fbpx
How to know if you’re talking to a wokescold: A scientific method for preventing IRL flame wars

[FYI: If you’re interested in data-blogging like this, I’m offering a little free course on it.]

We’ve all been there: You’ve had a couple drinks, you’re having fun talking with someone, then you blurt out a controversial opinion and everything goes belly up. Maybe your interlocutor scolds you, maybe they just walk away, or maybe nothing happens but there’s gossip a week later…

If you have controversial opinions, what you need is a method for knowing — in advance — whether your conversation partner can handle them. It needs to be simple and quick enough to be practical, but it needs to be scientific enough to offer real predictive validity.

It recently occurred to me that there exists a statistical technique that solves exactly this problem. It’s called recursive partitioning, and the practical tool it produces is called a decision tree. If you have data on public opinion and other demographic variables, you can use statistics to determine which chain of questions will give you the best guess about someone’s position on any given issue. If we create a decision tree to predict their position toward suppressing naughty opinions, then we have a simple, practical, and scientifically valid “life hack” for avoiding IRL flame wars.

Analysis

I did this last week and the results are very interesting. If you’re interested in the statistical details, or you’d like to run the code yourself (perhaps on a different outcome variable), you can find all of that here. In this post, I’ll focus on the social and practical implications.

Here’s all you need to know about the stats. In this analysis, “being a wokescold” is proxied by whether or not someone thinks racist speakers should be allowed or disallowed. For possible predictor variables, I included a handful of variables that are reasonable to ask someone about or easy to observe yourself.

Specifically:

  • sex/gender = variable named sex
  • race = variable named race
  • left/right identification = variable named pol
  • family income = variable named realinc
  • college attendance = variable named college
  • word knowledge or verbal skill (proxy for IQ) = variable named wordsum

I then conducted recursive partitioning, which breaks the data down into the sequence of branches giving the most predictive traction over the outcome variable.

Results

Figure 1 plots the resulting decision tree.

Figure 1

The graph is fairly intuitive, and if you’d like to understand the numbers better, see my more technical post over at jmrphy.net. Here I will give you a more concise and practical translation, resulting in a simple heuristic you can memorize.

If you meet a random person, there’s a 38% chance they’re a wokescold (defined as wanting to suppress racist speakers; one can debate this, but whatever, it’s a decent proxy).

The very first and most important question you can ask someone, to avoid a flame war, is: “Did you ever go to college?" If they say yes, the probability of them being a wokescold drops to 29% and that’s your best guess: They are probably not a wokescold. Nothing else will improve your guess from this point (at least from the variables we selected).

Now, many of you will say: But it’s the college-educated wokescolds one should be most afraid of! True. The limited utility of this analysis is also it’s primary social-scientific value: It reminds us that college-educated wokescolds remain a relatively minor anomaly, quantitatively speaking. Being educated still means you’re much more likely to support unsavory expression. It’s true that educated wokescolds are often the most dangerous landmines we’d like to tiptoe around, and unfortunately my particular analysis this week will not help you on this front. Fortunately, I have an alternative algorithm custom made for this use-case: If they went to college and they’re also a female with dyed hair, hold fire on your nuclear takes: They are probably a wokescold. Unless they’re Amber Frost.

If they never went to college, the next question you have to ask yourself is whether they're smart. You probably don't want to give them a vocabulary test, but conversation is pretty revealing. If they are smart, you infer they are not a wokescold (40% chance). If they are dumb, it's now a coin flip (50%).

Next, what is their race? This you can probably guess yourself. If white, this bumps them very slightly toward not being wokescolds (48%). If non-white, this bumps them toward being wokescolds (57%). From here:

If they are white and male, there's a 45% chance they’re a wokescold so you infer they are not — and that’s your final guess. If they are white and female, you should see if their family is rich or not. If rich, they are slightly less likely than a coin flip to be a wokescold (46%); if poor, they're slightly more likely than a coin flip to be a wokescold (54%).

If they are dumb and non-white, there is a 57% they’re a wokescold and that’s your best guess.

