Moral Reframing: The Key to Political Persuasion | Robb Willer

[00:00:04]so as part of the the talk this first talk we have rob willard to speak to us rob is three times a professor at stanford professor of sociology psychology and organizational behavior i don't know how like sociology and business kind of maybe fight it out a little bit but in my mind yes yeah that's right rob is a leader in the field of understanding polarization and social change and his work has really seemed to done me impossible so many of us study how to kind of increase respect across divides but rob has actually managed to persuade people across divides which is something that not many of us can say that we have done and so it's going to be great to hear about his research today i think about 40 50 minutes and then i'll leave it open for a question and answer and rob says he wants uh aggressive intellectual questions at least he told me personally and i'm going to tell everybody that so the more we can challenge rob i think the the more excited and happy he'll be to be here so without further ado uh thanks for coming rob and you can take it away thanks kurt i appreciate it yes yes please very aggressive uh unfair questions uh you know ad hominem attacks all of it please um let's see if i can get this slideshow going um oh yeah okay so uh it's a pleasure to be meeting with you all today uh big fan of the center big fan of kurt's lab which is one of the best moral and political psychology labs in the world and i see a lot of familiar names here uh in the in in this uh session so yeah so it's a lot of fun hello to everybody i i know and hi to people i don't um so my talk today is going to be about a technique of political persuasion that we call moral reframing which we've been working on uh for about a decade and what i want to try to do is lay out the basic logic of moral reframing and sort of tote up the evidence for it and identify limitations boundary conditions and so on and and then take your questions um you know because we're going to be talking about political persuasion i just want to lay the groundwork here of like why would we study political persuasion why does it matter uh well political persuasion is a major area of academic research a multi-million dollar industry and a critical mechanism of social change so hundreds of political strategists analysts in the public and private sector hundreds of academics hundreds of activists organizers spend spend months years of their lives millions of dollars trying to figure out how to change people's minds on political issues trying to win elections primaries ballot initiatives with a lot of the fate of governments and the trajectory of cultures and societies hanging to a great extent in the balance uh so you know if you've got you know some sort of social social change initiative you'd like to see happen in the world you're in the business of political persuasion whether you want to be or not and uh despite that being you know despite the fact that political persuasion is this major area of political activity i mean really like an industry unto itself in in america and beyond uh nonetheless persuasion is really really hard and there's very good reason to think that the yield on the vast majority of this activity is really really low or possibly you know null so for example there's this famous meta-analysis by david brockman and josh kala that showed that the majority of campaign embedded persuasion appeals you know in the form of you know especially ads uh fail to move the needle and have no detectable effects even in very well-powered uh designs you know that that should capture them they don't increase support for candidates they don't change the votes that the candidates get in the in the subsequent elections um and there's good reason to think that persuasion would be especially hard in the u.s context in particular why is that well for one the united states features really you know notable levels of cultural and demographic diversity both thinking historically and cross-culturally uh so relative to other countries you could make an argument that the us is the most or one of the the most culturally and demographically demographically diverse countries in the world uh and as culturally and demographically diverse as it has ever been uh which makes persuasion very difficult because you're trying to cross identity blinds group lines uh that are often very hard to cross worse yet the united states has really unusually high levels of polarization uh with you know strong partisan and ideological camps that don't want to agree with one another and have already staked out positions on most issues that you might want to persuade on and the u.s here again really sticks out internationally and in its own history in terms of its levels of polarization so there's these big group boundaries you have to cross and then something that's i think very silly this week with you know joe manchin's uh announcement that that he won't he won't be voting for for much of anything uh is the united states has a pretty sclerotic system of state and federal government that makes it very very hard to pass any significant legislation you look at the last you know 13 years of you know u.s federal government there's like one major piece of legislation that was passed arguably maybe two you know now you want to define that so as a result you know if you want to try to pass something on the basis of widespread popular support which makes sense in a democracy in the u.s system you've got to hit like 60 plus percent to really make a change initiative go through so uh high bar structurally very difficult to do and so it's perhaps no surprise that we find that most persuasive appeals fail so what could be an effective approach to persuasion you know what could we do how could we harness the academic literature to try to do this at least a little bit better um well one thing that we were thinking about about a decade ago when we started puzzling over techniques of political persuasion in my lab is that maybe the tight link that's been identified between political views and moral views in the u.s and beyond offers some raw material for forging higher levels of political consensus across you know political divides so the idea being that if you know that somebody has deeply held moral beliefs that are linked to their political views uh then that gives you something you could try to connect to in making a political you know a persuasive appeal to them on some political issue uh and the technique that we've done the most research on is moral reframing so this we define as reframing a political position to be consistent with an audience's moral values and the basic logic of this technique is is kind of cognitive dissonance based it's basically if you have deeply held moral commitments if your moral commitments are your uh as kurt would say deepest deep most deeply held beliefs is this is the coinage uh then assuming that you have political beliefs that are not as deeply held then we could go to you and say the actual association between your most deeply held beliefs and political position x is this like if you are gonna you know care as much as you do about patriotism then actually you ought to support same-sex marriage rather than oppose it as you do now and if you can make a credible enough argument that they think makes sense they're confronted with dissonance where they're like oh okay so patriotic people support same-sex marriage but i don't i need to move one of these things or i should change the strength of one of these beliefs in order to restore cognitive consistency i got to get in line here because apparently if you're patriotic you support same-sex marriage which one is easier to move it's easier for me to move my view of same-sex marriage which i don't feel as deeply about as i do about my levels of patriotism which is one of my most deeply held beliefs so this is kind of the basic logic is you present people with an argument that they ought to hold political position x given that their most deeply held moral belief is is why so in order to implement this and start to test this idea we're going to need some kind of a typology of moral values or those deeply held beliefs