Michael Macy and others write,
In each experiment, participants were first asked with which party they identify and how strongly. They were then randomly assigned to 10 parallel worlds of uniform size and administered a survey with up to 20 randomly ordered political and cultural statements (see table S2). In 8 of 10 worlds (the influence condition), participants could see which party was more likely to agree with an item, while in the other two worlds (the independence condition), they could not. In the influence condition, the participant’s own agreement was then used, in turn, to update the relative support of each party displayed to the next participant in that same world. Participants only knew about their own world and did not even know that there were other worlds.
Suppose that there is an issue where people can be either pro or con. In one “world,” people come to believe that Democrats are pro. In the experiment, people who identify as Democrats will take the pro side in that world. But suppose that in another “world,” people come to believe that Democrats are con. In that case, people who identify as Democrats will take the con side.
The authors conclude,
Our results suggest that partisan alignments across substantively diverse issues do not necessarily reflect intrinsic preferences but may indicate instead the outcome of cascade dynamics that might have tipped in a different direction due to chance variation in the positions taken by early movers. Public awareness of this counterintuitive possibility has the potential to encourage greater tolerance for alternative opinions.
The first sentence says that people are so purely partisan that their opinions can be socially influenced in almost a random way. That sentence is supported by their experimental results, although as with many experiments of this type there are legitimate reasons to wonder how much the results in the experimental setting carry over to the more complex real world. Also, note that I am very sympathetic to the theory of political psychology held by the authors, because I see it as consistent with my views in The Three Languages of Politics.
Can you think of any real-world examples of opinion cascades, in which Democrats or Republicans who took one side of an issue made a sudden shift to the other side? I sure can.
The second sentence says that perhaps people can be talked out of their partisanship once they realize that the effect of identity on opinion is so strong. That sentence is based on no evidence and is purely speculative. I would bet against it.
As Jeffrey Friedman has pointed out to me, it is obvious that there are differences of opinion on major issues. Yet hardly anyone sees this as evidence to doubt their own opinion. Instead, nearly everyone is sure that their own opinion is unequivocally correct. So I don’t think that pointing out problems of meta-rationality to people is going to get them to change their outlook.
Before any of the tribes can alter a group policy the tribal elites have to go through an internal e mail discussion to make sue all the priors are kept. They inevitably near a zero crossing and appear to violate a prior because of the email delay. Libertarians do not e mail, they always seem blunt.
Here are the 20 questions from the Supplementary Materials:
The only conclusion that I draw from this study is that vague/technical/unanswerable survey questions result in answers based on simple heuristics such as political affiliation. I don’t think this set of questions helps us understand opinion cascades or partisan polarization.
Most people don’t have well-developed opinions–or any opinions at all–about the above statements. I guess most people have an unconscious heuristic “the people I agree with are probably right about things I don’t know much about.” The obvious next step is to agree with them about those things too.
+1
Add another…
“I care about issue X. In order to get what I want on issue X, my coalition has to win power. As part of having people in my coalition support me on X (which some in my coalition don’t care about and some even disagree with), I need follow their lead on other issues that are less important to me or on which I am myself conflicted.”
So for instance, libertarians have long been divided on whether they should support Republicans for their market policies or Democrats for their social policies. And they’ve been divided on just what “libertarian” economic or social policy should be. In a recent thread there was a debate between the liberty gains from DNA registry (fewer individuals having their liberty violated by violent criminals) versus the potential liberty encroachment of the government having that information. Another could be abortion (some libertarians say the fetuses rights are paramount, others the mother).
It seems to me that in all political coalitions we have to prioritize importance of issues, or deal with issues we are conflicted on, and ultimately decide what the best compromise is. If we had parliamentary multi-party system that would just be replaced with party based tradeoff coalition building.
In a lot of ways “what kind of person/people am I supporting” is a lot easier question to answer than “what’s my stance on this issue.” After all, its a long multi-round game called LIFE, and the fate and relation to your most natural coalition partners is important in the long run.
The questions are the way they are as a consequence of the methodology chosen. Specifically, the study uses the famous “Music Lab” study approach, in which people were exposed to mediocre independent-scene music about which they had no strong initial opinion, and then told whether people in their same groups liked or disliked the song, in which case, their opinion tended to change in the direction of the group opinion.
So, as a general observation, we can say that when people don’t already have strong opinions about any questions, they rationally switch from considerations of direct personal interest to those of social / reputational interests, and use the heuristic of conforming to prevailing fashions among high status people in their reference social group.