A heuristic you can memorize

(This only applies in America, mind you, the land of the free.)

  1. If they’re a female who signals creativity or virtue (e.g., dyed hair, bumper stickers), don’t share any edgy takes (this is post-hoc to the model, just a precaution in light of data limitations and researcher experience).

Otherwise:

  1. If they went to college, they’re probably not a wokescold. You may gradually begin to share your edgy takes.
  2. If they did not go to college, but speak more intelligently than average, they are probably not a wokescold. You may gradually begin to share your edgy takes.

For all others, the safest decision rule is to not share edgy takes. Bonus rule only if you can master the above 3-step algorithm and you have an appetite for risk:

  1. If they are rich white people, you may gradually begin to share your edgy takes.

What about ideological identification?

The most intriguing result here, to my mind, is that ideological identification totally drops out — it appears to have no predictive power! As I wrote in my technical post:

[That ideological identification has no predictive power] is fascinating, given that many people today tend to think of speech suppression as a fashion on the educated Left! And it is, but that's only a highly visible minority. Political scientists would not be surprised by this result: We've long known that leftists and educated people are always more supportive of free expression (you just don't hear about those people in the media right now).

Limitations

Please note that the model here does not provide especially satisfying statistical discrimination. It’s better than nothing, but one must still proceed carefully. Always begin by sharing mildly provocative takes, and watching your interlocutor’s reactions. Do not advance to nuclear takes until several acts of mild edgelording produce only smiles, laughter, or excited edgy reciprocity. With additional data and more sophisticated modeling, we may hope to derive more confident predictions for more ambitious social maneuvering. Until then, be careful.

Democrats wanted strongman rule way more than Republicans — until Trump arrived

Democrats wanted strongman rule until Trump arrived

That graph is from the new book by political scientists Pippa Norris and Ronald Inglehart (2019), Cultural Backlash: Trump, Brexit, and Authoritarian Populism.

Just finishing the book now. I'd call it a rigorous validation of the conventional wisdom of the moment.  Certainly required reading for anyone with weirder theories about the culture wars. This will probably serve as the baseline model for some time.

Our theory argues that a cultural silent revolution has heightened polarization over cultural issues in the electorate, provoking an authoritarian backlash among social conservatives. We hypothesize that socially liberal values are spreading through intergenerational population replacement and demographic shifts, causing traditionalists (concentrated among the less-educated and older birth cohorts) to feel threatened, perceiving that respect for their core values and social mores is rapidly eroding. These developments have cumulated over time to reach a tipping point in high-income Western societies…

If you read between the lines there are some more tantalizing insights, such as the one dramatized in the graph above. I would like to write a longer review, but we'll see.

Patrons may have noticed this in my hard drive a week or two ago.

Are Politically Correct Students Better or Worse Students?

In my current book project, one of my goals is to provide the fullest possible empirical accounting of the strange new persona sometimes derisively called the "social justice warrior."

Although the main contours of my argument are pretty well developed, there are various sub-hypotheses that I've had for a while — but no data to test them.

Just last week, I was offered access to a goldmine of data collected by College Pulse. They told me I'm allowed to share my analyses, but not the data. They have an app that gives students various rewards in exchange for taking surveys. They've taken dozens and dozens of surveys including the widest variety of questions, with consistent respondent IDs for each survey. This means all of their surveys can be merged for all the individuals who took each survey.

The major drawback of these surveys is that they are not representative samples — so we can't really know the degree to which patterns identified in them generalize to university students as a whole. But the lack of representativeness is somewhat offset by the sheer size of the sample. Think about it this way: If you could survey 100% of the people in some population, you wouldn't much care how the sample was drawn, right? Many of the College Pulse surveys have quite impressive sample sizes, with quite a lot of them including 20,000 or more students. While this hardly approaches 100% of university students, it's more than enough to be quite interested in what these data reveal. All datasets are partial and limited, and need to be checked against other datasets with different virtues.

The wide variety of fascinating questions, and the large samples, make this an ideal, first-stab testing ground for any number of hypotheses.