and uh there's a variety of things out there that you know lend themselves to this there's you know schweiter's work on cultural ethics there's you know schwartz values typology we've used the moral foundation moral foundations theory which outlines five major domains of moral reasoning or moral you know moral thought i guess or moral judgment uh but you know it's not the only way you could approach this it's just the one we've we've used and it's been it's been helpful in our research so i'll lay out what this is really quickly so moral foundations theory is associated with jonathan height and jesse graham and it argues that there are these five moral foundations uh that people think through that people you know hold to different degrees and cultures emphasize to different degrees and i'll try to go through each one here and not assume people would know this theory because why would you uh so here are these five moral foundations so one is equality and fairness so people who would be very high in this moral foundation would be would uh be you know very concerned with whether a policy or an action they observe or a candidate promotes equality and and the fair distribution of rewards and punishments in society care and harm this is to this captures uh especially empathy and compassion based concerns about vulnerable groups harm being done to vulnerable groups and care and protection being extended uh to people who need it um and somebody who was high in this would be thinking to what extent does this business policy or this candidate or this action i've observed reflect showing that uh care and protecting people from harm in group loyalty this moral foundation is to what extent are groups that i care about being protected uh from attack uh betrayal to what extent are people showing loyalty to groups i care about it's associated with patriotism for example respect for authority so somebody who cares a lot about this would be concerned with uh appropriate levels of deference and respect being shown to authority figures in hierarchies that they consider legitimate and then finally purity and sanctity to what extent are are sacred pure things being protected from some kind of desecration or impurity and this this is the moral foundation that's most associated with religiosity so these theorists uh clump these together into two different categories the individualizing moral foundations which are these these first two and then the binding moral foundations which are these three with the argument being that these binding moral foundations are follow more of a a group based logic a protection of groups whereas the calculation of morality with these individualizing foundations is more at the individual level of whether individuals are getting their fair rewards whether they're being protected from harm i i we could have a debate about whether that is an appropriate categorization but it is helpful uh in part for the following reason which is that research shows pretty consistently that these two moral foundations tend to be endorsed more by liberals than by conservatives and conservatives tend to endorse these moral foundations more than liberals do and this should probably you know strike you as relatively intuitive like you hear more talk about equality and inequality and social justice amongst liberals for example you hear more talk about patriotism uh and purity and sanctity and sacredness amongst conservatives and amongst liberals so this is helpful because this offers us some way to know if i'm targeting a group for political persuasion what do they care most deeply about what are their most deeply held moral values so um okay so here are the two questions i'm going to try to answer in this talk so one does this persuasive technique of moral reframing work and then also do we do it because if we already do it then it's not really that interesting or important to to show that it works because well it is still important to show that it works but i so it's just a second question i think of interest whether or not we are doing this on autopilot already uh because if we are you don't really need to watch this talk you're already doing it right okay so the first study that we did to study this dynamic was actually we conducted this day like 10 years ago and another time and another place when in a time when uh people ran experiments often via craigslist postings to try to recruit participants to studies it's just a mythical land i'll just ask you to imagine and uh this is how we ran this study we were in the study with about 300 participants that we recruited through craigslist postings we would play sort of whack-a-mole with craigslist which didn't like that we were doing these postings and so we would have to put them up over and over and over again until we stopped doing it that way because we became concerned they were upset with us and uh we recruited 300 liberals and conservatives to a study uh where we would show them one of three essays before measuring their attitudes towards environmental protection and the three essays were the following so one of them was a pretty traditional message about the importance of environmental protection that was based on this moral value this moral foundation of harm and care um and that message said things like the dumping of waste and chemicals into our lakes and seas has resulted in the deaths of billions of fish and other sea life the result is barren soil across the world making it harder to produce food resulting in famine and starvation and had these images of you know ecosystems that had been harmed and not protected and it pretty much looked like what you would expect to hear from like the sierra club or something like that we compare that to a very different environmental message that we hadn't thought that we had seen before that was based on the value of value of purity and sanctity and so this message said things like it is important we protect our natural habitats from desecration and pollution keeping our forests drinking water and skies pure is of vital importance chemical particles end up everywhere in our food on our skin and inside our lungs dirty polluted air actually enters our bodies and becomes a part of us it had these pictures of these like kind of gross you know unpure uh impure you know you know city skyline this dump and then this very insane person drinking a dirty glass of water which i would not advise you do uh then we compare it to a third essay which is just designed to be neutrals is an essay about um the history of neckties so um let's see here so here's what we found so this is broken out by liberals and conservatives conservatives are the kind of reddish bar here liberals with the blue bars and this is uh levels of support for environmental protection according to what essay people read so some people read this control essay about neckties some people read the harm essay that we just talked about and some people read the purity essay we just talked about and this is the level of environmental protection support they showed afterwards and so here we see that conservatives this this you know indicates whether a difference between these average values is statistically significant or not and you can see that conservatives reported significantly more concern for environmental protection or support for environmental protection if they had just read this purity essay than if they had read either the harm or control essays which was our prediction was that this would increase conservatives concern for environmental protection and kind of the key differences here are all statistically significant and also of note that purity essay actually reduced the difference between liberals and conservatives support for environmental protection to statistical insignificance so that's noteworthy and i think probably not well not representative of future studies we don't tend to find such large persuasion effects and i have some thoughts on why the environment lends itself to bigger persuasion effects because i don't think i think people sort of misread conservatives views of environmental protection conservatives are not like anti-environment it's more they they generally value environmental protection they just don't