I kept harping on "But the economists are so much worse!" It felt like we were shitting all over the B students, when that other student over there is a lazy F piece of work.
Finally the third party present clarified to me that we all agreed that the F-student was a shithead, and that we were just critiquing the B student's failure to be an A student.
I used the NPR analogy, how I tend to rip into NPR's failings to broadcast excellent journalism, when they're far, far better than anything else widely available.
Also I heard some interesting critiques of a certain prominent sociologist that I know rather well, which I found fascinating and believable.
So we should be putting historians in charge of our left-wing politics? Will they give us non-clunky narratives? To the barricades, Von Wafer!
Context is just another name for covariates you could probably turn into dummy variables if you knew how to math it up.
Actually, the third party present alluded to statistics that was totally over my head, and which I wasn't sure what to make of, that went something like:
"Historically, the fundamentals of statistics include some presumptions which I find totally sketchy, or at least not obviously universal. This has led me to question how well scientists of all bents actually apply statistics and how robust their findings actually are."
We should put agronomists in charge of left-wing politics.
5: Certainly there are a great many big issues with how statistics are applied, but I have no idea what to make of at statement like "Historically, the fundamentals of statistics include some presumptions which I find totally sketchy."
I was assuming she took/researched a history of statistics course and was talking about some axiom that everybody takes for granted? I was hoping someone here could flesh out what she had in mind.
5
Actually, the third party present alluded to statistics that was totally over my head, and which I wasn't sure what to make of, that went something like:
"Historically, the fundamentals of statistics include some presumptions which I find totally sketchy, or at least not obviously universal. This has led me to question how well scientists of all bents actually apply statistics and how robust their findings actually are."
Defining an arbitrary line and saying results on one side are significant and on the other are not significant is sketchy. And then only publishing "significant" results is a recipe for disaster.
Also there is often insufficient attention paid to model errors.
8: Is there an assumption embedded in statistics on the nature of mathematical truth?
Or maybe an assumption that universe is mostly random with a few unusual exceptions?
8. I have no idea. If we were talking about a grad student, I'd assume it was another case of having read Kuhn while drinking cheap wine. In this case, it's either probably either Bayesianism or an advanced case of social constructivism.
an advanced case of social constructivism.
It might be this? She has a tendency to think everyone is full of shit and kidding themselves when they claim to know anything with any certainty. As a life philosophy, more broadly than just this statistics bit.
That sociologists get bogged down in theoretical jargon wars, while the right generates pithy memorable lies with large impact.
>a href="http://www.marxists.org/archive/marx/works/1845/theses/index.htm">11th Thesis on Feuerbach is serious fucking shit. It is exactly the dialectic between truth and lies, and lies should win sometimes. Or faith, or courage, or commitment, if you would rather.
Ain't no such thing as "history" since Marx. (Or maybe the French Revolution. Or the Enlightenment.) It is all fucking Social Theory Praxis. Historians lie, especially about their objectivity. "Since Marx?" Because Marx unveiled the ideology behind "science." There isn't a choice to bullshit toward a goal, it is the human condition.
Where was I? Oh yeah. Architecture and Authority in Japan Nijo. I need fucking breaks from Heidegger teaching Junger about nihilism.
(Watched Hiroshima mon amour last night, then read the IMDB thread where the younger generation said Lost in Translation was a better movie. We're doomed)
8
I was assuming she took/researched a history of statistics course and was talking about some axiom that everybody takes for granted? I was hoping someone here could flesh out what she had in mind.
Hard to say what she had in mind, it could turn out her objections are nonsense.
A historical issue is a (perhaps excessive) emphasis on ease of computation and/or mathematical elegance.
You ever calculated a regression model on paper? I have and it taught me about life*. Life and the importance of ease of computation.
*It's really dull and easy to make a mistake.
From the HMA threads
"Memory's struggle against forgetting is history's struggle against power."
Milan Kundera
13: It's funny you mention Marx in this context.
I was married to a Sociology graduate student, so I should have had more to contribute to this discussion. Her advisor for her Master's thesis was more of a historian than a sociologist, and one of his ongoing projects was to point out one by one all the errors of fact in the work of Marx and Foucault.
1
I kept harping on "But the economists are so much worse!" It felt like we were shitting all over the B students, when that other student over there is a lazy F piece of work.
Worse in what way? Apparently not at selling their work. And it seems likely you would like economists better if they were producing compelling leftwing narratives.
his ongoing projects was to point out one by one all the errors of fact in the work of Marx and Foucault
Typical social scientist, ignoring all the errors of fact in Newton and Laplace and Darwin...
independent and identically distributed is a very strong assumption.