The GPA of the SJW

I've long wondered if there is a relationship between attitudes toward political correctness and academic performance in the student body. In other words, are "SJWs" more or less likely to be high-performing or low-performing students? Or perhaps there is some curvilinear relationship? One could generate a few different hypotheses on this question, but for this post I will simply introduce the data (check) and share some basic descriptive statistics approaching this question. This is likely just the first of many hypotheses I hope to explore with this data over the coming weeks and months.

First, the univariate distributions require some comment. First, students in the sample seem to report questionably high GPAs. Or perhaps students are disproportionately drawn from schools with rampant grade inflation. Either way, there's something going on, because Figure 1 shows that the sample has a lot of students claiming to be nearly perfect students.

Figure 1: Self-Reported GPAs in the College Pulse Sample

Figure 2 shows that students in this sample are quite strongly opposed to the idea of physical no-platforming. The survey item says: "A student group opposed to a controversial speaker uses physical force to prevent the speaker from speaking. In your view, the students group's actions are…" Most find it very unacceptable. Note that this is just one particular way of tapping what we might very roughly call, for shorthand, SJWism. What's great about the College Pulse survey data is that they ask a variety of different questions revolving around moralistic, speech-focused political activism, so we'll be able to triangulate with multiple variables.

Figure 2: Views on Physical No-Platforming in the College Pulse Sample

Aside: One thing I'd like to do soon is a factor analysis of a few of these SJW attitudes. It would be good if we could extract the latent variable underlying, for instance, opinions toward physical no-platforming, "call outs," disinvitations, and the other related but different tendencies associated with SJWism.

Hypotheses

One of the arguments in my book is that the hyper-moralistic political activity of the "SJW" is, in many cases, a kind of thinly veiled economic activity. But there are a few different ways this might manifest, so we need to delineate different observable implications to make specific hypotheses falsifiable.

The omni-directionality of the following hypotheses merely reflects how little we understand the SJW phenomenon.

The angry runner-up hypothesis. Because today's political economy is increasingly a "winner take all" situation, individuals who once upon a time could enjoy relatively high income and status from a "second place" finish in the capitalist game, are now looking at prospects quite beneath their relative expectations. But the types of people who land in "second place" positions are still smarter and more capable than average — so they're not just going to accept outcomes beneath their expectations, rather some of them will seek to alter the rules of the game. "If you can't beat 'em, turn over the table!" If this is the logic behind SJWism, then perhaps we would expect SJWism to be most likely in the middle of the academic performance distribution. 

The shrewd winner hypothesis. Another possibility is that SJWism is a new kind of game, with emergent rules related to novel and complex socio-economic factors. Seen from this angle, SJWism might be more likely among the most intelligent and the best academic performers. If SJWism is the way to win cultural games today, and higher education is largely about signaling one's ability to win games, the best students might be most likely to become SJWs.

The brute force hypothesis. It might be the case that SJWism represents the vengeance of the intellectually dominated against the more intelligent. Seen from this angle, SJWism might be a way for the dumbest or most disorganized students to promote themselves through a kind of morally glorified brute force.

Figure 3 below shows a curvilinear relationship in which the lowest and highest levels of academic performance are associated with a slightly greater acceptance of physical no-platforming, and students in the middle of the range are least accepting. So it's not second-place students adopting SJWism to claw their way into top jobs or some such model as that (hypothesis 1). [By the way, I spoke too soon in one of my livestreams the other night; I reported this based on a too-quick look at this data, which I accidentally had backward — that will be the last time I foreshadow data-analytic findings live on Youtube before I'm actually done, sorry!] The differences here are pretty tiny, but that's in part because most students are strongly opposed to physical no-platforming, with the average level of acceptance quite low.

Figure 3: Political Correctness by GPA (Physical No-Platforming)

I was curious if Figure 3 might be a fluke related to that particular question, so I took another item tapping political correctness/SJWism to see if the same pattern holds. Figure 2 below shows that, yes, it does. The inflection point is lower on the GPA scale and the students on the bottom are not quite as PC as the best students, but again it's clearly non-linear: the best and worst students are less likely to think there's any problem with sensitivity, while the middling students are more likely to think students are too sensitive.

