Polling with Craig Burnett – Hofstra Votes

Polling with Craig Burnett – Hofstra Votes


[MUSIC] Hofstra votes is a civic engagement
campaign that was created to educate members of the campus and surrounding
communities about political and policy issues, and to encourage them to register
and vote. Hofstra has a robust tradition of providing academic and experiential
programs to learn about and participate in the democratic process. We would like to
welcome you to our Hofstra votes podcast. – Okay, hi everyone and welcome to the
latest edition of the Hofstra votes podcast. We’re coming to you from the
studios of WREG radio Hofstra University, our award-winning student-run
radio station, which you can find at 88.7 FM. Today we’re going to dig into and try
to demystify political polling and our guide to help us do that today is Dr.
Craig Burnett, an associate professor of political science at Hofstra. Welcome
Craig. – Thank you for having me. – Sure, so let’s start with a poll. Harris
did a survey back in the end of 2018 that found a majority of registered
voters don’t actually believe polls, which is, you know, really meta. So, what does that tell you? – Look, it’s hard. I’m not terribly surprised by that
number, to be honest. I think at least in part mistrusts that you’re
going to find in, say, 2018 or even this year and beyond is going to be
somewhat related to 2016 and that there was pretty good consensus among
polling outlets that said, “Look, it’s looking like it’s going to be Hillary
Clinton,” and obviously that turned out to be incorrect. So, for most people
who follow this somewhat closely or, you know, even those who don’t, the narrative
is changed in polling. But, there’s a couple of caveats there that are worth
at least mentioning. The first is that, you know, the types of polls that people
are saying they mistrust is because they were told that the polls had sort of
said Hillary Clinton was outside of the margin of error and it looked like
she was going to win. Well, those are national polls, right, and that
doesn’t necessarily translate to what happens at the electoral level, which is
a state-by-state column, which are a lot harder to get accurate, predictive
polling on. Outside of maybe a few exceptions, the very large states where
there are enough people that you can get many polls out there to build a
consensus. A lot of states are fairly small, and there’s just not that many people to
talk to, you know. One of the most difficult things you could do in the
world would try to be to run an accurate predictive poll in Wyoming.
– Because it’s just not a large enough sample size to get an
accurate reading? – There’s like 500,000 people there.
You’re going to have to be able to get a broad swath of that population and one
poll would not be sufficient. If you really wanted to have enough information
to say, “Here’s what we think -” – Are you saying that national polls can’t be trusted?
– Well, no, I think actually probably the opposite is true. It’s just about
what the information is telling you. A national poll will tell you, kind of,
where the entire country is at and look at the results, right, they sort of
match, you know. The election was won in the popular vote by a little over 3
million-ish, somewhere around there, and that more or less translated into the
kind of information the polls were telling us. Nationally, the poll ended
up being right. Hillary Clinton was the winner, but that’s not what matters
in a presidential election. You go down to the state level and some states
we kind of knew would be close. Some [states] we weren’t in enough, right, so we weren’t in
states like Michigan enough. We weren’t in states like Wisconsin really at all. Maybe a couple of polls were there, but nobody was sort of, I mean, I think we
were more in Arizona than we were in Wisconsin. Some of those are sort of a
surprise, right? I mean, there was genuinely a surprise on election night
that people just sort of didn’t know exactly what to make of it. The
polling in Ohio wasn’t great, but that’s not an easy thing to do. – So is it
essentially that because the way the presidential elections are decided, not
by the popular vote but by the electoral college, that national polls really can’t
predict a winner? – No, I wouldn’t make a prediction based on
the national poll. If I were trying to draw an electoral map, I would do my best to
get as much information of the state level to then make specific predictions
about what this state may or may not be doing in the election. That’s not to say
that the national polls aren’t going to pick up on national trends. They will, if
people have a preference for one candidate. I think they are generally
going to capture that, but not in terms of predicting, you know, a state-by-state election.