So, you could ask someone whether they were a “hipster” or “mainstream” not. You could ask them what they thought about beards or beanies or going sockless, but these things are already the subject of strong opinions. So you would have to find things that no one associates with hipsters at all, and so no one has any strong opinions on whether that kind of thing is hip or not. “Are you a hipster?” – “Yeah” – “Say, what do you think about signet rings.” – “Huh? I mean, I never thought about it. No one wears those”. – “What would you say if we told you that all the other hipsters now really like signet rings?” “Well then, I guess now that I think about it, signet rings are pretty hip and not mainstream at all, so, yeah, signet rings, cool man, I’m going to get a bespoke one right away.”
Meanwhile, “Are you mainstream” – “Yeah, very conventional.” – “Ever wear a signet ring?” – “My grandfather had one, so I’m ok with them, but otherwise, no one does that now, so I don’t really care either way.” – “What if we told you that mainstreamers hated signet rings, thought they signalled weirdness, and that dirty hipsters loved them now.” – “Well, damn, those hipsters ruin everything, down with signet rings!”
You might notice that this is how classic bandwagon advertising works too. Shocker! What this study shows is that the general fashion group-conformity instinct applies to certain weakly-felt political opinions that it does to arbitrary trends in dress or musical tastes, in which most people tend to follow the trends in their particular subculture, because those are the people with whom they must fit in, and who they are trying to impress and not stand out as a weirdo – which is a socially fatal condition with all kinds of serious negative personal consequences, and thus a totally rational impulse. But since fashions and conventions are used to distinguish groups from each other and signal belonging and loyalty, new splits look like “polarizations” over matters which were more evenly distributed before. Say, 1% of maintstreamers and hipsters wore signet rings before recognition of the fashion trend, and after that recognition, 0% of mainstreamers and 90% of hipsters wore signet rings.
Of course, the subgroup split existed prior to this ‘polarization’ which is just over one formerly indistinguishable matter.
One question is whether this trick could be used to encourage people towards moderation by, say, convincing progressives that the average progressive is more conservative or vice versa. My impression is that the impact would be highly asymmetric, working on firm conversatives, but not on radicalp progressives.
But see what I mean? The study design means that they reduce the set of potential questions to those confusing and uncontroversial enough to induce a fallback to a fashion-based heuristic response, where group influence matters a lot. Exactly the kind of things that don’t have much impact on people’s normal lives, or which don’t touch on their core ideological commitments.
So, for example, you might ask a conventional office-suit-wearing person what he thinks about neckties, and tell him that all the other COSWP just love neckties. He might answer, “I don’t care what they love, I have a very thick neck, and if I can’t have an open collar, for instance if I have to wear a tie, then I’m going to be really uncomfortable. Down with ties!”
Any question like that is off the table for this kind of study. Should it be?
The reason this is important is that it raises the question of how far this insight extends or can be fairly extrapolated to explain our contemporary political polarizations more broadly. For example, what about issues that are highly controversial because of dispute about matters that significantly affect people’s personal interests. How susceptible to group influence are people then? Which is cause and which is effect? Are they following the group, or does the group exist because of pre-existing commonality of interests?
The danger here is to extrapolate the context of petty and trivial fashion-prone matters too far and to conclude that all or most of out political disagreements derive from substantially the same dynamics, and so to dismiss those disagreements as merely arbitrary and pointless fights mostly fought by dumb sheep following their tribal gang herd (or their herd’s sheep dogs) without being aware of it.
There is a strong temptation for many public intellectuals to see things this way, and/or to chalk it all up to social media, because it creates a kind of hope that a different dynamic might reduce the increasingly intensity and severity of our political polarization. And it provides at least a theoretical possibility of transcending the noise of current debates without having to take sides in genuine conflicts of interest.
I just want to say, handle, that I very much enjoy your comments on this blog!
Agree with David completely- Handle’s commentary here is routinely the best on this site, and in the top 0.0001% I have seen on-line anywhere in the 21 years I have been reading comments online.
So do you think that set of questions are a good reflection of the Music Lab methodology? Are they the ideological equivalent of bland songs given away for free and actively listened to by the study participants?
I don’t think these questions are clearly less interesting relevant than the ones than many that are widely accepted as hot political issues. If the charitable interpretation here is people are incapable of forming their own opinions on things their political tribe hasn’t already made a decree on, its still not a particularly optimistic conclusion.
I agree that they should (have) tester some more overtly political questions of course, but these questions aren’t that technical. Many should call upon the same principles one would deploy in popular political questions. That many people can’t seem to make the connection does undermine the notion of that most people have value or principles-based opinions as opposed to merely following the herd.
Trump’s tariffs come to mind. You see people on the left suddenly more open to “free markets”, just so they can oppose Trump. Bernie Sanders has expressed similar protectionist sentiments. If we had a President Bernie Sanders, and he did the exact same tariffs against China, there’s no doubt in my mind they’d frame it as “look at how the President is standing up and protecting the common man”.