Figure 4: Political Correctness by GPA (Perceived Sensitivity)

These could maybe represent evidence for H2 and H3, in a kind of mixed effects model: Maybe the best students adopt SJWism out of their shrewd awareness that that's indeed how to win the game of institutionalized culture, and maybe the worst students adopt SJWism to turn over the table in a game they are unlikely to succeed in. Or none of these interpretations is accurate, which is very possible.

I was also curious if the curvilinear relationship is conditional on major. In retrospect there's probably a better way to do Figure 4, but it does the trick. I restricted the data to majors that had at least 200 observations. You have to be careful to not go fishing for patterns here, because of the multiple comparisons problem. Here's one thing that seems true, though: Most of the major/gpa-level combos that contain big outliers toward SJWism are on the lower half of the GPA scale (Nursing, Law/Crim, Econ, Education, Poli Sci, Chemistry). Whereas the positive gradient toward SJWism on the high-end of the GPA scale in Figures 3 and 4 appears to reflect a slight but more steady pattern in a number of majors (Comm, CS, Econ, Engineering, Pre-Med).

Figure 5: Political Correctness by GPA and Major

I wouldn't make too much of this but it's perhaps consistent with the "mixed effects" idea above. Better students veer toward SJWism because better students perceive that PC is part of winning in education and beyond, and so they slightly and gradually report higher levels of PC moving from 3.0 to 4.0. Then the really zealous SJWs are rare occurrences drawn mostly from the lower end of academic performers, perhaps as a kind of brute force strategy. But like I said, this was just a cursory exploration to dig into some new data. Let's see what else I find in future posts before I make any big claims.

Now, wars start themselves

Major wars have become less frequent, but a curious feature of the wars we still observe is that almost nobody starts them. When wars occur today, they appear to start themselves, or are started by some unknown entity. I learned this from a new article by Hathaway et al. Here are selections from the abstract:

This Article is the first to examine “war manifestos,” documents that set out the legal reasons sovereigns provided for going to war from the late fifteenth through the mid-twentieth centuries. We have assembled the world’s largest collection of war manifestos—over 350—in languages as diverse as Classical Chinese, German, French, Latin, Serbo-Croatian, and Dutch...

Examining these previously ignored manifestos reveals that states exercised the right to wage war in ways that would be inconceivable today. In short, the right to intervene militarily could be asserted in any situation in which a legal right had been violated and all peaceful channels had been explored and exhausted. This Article begins by describing war manifestos. It then explores their history and evolution over the course of five centuries, explains the purposes they served for sovereigns, shows the many “just causes” they cited for war...

Hathaway, Oona A. and Holste, William and Shapiro, Scott J. and Van De Velde, Jacqueline and Lachowicz, Lisa, War Manifestos (September 15, 2017). University of Chicago Law Review, Vol. 85, 2018 Forthcoming; Yale Law School, Public Law Research Paper No. 617. Available at SSRN: https://ssrn.com/abstract=3037538

Self-defense is the most popular justification for war throughout the period studied, but it's interesting that its prevalence steadily grew from the middle of the twentieth century. Most traditional justifications for warring have become obsolete. Religion was once a fairly common reason for going to war, but now explicitly religious wars among states are virtually extinct.

https://ssrn.com/abstract=3037538

Don't be fooled into thinking that interstate aggression has been humanized. Quite the contrary, these data only suggest that war is an increasingly algorithmic process, increasingly devoid of human agents: When every player in the game invokes "human rights" to blame it on some other guy, this is not evidence that human rights have been normalized; it is evidence that humanity has been evacuated from the underlying process, through the cold and calculated manipulation of human emotions for ulterior purposes. 

Educated Errors? On Perceptions of Trump's Ideology

More educated individuals generally know more than less educated individuals. If you test what they know about any random political issue, for instance, uneducated individuals are more likely to give an incorrect answer, or say "I don't know," relative to more educated individuals. At first this sounds obvious, but what if the correct answer to a question is that we don't know?