No, I wouldn’t use information from what in many of these polls are, equate to
about five or six hundred people across the country to say what is going to
happen in Florida. That would not be an appropriate tool to use. – So, that sort of comes to my next question, which is can a poll really
tell us who is going to win in 2020 18 months out, a year out, or, you know, a
month out? Is essentially that you shouldn’t necessarily trust national
polls to tell you “Yes, Hillary Clinton is going to win or Donald Trump is going to win,”
whoever, and, you know, that you really ought to look at polls from swing states? – Absolutely. Focusing on the swing states is always going to be the way to try to
get the best guess about what the election is going to do. But, you know, at
this point in the cycle, I think the best you could hope for,
especially given where the contest actually is, is who is going to be the
Democratic nominee? Who has momentum, right, if you’re going to talk about it. Right now as we sit here in August,
right, Elizabeth Warren has been has been gaining in the polls, Biden
has been more or less flat, and so that’s where we’re at, but it’s still really
early to tell. I wouldn’t do any head-to-head matchups. I wouldn’t take
much stock in that. – So all the polls that we’ve seen in the last few months that are saying that, you know, Biden could beat Trump or Warren could beat Trump, those
have very limited value. – They’re fun. I would say take them is
what they are, which is a horse race poll, but this isn’t like a horse race where
you actually can determine the winner. This is just a predictive horse race.
Take it for what it is, that oh, well, so-and-so could do better relative to the
President. Well, as of right now, perhaps, but this is not framed in the way that
the question is actually going to appear to voters come November 2020.
There has been a whole campaign, right. The nominees for the Democratic Party will
emerge and then the question will change. This person versus the President and
what hasn’t transpired over, say, another six months of time or even just, you know,
for most voters the last couple months of the election. That’s unknowable
at this point. – Well, that’s interesting you say the last couple of months of an election, right, because the conventional wisdom is that people are not really
paying attention. You are paying attention, I’m paying attention, reporters are
paying attention, but that your average voter is not paying attention a few
months before. Things have really changed
in the last two years or so, or really more than that since the 2016 campaign. The
news cycle has gotten so fast that it doesn’t seem as if, you know, I
wonder one event or even two events don’t necessarily change people’s point
of view. Also, people seem to be shutting down to some extent in general,
not related to polls, but they’re shutting down because they can’t deal
with the stress of sort of this constant roiling of the news cycle. So, how did
that impact polling if people aren’t actually paying attention, perhaps, as
closely even a few months out? Does that exacerbate that trend, that
dynamic? – Yes, it does feel like information overload. I think you’re right in that more people are probably paying attention than happened in the
past, whether or not that translates into a different action on Election
Day. It is a bit harder to predict, but I think right now we’re at
one of these points where the information landscape has changed pretty
fundamentally and that we do have a President who prefers direct
communication with people via his Twitter account more than just about
anything else. For some people this is great. If you agree
with what the President is saying, then this is music to your ears. For a lot of people, including some of his supporters, this has
not gone this planned or as the way they would probably have planned it and some
people really dislike it. So, you do have a President who has a megaphone
who has been more than happy to use it. Then, you’ve got a lot of new polls out
there. It has become cheaper to do this, so you’re inundated with
information and how do you sort through that? – Yeah. – You’ve got that kind of core
political junkie that has maybe picked up a few percentage points. Then you’ve
got people who maybe, as a matter of course, try to stay informed about what’s
going on and maybe that’s grown a little bit too, but it’s certainly, I think, still
the case that the voters will end up going and sensibly deciding the
election. They haven’t quite tuned in in the way that you and I and everyone else who, this is our job or we just really enjoy it. For some
people this is their hobby. Those voters haven’t quite tuned in, so the
information that they’ll receive will be, you know, it still will matter
what’s going to hit them right around that two months before the election and
for some even later. A lot of people report still deciding more or less the
day of the election how they’re going to vote. It is weird. It’s that, you know, as
weird as it is to think about that because, you know, you would think that, well, you’ve had enough time to evaluate any President after a full term.
But, if you even look at the public opinion polls about how is
the President doing, there’s still a couple of percentage points that go,
“I don’t really know enough about it,” and for us that’s
like well how do you not know enough about it? Well, some people just, you know,
they’re just not that interested in politics. That doesn’t mean they don’t go
vote but it’s not something they think about. – They’re not the day-to-day. They’re not
following the day-to-day. – No, and that’s not a pejorative saying. People are busy. They have lives, they have kids, more to do than
they probably can fit in a day, and when it comes time to, you know, if they have
some time to relax, they don’t necessarily pick up a newspaper and turn
to the politics section. – So, what’s interesting about that, I wonder, is are those folks, the folks that are not necessarily following things even a few
months before, are those your swing boys? I would presume that the people who are
following closely, those may be diehard, diehard supporters one way or another, I
wonder. Are those people who are not paying attention as early, are they your
swing voters? – Typically, yeah, that’s typically true. – [Laughing] How do you possibly know
anything if the people who are going to decide the election don’t know? – Well sometimes those people don’t matter. So, you know, sometimes – There’s not enough of them. – Yeah, in 2008 it didn’t matter. The national trends were pulling
everything toward it’s already going to be a positive Democratic year, and
then you throw in the economy melting down right before the election, and it
was obvious, right? It almost didn’t matter. It was a question
of how big of a victory for Barack Obama, right, and there was a couple
surprises. Nobody had North Carolina on their map, for
example, so sometimes it doesn’t matter. But, when it’s close, it would be a gamble. – Okay, so knowing that and, as you said, there’s lots and lots of polls out there,
it’s gotten less expensive to do them, as you noted, and so, you know, we’re
inundated. So, if I’m sort of either the average voter, I’m sitting on my couch,
I’m online, or I’m watching CNN or Fox, and they’re talking about some new poll.
What would be your advice for how that person should process the information
that they’re getting from cable news about a particular poll? How do you know
whether or not it’s reliable? How do you decide? What do you look for? – I probably wouldn’t recommend going to cable news. – [Laughter] – That would be my first
recommendation and it’s not, yeah, I’m sure they’re very nice people but
there’s certainly, you know, their model is to make money and to do
that you have to get viewers and so you if you’re going to discuss the contents
of a poll there’s actually a pretty good case to be made that cable news would
not be a good place to do that because they’re going to sensationalize certain
parts of the report. The narrative isn’t necessarily going to be about
accurate. But, I understand a lot of people watch the news, although- – That’s how people their information. – Well, I mean, some people, not everybody, right. The
older voters do, not necessarily younger voters. To get to the general, like,
how do you evaluate, you know- – Yeah, how do you evaluate polls, how you’re getting your information. First start with who is sponsoring it. [That’s] the first question I would ask. I know it has been one of the political dividing lines is the
mainstream media, whether or not they’re biased, right. I still generally think
that it’s true, and hopefully it would be generally accepted, that those outlets or
sponsor polls to try to be accurate. This would be true of New
York Times, Wall Street Journal, the big three television outlets
and Fox News, too, which a lot of left-leaning people would be like
“Well, I’m suspicious of Fox News.” Know they hire professional pollsters to do
these things. The way they talk about the results, you know, might be different
right and that’s what I was saying. Maybe that’s not the best place to get your
information, but the polls, hopefully, they’re following best
practices. There is a professional, Like anything, there is a professional
organization that sets standards that-. Yeah, AAPOR, the American Association of Public
Opinion Researchers, they actually have awards for-, I mean,
there’s a whole conference you can go to if you’re really into it. If
you run public opinion polls you do kind of have to go, but there’s a whole new
outside agency the sort of set standards about reporting about how data
collection, how you should talk about revealing how you weight your data, and
that sort of stuff. – Does that organization do a Consumer Reports thing where they
sort of rank the, you know. I’m guessing not, but, you know, something where you can
look at this and say, “Okay, let me see this one,” like for the best baby
carriage, the best poll, or the most accurate poll. There’s no outside
organization that’s going to give you any guidance on that or is there?
– Well, 538 does their ranking, you know, that’s an out they have. They’re also trying to make money too. I think generally they try to do a
good job. Everybody’s going to make assumptions about what they think
accuracy is, but if you were interested in looking, 538 is useful for a lot of
reasons, but one of them is that they actually try to match up what the
poll says and what the outcome was and give them a percentage of how many
times they were right, which is essentially what we
care about the most, right. Were they able to predict the outcome correctly, which
then indicates, at least on some level, that they probably did a scientific poll.
