Study: Lefties assume divergence, Righties assume consensus

[madd0ct0r]

http://www.psychologicalscience.org/ind ... think.html

Liberals Aren’t Like the Rest, or So They Think

Tags: Attitude, Motivation, Personality/Social, Political Psychology, Social Perception
Liberals tend to underestimate the amount of actual agreement among those who share their ideology, while conservatives tend to overestimate intra-group agreement, according to new research published in Psychological Science, a journal of the Association for Psychological Science.
These findings may help to explain differences in how political groups and movements, like the Tea Party and Occupy Wall Street, gain traction on the national stage:
“The Tea Party movement developed a succinct set of goals in its incipient stages and effectively mobilized its members toward large-scale social change quite quickly,” says psychological scientist Chadly Stern of New York University. “In contrast, despite its popularity, the liberal Occupy Wall Street movement struggled to reach agreement on their collective mission and ultimately failed to enact large-scale social change.”
Stern, with co-authors Tessa West and Peter Schmitt, recruited almost 300 hundred participants to complete an online survey. The participants read political statements (e.g., “In general, I support labor unions,”) and non-political statements (e.g., “I enjoy coffee”) and were asked to indicate whether they agreed or disagreed with each statement. They were also asked to indicate how much others of the same political persuasion would support their own attitudes – a measure of perceived in-group consensus.
Liberals showed what the researchers call “truly false uniqueness,” perceiving their beliefs as more divergent from the beliefs of other liberals than they actually were. Moderates and conservatives, on the other hand, showed evidence of “truly false consensus,” perceiving their beliefs to be more similar to those of other members of their political group than they actually were.
Data from a second study suggest that the relationship is driven by participants’ desire to feel unique: Liberals reported a stronger desire for uniqueness than did moderates or conservatives.
Surprisingly, these trends even emerged among nonpolitical judgments, such as preference for coffee: Liberals believed their preferences were more different from those of other liberals than they actually were, while conservatives believed their preferences were more similar to those of other conservatives than they actually were.
Given that perceptions of in-group consensus can be an important motivator for social change, these new findings may help to explain why liberal and conservative movements develop different political trajectories:
“Liberal social movements might struggle to develop solidarity and formulate shared goals within their ranks, both because liberals want to maintain unique beliefs and because they underestimate the amount of agreement among their members,” Stern explains. “Conservative social movements might initially capitalize on perceiving agreement to galvanize their ranks, but their inaccurate perceptions could impair group progress when actual agreement is necessary.”

I'm not really sure how legit their extrapolations are - Occupy came together as a bunch of people opposed to the bank bailout and opposed to goverment policy in general, but without a shared idea of what to replace it with. The tea party came together on a shared idea of a solution (lower taxes, freedomTM), thoughts?

[Formless]

Its an online survey. That says everything I think you need to know. Its untrustworthy data until they replicate it in a manner that isn't anonymous, and you would still need to know what the survey questions look like to eliminate the strong possibility (present in all online surveys) of leading questions. Start with "don't use the words liberal or conservative" and "don't automatically trust self identification", and end with "the survey taker's interpretation of Liberal vs Conservative must be taken into account too, being that they are psychologists and not political scientists". Everyone has their own interpretation of what "liberal" means, for starters, which is the thing I think most likely to skew the results towards everyone thinking of themselves as a unique little snowflake (note that they got the same result even on a-political questions. That to me strongly suggests something in the survey was priming the survey takers).

This would be one reason I think the spectrum method of political identification is problematic, because its so easy to mislead people with it. Even if the study is correct, it means you have one political movement that is clearly defined, and one movement that is only defined by not being (or opposing) that movement. And notice how conveniently it fits the political agenda of a two party system... anyway, it would be an ironic result considering how contrarian the Right likes to be when its still extremely easy to figure out what it stands for.

[Simon_Jester]

Assuming for the sake of the moment's argument there's something to this-

The premise here would be that part of OWS's problem was that the protestors said "look at how many different things we want" and decided NOT to push a unified agenda, even when in fact they could easily have hammered out a shared platform that 99% of them could agree on, albeit not a comprehensive one.

Conversely, they would allege, the Tea Party routinely campaigns for "small government" despite very divergent and muddled ideas among its own membership about what "small government" means. Which is why you have a serious political movement whose adherents include people willing to say stupid things like "get your government hands off my Medicare."

[Formless]

Also, on that note I suspect that Occupy was a more inclusive movement than the Tea Party. Few people could ever stand to hear the lunacy of the latter, while the former pretty much defined themselves as being the average person vs the highest earners in the country.