If you think that the purpose and main effect of education is to increase a person's store of true information, then an educated person should be more likely to say "I don't know" when asked what they think about an objectively unknown or uncertain matter.

But what if education does not increase one's store of objectively true information so much as it increases one's familiarity with respectable or high-status opinions? That is, education may primarily educate one about what educated people feel and believe. In this model, education is mostly about gaining familiarity with the ideas and gestures that are valued by the already educated classes, and having some dedicated time to practice mimicking them. Getting educated is not really about developing a fuller or more accurate model of the world, it's about learning to pass as a member of the education club, to get hired by employers who obtained their employment power through their own educated credentials.

If this is the case, then it's possible that more educated people would be less likely to admit "I don't know" in the face of really-existing uncertainty, and would be more likely to say whatever they think other educated people would be most likely to say.

It recently occurred to me that perceptions about Trump provide a nice opportunity for exploring this question empirically. For someone who is so ideologically polarizing, Trump is ideologically ambiguous. Yet, educated opinion seems to be that he is a major reactionary — even fascist perhaps. Either I'm incorrect in my sense of what educated opinion is, or this is perhaps a case where the educated tend to be in agreement — incorrectly.

What is Trump's Ideology?

First, we should establish what is objectively known about Trump's ideology (especially what was objectively known around November 2016, because the survey data we'll explore were gathered at that time). For the bulk of his adult life as a public figure, he was a moderate Democrat. As discussed in Bob Woodward's recent book Fear, this was taken very seriously in the early days of his campaign strategy discussions.

“You’ve got some problems on issues,” Bossie said.
“I don’t have any problems on issues,” Trump said. “What are you talking about?”
“First off, there’s never been a guy win a Republican primary that’s not pro-life,” Bossie said.
“And unfortunately, you’re very pro-choice.”
“What does that mean?”
“You have a record of giving to the abortion guys, the pro-choice candidates. You’ve made
statements. You’ve got to be pro-life, against abortion.”
“I’m against abortion,” Trump said. “I’m pro-life.”
“Well, you’ve got a track record.”
“That can be fixed,” Trump said. “You just tell me how to fix that. I’m—what do you call it? Pro-life. I’m pro-life, I’m telling you.”

It was also mentioned in this conversation that Trump only ever voted in one primary, suggesting that his ideological preferences are probably just not strongly formed. Here's another choice selection from the Woodward book, revealing Trump's profound ideological ambiguity going into his campaign:

An hour into the meeting, Bossie said, “We have another big issue.”
“What’s that?” Trump asked, seeming a little more wary.
“Well,” he said, “80 percent of the donations that you’ve given have been to Democrats.” To Bossie that was Trump’s biggest political liability, though he didn’t say so.
“That’s bullshit!”
“There’s public records,” Bossie said.

Even when you look at the positions that define his Presidential agenda, he's ideologically ambiguous. As Nate Silver has put it, “extremely conservative stances on issues such as immigration with surprisingly moderate (or even leftist) ones on other issues such as trade — with a lot of improvisation (and inconsistency) along the way.”

In a November 2016 article entitled, "We’ve Never Known Less About An Incoming President’s Ideology," Harry Enten of FiveThirtyEight presents the table below. Donald Trump is the second-least conservative Republican President in the last 40 years (only GHW Bush was more moderate than Trump).

FiveThirtyEight

I went and wrangled the 2016 Cooperative Congressional Election Study dataset, which has been one of the most common sources of data for many of the buzzy analyses you might have read about in the past year or so. As the figure below shows, the most common perception of Trump's ideology is "I dunno." That's pretty interesting. The second most common is "Very conservative."