18: Worse in the degree to which they promote agreeable left-wing policy.
17.2: People like that are great at parties. You can always get them talking about their pet peeve and then just leave while some more polite people get stuck listening.
Possible fundamental problems with statistics:
1. the widespread assumption of normality?
2. the arbitrariness of significance testing?
3. the presumption that frequentist statistics are analyzing things in the absence of a prior?
4. the validity of factor analysis?
Not sure how much the last applies to sociology, and it's not really that universal. The first three seem like reasonably good candidates, though.
2: oh, see? Now everybody really wants to know about that.
the right generates pithy memorable lies with large impact.
Speaking of which, I recently perused McMegan's blog in The Atlantic, and was astounded to find it full of worthwhile, non-stupid content. I was beginning to wonder whether some parasites from cat feces had colonized her brain. Then I came across the following passage:
Under the old model, as long as you have a modest antitrust policy in place, the market will sort things out so consumer protection can be limited to outright fraud. Adding bounded rationality into the mix suggests that consumers really can sign up for a bad deal. This then makes it somewhat facile to claim that by definition any freely entered exchange is mutually beneficial and from this we can imagine a variety of consumer protections. You may still have reasons to be skeptical of intervention, but it's not tenable to say "markets work just great, thank you very much."
There's no way 4 is fundamental and 1 isn't that big of a deal. Plenty of statistics don't require the assumption of normality and many of those that do are still unbiased (if less efficient) when normality is violated.
15
You ever calculated a regression model on paper? I have and it taught me about life*. Life and the importance of ease of computation.
Sure but it can lead to a tendency to mistake ease of computation (or mathematical elegance) for correctness. (See for example various financial models.) And with computers it is now feasible to do much more extensive computations but much of the theory of statistics was developed before computers and may not have caught up.
the presumption that frequentist statistics are analyzing things in the absence of a prior
In what sense is this wrong?
27: I didn't say I believed it was a problem. Only that it might be what somebody might think was a problem.
When I think of historians that hate sociologists I think of Wallace Stegner's Angle of Repose.
The hero of the novel is like Stegner a historian, and his arch enemy is a sociologist. Stegner sees sociologists as ahistorical utopians -- they believe they can create a perfect society as if from scratch.
This did not match my experience of actual sociologists.
much of the theory of statistics was developed before computers and may not have caught up
I'll say. It was fairly astounding to read lines (in the textbook for a stats class taught in 2012) discussing how certain ways of computing statistical results (while better) might not be useful because they are too computationally intensive.
I just read that too, though I was reading the bizarro McMegan.
In what sense is this wrong?
In the sense that there is actually an implicit prior no matter what kind of approach you take.
N.B. 35 may not be uncontroversial among people who know a lot more than me about statistics.
But I did pass both my statistics classes so I clearly have some expertise to bring to the table.
26 is on-topic, as the actual non-McMegan author of the quoted text is a sociologist.
35: There is an implicit prior, I suppose. I don't usually think of it that way. But you can't really avoid that implicit prior if you make any sort of causal or correlational* statement even if you avoid quantitative work.
*There's probably a less statistical word that I don't know.
Stegner sees sociologists as ahistorical utopians -- they believe they can create a perfect society as if from scratch. This did not match my experience of actual sociologists.
This is a somewhat accurate extension of 2.
If we were talking about a grad student, I'd assume it was another case of having read Kuhn while drinking cheap wine.
I meant for that to have a heart after it.
But you can't really avoid that implicit prior if you make any sort of causal or correlational* statement
Well, right. Hence the Bayesian-ish argument that if you can't avoid it you had might as well acknowledge it and think about the best way to use it.
27
There's no way 4 is fundamental and 1 isn't that big of a deal. Plenty of statistics don't require the assumption of normality and many of those that do are still unbiased (if less efficient) when normality is violated.
I think 1 (which is an instance of the preference for mathematical elegance) actually is a pretty big deal. It led for example to IQ being defined as normally distributed (although IIRC the empirical evidence for this is not all that compelling). And lots of financial models inappropriately assume normality (the fat tails issue) with potentially serious real world consequences.
43: Sure, except that in my experience 95% of the people who want to find the "best way to use it" are looking at statistics the way GE's lawyers look at the tax code.