With the Right becoming more and more extreme, and the "middle" (read: Overton Window) of the spectrum in this country shifting more and more towards their side, this could be taken as a sign of that political strategy's shortcomings. Instead of actually shifting the country's opinions to one side, its possible that its been alienating people whose political opinions aren't simply a sense of belonging towards their party (read: the Republicans). The "liberals" in the survey may actually include conservatives, and on some level the people being surveyed are keenly aware of it! Its just that the Right no longer represents the vast majority of people any sane country would deem "conservative".

It would make sense given some of the things I've heard from this board's self identifying "moderate conservatives".

[Magis]

Its an online survey. That says everything I think you need to know.

Actually it doesn't really tell you anything at all.

Its untrustworthy data until they replicate it in a manner that isn't anonymous,

Why?

and you would still need to know what the survey questions look like to eliminate the strong possibility (present in all online surveys) of leading questions. Start with "don't use the words liberal or conservative" and "don't automatically trust self identification",

The whole purpose of the study was to investigate the views of people who self-identify as being either liberal or conservative. By removing the self-identification part, you would be defeating the point of the study.

[Formless]

Surveys in general are problematic data, anyone who has taken basic psychology or sociology knows that. They get used anyway because they are cost effective, assuming the survey was designed well. But from my experience, online surveys where the participants are anonymous (meaning that teh surveyors have no idea who they have chosen) run into other problems. Is the survey representative? How many people opt out because they feel they have no time to take it, or hate taking surveys, or are jaded? Is there a selection bias going on, where the survey might have been passed on to friends, family, social networks, etc.? If its anonymous, its that much harder to figure out which of these might have happened. That's not limited to the internet, but also to surveys conducted by phone, for instance. The internet just helps exacerbate the problem because of how easy it is to forward it to others.

The whole purpose of the study was to investigate the views of people who self-identify as being either liberal or conservative. By removing the self-identification part, you would be defeating the point of the study.

No, at no point does the article state that it is purely about self identifying liberals and conservatives. In fact, the whole point in asking questions like "do you generally support labor unions?" is to try and get at what their real political persuasion is, not the one they think they have. My point here is for any future study, so that those don't get biased, and for people who want to know what to look for in a good or bad survey. However, even with the survey trying to eliminate self identification bias, that doesn't ensure that they will necessarily be identifying actual liberals and conservatives, because the survey writers may be influenced by their own prejudices and stereotypes of what a liberal is or a conservative is.

On that note, two more things I thought of after reading it last night was that if there is such an effect (assuming no bias in the survey itself) is that they may be misinterpreting the data when they conclude that liberals have "truly false uniqueness". One, the surveyors may not have taken into account why the people being surveyed think they are unique. It may not be just about what their beliefs are, but rather why they hold them that makes people think of their beliefs as different from others on the left. Someone with more education, for instance, will be more likely to feel that this is an important distinction, but its one that is hard to elucidate in a question and answer survey. Alternatively, it may be that there are other political issues that they feel differentiates them from other leftists but aren't asked about on the survey. This could happen in an environment where Right wing issues overcast Left wing issues. Many people here are definitely left wing, but there are still tons of political arguments on this forum nonetheless. It may be that Right wingers consider these to be trivial, or else have never been introduced to alternative viewpoints on them to consider.

It reminds me of a discussion I had recently with my dad. I consider myself a progressive, and was trying to explain the concept of a regressive to him, but he didn't really see the difference between that and a conservative. He considers himself basically a left winger, but is a bit old fashioned about it, and on the progressive-conservative spectrum (not to be confused with the liberal-conservative spectrum-- I.E. Left vs Right) he seems to think that there were aspects from the past that were actually more progressive than today. So from that, he made the argument that a conservative could be for (in his words) "status quo ante", that is, wanting to return to a previous status quo that was better than today. Not necessarily incorrect reasoning (though still motivated by a similar nostalgia factor as a Regressive), and supported by the fact that not all societal changes are good, or without tradeoffs. All I could say in response was that globally, I thought progress has tended towards better quality of life, which in turn he could only shrug at. Point being, it appeared as if we mostly agreed, but whereas I took a global perspective on the good progress brings, he took a more narrow focus and concluded that he couldn't necessarily consider himself a progressive (at least, not the same kind as myself). Its probably the same with Right wingers-- from my experience they often seem like they only care about a very small set of issues, and ignore the rest entirely. Then when they are dominant in political discourse, that's all that gets talked about.

[Alyrium Denryle]

urveys in general are problematic data, anyone who has taken basic psychology or sociology knows that. They get used anyway because they are cost effective, assuming the survey was designed well.

They are problematic. Every form of social science data collection is problematic. All of them. This means the researcher needs to understand the limitations of their data, but a lot of that can be ameliorated with sheer sample size.

But from my experience, online surveys where the participants are anonymous (meaning that teh surveyors have no idea who they have chosen) run into other problems. Is the survey representative?

All Surveys Are Anonymous (in the sense that personally identifying data is never preserved). There are four basic ways to conduct a survey, in order of increasing sample size that is feasible per dollar: In person, Phone, Mail, Online.

In person interviews have small sample sizes, and while you can get a lot of detailed information that way, they have Serious problems with their data. First is selection bias by the researchers. Sample location, who they elect to ask. The second is selection bias on the part of the respondent. The third, and potentially the worst, is the fact that you are up-close and personal with a subject and unless you are using deception, they are likely to tell the researcher what they think they want to hear precisely because they are NOT anonymous.

Phone interviews have similar problems, though selection bias by researchers is reduced because they select participants through phone directories at random using a computer algorithm to generate a call list. Though, because the interaction is social, you still run into the Social Desirability Bias hardcore. You also run into a wall when it comes to self-selection bias on the part of the participant, because people hate to be called at random and they will often only participate if the subject matter is important to them.

Mail is better with respect to Social Desirability bias, but worse in self-selection because it takes effort to mail back the response.

Online surveys can reach massive massive sample sizes. Effort on the part of respondents is reduced so the self-selection bias is reduced. Selection bias on the part of researchers is also reduced, because they can systematically avoid it using cluster sampling or simply rely on massive sample size to make it irrelevant. The interaction is not social, so Social Desirability Bias is reduced. You can easily ask demographic questions at the front of the survey, including income for example, and people are way more likely to be honest about it precisely because they are anonymous, and they are more likely to give honest responses to controversial questions because they dont feel as if there is a person there. Judging them.

How many people opt out because they feel they have no time to take it, or hate taking surveys, or are jaded? Is there a selection bias going on, where the survey might have been passed on to friends, family, social networks, etc.? If its anonymous, its that much harder to figure out which of these might have happened. That's not limited to the internet, but also to surveys conducted by phone, for instance. The internet just helps exacerbate the problem because of how easy it is to forward it to others.

No. What you end up getting is a natural cluster sample, which is just fine provided the sample size is large enough, and the tendency to forward is the same across the groups of interest (across a political gradient, in this case). Plus, if you are recruiting by way of a social network, the interconnectedness of those networks ensures that you cover a representative sample of the population. You can even check the data out mathematically. With a huge sample size, a question about the political leaning of your respondants should come up as normally distributed (or close to it) because the question is asked on a 1-7 agreement/disagreement scale, or some other similar but quantifiable way. You know you have a bias if you have a bimodal or heavily skewed distribution. This is something that is harder to do, or less likely to work with other methods, because your sample sizes are smaller and so your ability to mathematically detect biases in your dataset is reduced.

The other things you mentioned in this post... frankly, you are criticizing a study for not including something that is beyond its scope. "Is this potential effect observed, or not" was the question. Later, someone can design another study to figure out the cognitive and social Why.

Then there is the fact that research in other subfields of psychology tend to back up the results of this particular study. Liberals and conservatives tend to have different value sets on a cognitive level that translate to differences in brain activity. Or rather, everyone has the same values, but the emphasis is different. Liberals tend to value fairness and empathy more than conservatives, who value deference to authority/tradition and purity. Again, these are differences in cognitive emphasis that are shown in fMRI as well as well as survey and experimental data.

This study makes sense in light of this. A person who values fairness and empathises with others is more likely to assume that people have different views, be more willing to listen to those views and get stakeholder input before putting forth a concrete policy suggestion than someone who values purity and deference to perceived authority.

Liberals: OK folks, we know there is a problem, now what do we do about it? *proceed to herd cats for months in Central Park*

Conservatives: Ok folks, we want small government, lower taxes, and a return to traditional values *proceed to march in lock-step to political and ideological authority figures, shunning anyone who deviates from ideological purity by way of primary challenges*

[Formless]

They are problematic. Every form of social science data collection is problematic. All of them. This means the researcher needs to understand the limitations of their data, but a lot of that can be ameliorated with sheer sample size.

This had a sample size of 300 participants. What does that say to you?

All Surveys Are Anonymous (in the sense that personally identifying data is never preserved). There are four basic ways to conduct a survey, in order of increasing sample size that is feasible per dollar: In person, Phone, Mail, Online.