So given that "we’ve never known less about an incoming president’s ideology," and that one could make a decent data-driven case that he is a moderate Democrat or a moderate conservative, it seems to me that "Not sure," was arguably the most objectively correct answer. "Somewhat conservative" and "Middle of the Road" also seem like arguably correct answers. But the fashionable perception of Trump as "very conservative" is, as far as I can see, basically incorrect. If this was a quiz and I had to grade it, I would be inclined to mark the answer "conservative" as wrong, too, given Trump's significantly left-leaning policy positions and his history as a pro-choice Democratic Party supporter.

When you look at perceptions of Trump's ideology by education level, as shown in the figure below, you can see that education is negatively correlated with uncertainty ("Not sure," the pink bar all the way to the left). This would be unremarkable for most survey questions, but given Trump's unique ideological ambiguity in November 2016, this pattern may reflect how education has more to do with demonstrating knowledge of respectable beliefs than with analytical or empirical sophistication. 

In the above graph, it's difficult to eyeball the other proportions. So let's zoom-in on the arguably most correct, uncertain answer and what I take to be the respectable/fashionable but incorrect answer (that Trump is either plainly "conservative" or "very conservative.") In the figure below, it's easier to see that education is not only associated with a clear decrease in reasonable uncertainty but a modest increase in the fashionable view that Trump is clearly conservative. Post-grads are not much different than those who have minimal contact with college life, but finishing high school and having some college experience are both associated with clear bumps toward the view that Trump is clearly conservative.

Educated overconfidence Trump's ideology

Overall, these data are not very strong evidence of anything, but they do give a little indication that I may be onto something with my hypothesis — that education may, in some cases, be associated with less correct beliefs. A larger hypothesis might be that, in any context of increasing uncertainty, there will be more and more margins at which the less educated are accidentally more sophisticated than the educated.

Occupations and Their Ideologies

Occupations are strongly sorted by ideology. Political scientist Adam Bonica has produced reliable and consistent estimates of ideological placement for a huge number of individuals, politicians, and organizations. As he writes here, he was especially struck by how extreme are the mean ideology scores for various occupations:

Although the ideological orientation of these industries is not much of a surprise, the extent to which these industries favor the extreme, rather than moderate, wings of each party far surpassed my expectations. Some of the distributions more closely resemble what I would expect from occupations that were subject to the spoils system–for instance, US postmasters prior to the Pendleton Act–than major contemporary industries with no official partisan ties.

In Bonica's AJPS article, he makes a similar observation:

In some industries, ideological sorting easily exceeds the levels of sorting observed along geographic or economic lines.

Ideological Ranking of Occupations and Industries, by Adam Bonica
Workplace Ideology: Selected Firms and Institutions (2004-2008), by Adam Bonica

Left-wing occupations are farther to the left than right-wing occupations are to the right

An interesting wrinkle in this data is that the left-leaning occupations are more left-leaning than the right-leaning occupations are right-leaning. Of the left-wing occupations and organizations, there are many firmly below -1, but for the right-wing occupations there are very few above 1. Indeed, even iconic representatives of evil right-wing military-industrial capital — such as Boeing or Exxon Mobil — are essentially centrist, at least with respect to their staffing.

You can see this again when he places industries into three buckets: Left, Right, and Divided. In the graphs below, it's easy to eyeball that the peak of the left-wing distributions are further to the left than the peak of the right-wing distributions are to the right.

Industries aligned with the Left
Industries Aligned with the Left (2008)
Industries Aligned with the Right (2008)
Ideologically Divided Industries (2008)

Social performance for status vs. mundane thing manipulation?

What are the underlying variables that explain how occupations sort into these three baskets? Just speculating, the left-wing occupations seem to be mostly about social performance and they garner high status. The right-wing occupations are mostly about mundane things and garner zero or negative status. And the divided occupations are those that call for ambiguous combinations of these things (person-facing but socially unimpressive).

And because some of those graphs are a bit old (2008), here is some confirmation that the basic patterns have not changed, at least as late as 2012.

Mapping the Ideological Marketplace (Bonica 2014)

The content of this website is licensed under a CREATIVE COMMONS ATTRIBUTION 4.0 INTERNATIONAL LICENSE. The Privacy Policy can be found here. This site participates in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites.

linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram