Mike Huckabee but they were about two to three weeks behind anything that we saw
the what was going on here was determining what people saw the natural
guess what happens if you see somebody going up in the polls in New Hampshire
John McCain surging did New Hampshire there’s going to be a national story
about that then people naturally say interesting and so there’s a couple weak flag in the
national polls part of the question is the voters asked
is why you sign now listen this is what I get from journalist a lot Kevin no the
better than visible tight journalists from other parts of the country when I
see these big numbers of people that are undecided why should they it’s it’s
October now the elections four months away why would you make up your mind two
days before the election why should I make up my mind so this is the exit
polls going back to 2000 this is asking people and this is New Hampshire this is
not natural exit well this is New Hampshire primary exit polls when did
you decide we were been a vote and look at the percentage of people who said
they made up their mind on Election Day rated for 15 percent of 27 percent on
the Republican side of 2016 we typically are done interviewing on
Sunday Monday and Tuesday for people to make their life you look at the last
three days anywhere from 25% of the 40% 27% of 2016 said they made up their mind
in the last three days we’re asking people who they’re gonna vote for when
the right answer is not sure it’s very challenging so I just want to use a
couple examples of how this becomes problematic not only for us with a
pollster but really if the narrative that’s deciding what’s going on in the
campaign this is 2008 election these are polls that we did at the
survey center and we’re seeing that Hillary Clinton’s kind of leading
throughout Barack Obama pops up here at the edge and words an undecided the
people who really say I have no idea vote for two days before the election
they have to get it right but they could have been their methods shouldn’t blank
block that they got it right the way you calculate
so it’s not so good yes you’re going back to when you were talking about the
dollars involved yeah a nestling does one course UNH for
the college experience he went down to committee at that time and said we can
boost our visibility by conducting polls they’re close enough to New York City
they build a visor E Committee media people from all various different
television and newspaper medias and they didn’t do work for any one of them but
the university spent million a million half dollars on Poli
it was the university’s advertising budget and since that time it’s called
from a school that nobody could pronounce to a school that now has
Division one sports and it’s now when Fe Act University they that’s the way they
build that University is the advertising they got from there it was a smart
decision of a part of that Dean but universities I’m doing because I teach
courses and have done research on Survey Research holes are the one thing that
you can collection polling is the one time in which you have a real world
event that you could try to predict and try to figure out the people that you
talk to match the people who showed up so it’s totally academic survey
researches of relying on this for years because it’s one thing that you test are
our methods don’t work we go back and look and figure out what’s wrong with
our methods and how we change them but so there’s a lot of reasons to do it but
the visibility is certainly one oh the students yet we continue yeah for us we
we typically average $350,000 that page students every year so it’s a
significant amount of money and students can they can learn just doing the
interviewing many were supervisors we have a lot of our former students go to
the kind of research back end of it and work in there
so yeah it’s a lot of training and they get out of that so it’s very much I would say no Bill Maher founded
northeastern getting more space up in the front a lot of states do that thing
that they’re gonna get more ability to influence who the nominee is the
research shows the more front-loading there is more important I went to the
interview come because that winners win losers lose makes a big difference
because the timeframe for you to run a campaign outside of New Hampshire where
there is structural campaigns that their staff hire and all this or stuff most
campaigns don’t have anybody in Texas or California because what happens here it
can be advertising dollars what happens if you win the New Hampshire primary how
much of an ad buy that would be why hundreds of billions of dollars
worth of advertising that you get from that that’s why you gotta remember it’s
a sequential process and a good number of the candidates that they may like in
Texas or California right now that won’t even be on the ballot that so it’s the
winnowing process that Iowa New Hampshire really served we don’t pick
the winners necessarily but we determine whose can be the front-runner who the
challenge the main challenger is who starts but early voting and I made although
even all of these things are designed to increase turnout they don’t work the
states that have gone to lots of early voting or Oregon where you everybody
votes by mail their turnouts the same or down from when they go game is not a
hard act the bear even in states which have barriers to voting it’s really
pretty simple you might have to wait in line ten minutes longer than you did
otherwise the barrier to voting is here it’s not any any other things it’s in
your head so I don’t think it’s gonna make that much Andy what can you say
about the difference between the political election polls and then other
kinds of public opinion polls and given all of these challenges and obstacles in
the in the political realm would you say that they exist
some of these questions that you asked as a Granite State poll and whatnot
absolutely we said the same barriers the bigger issue that we have with any sort
of public opinion polling is the assumption on the part of the researcher
that the person that we’re talking to knows or cares about the issue that
we’re asking about so I think out of something never back in 2008 we’re going
through the financial crisis and they kept this plan in Washington that we do
rejigger the banks somehow and fund the banks people are asking polls about
whether did they did they approve or disapprove with the financial bailout
plan to banks the real answer how big of an impact is
good news they’re going to improve thanks to the same thing
the real answer that is I have no idea that’s what the true answer for most
people is I would never may have heard of it I have no idea what it’s going to
do and I have no idea or not but they still have you know 60%
favorite 35% of publisher you ask people to question they’ll give you an answer
because what we have is it but I started off saying to the issue we have in
Survey Research is a psychological issue there’s some issues social desirability
you don’t want to look stupid in life so what somebody can ask you a question
because if you think about a survey as a kind of an artificial conversation I ask
a question you’re expecting give an answer and if you don’t give an answer
if you say well I’m not sure I thought about it
I’ll say well do you flee more towards one side or the other so I can force you
to give an answer and reality for most issue based polling is never heard of it
don’t really care what we try to do in all of our questions but polls that
force people to choose especially just just think about if you see a pulling on
any sort of a policy issue ask yourself what do I know about that issue and even
if you’re a policy expert and think about the person who’s cooking
dinner they got two kids running around and you’re asking about some arcane
issue and ask them whether they favor opposes their answer the Truman’s are
coming mmm that’s why presidential election ploy or general election point
is typically much more accurate than primary polling because we provide
information we say are you gonna vote for Donald Trump the Republican or
Hillary Clinton the Democrat and people here
Democrat Republican any vote for that and especially if you get down past
presidential past governor maybe you get down the congressional seats or there’s
elected officials here they need to say it people vote for their state senators
of their state house represent is not because they know you meant the New
Hampshire’s we have 400 state reps so everybody knows our state rep truth is
building those who their state rep as they vote for R or D that’s that I think is why those are for
actors but policy holiness there’s a whole range of issues just get one good
methodological problems we were doing some pulling for the American Cancer
Society this is a long time ago some people what arguments are gone but they
wanted to shift from their traditional role of funding research and education
in cancer – pushing non-smoking campaigns they have done a poll that SF
do you think the American Cancer Society should continue its traditional mission
of education research on cancer or should it get into public policy these
cops to stop smoking well it’s 60% said get into public
policy about stopping smoking I said well that looks like what we call a
recency effect more unlikely people are more likely to choose the last thing
that they heard so we did this in six states where we rotated just the
responses same questions same words we just rotated the response so you got
traditional mission of education first and policy second or policy person
education stay whichever was read second got 60% that the 40 percentage point
shift in whether you support or oppose something it’s an art event the real
answer is people saying American Cancer Society I think I’ve heard of them I
don’t know what they do so it is a real problem
I’m looking at the time here I’d be happy to answer any other questions for
you but I think it supposed to end at 9:30
yeah one more yes sir when I look at your chart about you know you no longer
can pick a random sample and get the people to so you made out trying to find
that but in addition to that because we’re putting out that poll you are
affecting the results by just doing any measuring system in it it seems that
you’re never going to get there because you know the reason you’re doing it is
to put out that information but the more you put out that information the more
you in fact the answer it’s comforting to know that most people don’t bother to
read the news or wife Sydney who’s score not academic but actually a lot of
research has gone into trying to look at the impact of hold on elections and it’s
very minimal there’s one bit looked at this is a German professor
Elizabeth well employment that came up with a theory that she called the spiral
of silence and that she was looking at mainly campaigns in Germany in the 1980s
and 1990s where the the theory basically comes down to is if you think that the
candidate that you mike is less popular you are less likely to wear a button
that’s quite a new foot a bumper sticker on your car
less likely to talk about that candidate or support for that candidate at a
cocktail party and less likely to tell a holter that you’re going to vote for
that candidate but when it comes to election day they still vote
and you can make that argument many people made the argument that’s what
happened in 2016 Donald Trump was certainly seen as the less appropriate
candidate you didn’t see as many Donald Trump stickers and signs you did Clinton
signs but they came out of it anyway so most of the research has shown that
polls have had very little impact on election so what Time Lord did have an
impact that countless in 1980 exit polls were released before the polls were the
poll were closed in California and Jimmy Carter conceded the election before
polls were closed in California many Democratic congressmen said they lost in
California because Democrats said a helmet that we’ve already lost the
presidency I’m not gonna bother to go vote but there’s very little evidence
that it actually

Add a Comment

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