45: huh. I believe it, but for whatever reason in the worlds I know about it's something close to the opposite; the people who use traditional methods are sloppy, wishful thinkers much of the time and the people who put in the effort to figure out a Bayesian approach seem more serious about finding a less fraught solution to some of the largish current analytical problems.
46: Nobody* in my world ever uses a Bayesian approach explicitly. They just ask questions that would be Bayesian if they were asked before I finished the analysis instead of after I finished and found no relationship.
*Except one guy.
I recently perused McMegan's blog
I just read that too
I only hope you shield your children from knowing you engage in this kind of risk taking behavior.
47: oh, right. That's the worst of both worlds, certainly.
I was having a discussion recently about whether the reporting of exact p-values (as opposed to just saying p less than 0.05) was essentially incoherent, as it treats a p-value (which is a frequentist concept) as a degree of belief in a hypothesis.
Another problem with statistics (in social science research), not fundamental but as currently used - the privileging of significance tests over power tests, that is, being much more careful about catching false positives than false negatives. A professor of mine harped on this as a very common error, including among professionals.
50: ooh, yeah. That's a good one.
49: Depends on how much space the journal gives you.
50: That's not always an error so much as a choice based on what is of interest and what it costs to get cases.
52.2: True, but it's ossified from a choice to a default.
Huh, for some reason I read "power tests" as "effect sizes" in 50. I don't know why I thought that made sense.
Oh, effect sizes is another one, of course.
Anyway, we've had a great deal of trouble getting a trial with a negative result published. We had power but everybody says we should have done a longer follow-up for it to be of interest.
I want to sent this OP to all of my students.
I only hope you shield your children from knowing you engage in this kind of risk taking behavior.
They have no idea. I tell them Daddy is going upstairs to smoke meth.
58: Really? What's the aspect of it that underlines what you tend to say to your students?
English students have a hard time learning that we are not analyzing individual texts against the backdrop of widespread "societal" phenomena that we can take for granted. And certainly we don't know from having read Moll Flanders what that backdrop "actually" was. They extrapolate from the tiniest little shreds of fictional evidence that "societal attitudes toward women" were this or that, and are almost entirely irrelevant to doing literary analysis. I keep saying I don't even know what the fuck a "societal attitude toward women" is. Who is having the actual attitude? Does "society" include women themselves when they write this? It's just junior-high-level sociology that creeps into analysis of texts.
This semester I had a student who took two months revising a first paper (about how all seventeenth-century people definitely thought women were scum or something) to basing her argument exclusively on the words in the book she was reading, and she ended up making a really amazing argument about it. But I think we met about ten times, just for the purpose of me saying, no, find it in the book. You can't tell me what real people all over the earth 350 years ago thought about women.
So I checked the by-line, and sure enough, McMegan is still on leave, and her blog content is being produced by a rotating cast of guest contributors.
Not only that, but her guest contributors are, appropriately for this thread, mostly sociologists. (Gabe Rossman and Scott Winship.)
32: It's his son, who, while something of a nemesis, is not perspicuously characterized as his arch enemy. Also, Lyman Ward is not Stegner.
This Historian has a slightly weird perspective on sociology, it has to be said--although (3) is more or less what I'd expect a certain sort of historian to say about sociology, for the same reasons I'd expect a certain sort of sociologist to say exactly the same thing about economics.
64: I couldn't really argue knowledgeably, but I wondered what sociologists would say about historians, in kind.
Moby is clearly a front for Big Stats.
61: That's interesting. I could see doing that, as an undergrad. Or as current-me.
A historian, a sociologist and an economist walk into a coffee shop. Barista says, "We don't get many assholes in here." The historian says, "Since when?", the sociologist says, "Is that a social fact?", economist says, "And at these prices you won't get many more."
63: I was hoping someone else read the book! I read it a long time ago, and my memories of it are vague, so it's very stupid of me to try to argue about it.
Anyway, arch enemy, nemesis, what's the difference?
And, Lyman Ward is not Stegner, yes, but I'm thinking their philosophies are pretty much indistinguishable.
68: Even if in other classes they're learning about "societal attitudes," whatever those are, it's certainly not appropriate English methodology, just as it would be inappropriate to give your philosophy professor a narratological reading of Hume.
70: Well, so, his son wants to move him out of the house and into assisted living, or something like that. And there's a generational divide thing going on. But 'enemies' just sounds too intentionally hostile.
Maybe I don't know enough about Stegner, but I don't think he was nearly the oldster crankypants that Ward is.
Or as current-me.
You can't step into the same mathematician twice. Unless she's really sick or just not moving to be stubborn.