I mean, of course it just gas sometimes. It’s 50/50, but, you know, then it truly would be a coin flip in some of these contests. – So, were there any red
flags that the average person should look for when it comes to methodology
like something where it’s just like, “Yeah, when they do this, keep moving.” – Yeah,
the polls that are conducted on websites of news organizations or a campaign
are junk. If Fox News says, “Oh, we have an online Fox.com poll. Please
go online and fill it out.” There’s nothing scientific about that. You know, that would be true if CNN or if one of the
campaigns said, “Hey, come fill out this poll,” you know, if Joe Biden says “Go
to Joe Biden.com,” or whatever. – Yeah, because there’s no science behind how they’re asking the questions
or what type of voters are trying to get a representative sample. – Absolutely. There are also websites dedicated to, like Reddit and places like
that, to go in and mess with everything. Not only is it a bad sample
to begin with, you actually have- – People can stuff the ballot box. – [you have] people who are trolling
them. – So, we’ve talked quite a bit about polls related to candidates
that are “calling races.” Are polls about issues? There are a lot of
polls out there that sort of gauge-. I mean, we do one here at Hofstra. The National
Center for suburban studies does polls that really kind of gauge what issues
are important to people in the suburbs, for example, but are polls about issues
more accurate than polls that are straight up who are you going to vote for? – Yeah,
I mean, when you get to the horse race poll, which was candidate X versus candidate Y, there is always going to be movement on
that. Some people change their minds or just the sample you
got with somewhat different. They’re trying to do different things, so
there’s sort of a fundamental question difference here. One is can I predict
who’s going to win? The other is what do people care about? That’s less
volatile and generally, I think, you can capture what’s going on with the mood of
the country, the issues that are on people’s minds, and then try to paint
a picture about how these things might be related. The last
suburban poll, which actually ran was-. It came up pretty clear
that people were concerned about NAFTA, free trade, and immigration.
When we were looking at this, my guess is this is two polls, [and] it was
2016. I don’t even know that we actually had-. It looked like Trump was
going to be the nominee. But, when I looked at it I said, “Wow, this is an issue that’s coming up quite clearly. If he is able to
harness this, immigration, and NAFTA and free trade,
which were the two things he was talking about constantly during the campaign, it
certainly was something that was showing up as people were concerned about. It’s a
little hard to disentangle how much of that was already sort of going to be an
issue and if that’s something he just tapped into or if that he brought it to
the forefront and captained something that may have been in the undercurrents.
That happens a lot. Issues wax and wane in terms of their importance all the
time. That just happened to be an election where he either made it appear
on the agenda or it was already going to be on the agenda and he just happened to
be the person who was talking about it. – So, if you’re looking at a poll
historically, so a poll like the suburban polls that Hofstra does that
they are over time and you can gauge sort of where people are going
over time, how opinions have changed, should a red flag or something
that you should pay attention to-, do you see significant shifts? This would be really historically based on candidates or based on issues. If
I see a swing of like twenty or thirty percent in one direction the other
between a couple of polls, does that mean it’s wrong? I think about this in
relation to-, we said, “Well, if your SAT score swings by 50 or 60
points between the two times you take it,
that must mean you’re cheating.” Statistically, it’s very rare
to see those kinds of swings, so I wonder if you see something like that does that
actually indicate a real change or does that indicate a problem with the poll? – It could be both. Some issues that move rapidly it actually is-. I think
historically it has been true to be very suspicious of polls that would change
very much in a very short period of time. It also depends on how. Are we comparing
apples to apples or apples to oranges? One poll’s methodology will never be
quite the same as another. If you were, say, taking a poll year to year to year
to year and you’re tracking an issue and that poll has kept the same
methodology, you probably should trust that. Now, that’s not to say you
can’t get an aberration. Certainly you can. Again, we’re talking about, in a lot
of polls, several hundred people that are going to present
the views of some sort of sample, some sort of population.
Sometimes it’s the whole country, sometimes it’s a state, and sometimes it’s a
more specific population than that. Can you get a bad sample? Yeah, absolutely you can.
That’s when you hope that you’re able to find some movement in other polls, too.
That’s not to say that you should be desperately trying to find that because
there is this sort of herd mentality in polling that if you’re an outlier, that you’re sort of pry at that point and that you did a bad job. When that
may not be true, sometimes that’s just the tip of the spear of what’s to
follow. Some issues do change rapidly. Candidacies do rise and fall rapidly. It
can happen. Same-sex marriage is an example. – Yeah. Marijuana is used toward, legalizing marijuana, is another example where you
can track over just a couple of years pretty monumental change. The population hasn’t changed all that much in this period of time. Clearly people are changing their minds. The Republican nomination of 2012
where everybody had their day in the sun and then finally they
settle on Romney. It literally was just about everyone else for a
while. Newt Gingrich was going to be the candidate. Herman Cain looked pretty
good for a while. – Did you see a similar thing happening with the Democratic field in 2020? – Possibly, although, it has been pretty rigid actually until this point.
We’re not really there yet, so it could very much undulate like that as we get
into November and December. They start debating more, and
they get down to a smaller field of candidates on the stage. – Amen to that. So, did it make sense in the early days of the Democratic primary for 2020
to even do any polling with that many candidates or does it
make sense to sort of wait as the field has been
narrowed? – Well, yeah, they have to sell advertisements and newspapers and all
that stuff so, no, it always makes sense to do the polls. Should we be taking much
stock in them? No, probably not. I mean, except when you see something durable, right? There’s a couple, even at this point, of
durable things that we can take away from this. Biden is still the front-runner.
Sanders has a base of support. Warren seems to have now captured some
percentage. We’ll see if that is durable, it’s a little early to tell at this point,
but those things tend to be true still. – I wanted to ask you, too, a little bit about
the difference between the likely voter polls and the registered voters polls.
Are either of those more accurate and is it that you do one earlier in a cycle
and one later in a cycle? Should we trust one more than the other? – This is a “it
depends” answer. In 2018 we had a pretty big wave election. There were a lot of
people who voted in that midterm election who probably didn’t fit the classic
definition of a likely voter. If you’re using a likely voter model to
predict the 2018 outcome, you’re probably going to be wrong and you have to just
know that going into that that there’s-. I mean, it was almost palpable in how much
energy there was on the Democratic side to go out and vote and on
the Republican side there was a sort of a depression, right. These sort of
things tend to follow. Like, you go back to any wave election in the past 20
years and there’s sort of-, it’s kind of obvious what’s going on, you know.
In 2006 it was obvious. In 2010 it was obvious. Democrats knew they were
walking into a bloodbath, basically. There’s leading indicators for all of
these things. The biggest one is retirements, alright, congressional
retirements. When people start to retire it’s because they don’t want to run for
re-election. It’s not because they really want spend
time with their family, it’s because-. I mean, I’m not saying they’re not family
people, but what I’m saying is- – He’s not saying they’re not family people, members of Congress who are retiring. It’s more likely that they know how hard it is to run an
election. It’s a lot of work. They have to raise money, they have to raise a ton of money and then you
have to go out there and actually do all campaign stops. If you think you’re
going to lose, well, what’s the point in doing all that? So, the leading
indicator is always retirement on these sorts of waves. A likely voter
model is probably not the right one to use- – In midterms. – Midterms or in any election
that you can kind of sense there’s a wave coming. – Okay. – Right, so in 2008
that wave was still there, right. It actually built even more. They had super
majorities in both houses because of it. So, a wave is going to bring out
people who don’t typically vote or aren’t reliably going to vote, so, likely,
voter models will be more incorrect. A registered voter [model] might be better. The
truth is you kind of do need to just figure out some sort of model that
predicts, as best you can, who’s actually going to vote. There’s
also nothing wrong with reporting multiple numbers. – Meaning? – Saying here’s
what the registered voter model says, here’s what the likely voter model [says], unless your
ex-ante is saying we’re only going to survey these kinds of people and then you don’t
have that option. There’s somebody out there. You, yourself, as a
consumer of information, can go and sort of look at them all and see what is
the difference here. But, knowing another person who’s just following this
can also figure out what is some push and pull factors that are happening and
when I would talk about the 2018 election and they said, “Well, what do
you think is going to happen?” I go, “Well, if it’s close, it’s going to break
Democratic.” We saw it in this state, right,
a massive turnover in the state Senate- – Yes. – and a lot of those seats weren’t
really on people’s radar even because why would they be? You were talking
about state Senate races, right? So, not- – No, but that was a big deal and it had
material impact almost immediately in terms legislation. – It did and when people would come, especially from the median, say well
what do you think’s going on the state Senate? I said, “Well, getting accurate polling just forget it.” But, if it’s close, it’s going to break Democratic and
they all did. They almost all did. I believe there was only one seat or there were couple
seats. One of them was in Texas, congressional seats, that voted for
Hillary and then sent a Republican to Congress.