There is some of that going on but we have to remember:
1) Free trade is more open ended policy that can be campaigned on in variety ideas. Obama campaigned more trade restrictions than McCain in 2008 and yet after 2009, he was a free trader. Also Democrats are moving to the urban and suburban voters.
2) There is a degree of once policy is set, the winners and losers feel the changes more. I bet a lot of farmers thought Trump would not hurt them in 2016 with trade wars. (For instance I thought the corn markets would be effected by Trump’s breaking NAFTA in 2016 not soybeans with China.)
3) There is always a degree of exaggerating the impact of policy changes. Our business in 2010 blamed Obamacare for employee healthcare changes, that the majority of the company made in 2007. And it becomes the built in excuse for everything. I bet retailers in early 2020 start howling about decreased profits from trade wars.
Well political identity is hard to break though and campaigns have always tried ‘identify’ with their voters.
We forget Ronald Reagan’s 1980 Presidential campaign was not an automatic win and I believe the Moral Majority used religious identity to assist their chosen candidate to win the Presidency. (The fact that Reagan was an actor, and especially since he was a Bad Actor, created a lot of concern that he was not able to be a functioning President.) But why did the Moral Majority help carry Reagan:
1) Reagan had to beat Bush Sr in the Primary and they organized to make sure Reagan took the Primary in 1980.
2) The Moral Majority campaigning for Reagan weakened one of Carter’s strength which was he was an extremely religious man that Christian voters could sympathetic to in 1980.
Anyway, I still hold the conservative movement issue with the modern economy is young people are not identify as much for their religious beliefs or identifying themselves for their career.
(And yes Carter primarily lost the 1980 due to the Iran Hostage Crisis but in January 1980, the crisis could end at any point, and might not be a deciding factor.)
One wonders what proportion of respondents would have “I don’t know” as the response most accurately reflecting their initial position. And then how many would prefer to know why one party tended to support a particular position. One wonders if the results would hold up in countries with proportional representation. In the winner-take-all USA, there is a rational reason to oppose whatever the other party supports – politics there is a zero sum game – as well as “weak horse, strong horse” strategic concerns. One wonders if it is really the participants who are defective or whether it is the political system. One wonders if the music results just suggest that there are advantages to just going along with the crowd, keeping a low profile, and not attracting attention to oneself seem to be fairly successful adaptations and fairly well ingrained.
Roger Sweeney and asdf have identified the ingredients of the secret sauce: efficient cognitive deference to the Party + strategic solidarity within the Party. (Not cascades.)
Thanks. Don’t forget RAD. I was trying to be William J. Brennan to his William O. Douglas.
As noted, these questions have to be based on things most folk don’t have such a strong opinion of.
If this is trying to suggest some way to reduce polarization, it fails to get to the root cause: Democratic Party dominance of academia which has morphed into a hatred of Republicans — to the point many educated/indoctrinated people think Reps are evil. Even those who claim to be atheists (evil is anti-God, without God there is no “evil”).
Comparing the amount of prime-time news about Presidents Bush, Obama, & Trump, and how much is positive, negative, or neutral, would be more interesting, and probably show much more positive & neutral coverage of Obama, even on the economy.
Most folk, rightly, don’t follow politics so closely, so going along with others is pretty good.
Abortion – social programs is a good counter argument. 40 years ago, lots of Catholics were against abortion and in favor of more social programs; Democrats. Many still are, but stay Dem despite the Dem promotion of abortion. Many others, due to their opinion against abortion, changed their identity to become Republicans. Often feeling cast out by the Dems. I don’t think so many of these ex-Dem pro-life folk favor smaller gov’t spending on social programs, despite now being Reps.
Channeling Taleb, I think the variation in opinion that the authors are researching arise from questions of skin in the game. It’s cheap to have a personal opinion, but when you are exercising political power (as a voter in an identity group) then coordinating your answer with or against others swamps the value of your individual opinion.
If this is true, then political opinions expressed outside the context of a political competition are not worth much.
“Yet hardly anyone sees this as evidence to doubt their own opinion”
Well, I’m definitely one of “hardly anyone”. Any belief that has a genealogical basis (political, moral, religious) should definitely be suspect. Given that a belief has an origin that both explains A) why we believe it and B) why is feels so objectively true, it should at the very least get us to admit that we can’t be so sure that our belief is justified
Two things come to mind here:
1. Stunts that involve going to university campuses and asking students on video whether they agree with a particular statement by Trump. Of course most strongly disagree. Then it’s revealed the statement was made by Obama. Or vice versa.
2. This old Scott Alexander post about the politicization of Ebola (and everything else):
https://slatestarcodex.com/2014/10/16/five-case-studies-on-politicization/