Donald Green: The Science (and Pseudoscience) of Elections


Hi thanks so much my name is Quentin Palfrey i’m the executive director of JPAL north america and as we draw to the close of of a historic election it’s real honor to be here with the man who literally wrote the book on get-out-the-vote efforts to help us understand the science of what we know about what works and what doesn’t in elections he’s a professor of political science at Columbia and his research interests let him describe but they include voting behavior and political participation he’s written a number of books including the seminal book on what works and get out the vote operations and he’s also really fun and interesting guy who’s really interested in designing board games and woodworking and raising seeing-eye dogs and all sorts of things so without further ado I’m going to turn it over to Don the such a pleasure and an honor to be good i have a lot of material to run through 35 minutes but i’ll have plenty of time for questions and happy to stay way after to talk about anything you care to talk about louisiana to the light-hearted talk that I haven’t planned today is kind of a romp through a vast literature on what works and doesn’t work in campaign craft but also kind of a send-up of the kind of phony claims that you’re likely to encounter if you like me to get about three emails a day from campaign consultants Hawking their their wares of all sorts in fact that I literally got three on the train ride here for last-minute things I can do with phone calls i thought i would share some of those with you today I know that most people who talk about voting we’re talking about no damn exciting talk about the big picture you know how would turnout rates be affected by fundamental changes in our constitutional system you know what if we add up what if we had proportional representation or what if we had a very different kind of voting environment what if we didn’t have the electoral college i’m going to have to talk about those questions of Abkhazia perhaps but we’re going to ditch them for today because we’re going to go down to a very very narrow set of tactical questions you know imagine the election is going to happen say in a week say and and now you have to decide how you’re going to allocate your campaign resources so imagine you’re running one of the campaign’s literally one of the Hamptons and its really too late to change the tenor of your campaign boy is it too late for that and and and and now it’s time to figure out you know what specific things you’re going to do so which voters in particular will you be targeting and how how in particular will you be communicating with them and i would say that you know the way to sort of think about the trajectory of American presidential elections over the past two decades is that a prior to 2000 for most of the campaign strategy and tactics was focused on persuasion and then from 2004 through 2012 the focus was on mobilization and 2016 to really felt so what is the distinction in the world of mobilization you’re trying to bring a bigger army the battlefield or simply trying to get more of your supporters to vote not really necessarily could change any Minds you’re simply going to to get the people who already support you or your candidate and and get them to vote in the United States that’s a very real strategic option because our turnout rates are famously we have the second lowest turnout rate of any Western democracy did used to all be that way as I’ll describe it later on but it is that way now the alternative is persuasion to try to change minds not necessarily change who’s going to vote but change what people do when they cast about and those two different approaches have very very different strategic implications for targeting and outreach so what do you think about you know mobilization you’ve got to be careful because if you’re going to be mobilizing someone who ultimately is like a coin toss between voting for your voting against you on average you’re going to get nothing out of it right so you so that is why on the right you’re trying to figure out who your conservative christians are your gun rights advocates are your home schoolers the kinds of people who will vote disproportionately for the right as opposed to the left where you would would focus disproportionately for example of african-americans are environmentalists or people who have signed petitions for a minimum wages increases and things like that so targeting your your your particular partisans is super important if you’re going to be doing mobilization now on the persuasion side targeting is a little bit different because they’re you typically want to focus if you’re doing a persuasion oriented campaign that people who are so-called high propensity voters and that it’s this is jaipal has important distributive implications why because of if you’re going to focus your efforts on high propensity voters that typically means that you’re going to focus your efforts on older voters more residential lease table voters and more after involves and people who are outside the scope of that kind of mobilization channel will receive very little campaign communication directly from campaigns and if voting is a habit forming activity they’re essentially going to be forming a habit of not voting so the the implications for the kind of bifurcation of the electorate of of campaign tactics that focus in on this kind of targeting strategy are often profound ok and and by the way because campaigns are are definitely afraid of mobilizing their opponents the very often say nothing to their opponents so if I’m a Republican and I know I’ve targeted you as are tagged you as a Democrat i’ll make sure not to say anything to you lest i get you all riled up get you to vote when I when i’m probably going to bear the brunt of your your anger so then what is micro-targeting well this is a good topic for people at MIT you know because some of you have had a semester or two of regression and I just want to say that it’s a big world out there and you can make a lot of money fast without having to know that third semester stuff pretty much two semesters it’s good you’re good to go so what do you do the idea of micro targeting is you’re going to sell a product and the product you’re going to sell is a database and how are going to create that database you’re going to do a survey about say 2,500 or 3,000 people and you’re going to ask them every question under the Sun but those people are going to be sampled from an existing voter file public file that you can pretty much purchase in any state so you’re going to survey people and you’re going to get their survey responses and you can you can ask you know who prefers trumpet who prefers Clinton you’re going to build a regression model to forecast that support and then you’re going to use those very same background attributes that you find in the voter file to get predicted values for all the millions of people who are not in your survey and then you’re going to try to sell that for about a hundred thousand dollars pop and that’s that’s micro-targeting and it’s it’s a big business in a very lucrative business i guess one of the questions then is doesn’t work out this is too small for you to read perhaps but um I don’t I am kind of addicted to reading the advertisements of that are associated with the emails that I received it i just i get such a kick out of it because it’s a totally unregulated business and you can say anything you want it’s it’s the Wild West so so you can find lots of testimonials like this one if you want to know whether micro-targeting works you could ask the experts and they’re happy to give you lots of examples of things that work so here’s it here’s one particular website it says you know what does micro-targeting do it it utilizes advance the information ie the predictions in your regression and uses all of that data and much much more to develop a custom and proprietary segmentation the keywords proprietary because they’re gonna sell it to you and now they’re going to tell you a story about a campaign that fell just short of victory until they figured out how to micro target their Republican base while not annoying the people are going to vote against them by targeting them inadvertently and that it’s proof positive because although they lost narrowly by only 525 votes in the last time out this time they want by 30,000 votes QED that their product works so so interestingly you know I read this before i was doing some randomized trials and collaboration with the Rick Perry campaign and i justjust say that you know I i work with both sides of the aisle in the middle you know basically anybody wants to run a randomized trial and wants to to do public science I can’t be buried in some proprietary you know embargo thing forever but it wants to go to contribute to knowledge i’m happy to do an RCT with them so we did a series of rcts with the Perry campaign and one of the interesting things was it we gathered a whole lot of survey data and you could compare the micro targeted Perry supporters to the actual people who were surveyed and said whether they were Perry supporters and here’s a cross tabulation of that so if you look on this side you see that the most pro GOP predicted people well about 60-percent of them in fact predicted the in fact supported Rick Perry but on the other side you know if you said well i’m a democrat i wanna i want to target the people who are going to be pro Democrat well actually thirty-nine percent of them are going to support the Republican anyway so you know you paid a hundred thousand dollars in the middle person you’re actually contacting is not somebody sympathetic to you so it’s a it’s a big industry so the most hilarious claims are in the world that is now at the forefront of fun you know getting you to spend your funds on them and that’s that’s phone calls for the end of a campaign the phone call vendors basically like extend the bushel and money just pours in because it’s too late to orchestrate things like direct mail which require a few weeks of lead time you could maybe orchestrated canvassing operation if you’re really quick about it but that window is going to close pretty pretty soon too but phone calls happen in abundance in the next few days so what then is this is all about it’s about a testimonials and the testimonials are hilarious so we had dick van dyke you’re probably too young to remember dick van dyke kind of he was he was kind of like the Disney actor of his ERA record a message so it’s a robocall ok robocall which is i’ll demonstrate a minute using several rcts never works as a perfect record of never working on a referendum that was expecting a close vote and less than forty percent turnout had record turnout we want by fifty percent of vote say no more that’s that’s quite impressive proof to me so this one came to like three days ago figures don’t have to open that one with figures don’t lie let’s see whether figures don’t lie so this is their evaluation of the effectiveness of their own phone calling operation and the outcome is whether or not people turn out to vote so what are these bars let’s take a closer look this is good okay so so what is the idea here they’re going to show you a bar graph of people who had no completed call so these are the people they may or may not have called but whatever they they didn’t reach them and with one completed call turnout goes up by about six percentage points notice that they say twelve percent lift because they’re going to use the biggest numbers they can so they’re going to use percentages rather than percentage points that’s okay and they’re gonna go to complete three complete for completing your interrupt the mid-sixties so these calls are enormously influential and if these costs roughly a dollar per completed call that would be rather cost-effective by the standards of campaign craft but now we have to ask yourself if you were in a research design class what kind of grade would you give this comparison probably enough right because you would say are the people who get a call to complete a call comparable to the people who don’t complete call well in order to adjudicate that and this is foreshadowing the second half of the talk maybe we should do a randomized trial where we randomly assign people to either get a call or not and then compare how different the voting rates are those of those people have been randomly assigned and this is an experiment we did in 2004 in in in illinois right before the hotly contested presidential race it’s you know it’s it’s kind of a timeless thing it doesn’t really matter when you do it but that the basic idea here is 64.4 percent of the treatment group ends up voting according to public records and 63.4 in the control group that’s about one percentage point gained about half those people were actually contacted so if you take one and / point five you get about a 2 percentage-point gain on the people who are contacted that’s a lot less than six percentage points so where’s the six come from well no again just kind of skipping ahead for for fun let’s compare the people who get a call that is absolutely nothing to do about part with politics these were people who were encouraged to buckle up for safety I mean maybe that makes vote but i kinda doubt about it and notice that the day to vote at a rate of about five percentage points higher when they’re contacted and notice that the people who are uncontacted voted a lower rate and it’s not because they’re just disappointed that no one called them it’s it’s all got to add up to 0 right so so the whole point is that people who are contacted our are different now why would they be different well they’re not dead for one thing and another thing is they haven’t moved away and they might have been contacted by other campaigns but probably most importantly they’re sufficiently interested in what you’re saying to actually answer the phone and and you know complete the call so it’s probably not surprising that you could find the spurious relationship even when you know greatly exaggerated the truth now you could say well alright we’re not gonna listen to campaign consultants and let them do the evaluations let’s let card-carrying professors do the evaluations but for a long time and even to this day a lot of the evaluations that are done of campaign crafts are done based on survey research and in those cases you ask people whether they were contacted by a campaign so for example the american national election study famously asks people did someone call you up or comment come around to encourage you to vote for a particular candidate or party and you’re giving them them a self-report of what you’ve done and that may or may not be accurate to do the vagaries of of memory but also your yeah i’m interested in politics you know they’d be the kind of person i do the kind of person they would contact and so you’re you’re kind of giving a characterization of yourself and then of course the contacting and the targeting is not done at random especially if they’re trying to target high propensity voters so even if these contacts had no effect you would see a strong correlation between saying you were contacted and saying that you voted so that’s unfortunately all a lot of the political science literature on this question from you know roughly the nineteen fifties through the nineteen nineties so since the 1990s late nineteen nineties a burgeoning new field has been generated which one really uses randomized trials in a kind of jaipal sort of way to get the question of what works and so how did these trials work well they’re very much like pharmaceutical trials you take groups of voters and they could be individuals that could be household they could be media markets or precincts you randomly assign them to treatment control and then you see using public records how votes actually came and the idea is you’re going to treat the treatment group and try not to treat the control group diff there there’s crossover you have to make statistical adjustments which you’re gonna have to take classes to hear about um but the the key thing is that we’re not going to ask people with a vote we’re just going to use public records because we don’t want any cross-contamination between you’re getting the treatment in room and you know thinking that you voted ok so how does a typical experiment look you know just for fun i I’ll pick one that we did in 2014 this was Carly fiorina’s super PAC was in New Hampshire and here’s a client kind of classic case of a two-factor experiment they either get a phone call or they don’t they either get 0 1 3 5 or 10 pieces of direct mail encouraging them to vote ok and particularly the voting the courage being encouraged to vote for conservatives and you know just kind of put this in perspective after a campaign consultant takes a take of of the of the production costs and what not mean that sending 10 pieces of mail cost something like four dollars and fifty cents that’s like four dollars and fifty Cent brick that you’re sending somebody and the question is you know what you get for that amount of money and the answer is well you know this is the phone effect if you didn’t send an email and you you did a phone call well that control group looks a lot like the treatment group in terms of the the actual outcome if you look across the the columns though you see that the turnout rate for ten pieces of mail 67.3 vs 67 point out he got three tenths of a percent percentage point which is to say you know if you said a thousand pieces of mail you get three additional votes in each one of those chunks of 10 cost four dollars and fifty cents so that’s that fails what i would call the trail of dollar bills test you know you’d be better off just leaving a trail of dollar bills leading the person to the polls because this is this is just not it’s not working no that’s going to be typical of a lot of a lot of these kinds of tactics and so that’s why rcts are very important because if you didn’t do an evaluation everybody has a war story about how they won or lost to give an election and what they did or didn’t do ok so what what then what do we say about all of these experiments you know you see that you know if there’s there’s a book that I i co-authored alber republished by 3rd edition last year and the Ivy League joke is the distinction between books you’ve read in books you’ve read yourself so this can be kind of like a book you’ve read and and so i’ll just give the kind of quick summary here there’s basically a distilled into one sentence you would say quality matters the the quality the kind of authentic personal interaction associated with campaign communication makes a big difference when it’s impersonal when it’s a robo call when it’s a male reminder it tends to be either ignored or forgotten or whatever but it but if it’s a if it’s an actual heartfelt appeal like it’s very very important you know Ben that you come out to vote you know Ben can I count on you to vote you know average you know this election coming up don’t you know we can’t lose this one it’s a little like going to a social event like a party you know you could get a card and it could save the RSVP or an email but if somebody actually calls you up or or buttonholes you in person it’s a lot more effective that said there’s very little evidence of synergy when you talk to campaign consultants the very often will talk about the number of touches of voters you gotta start off with a mail piece of mail that maybe a phone call that maybe three more pieces of mail and then you know the program can do the program continues with a door knock and blah blah blah and i’m not saying that you know more isn’t necessarily better but it’s just you reach you typically reach diminishing returns where’s the way they characterize it is you know it’s the magic of the whole package where we’re having an ongoing conversation with the voters well probably not so so what then do the do the RCTs randomized control trials suggest about the effectiveness of these kinds of tactics well it’s a polyglot you know group of randomized trials many of them are but not all are done in Los alien selection so you got to be careful about reading these things you know closely but you know basically you know these the the characterization i’m going to give you is probably about about right even further for high ceilings selection like the one we’re in typically for people who are canvas at their doorstep the people who answered the door and talk to a canvasser and see a bump of of something like four percentage points sometimes it’s more like seven percentage points when they’re low propensity voters or when the canvassers are especially good but Canvassers may not be especially good you know it all it all sort of depends but in typically a high ceilings election like a presidential election the people were influenced by canvassing are the people whose base rate of voting are you know somewhere between 25 and fifty percent something so that when the base rates are low you tend to get more impact if they’re already you know up there with a ninety-five percent chance of voting you know you’re not gonna move that much and so you have to kind of keep in mind that you know typically in a presidential election where turnout rates are in the roughly sixty percent rate and roughly seventy to eighty percent of the eligible public is registered you know most people have a very high probability of turning out and so if you want to be effective you’ve got to focus on young people are people who have recently moved or people who have been registered but to vote and only intermittently and so on okay so I’m gonna I’m gonna really move fast as I got the 10-minute warning so i really gonna go fast so the one dirty secret about canvassing is very often campaigns are incentivized to do door knocks rather than to actually have conversations and that runs afoul of the quality versus quantity trade-off robotic calls almost never work except with special exceptions having social pressure which I might talk about later volunteer phone banks however seem to be pretty effective you know if you can go off to a battleground state calling the people you know or could talk it and having a heartfelt conversation it’s probably pretty darn effectiveness as a as a campaign tactic and certainly cost effective on the other hand delegating that to a commercial phone bank that’s going to do a lot of perfunctory calls cranked out a 13-hour that’s going to be fully minimally effective interestingly there’s lots of political science theory that say it’s the heat of the election that matters you know kind of getting a sense that there’s really some contestation going on actually there’s no evidence for that whatsoever when you send people a truckload of advocacy mail about the importance of the elections the issues that are steak blah blah blah blah has no effect on on turnout and all that will be amplified when we look at a quasi experiments on presidential advertising which almost never say the word vote by the way and have no effect on turnout so that’s kind of interesting but non-partisan you know encouragement to vote you know they’re not super effective but they’re they’re mildly effective an email is a perfect 0 or even worse with the exception of person-to-person email but you might work because again it’s more personal this is just kind of quick snapshot of several experiments that send up to 12 pieces of mail the red line is a is advocacy mail four Republican candidates in 2012 done by Cubbison finds nothing or less than nothing this Virginia thing was done in 2005 by a democratic operation they find nothing this one was the Carly Fiorina super PAC which is more encouraging than just purely advocacy encouraging turnout and it finds a little something-something and this is a pure kind of non-partisan do your civic duty type thing but you can see in all cases you reach diminishing returns so would kind of a kind of interesting that you five is about the number that people can stand before the backslide ok so so just because we only have a few minutes i’m just gonna talk a little bit about what’s on tap for 2016 or what’s going to happen next few weeks much of what we are thinking about now draws its inspiration from 19th century America one of the interesting things about United States Army has famously low turnout right but it wasn’t always that way we used to have very high turnout turnout rates in the eighty percent or or above rate and we were positively you know northern European in those days and um and what was different well you know it was you it was different in many ways one of the things that was different if you can sort of see in this picture the guy in red he’s voting in public right is there’s nope no secret ballot the other thing is that all men the other thing is that this guy here is is he’s really drunk about the keel over this guy’s pouring whiskey right um it was it was a whiskey sadaf Eric that’s part of the reason why i was so hard for women to get the vote because so many polling places where in saloons Victorian women that was not you know beauty is an appropriate thing for women to do so so what would change once progressive reforms came in the late eighteen eighties the the ballot was secret there was the 50-footer 75-foot rule that distance party workers from the actual act of voting and you know no no whiskey so all of those things come diminished turn out so that by the nineteen-twenties people were concerned about low turnout are our kind of flipping out because turnout has declined at roughly by 40 percentage points so we actually did some experiments in not particularly good experiment we did them 2005-2006 where he took match pairs of precincts and flip coins and put one of them into the Election Day Festival where we had kind of a 21st century festival with no whiskey was you know family-friendly we were making cotton candy and flipping burgers and so in each case we would flip a coin and one of the precincts would be in the festival case the other in the control case and we got about a two and a half percentage point increase in precinct level turnout but it was it was it was considerable but you know these were lame professorial parties that we could do better so so the idea is finally get around to 2016 been trying to plug this for years but now we actually are going to have a serious later part is you can actually see where they are at this website and they’re going to have you know Univision parties all kinds of things you know food entertainment who knows and they will feature puppies no no whiskey but there will be puppies and then that sounds like a big draw to me so we will see ok so so to wrap up if you’re not going to talk to people face-to-face and you’re not going to have these kinds of social events you know what are you going to do well one of the things that is often done especially in the wake of some recent research which I’m connected to is to enforce social norms pre-scripted social norms norms about how people should behave and even when people are not voting in vast numbers they still Harbor a strong sense that voting is inappropriate activity and it’s a matter of their civic duty the problem is that if you use heavy-handed tactics like scolding people there’s going to be a public outcry and backlash and possibly reactants this tactic it turns out is incredibly old the very first get-out-the-vote experiments done in the nineteen twenties but were were done using exactly this tactic this cartoon compares the slackers who won’t defend their country and times of war with the slackers who won’t take time out to vote ok so you know you’re basically shaving people have noticed they were saying hello brother and this guy is reacting negatively so so that the old days in that now this has become quite current so campaigns routinely send out shaming male that has that has changed however in recent years and I think that a growing literature has suggested that you can get many of the same effects without the reactants if you express gratitude if you think people for their past participation and you can still let it be known that voting is a matter of civic duty without necessarily going to the point of being creepy and scolding meant enforcing social norms um what are the effects of these kinds of tactics well they’re not as big as they were in the low salience elections that we originally test them in typically there are much much smaller but there’s still probably cost effective so what then about TV commercials with this is a very interesting feature of the American electoral college map that for example is the 2,000 map but it’s basically the same idea every election there are media markets in our country that straddle state borders and sometimes one part of the media market for example Pittsburgh beams it’s commercials into Western Maryland right so Western Maryland is never compare alone is never competitive but you know if they’re in the Pittsburgh market they’re going to see 10,000 adds a typical election this time with only one side advertising agency 5,000 out so what’s interesting is you might ask well is there any evidence of the places at the kind of fringes of battleground states for voted higher rates the answer interestingly is now which I think it’s theoretically quite quite relevant of course many things are untested so there’s been about 25 experiments on persuasion in the basic bottom line is it’s really hard to do it’s hard to do the studies and it’s hard to find effects that are doing studies because you can no longer use public records in the same way to gauge outcomes if you want to do it in a kind of large-scale way you typically have to treat an entire precinct and that’s expensive to do alternatively you have to do a survey at the end to figure out whether the effects of the effects are expressed in terms of post-treatment surveys in the treatment control what we know is that you know these effects tend to be fairly small and often transitory we did a series of experiments basically exposing people to real newspapers that there was sort of like the New York Post they were basically scandal sheets only left thinner than the New York Post Christina for folded pages and the treatment effects on the right are fairly large on the left there fairly small one the right this is the share of people who don’t know about the candidates in the control group so they have no idea so when people are hearing about people for the first time in these candle sheets when you interview them weeks later you know unbeknownst to them there’s connection between the surveillance in the newspapers there’s a substantial effect but for better-known candidates there’s actually no effect and when you look at actual field experiments that try to persuade very often they came up they come up short so you know for example this is that covers an experiment where they did kind of standard Republican direct mail in three legislative races this is micro targeted mail and as you can see it had zippo effect we did some TV experiments whoops time’s up well you’ll never hear about them sorry the bottom line is this is the effect within the week that the advertisements were deployed and that was the effect that week after and so the the concern is you might move opinion but you’re not going to move opinion in an enduring way and that was one experiment and we obviously need to do more but it’s a little hard to do more because campaign consultants are not so eager to do experiments on television because the the mark-up for them it’s really quite considerable okay so why don’t we just kind of cut to the end ok so what would you do so now what are your your the audience and you might think well greens kind of convince me that maybe I like there are some things that I can do to increase turnout but I’m just a person I’m not a whole campaign what can I do in say the next 45 days well you know i would say that if you wanted to raise turnout appreciably and I think we have to be realistic to to produce a five percent in 4 percentage point increase in turn out which is say one extra vote for every 20 people you you interact with that’s a pretty big effect in a presidential election if you could do that that would be pretty impressive what kinds of things could you do well if you know some people in battle ground States you could skype with them or you could do something you can go visit them home you could say I’m not leaving until you vote early for people that I perfect or you could get others to pledge to vote and then you can hold them to that pledge in places like North Carolina believe that the the voter file is actually online so you could say to your your pals in North Carolina can I count on you to vote and then say good because i’m going to be checking every day to see whether you’ve done it and and then there will be social norms and accountability but I think maybe the most fun thing to do would be to you organize a house party you could organize it i guess in somebody else’s house and milk let’s clean up that way and and and try to raise turn out that way but that’s essentially 19th century version of a the votech because you can still serve whiskey ok so then just kinda end I would say you know campaign professionals are kinda interesting lot because they’re very very savvy very smart they often have tremendous on the ground knowledge of how politics works they often are also very partisan so it’s not as though they don’t want their side to win they often do very very badly but they’re beset by conflict of interest because they tend to have an ownership stake in the mail house or in the phone bank or whatever they’re doing and so it’s very hard for them to kind of take a detached view of the effectiveness of their operation and you almost never see homegrown randomized trials by the vendors themselves for political scientists or you know those who do social science more generally you know I think that that this line of research has gradually transformed the study of of campaign crabs and camp and campaign crafted and also the study of the electrical effects but it has also opened up new avenues for theoretical explanation such as interpersonal spill overs or habit formation you know what happens if you encourage somebody randomly to vote in one election what does that mean for their participation in subsequent elections you know in the united states where we have arguably the most frequent elections in the world sometimes our electoral calendars like into a gateway drug where we give lots of people an opportunity not to vote in and then guess what they’re voting out habits are often severed as a result so one common suggestions that we should consolidate our electoral calendar but of course it’s not consolidated for political reasons there are some things that we don’t know about and I think rather than me ask the questions you can ask questions when we pause and open it up to the light ok that was terrific gone thank you so much i’m going to take questions both from the audience here and from the audience online before you let me just ask one question myself which is this has been the strangest election I’ve ever seen involved in a number of them what are the things that surprised you about the 2016 election and what are the things that you think we may learn from this one sort of going forward that we didn’t know before this particular content because it’s a great question it’s a big question there’s so many surprises in this election you know we’ve never seen a presidential campaign operated this week or a candidate when was the last time that a presidential candidate routinely did not use a teleprompter or or or tweeted things himself right usually you have circuits doing the dirty work but the attack directly yourself is a bold move and we will see whether it becomes part of the routine playbook i would say perhaps the most interesting thing especially for the Trump campaign I mean on day one he announced that he is worth 8.7 billion dollars but the amount of capital intensive campaigning that he has done has been relatively surprising fairly small he apparently has only put in something like 33 million dollars into television advertising through through the early part of of october and that’s less than many self funded candidates put into their senatorial campaigns so what we’re in some sense learning is what the electoral map showed would look like in in this weirdly asymmetric case where the Clinton campaign more or less is business as usual the usual amount of spending on TV and usual amount of spending on the ground and the Trump campaign is doing something askew so that so I i wish that he were more on the regression line kind of candidates that we get a more controlled experiment that you know you get what you know they’re it all comes as a package ok many many surprises i mean they have been watched every presidential debate tonight I’ve never seen one where the candidates didn’t shake hands so it was a new moment in some sense in terms of the disability of presidential interaction renuka we’re going to take a couple of questions from the from the internet from folks who have that’s running okay to speak up already the question ok ok so let me let me just a second one first because it’s an easier question so yes why are some people surveyed i remember hearing many many funny speeches about this because I sort of grew up and I kind of survey environment and the routine question was who you pulling if you’re not calling me who are you bowling and and i think that the fact is that there are lots of reasons why somebody would be contacted for a survey some having to do with just random chance and some having to do with the fact that their particular demographic profile is rather abundant and so there isn’t much uncertainty about what people like them are going to do even if if their unique as can be nobody cares about them because they’re in a boring group so that’s a that’s the survey question on the on the script question i think one of the things that has perplexed academic researchers who would love love love to write articles about how the message matters is very often the message doesn’t matter that much in part because you know so much that the people remember about a phone conversation is about you know the subtle things you know was the person reading in a mechanical way did they seem enthusiastic did they seemed as though the candidate was their candidate you know if they were to be asked the question could they actually answer knowledgeably do they have an accent suggesting that there from way way out of town and all these things are are are not about the particular content of the message but I think what’s kind of interesting about a lot of the head-to-head matchups between messages that make partisan advocacy appeals or make ethnic group appeals or make general civic duty appeals is that they all work about the same and if anything the civic duty appeal maybe works the best because it’s not immediately challenging the the listener great so now we’re going to take a few questions from the audience great basic questions how does early voting change the manner in which campaigns do outreach for persuasion mobilization early voting is a real boon to any campaign that has a mobilization strategy why because they can allocate their resources much more efficiently over a number of weeks as opposed to all the final 72 hours it’s true that campaigns often even even in places with traditional electrical systems start weeks in advance just to be able to cover as much of the voter file as possible but they can do so much more in a situation where they have early voting because in it’s typical for people who are very very charged up about voting to vote almost immediately at the first opportunity this is especially true for example in states like Oregon Colorado that have all by mail voting typically half the people who are you know very high propensity voters will vote immediately upon receiving the ballot and when they do the registrar notifies the the you know the other part parties by by communicating public lists and that allows the parties describe those people off the target list so it makes makes their they’re canvassing resources go much much farther so that’s good but that in addition to the extent that early voting involves say citywide voting centers that are open on weekends you get out from under the problem of getting people to vote on a tuesday which is ordinarily working day so you can in principle orchestrate much more effective mobilization strategies around that using festivals or some other gimmick or nothing at all in that proved to be rather effective for the Obama campaign apparently in 2012 getting people from churches to go vote you know it’s a kind of routine social matter that said i would say that in a non battleground state where there is a whole lot of mobilization don’t expect the the simple institutional fact of early voting to make much difference I think it’s been kind of a disappointment to reformers because it hasn’t really changed the composition of the electorate as much as hoped because it’s essentially a a way for mobilizing agents of parties and campaigns to leverage their resources more effectively not necessarily something that would dramatically change the cost-benefit calculation of individual photos the question is have they been get out the vote attempts to that focus on the homeless and other people who are gues you said call the political dropouts but really it sounds as though they’re just marginalized people of any description i think the answer is typically know typically no because for well for two reasons three reasons one is in places that have voter registration systems that require some advance planning typically those systems require some kind of documentation and almost people tend not to have it another another issue is of course you know you’re not sure about the partisan leadings of those people and then very often campaigns think well I’d rather go with a known quantity I’d rather go with the person who has higher vote vote propensity than such a person because i have to allocate my resources efficiently and rather than go to that person maybe I’ll go to a college campus and find a 18 year old who’s likely devoted mobilize but not likely to vote otherwise i would say the closest thing to that kind of line of research is the research on felon disenfranchisement and the extent to which get-out-the-vote efforts directed them increase turnout and the answer seems to be the effects are small faction got that were associated with this follow-up on that just a couple dollars so which were searching for it which one to my left for that yeah there was a there was a close interest research in quotations the there was an experiment but you know no real outcome measure so the so the pair of researchers David brockman at Stanford and Josh Calla at Berkeley did a follow-up experiment with the same group this time in Miami and they conducted a randomized trial using the same general approach which was intensive canvassing and they they reported their results in science i think back in May and interestingly they were canvassing not on the issue of gay marriage because marriage it no longer become an issue but rather on the transgender equality and they did find effects so it’s kind of a kind of interesting postscript to that mess get three months so the results from the baseline hold my question what did he do but you always let me know that’s a really good question so the question is what does matter in campaigns and I really don’t want to seem as though i’m arguing that nothing matters what I’m suggesting is that you know if you’ve got high quality and high quantity that’s going to matter i’m in quite often that’s what you do have in an energized campaign you’re referring to the brakes at vote that was a case where there really was quite a lot of discussion in social networks and otherwise about this issue and so you could imagine cascades of of influence going in all directions and because it was so heavily covered i can’t imagine that it that all that communication had no effect but whether it was the kind of effect that can be picked up in polls is a different story right because the polling industry not just in the US or the UK but worldwide this is