He’s retiring. – Because he sees a wave. – I don’t think he thinks he can win in
2020, but he’s also probably-, he seems, from the interviews I have seen, like
a very kind of moderate middle-of-the-road person and a lot of
those guys have become somewhat fed up with Congress and just don’t want to do it
anymore, anyway. I wouldn’t necessarily say that- – Right. – One retirement does not
mean, but there’s actually been several and several of them in Texas, which
is interesting. If you’re trying to determine where the
national trend and the push is going, it certainly looks like the early
indicators that’s the case. – So we talked about likely versus registered voters. I
wonder, in terms of how early voting, which, by the way, starting this year will
be available in New York State, we try to be nonpartisan here, but I am going to
express opinion to say thank you, finally. Early voting exists in lots of
other states, particularly in those states, and has for many years. I’m also
reminded of my experience when I was working in Florida during the 2000
election and how exit polls during that Presidential election called Florida for
Gore without realizing or without registering the fact that there was a
swath of the state that broke conservative in the Western
Panhandle that had not yet voted in time, you know, because the polls closed
later because there’s a time difference. – Yeah, I mean, Florida was
essentially a tie. – That’s right. I think about that and,
you know, the argument being that do you have to create a new model for people
who voted early like have you already voted? – Yes, and they do. They do ask that
and still tend to. Most polls [do] now, especially if they know where they’re
going. The question is not just are you planning on? It’s have you already? It’s true, a lot of people do opt for that and that, states like Washington and
Oregon, it’s all mail-in. There’s no voting. – Right, nobody goes to the polls. They mail a letter or press a button. – Yeah, exit polls are not a thing there.
The whole country is a patchwork of election laws and it
depends on what can save you time. But, increasingly, a lot of states have some
version of this, whether it’s early voting or no-fault absentee, where
you can just request a paper ballot and you can send it in or drop it off on
Election Day. Yeah, it does present a challenge. I’d say it presents
probably more of a challenge to campaigns than it does to pollsters
because at least you can ask people how they voted.
Campaigns are, well, if you’ve submitted your ballot a week before the election
well you’re kind of ungettable at that point, so trying to figure out who
those people are and which ones you should be trying to push is going to be challenging. – Well, I guess I wonder, too, with early voting if you’re asking early voters who you voted
for how reliable a barometer is that? Can you
extrapolate from early voting, if you were polling just early voters, how
things will break ultimately? There’s certain demographics a tend to behave in
certain ways and the older voters tend to vote earlier or through absentee
for various reasons. Something [like] that they can’t, maybe, get to the polls on
Election Day. You tend to-, the way that the current dynamic is, you’re
probably going to get a more rosy Republican look with some of the early
voting states and you would. But, as is always the case to decide any election,
it matters who turns out. It still matters. There’s a whole cottage industry
in political science that looks at early voting and tracks who’s
returning a ballot is a plus D, is it a plus R, and they track it by day. This
is how we know like Sunday tends to be a big plus Democrat day in the South
because it ends up being a lot of churches will gather together
and then they have movements that’s called Souls to the Polls and they all vote early. It was one of the reasons why when some states tried
to actually remove Sunday early voting days it was sort of- – An outcry. – Yeah, it was a big uproar over that. So, it’s tough because people vote at different times
and so, yes, it’s useful, I think, to get a general trend of which way it’s
going. If you have many days worth of data or if you have the whole two weeks or whatever how long the early voting period is, you probably can
gather some sense of that, but that can all be wiped out on
Election Day. – And that is essentially when most people vote, still. – Yes. – So,
I wanted to ask you my last question, and then we can kind of move on if there’s
some other things that you want to talk about, is the difference between online
versus phone polls. Not the online polls you were talking about earlier where it’s like “Come and vote,” like- – Right, no it’s very much important
to make a distinction. We talk about online polling- – Ones that are done
scientifically in the same fashion as a phone survey poll, but just the method of
collecting the information is different. Is there a difference in their accuracy?
I know Pew had some studies that talked about the fact that the mode of
questioning can affect how people answer it and that people might be more likely
to answer questions more honestly about difficult topic, like race, for
example, online versus over the phone. Do you think one of these is more
accurate than the other? Which do you prefer? – So, the old pollsters would say neither is good, right? They would
actually demand you do face-to-face interviews, so it’s really, you know- – That’s true. I
didn’t even-. Does anybody do that anymore? – Maybe outside of a few academic and those are not polls
designed to make predictions they are very much polls designed for academic
research and so you wouldn’t be able to harness it real
time, anyway. But, there’s some truth that at one point the random digit dialing
phone call poll was the new kid on the block and that was the disruptor because
it was so much cheaper to do that than to go door-to-door or however your
sampling scheme was. If you’re doing cluster sample it might be certain
neighborhoods in a certain city or something like that. Phones
got a lot cheaper, but they have their own issues, right. Especially in
the beginning, not everybody had a phone or certain populations didn’t have a
phone. It could be a lot harder for a long time in this, well,
even today, it’s still harder to reach minority groups with any method that you
that you’re attaching and you’re using to conduct your survey.
Now polls have their own problem, which is that there’s
a lot- – Nobody uses phones anymore to talk? – Yeah, there’s no landlines. It’s that most people have
cell phone only or if they have a landline it was sort of forced on them
because the cable company says you have to have a landline. – I mean, I certainly have a
landline, but I don’t use it all that much. I wonder, so even on a cell
phone, though, if you had access to those numbers and you can do a phone survey
that way, is it more effective to do online because people are more likely to
answer because that’s sort of become the most common mode of
communication? Are there such things as text polls? Can you do- – Yeah, sure. Texting seems to be more likely even than traditional online polls. – I mean
basically any method you can think of now that you can will be used. Cell
phones is the big problem with polling on the phone. – And yet let me know how marketers find my number and call my cell and offer me cruises every
week and yet I’ve never gotten a polling call on my cell phone. Why is that? -Well, if you’ve done business with a company and then you’ve invariably agreed to sell your information about 50 million times over.
So, it’s because it’s always with us and our business partners. – Right. It just seems like for the random companies, cell phones aren’t a problem and yeah. – Yeah, well they have a list of numbers. For a polling organization-. These are
telemarketers, right. There’s money to be
made on those cruises. There’s not money to be made in a public opinion poll and
you do have to pay for those numbers, but the difference in the law is very clear.
Whereas on landlines you can actually just have a computer or an old-timey
machine, if you want, randomly punch in the numbers and you get somebody- – Sure. – That’s not the case of cellphones it’s against the law to random digit
dial cell phones. You have to have somebody punch it in manually and so
that means a real person has to be there. – Oh, that’s interesting. I did not know that. – and [the person] slows it down. What you do is
you buy a list. You buy a list of cell phones. The other problem is the
response rates have plummeted because it used to be when Gallup called
you in the 1960s it was like “Oh, wow. My opinion? Absolutely.” Now it’s “Well, let me ask you a question: Are you really happy with your credit card
service?” Like, what? I don’t want to talk to you about my credit
cards, so we’ve been inundated with the stuff. It has absolutely plummeted.
Phone surveys are hard. It doesn’t mean they’re bad, it just means that they
have a whole new set of challenges that have been building for the last 20 years
that, I think, we’re solidly in that point where it’s not only more
economical to do online polls, but it actually might be crossing that
threshold of better. It still comes down to who’s in the survey and how
accurate do they reflect the rest of the country. It’s certainly the case, and
you don’t just want everybody who wants to be in the poll. It’s sort of-.
Everybody says, “I’m always skeptical of the person who wants to answer
a twenty minute political survey on the phone.” – Yeah. [LAUGHTER] – Like, who’s doing that? Well, you know, we need the people who don’t want to do
it the most, actually, because their opinions are probably more reflective of
the average person. So that’s true of the online nature, as well. You have to
make sure that the sample is good and there are ways. All of the survey
companies, phone [and] online, have developed pretty complicated weighting
schemes to come up with a way to make what are now, for everybody, imperfect
samples into better samples. Better, more reflective samples of what’s going on. – How do they do that? – It’s mathematically done with percentages of what we know the population looks like.
This is why the census is so important and while we’ve been- – The ultimate in polls, really. – It’s right, it’s really the most accurate poll we have. The value for research is immense because this is where you get into the difference
between a population in the sample. We never can really survey the population,
with the exception of something like the census where it’s such a huge endeavor.
You’re not going to be able to survey 300 and something million people
every time, so you have to take a small sample the population and the
reason that we can do that scientifically is because we know what
the population is supposed to look like. So, once you know those parameters,
what percentage is White, what percentage has X amount of education,
then you can look at your poll and re-weigh it to match what the population is
supposed to look like. They’ve come up with even more sophisticated ways to
do that knowing certain things about how likely you are to answer the
survey. [They would] maybe down weight somebody who’s a little bit too gung-ho about it. Knowing that not everybody is a democratic or republican activist, right.
Most people aren’t, really, and so those are the people who are most likely to want
to answer your polls, so you have to figure out who they are and actually down
weight them a little bit so they’re not over-represented. – How do you figure that out? – Well, it’s pretty clear. [Laughter] If you’re out there donating money to campaigns and you’re volunteering on campaigns, you’re an activist. – But then how do you, the pollster, identify that person if they’re answering it on online? – Yeah, you ask them. People are happy to tell you something. – I want to end this with a bullet-hole of sorts, but I wanted to give you an
opportunity to waive any issues that we haven’t covered if there are any things
you want people, that you feel really strongly, that you want people to know about
polling that I haven’t asked you about. – The one thing I would keep
in mind if I were thinking about how to interpret polls is polls are a snapshot.
They’re a snapshot in time. They are almost obsolete the moment
they are released. The public is constantly changing. – Well, that
increases my confidence. [Laughter] – Well, it’s true. Focus on the broader trends, what is
durable, [and] what seems to be kind of sticking. If one poll comes out and says
that Michael Bennett is now the new frontrunner. Well, maybe, if you can see it
coming and it builds and you can kind of see that trend with multiple
polls. But, if there’s a single poll that said that, you have to wonder about maybe
they just call his house too many times. – [LAUGHTER] – Sorry, senator. – All right, so we’re going to wrap up with a quick poll here. This is just
going to be a little fun, so don’t be nervous. – Alright. – Dog or cats? – This is tough because I have a cat. I have a cat. – That’s good. – But, I’m thinking about getting a dog. – Okay. – I have a four-year-old. I think he
wants a dog. I’m not sure. – Okay, fair enough. – Yeah. Maybe both! – Okay, you’re not being a really good polling subject here. – I’m sorry. – Coffee or tea? – Coffee. – Okay. Alright, I’m with you on that. Mets or Yankees? – You know I can never forgive the Yankees for what they did to my A’s. – Yes. I wonder if that’s why you’ve become an expert on polling because
you’re an Oakland A’s fan. After all, one could argue that-. I know.
Craig’s looking at me for those of you who can’t see it Craig’s giving me a side
eye here. Think about it, Billy being the general manager of the Oakland A’s
pioneered Seaver metrics, which is the use of statistical analysis to try to
bring some order to predicting how baseball teams will perform. – I don’t know
that I’d make that connection. – [LAUGHTER] I’m kind of polling with that research, you know? I’m trying. – I don’t know that you
know 17 year old me skipping classes to go watch A’s games was thinking about that.
That’s really pre-Billy Beane era, so yeah. – Alright, so thanks to Craig Burnett for a
great conversation and, as always, thank you to WHRU radio, Hofstra University, 88.7
FM for providing us with space and technical support to make this podcast
happen. Just a few reminders if you want more information about events that
will be happening on campus associated with Hofstra votes, go to hofstra.edu/votes. You can also hear other podcasts that we’ve done in the past in
addition to listening to this one again. If you have a comment or a question
about this podcast or any suggestions about the podcast people we can talk to or about events that we can hold related to Hofstra votes, give us a shout
out on Twitter using #HofstraVotes. In addition, we want to remind
people you can follow Craig on the Twitter at @burnettpolisci. With that, thanks very much everybody
and don’t forget to vote. [MUSIC]

Leave a Reply

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