care as much about it as liberals do they don't moralize it as liberals do and so they're unwilling to you know suffer perceived economic trade-offs or expansion of government which they don't like in order to do more environmental protection so i don't think yeah there's a reason we don't usually find these uh these strong of effects conservatives are open to persuasion on this topic we also measure people's concern about global warming this is like 10 years ago so these concern levels would assuredly be higher now um or at least be at least a little higher now and what we found was conservatives reported significantly more concern about global warming if they read the purity essay then if they read one of the other two essays uh though this did not reduce the you know robust ideological difference in views of global warming to insignificance but it's noteworthy that the essay didn't even mention global warming that's just like a related environmental issue so this would be like a spillover effect to a a not mentioned environmental issue okay so coming out of this study we had some initial evidence that you can leverage people's moral values to make new political arguments that might persuade them on positions to hold positions they don't currently hold or at least hold them more than they currently hold them but we wanted to show that this was robust to you know different topics different moral values you might connect to so the next topic we studied was same-sex marriage and what we did here was uh this was based on a sample of university alumni 250 or so university alumni and what we did was we we have this method right where we're like okay so same-sex marriage who doesn't support same-sex spirits okay conservatives what are the moral values they hold more than liberals these three moral values okay so can we make uh an argument that makes some sense that you should support same-sex marriage because of purity and sanctity well we couldn't really come up with one what about respect for authority couldn't really come up with one what about in-group loyalty we thought we could come up with that so that's what we did to see if that would uh be persuasive to conservatives so here's what we did we randomly assigned people again there was a control essay which i believe is about the the history of chairs you may find a theme you you want the you like to find an essay that doesn't particularly have an effect on people but nonetheless you know gives them the experience of reading something uh so that's why we're always studying these we're using these control essays that are kind of ridiculously neutral and benign um and then we compared it to a kind of a traditional marriage same-sex marriage argument in terms of fairness so that essay these are just excerpts it said things like gay couples deserve the right to get married the law requires that all citizens be treated equally and have the same opportunity to pursue happiness uh legalizing same-sex marriage is the only fair course of action in a country founded on the principle of equality and then we compared this to a very different argument for same-sex marriage that we hadn't heard it was based on on the you know the moral value of loyalty and and it specifically invoked patriotism loyalty to the country so some excerpts from this that said things like same-sex couples are proud and patriotic americans they share the same basic hopes and desires as all americans like other proud americans gay couples peacefully build lives together buy homes and contribute to the american economy and society and here are the results uh we found that conservatives who were presented with uh the loyalty argument reported significantly more support for same-sex marriage than either of the other two essays um again this is well here the effect size is smaller uh which is this is more typical of what we found in research since then and also of note with something we don't tend to find but we did find here so i want to call it out is we actually found a falloff in support amongst liberals from when they heard you know the argument for same-sex marriage in terms of fairness a value that uniquely resonates with liberals versus group loyalty a value that you know that liberals endorse much less or significantly less so that's noteworthy we i believe have never found this in any other study this sort of fall off in the group that is not targeted for persuasion um but especially if you were trying to implement the strategy in the world you want to be very mindful of backlash effects in the group that you're trying that you're hoping continues to hold the belief that you're trying to persuade other people to hold if you're just moving some people up and other people come down then it's not it's not a very helpful method um so anyways just wanted to call that out so this is a couple of our you know early studies we did on this and i just wanted to talk a little bit about successful and unsuccessful replications that happened since then um it's gonna linger on this page for a little bit probably the biggest area of research using moral reframing or really the biggest area of research on political persuasion in general is on environmental attitudes and there's been multiple replications of of this technique there including a direct replication of the first study that i showed you that was done by another lab with a with a representative sample of americans um there was a really interesting study on uh universal basic income that was done by some uh some folks here at stanford who showed that framing universal basic income in terms of its ability to provide workers liberty and freedom was resident liberty and freedom are not in the moral foundations but they're i think clearly values that conservatives care a great deal about which uh which made this i think a smart smart study to do uh there's been multiple studies as you might imagine of trying to extend this to support for political candidates and um let's hear one of these was unsuccessful and one of these was successful i have thoughts on why the peterson and siminovitz study was unsuccessful but i'm not going to linger on i'll just report it here we have a recent study on immigration attitudes where we found moral reframing can increase openness to to legal immigration in the u.s we've also studied this you may see a preponderance of like liberal social issues being tested for political persuasion uh we've also studied this on conservative social issues like support for military spending and making english the official language of the united states and found that the technique can work there as well um it's been applied a little bit to behavior to charitable giving but not a lot of studies of behavior we tried to apply it to cover behavioral intentions early in the pandemic to see if you could use moral reframing to get conservatives to do more mask wearing or at least report the intention to do more mask wearing and social distancing this was not successful other people have found patriotism framing of public health behaviors during the pandemic can be can be successful in changing intentions at least i think that our study here looked too obvious it was too much of a direct persuasive appeal and people felt um potentially manipulated in an information space where they're just being assaulted with persuasive appeals but that's post hoc we thought we thought this would work so um and then there's been some really interesting field experiments done using moral reframing uh one was done by the analyst institute in 2016 this is a progressive political research organization in dc and they sent out direct mailings that were uh anti-trump or pro hillary they're left of centers they would they were only testing how to get hillary more votes in 2016.

[00:22:50]and uh they used moral reframing or didn't in these direct mailings and found a three percent increase in conservatives subsequent voting for or reported voting for hillary clinton uh relative to donald trump in a set of ten or eight swing states and they they actually have strict confidentiality rules around their results that's that's actually all i'm allowed to say about that study i have like permission to say three sentences about this study uh they also did a study in 2020 testing anti-trump tv ads and their effects were null uh for moral reframing although other research showed that all tv ads well another study suggested that anti-trump appeals in general were unsuccessful at that stage because views of trump had been so crystallized after five years of tons and tons of trump exposure of you know arguments on both sides um and then finally there's a really interesting field experiment that david brockman and the ubiquitous david brockman and josh calla did uh just i think about a year or two ago where they trained door-to-door canvassers to use moral reframing in contacting people about abortion uh and trying to persuade people on their abortion attitudes and found some you know pretty promising effects here of uh moving people in door-to-door contacts using moral reframing relative to not relative to this kind of standard door-to-door canvassing standard of care uh canvassing approaches and with the biggest biggest effects being like intentions to take action to support for uh support like a pro-choice ballot initiative or referendum uh but also significant effects on on views of planned parenthood and there's some some evidence for persistence of these over time okay so i want to tell you about one other study that we've done along these lines um but maybe i'll stop for one second just is there any questions any clarifying questions about what i've been talking about um i'd like to ask yeah um so a lot of the stuff you've been talking about seems to be like geared towards persuading people to adopt support for a certain position so like you were just talking about like pro-choice attitudes um does it only move like does your research and like what you reviewed suggest that like things only move in that direction or do they also move in the opposite direction where like you can convince people to now believe who used to be pro-choice that actually abortion's always wrong or something like that you mean and is is are you saying is it effective for moving people towards conservative views or just moving people down in levels of support for something like i'm not sure yeah if it's just about like moving to towards the conservative side but like a lot of these are about getting someone to agree with something they didn't used to agree with what does it like can you also get someone to disagree with something they used to believe or something like that yeah i mean it's a couple different answers to that so a lot of these effects are on populations where they they could have been considered to have had the opposite view before the persuasive appeal so that's like one thing i hear you asking is like can you get people to like change their minds and in general i would say the effect sizes are too small to claim that there's a lot of actual mind changing where people are going from like oppose an issue to support an issue it's the effect sizes on average or smaller than that but there are there is persuasion amongst those people who currently oppose something if that makes sense but small effects and then the other thing i hear in your question is uh can you increase people's opposition to something and the only study i can think of that really tried to do that would be the donald trump field experiment that didn't work although i kind of think for different reasons just that people had made up their minds about trump so it's just hard to move uh so yeah i guess that's a gap and sometimes political projects are about stopping things especially conservative political projects so uh where you want to protect the status quo so it's a it's a gap yeah rob i might as well hop in here since my question do it can uh can go now and sorry if i'm chopping my wife is on an nih panel right now so we're both using a lot of wi-fi oh yeah you seem great yeah so it seems like there's a fair amount of heterogeneity in the in the effects a little bit and one thing i wonder is whether it seems that moral reframing often works but i i wonder if it's kind of like a swiss army knife in some sense where like sometimes it works for different reasons and perhaps the locus of it working is not a value per se but things that sometimes co-vary with the value so in the first example of um of harm and purity the harm was to fishes really and the but the purity example was almost like harm to oneself right so that's true yeah it really highlights the kind of like you're going to drink fluted water and i think everyone is afraid of harming themselves in some sense of drinking polluted water which is maybe why everyone goes up and then in the in the gay marriage the the loyalty is maybe conservative seeing similarity with someone and perhaps the liberals go down because they see gay people as now conservative somehow so it's not saying that it it it definitely seems to work but but it might seem to work for different reasons across the manipulations if let's say there's not consistent you know foundation values or something like that right right yeah yeah i take your point um to the for the environmental example i want to i'd want to look at the essays again to see just how bad that confound is because i i'm not i just need to read them again because i'm not confident that the hard one didn't identify harms that would apply to to the self as well as to ecosystems or whatever um but that yeah that would be that would not be our compound that's present in many of the studies the other point that i mean there's kind of two big mechanism stories here two big reasons why the effect might happen so one is that the argument just makes sense to you because it is persuasive to you because it's in term of in terms of first principles you believe strongly and so you find it convincing you know um if you tried to convince me to go see a band and you were like they're a lot like the pixies and pavement i like pixies and pigs so that makes sense to me and i'm more likely to see the band you know um that that's kind of the one we've ridden with and we have some evidence for it but we don't have like knock your socks off evidence for it because well yeah because perhaps because it's not the only mechanism but also because it's kind of it's pretty hard to study um and then another would be that something about the argument signals that the person is in like a common group with you and you ought to agree with them because they're credible to you so you share a moral community with them or a political community with them or they're conservative like you or they're patriotic like you or they're a republican like you you know or religious like you something like that and you're like oh okay i'm conservative republican slash religious slash patriotic i will agree with this person more because i have something in common with them and we have evidence that that isn't the whole story but i but we don't really have a strong understanding of how much the effect is due to these two causal pathways which i think of as probably the two primary candidate pathways but speaking of you know signaling the source of the message this and speaking of candidates uh let me show you this study and we'll get a little bit more data to chew on in in thinking about this question of why these effects are happening um so this is a hypothetical candidate study where we were interested in whether or not you could uh reliably change levels of support for a political candidate so we showed people a hypothetical 2020 democratic candidate who was ostensibly a challenger to donald trump uh the person was portrayed as having either an economically progressive or moderate platform which was just something of substantive interest at this time um we had four different framing conditions so we had we gave speech excerpts from this candidate where they talked about their platform in different terms so in one of the conditions they talked about their platform in terms of these individualizing foundations that liberals hold more than conservatives in another condition they talked about their platform in terms of these binding foundations that conservatives tend to hold more in another condition which uh was interesting to us because it's a a non-moral way that candidates often talk we had the candidate talk in more like technocratic language i'll show you in a second what that what that means but we wanted to have a persuasive appeal condition that was non-moral so this was our effort to do it and then we have just a no persuasive content condition where you don't read any speech excerpts from the candidate uh and instead you just read some like boring stuff about what the what an election is and uh but in all these conditions you found out that the person was a democrat and what the details of their economic policy platform was and part of what's nice about that theoretically is we should be holding constant the source to a great extent here you know the person is if you're a conservative you know they're not in your identity group you have policy information you have party affiliation you might still make some inferences that they're a supra a relatively conservative progressive democrat but you know there's we've constrained how much you can make those inferences a great deal here um and then we ran this two ways we read on on a mechanical turk sample and we replicated it on the national probability sample uh which we which we recruited through through newark so here here are excerpts from the materials um so this one was the individualizing foundation and i'll just i'll just speed through this i've highlighted these were these these kind of buzz words that especially were meant to signify that this person is speaking in terms of the individualizing foundations we're not bolded in the original piece but you can see here this candidate would talk about economic justice fairness compassion uh stopping corporations from exploiting people discrimination economic justice elsewhere they talked about social justice so this person's talking in terms of these more uh individualizing moral foundations we compared that candidate to a candidate who spoke very differently about their progressive or moderate economic platform this person emphasized like loyalty and respect america loyalty to country freedom the american dream family and community patriotism sacredness degradation dignity etc america america america so uh lots of invocation of binding and moral foundations this is the technocratic the aforementioned technocratic candidate and this is a way that candidates talk especially democrats we actually base this on an essay or an op-ed that tim kaine wrote and um you know it emphasizes like growth productivity you know gdp per capita good paying jobs small businesses prosperity jobs jobs jobs et cetera et cetera and uh so here are the overall results this is just collapsing across the whole sample liberals conservatives moderates everybody and you can see here that the two better performing candidates were the ones that talked in terms of the binding moral foundations consistent with moral reframing and then the candidate who spoke in terms of the the technocratic language which we did not expect we thought that the binding moral foundations would outperform the technocratic language uh but we were wrong about that um and then breaking this out which is i think more interesting bring this out by the ideology of the people in the study looking at liberals moderates and conservatives separately we see that all the action is amongst the moderates and conservatives so the liberals are like democrat with an economic policy agenda that i like i'm i'm going to vote for this person against donald trump regardless of the you know how this person's talking you see a little bit of a falloff but it's not significant when they talk in terms of these more like conservative values if you will um we never really find effects amongst moderates you know we usually only find effects amongst conservatives because we're targeting conservative values here so this is kind of noteworthy and interesting to think about that that we found pretty pretty robust effects that replicated in the nationally representative sample i should say that two samples have all the same interaction effects are pretty strikingly similar results and yeah so you see here for moderates and conservatives that they're the subsets of the sample where the binding and technocratic uh frames are are more effective for driving candidate support and then you also see that the binding frame is actually outperforms the technocratic frame amongst conservatives which makes sense because the binding frame is really targeting conservatives morality uh so they're really being sort of micro targeted here um let's see here oh this is one thing that we included that we thought was interesting um more fodder for future research is uh the extent to which people's let's see if i can articulate this right the extent to which people's support for this democratic candidate was correlated with their racial resentment their levels of racial resentment which is a correlation that uh like people's presidential support or presidential preferences have become more correlated with their racial resentment in the united states or at least whites have over the last five years basically since donald trump showed up and uh and so this is just among whites we were interested in would one of these frames reverse that a bit and bring that association down would it reduce the role if you will of racial resentment in uh deciding how much you support in this case a challenger to donald trump and what we found was that the technical and binding frame conditions both reduce the role of racial resentment in uh people's candidate support and this is i think that for a number of reasons this is really just an exploratory finding but i think it's it's interesting my my sort of interpretation of this would be that here you're giving uh policies that benefit most people and moral frames that are persuasive to conservatives who tend to be higher in racial resentment and moderates who are higher than liberals and you're crowding out the role of racial resentment in their in their candidate choice um you're giving them another reason to to base their candidate support on or two other reasons okay so um one thing we get often with this research is it's very intuitive especially after you've heard it uh like why why wouldn't you think to articulate an argument in terms of the things that the person you're speaking to believes the most in or values the most for example if you're trying to sell a used car to someone trying to sell them your car you're not going to talk to them about what's in it for you you know and how excited you are to get their money and all the cool things you're going to spend it on you would attend to their perspective you would think like what is good about the car and why is it a good deal for them um and you know but our argument is that this is not actually the way people approach you know persuasion the political domain like our first thought is to render our own reasons for holding the political positions we do and we really don't intuitively think about other people's perspectives um but as for whether this idea is intuitive or not i found this interesting quote from blaise pascal a couple years ago where he said it is necessary to have regard to the person whom we wish to persuade what principles he acknowledges and then observe in the thing in question what affinity it has with the acknowledged principles which is just like a more elegant and also more general articulation of the idea that i've been talking to you about today which is just like the plight of the social scientists do you think you have a new idea and then it turns out that a french guy with terrific hair had that idea you know half a millennium ago um but you know just while in the abstract the idea might seem really obvious it's not very obvious to people when they're working on autopilot and making political arguments uh so for example we recruited a sample of liberals to make pro-same-sex marriage arguments we even said we'll pay you if you're effective and you persuade a conservative with your argument and they overwhelmingly used individualizing moral foundations not binding moral foundations uh that our research suggests would be more persuasive uh conservatives same pattern when we recruited a lot of conservatives to make persuasive arguments for why english should be the official language the united states they make arguments in terms of binding moral values not individualizing moral values we coded environmental rhetoric from op-eds and major american newspapers and public service announcements you know that most of that rhetoric is in the harm care moral domain some in the fairness domain very little in these uh binding moral foundations very little rhetoric of that sort presumably because it's mostly liberals making pro-environmental protection arguments and they're using their own moral values um this is a this is a possibly overly complex analysis to render in the next few minutes but this is a graph we we uh analyze the moral word usage of every political every presidential and vice presidential candidate since 2000 uh from every debate that they did and grab them in terms of their economic ideology which is like pretty representative of their general political ideology measured via uh inferences of political ideology inferred from uh their donor base which is you know publicly available and all donors have been assigned political ideology numbers based on which congress people they support the most and so you can sort of backward you can reverse uh this to figure out political infer people's candidates political ideology even if they've never been in congress based on shared donors so that's the x-axis and then the y-axis is the ratio of binding moral words used in debates relative to individualizing moral words used i mean one thing that's just kind of interesting here is that republicans you know consistently use more of these ostensibly conservative moral values than liberals which they ought to if that theory is right um but the thing that was interesting for our purposes is that there's just like not a lot of folks over here in this zone which our research would suggest would be very a very effective zone where you use binding moral foundations to articulate a a more progressive economic policy progressive economic policy is really really popular um and then binding world foundations can add support if you're a democrat they wouldn't if you're if you were a republican uh but this would be the zone with like you know john edwards uh dennis kucinich if you remember him eric swalwell a lot of sort of economically populous progre economically progressive democrats um but it's not super represented even though it's the one that our research would suggest would be would be really effective um one thing uh yeah so this has also been a technique that's been widely applied for example uh just one example that sticks out because it was very on the nose application was uh elizabeth warren had a had a job kind of basically a green new deal plan that she rolled out during her campaign which she called the plan for economic patriotism and she talked about it in terms of more conservative values and this was very much a play to get white working-class voters who were probably conservative but potentially democrats sort of former union upper midwest uh demographic she wanted to build support in that group so she used royal reframing um and i think the various you know positive effects we found in prior research and some of the applications have led people to describe it as sort of like one quick trick for political persuasion this is like literally described as a as a simple psychological trick for political persuasion um this is not i'm not is it that it is not that i wouldn't say and i'll try to explain uh a better way to think about it and i'm not throwing shade at olga kazan is a very good science writer and probably didn't write this headline so let's talk about limitations and boundary conditions so one is you know the research is pretty clear that reframe messages can also be unpersuasive there are examples of this the clearest example is that very very short messages based on reframing don't work or i haven't seen any evidence that they work in a well-powered study one of the earliest tests of this was by anyway day at all in 2014 they did just like one sentence statements of like supporting abortion as an issue is a matter of patriotism or something like that and that is not enough to change people's thoughts about the association between their values and their politics really that's my interpretation of why these short messages don't work and so it's a difficulty of if you're going to make a credible argument for a new kind of consistency between moral values and a specific political position you need some real estate you're going to need at least a couple or three paragraphs and a good argument to convince anybody that those things are consistent you can't just say it that doesn't convince anybody but then when you do a couple paragraphs of persuasion you inevitably wind up creating some confounds in your in your treatment like kurt pointed out with the environmental study which may have emphasized effects on the self more in the purity condition than the harm condition even if you take that out there's going to be other compounds there but if you try to make an argument for why the environment's a matter of purity in one sentence that's tightly controlled you just you're not going to have any effect at all so this is the difficulty uh also i think that reframe we haven't actually seen a lot of examples of this except maybe the same-sex marriage study i showed you but reframing backfires i expect if it offends base value so you could get backlash effects amongst the group that already holds the political position if you make a reframed argument that that actually offends their value so what would be an example of this one thing if you look at this whole literature on reframing one thing you notice is that um people are researchers are constructing these persuasive appeals in a way to not elicit backlash in the base so like you could make a group loyalty based argument for same-sex marriage that's super offensive that's in turn or two liberals moral sensibilities that's about america being the most important country in the world being clearly better than everybody else like a hyper-nationalist or ethno-nationalist argument that i don't know how you connect that to same-sex marriage but say you found some way to do it it would turn off liberals you know and we didn't do it that way even though that is within the loyalty foundation so this is important to articulate as a boundary condition because i haven't yet is that there's a way to do a loyalty argument for same-sex marriage that would be effective and there are ways to do it that would be ineffective or even could elicit backlash effects and and so that should all be fleshed out better but this is my going read on it is don't offend base values if you don't want backlash effects james we'll take take your questions uh yeah that ties in really quickly with something i was thinking earlier do you know how maybe this isn't the right word but like how sticky these changes are like if you if you manage to change someone's uh mind on an issue does it stay do they keep thinking that or do they revert back to some base value after a given amount of time without like a reinforcement because i could see it being effective if it is not very sticky to hammer this for a short period of time and reinforce it enough in people who are um holding the uh the value that like uh same-sex marriage should not be supported you hammer it in enough that it changes them to hold a softer position on that but then the the stickiness will bring back the people you've lost yeah especially if they go back to their political community and hear other arguments the other way right yeah i think it's i i love the question it's terrific question it's a great point i think there are three stu it's basically just the field of experiments that have tested for durability effects and for the most part those experiments also had uh so two out of the three of those did show durable persuasive effects the one of the analysts institute studies not the other one and then the candace in study and notably uh one of the two that showed durable effects the better powered one which was the canvassing study also had easily the most impactful intervention of any of these so i wouldn't expect i don't know maybe with a ton of power you might find that reading a three paragraph essay could have a persistent effect like a week later but i wouldn't expect it like three months later to have an effect but uh but a door-to-door canvassing interaction you know where somebody talks through it with you and makes a credible argument re-articulates it a few times i'm not surprised that they did find you know durable effects for that kind of a more meaningful encounter jonathan yeah uh rob i think you definitely just now got to part of what my question was going to be which was um you know was there ever i guess in your like you know in the literature or in some of the older projects you worked on um was there like more consideration at least into um who the um you know person was uh giving them this message for example a family member or a friend um i know that you know those could be potential other factors that might um for example affect the durability or um affect just how long this um is going i know that you mentioned that you know the longer you get the more compounds right um but those are also potential variables like look at that and i was wondering if the literature kind of pointed towards that or was getting there to that point no it's you know it's hard to do it's a great question and it is a path to application um in fact this is uh pure contacting has emerged as like a major political persuasion technique and in general in early field experiments is shown to be more effective than a stranger texting you right um so i think there's a lot of potential there i think the challenge is training people up you know to do this effectively but right now there's like basically no research on on your question which i think is a good question the closest would be that candidate support study i guess where you know some stuff about the candidate and so you know it's along the lines of what of what you're asking um but it is different let me just wrap this up and then we'll take these questions um oh yeah so another limitation small effect sizes if you don't see if you see large effect sizes with political persuasion research you should be suspicious um and and you don't generally see them here and it's much more often tested on liberal issues and candidates than conservative ones which might make you worried about whether the theory is really the theoretical ideas behind it are really robust uh so just to reflect on the motivating questions does the persuasive technique work and i think the answer is yes it can um but it can also not work so there are necessary conditions for working and then do we do it they're more confident that the evidence is pretty clear that we don't do it um so i'll speak to these last things and just not not show you that and just acknowledge my collaborators on this research and and some of the folks in my lab that have contributed to this and thanks for your time and attention and we'll do we'll do more questions do i feel them myself if you feel comfortable then happy to have you do it okay thanks kurt uh so i saw chase's hand next yeah thank you so much for that presentation professor uh my question is is sort of a two-part question you talked a lot in the beginning about how there's just a lot of diversity in america and that cause and a lot of political polarization which changes things um i just personally um have learned that uh american politics i've just i've heard is very much kind of skewed towards the the center or center right it compared to a lot of other countries i know like for example in a lot of countries in south america or europe there's everything from on the right like christian nationalist parties and on the left like communist and anarchist parties which would be sort of unthinkable in america on both sides and i just so that in this two-part thing i just want to know when translating that to morality because you know uh you said that you know political beliefs often translate to morality does that mean that americans just sort of moral oppositions are actually a lot closer to each other compared to a lot of other countries and you know accordingly does that change like how moral reframing works does it make it easier for countries with political systems like america does make it easier for people to have these moral slash political shifts anything like that yeah i think it's a really interesting question and you know there is cross-national research on just you know levels of endorsement of these moral foundations and i i might i don't remember in enough detail to give you a good answer what it shows my thought was that it showed pretty consistent left-right differences in levels of endorsement of moral foundations across cultures but you can't even administer the questionnaire in some cultures because it's you know it's like it's in english and it's somewhat it's got some culturally specific content um i guess one thought i have is that i i mean i'm sort of answering a part of your question of like is there more moral polarization in the us than elsewhere and i my instinct would be to say yes just because there's such a strong religious community in the us there's such a strong like evangelical community in the u.s and then there's like a hyper educated very very progressive community in the us on the far left and uh and these groups go like really far away from each other and you see these groups you know in in other other societies of course in the west especially but not as strong we just have more religiosity in the us um but i didn't really do your question justice i just i just chewed on a part of it but yeah i'd like to know better i'd like to get a better answer to it thanks chase i saw morgan's hand next hi thanks so much for the talk is fantastic um i actually have two questions that are unrelated to one another i suppose um the first question is methodological um have you or have you considered doing some sort of linguistic analysis um opening having some open-ended questions after you ask you know whether you would support this policy and saying you know here's maybe some underlying mechanisms or here's how their arguments change based upon the argument that they're presented yes yeah yeah that's very cool we've not done that um but that's that's a very good idea it also would be interesting just to have people write um about their new view you know about their view after seeing the appeal because i would bet that that would also help with persuasion um because they would sort of in the moment of persuasion you know exercise the new view a few times and that's actually probably part of why the canvassing works is because part of it is you sort of talk with them about their view their views getting updated in the interaction and they're telling you about it and and research prior research shows that articulating something can kind of make it land more deeply cognitively um yeah so that's a good idea partly as a persuasion catalyst yeah definitely some more deliberative type processing right right and so my second question is um in a second world of mine um i am working with some intel analysts for the military and part of what they're doing is open source information analysis including you know twitter data for example do you think that you could almost reverse engineer this sort of persuasion technique and identify others attempting to persuade you or persuade a specific community based on the moral framing for example that they use yeah i think you probably could do that um i i think that so a challenge is that text analysis around the moral foundations is is really weak right now so i you know like i've showed you this this graph where we uh looked at word usage and whether which moral foundation it reflected um and we the only reason we're powered enough to even find an association there is because we looked at a huge amount of text uh between people that are very ideologically different and we uh collapsed across all the different moral foundations um into binding and individualizing but in general use of this dictionary of words associated with the moral uh foundations it it shows you don't find very robust associations like conservatives don't really clearly use the binding foundations more than liberals and and so on um so we need better text analytic techniques and right now honestly if i was to do one thing i would just do trained human coding still because it works on smaller chunks of text you know then then then quantitative text analytics techniques would um so that's that's still i think the best way to do it but hopefully we'll beat that at some point yeah i think there's an article that came out this year that showed i think it was text the embedded nlp technique was a little bit better than the bag of words dictionary-based type technique for language analysis so yeah the only problem with trained human coders is just they can't they can only do so much text you know but like most of the techniques that you would do would involve some kind of training ai to do what the humans did you know and so you couldn't really do better than the humans you're just trying to approximate the humans so that would be the next thing i would do yeah right thank you yeah thank you morgan all right maybe two more questions and then we'll let rob go to things and everyone else who didn't get asked can write rob um aggressive emails since it's in private very aggressive in front of other people i won't understand them if they're not aggressive yeah uh so i saw sarah's hand next hi so i don't know how well this question is going to come out because i'm still kind of like working through it in my head but beyond just the question of does the persuasive technique of moral refaming work is the goal of moral reasoning to teach how to best extract the truth from the lie or extract the right way from the wrong way or how to best beat the other side out because both believe what they believe is the truth and what the other side believes is the lie or the wrong way so in this like battle between weighing the morality of one side and the morality of the other side like what is the essential truth that we're trying to work towards that will unite both sides because if you're just trying to beat out the other side it's like how do you know that you're right just because you have the best like strategy to the other side conform confirm yeah conform yeah yeah it's an excellent question and i guess i so this is full on in relativist subjective moral land you know where you could use moral reframing to bring people closer to uh political opinions that are more in line with evidence or further from them you also could use it on just sort of really pretty purely morally subjective topics that you know where evidence isn't really the primary basis of somebody's position i would say most of our application is to things where if you have a certain moral value you'll tend to have a certain political position regardless of your read of the facts of the matter um so we're like pretty fully in like subjective land here and i guess this is this offers me a good opportunity to put down put note the caveat that there's there's nothing intrinsically morally virtuous about this technique um unlike say you know anyway there's other other ways of bridging political divides are i think you could argue they're sort of intrinsically virtuous because they reduce political conflict or you know social conflicts this is not really like that you could use moral reframing for very bad ends you could use it for what i would consider more virtuous ends uh and history is replete with examples of moral reframing being used to you know to do terrible things from the loyalty and you know purity based arguments of the third reich or you know fascists in italy in the 1930s to purity and patriotism arguments used by the khmer rouge you know like you can find all sorts of uh terrible examples of moral reframing being used to do really like the worst things in human history but also you can see examples of it being used to pass you know political initiatives that you would probably really really care about um so just wanted to highlight that that it's it's it's no more virtuous than the end that it's put to um which is sort of answering your question i have yeah thank you thank you sarah just add on a tiny bit if i can um sarah sorry oh yeah go ahead time for one more question so we'll get one more person uh to get in sorry to cut you off just to for equality and fairness i guess values i do not hold okay sorry i saw francisco next yeah so first things first thank you for for the great talk it was really informative as per my question um so i was thinking about how you interpret the results of the study where you asked liberals and conservatives to provide arguments uh as to convince someone else of something and i was wondering so did you actually provide participants with specific arguments and they had to choose or they were free to to generate their points because well imagining this because people on one side might not really be able or know how to connect with the other side so people only know so much as about their perspectives and you know all about eco chambers i guess so they mostly know how to support one's stance because they interact with like-minded peers so it might be the case that they're simply not providing arguments that are somehow reframing their point because because they don't really know how to not necessarily because they choose not to do it yeah yeah okay so outstanding question i'll try to this has been a whole area of interest to me um so i'll try to answer briefly in deference to the time the so i see like maybe three possibilities here i think you just listed them is like people people might not realize that the morally reframed message on average will be more persuasive they might realize it but be unwilling to do it for sort of moral or political purity reasons you know um and then a third possibility i think is the last one is they might not be able to do it because it's technically it's hard it involves perspective taking you you might it would help to know about this moral foundations theory perhaps um and then uh yeah okay so and then there's multiple reasons people might be unwilling to do it like they actually strategically think it's a bad play or they think from a purity they think it's impure you know like they want to stick to the purity of their ideology they don't even want to exercise views in terms of values they don't hold so we tried to study this with a series of kind of small sort of kind of studies that are not really tight they're just sort of exploratory and interested so one we showed people moral you know like the arguments that we had tested in prior studies and said okay you need to send one of these to say a conservative and we'll pay you if you're persuasive you should try to be persuasive and on average liberals and conservatives were good at seeing that the reframed message would be more persuasive they also so when we ask them which one to be more persuasive they're pretty good at that we asked them which one do you want to send they mostly were willing to do that so in both cases conservatives were like a little better than liberals conservatives are a little better liberals it seemed that the reframe message would be more persuasive and uh and might say better they were a little more willing to send the reframe message than liberals but the majorities for for conservatives and liberals were able to see the refrain messages more persuasive and were willing to send it um and so that leaves us with kind of the last explanation so why don't people spontaneously do it i think because it's hard and they don't have the knowledge base to do it necessarily but then the canvassing study suggests that at least politically engaged motivated and informed people could be trained to do it quite effectively in the field um so that's that's where i think that all all lands yeah perfect thank you yeah thank you all right so let's give rob one more fake zoom hand [Music] and thanks to all of you for asking such wonderful questions and for showing up at this talk today the first of eight so stay tuned next week for another delightful talk i'll be tweeting about it soon and for those of you who are part of the summer seminar you have that link in the syllabus and we will meet you there so thanks everyone have a great week stay safe and we'll see you very soon thanks everybody thanks for the great questions thank you