Yes, I know this, and the specifics of each. And I said that the issues I identified aren't unique to internet surveys. I just happen to find internet surveys particularly poorly handled on most occasions, based on personal experience.

The interaction is not social, so Social Desirability Bias is reduced. You can easily ask demographic questions at the front of the survey, including income for example, and people are way more likely to be honest about it precisely because they are anonymous, and they are more likely to give honest responses to controversial questions because they dont feel as if there is a person there. Judging them.

I don't buy that people are necessarily more honest online simply because its not a face to face interaction. Its quite easy, especially if a survey becomes viral (which you kinda hope for since that's how you get the huge sample sizes the internet can provide), for people to just troll it and throw random answers in for the lols. Also, there are demographics that you can't study this way because they lack a computer or free time. That's what in person interviews exist for.

Besides which, the cheapness of internet surveys has a problem on the researcher's end. Since you can throw more of them out there compared to other survey methods, its tempting to slap them together lazily and with less care towards eliminating leading questions. This is particularly an issue when the topic is political (like, you know, politics). That's where my suspicion comes from whenever I see internet surveys specifically.

No. What you end up getting is a natural cluster sample, which is just fine provided the sample size is large enough, and the tendency to forward is the same across the groups of interest (across a political gradient, in this case). Plus, if you are recruiting by way of a social network, the interconnectedness of those networks ensures that you cover a representative sample of the population.

That depends on the social network. Compare Facebook, where so many people across social classes congregate, to something more like Reddit or others that are self segregating. Obviously, you want to make sure the survey is forwarded or put up on one that is of more general popularity, unless your goal is to survey a very specific demographic.

Which is just not what I've come to expect, unfortunately.

The other things you mentioned in this post... frankly, you are criticizing a study for not including something that is beyond its scope. "Is this potential effect observed, or not" was the question. Later, someone can design another study to figure out the cognitive and social Why.

But, I think it is premature to conclude that conservatives suffer from some sort of false consensus effect and liberals the opposite. To see what I mean, let me jump ahead in your post for a moment:

Liberals: OK folks, we know there is a problem, now what do we do about it? *proceed to herd cats for months in Central Park*

Conservatives: Ok folks, we want small government, lower taxes, and a return to traditional values *proceed to march in lock-step to political and ideological authority figures, shunning anyone who deviates from ideological purity by way of primary challenges*

These two movements make for a great story and great journalism, but they may not be the best examples due to Occupy being so inclusive by design. An alternative interpretation might be:

American Liberals: value discussion and diversity, and so when attending an event like Occupy they assume (correctly!) that others are there for a multitude of reasons. Note that this is in contradiction to the survey, because there really was a variety of grievances on display at Occupy, tied together only by the barest of common themes. Its not false uniqueness, its a failure to properly structure Occupy into something actionable.

You got the conservatives right, but I must still add that their stranglehold on American politics can stifle even their opponent's conversation by their presence and preference for immediate action at all times. The healthcare bill is a perfect example-- options like public payer aren't even allowed into the room, and even outside Washington a Right wing person can easily hijack a conversation to be about their concerns and conspiracy theories, taking airtime away from everyone else who now has to acknowledge them and debunk their stupidity. Also, just because they are shunned by other conservatives doesn't mean that they are suddenly liberals. It just means the group has shut them up.

Then there is the fact that research in other subfields of psychology tend to back up the results of this particular study. Liberals and conservatives tend to have different value sets on a cognitive level that translate to differences in brain activity. Or rather, everyone has the same values, but the emphasis is different. Liberals tend to value fairness and empathy more than conservatives, who value deference to authority/tradition and purity. Again, these are differences in cognitive emphasis that are shown in fMRI as well as well as survey and experimental data.

I would very much like to see those studies, because to be blunt the part where you went to fMRI for extra credibility raises one of my skeptical brain's red flags. That's a field that is still very prone to exaggerated reporting of results and inferences.

[Kingmaker]

This had a sample size of 300 participants. What does that say to you?

That the sampling distribution of the mean will be approximately normally distributed? (Assuming they used proper sampling techniques)

Let me ask you: what dictates the sample size necessary to draw useful conclusions? Why is 300 inadequate?

[Purple]

It is a measure that you can calculate. As in there are formulas out there that will give you the appropriate sample size for a desired certainty factor. Now I don't know exactly how, given that this is a sociology/psychology problem and all my training in statistics was focused in computer science. But I distinctly know that it is calculable.