kind of in disarray you know in the old days it was about random sampling and now it’s I mean the non statistical uncertainty overwhelms this disco uncertainty because cell phones and everything else not massive rates of non-response make it very difficult to 22 guess properly as to who’s going to vote and and how they’re going to vote so in the United States we often have the happy accident that are poles for reasons that nobody understands not even the pollsters are accurate because the biases associate with the polls are correlated with the biases associated with turnout but in the UK where you had famously inaccurate polls going back to the nineteen nineties there doesn’t seem to be that happy accident so maybe you shouldn’t be so surprised what water well i would say you know with respect to voting days that are non-traditional days of the week remember that Louisiana for example typically in its own elections both on saturday it’s not known as you know a spectacularly high turnout state even under those conditions and you can understand why that might be that if you just have the same old electrical system but on a different day you know you haven’t really fundamentally altered the way in which people relate to elections and I’m not sure that people do or don’t vote so much based on the cost of elections that was that was after all that patha sins that was tested when there were so many changes to make voting easier to make you register later to make you vote by mail there are lots of things that would make building more convenient that didn’t have a profound effect on on turnout with respect to these broader institutional questions you know it’s it’s hard to say it’s hard to say based on the kind of rigorous evidence-based you know foundation you can I think you can speculate but very often when when people who have waited on on questions have speculated about big institutional questions that kind of gotten wrong apparently when countries have adopted the you know the the method that was recommended so you know New Zealand being a kind of classic case of adopting a new electoral system and not seeing a surgeon in turn out so i think we’re going to do two more questions one from the web and one from the room so this mandatory believe actually um was i think aloud initially into of the state’s constitution but it was never actually enacted in one of the more interesting things about mandatory voting is we have other things other specific things that are mandatory in the case of jury duty for example mandatory but we know they got a path dependent thing we made jury duty mandatory minute but not voting and now of course it’s it’s highly unlikely we’ll ever see a change but one of the interesting things that happen for example in Australia is they that both parties agreed to have mandatory voting so they could take you off the table the mobilizing aspects of campaign craft it was a little bit like a an arm’s negotiation I really can’t see that happening now because you know there seems to be a bit of an asymmetry between the parties in terms of what they think they would get out of having more people vote I would say however that political scientists are very very divided on the question of whether if everybody voted the election outcomes would be different and they typically conclude not very much not not as much as people might imagine when they think of a kind of progressive tilt to the people who are not voting routinely ok last question very and got it that’s the question that is a very big question it’sit’s question has many many levels do they feel beholden to do an office what they say there’s so many famous examples of them doing or not doing it for example george w bush declares that he’s not going to engage in that nation building in that I think the second debate with alcohol and that that you know it’s certainly led to surprising results but there are many instances where where candidates do for shadow what they’re going to do and they actually take their their election has a mandate so it’s a it’s kind of a mixed bag i would say that I did the very often the issue is can’t campaigns have a huge incentive to promise things to their base to energize them to mobilize them get them to work even when they know that there’s really very little hope especially in a in a divided Congress to get anything like that look what they’re dreaming about past and so it seems as though they’re really trafficking on is sort of a vision that will be the beginning of a evidence of negotiating agreement but you don’t say that right now it would be be very difficult to to say what what say the the Trump administration would do based on on the the specific policy position that he’s laid out either because he’s been vague or because they wouldn’t actually be tenable as his policy proposals in Congress with Hillary Clinton it you know because it’s pretty much kin to the status quo under the Obama administration maybe there wouldn’t be that big a change but many of her proposals probably are not actionable in Congress given its current configuration so in a way both of the candidates have an incentive to say things that would be appealing knowing that if they don’t really have to answer to these promises down the road because they’ll be blocked by there moments I will say some of these work to move the White House and working on a number of campaigns that the argument that the office holder made of particular commitment during a campaign is often a trump card internal policy conversations that people are really very reluctant to suggest to somebody that they go back on the campaign much so that I think it does have a social norm in that sense and i’m gonna give you the last word are their final thoughts final thoughts i would say you here you are at MIT home to you know great departments in all the social sciences and you know I think that taking some time out for coursework that would expose you to not just the big picture but very often the little picture like how do you know whether something actually works that is a kind of i like the big picture love the big picture but i would say trying to understand how you’d sort out good evidence from bad it’s probably a worthwhile venture at least once in one class you to thank you we’re still

Add a Comment

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