Anyway, arch enemy, nemesis, what's the difference?
Nemesis, by definition, wins?
but I wondered what sociologists would say about historians, in kind
If we're trafficking in caricatures, something like "You say my argument doesn't quite apply in its details to the situation in the Swabian village whose structure and development c.1420--1435 you have devoted thirty years of your life to studying? I'm willing to bite that bullet."
It's a truth universally acknowledged that societal attitudes aren't manifested in fiction.
It is a truth universally acknowledged that a philosophy professor in possession of a great book must be in need of a narratologist.
I want a container of smooth hummus with great big heaps of garlic in the middle but I'm so hungry and the Arby's is so much closer than the grocery store.
Isn't 24.1 addressed by nonparametric methods (which nobody uses)?
I used to be worried about the "ease of computation" thing, but my understanding is that when used for prediction, simple methods often outperform more complicated ones.
Though in practice I have seen gains from using sophisticated techniques *correctly*, sometimes.
78: I also neglected to comment on the hummus thread.
My wife trained me not to put too much (that is to say, almost any) garlic in my hummus, not merely by refusing to eat it, but also by refusing to get near me for days after I ate it.
It is a truth universally acknowledged that a universally acknowledged truth is neither acknowledged, true, nor universal.
WHAT IS A SOCIETAL ATTITUDE? This is making me nuts.
83: Be careful. Society looks down on angry women.
They extrapolate from the tiniest little shreds of fictional evidence that "societal attitudes toward
This describes a significant fraction of conversations here, mostly the ones with lots of comments.
85: That's right! AWB, are you daring to insinuate that our work here on Unfogged isn't valid scholarship?
Judges have to make judgments about societal attitudes all the time, and they are often quite cogent. For instance, in Roe v. Wade, Blackmun had to decide that at the time the relevant amendments to the constitution were written societal attitudes toward abortion were relatively lax. I think he even used the phrase "societal attitudes."
12
an advanced case of social constructivism.
It might be this? She has a tendency to think everyone is full of shit and kidding themselves when they claim to know anything with any certainty. As a life philosophy, more broadly than just this statistics bit.
Ugh, this sounds annoying. I'd try to change the subject when she gets talking like that if I knew her, or, failing that, try to avoid her.
I have a friend who's like that about some things. He's worried about the future of his current relationship because he's a committed agnostic, and his girlfriend has said she'd want to raise children in a churchgoing family - apparently she has no denominational preference and isn't the type for something too strict, it's just the good-environment-for-children thing - and he doesn't want to do so because he can't find a church that is open enough to the possibility that it's wrong. He's endangering the future of his relationship with a hot blonde doctor with whom he apparently gets along fine otherwise because she wants some kind of church, but no churches are insufficiently uncertain for him. He's a know-it-all in general in some ways, which is also annoying, but it's particularly baffling to see a know-it-all about uncertainty.
As for the general sociology issue, at this point I'm inclined to blame the media (what a great scapegoat in general) and/or modern culture in general, rather than anything about the discipline itself. It's just that even if sociology was done perfectly, it would still lend itself to overly simplistic narratives and overly broad applications and politicization. Like evolutionary psychology.
88.2: The Unitarians are the go-to for this. (Although CA's father became Quaker in the 80s for the purpose of giving CA cover for a CO claim in any potential future war started by Reagan, and they told him they were cool with his agnosticism. I think that was just that particular congregation though.)
he doesn't want to do so because he can't find a church that is open enough to the possibility that it's wrong.
Not even the Unitarians will do? That is extreme.
As an example of how screwy statistics are in practice, note that the average error in election polls from the last 3 weeks prior to an election is larger than the typical reported margin of error. A lot of statistics seems to be a game that people play to make their results appear stronger than they actually are.
Or something changes during that last three weeks. I assume people keep campaigning for a reason.
88. Sounds like he'd be well shut of her. She wants her kids brought up in an environment which will teach them the meaning of life, the universe and everything but she doesn't care which? WTF? I think he should go for the First Church of Beelzebub Rampant.
88: He sounds perfectly reasonable to me. Would it bother you if he just said he didn't want his kids raised in a church at all? Also, I think her wanting to raise kids in a religion she doesn't herself believe in because it will be "good for them" is itself a perfectly plausible dealbreaker. And "hot blonde" as some kind of counterweight? Really?
an advanced case of social constructivism.
It might be this? She has a tendency to think everyone is full of shit and kidding themselves when they claim to know anything with any certainty. As a life philosophy, more broadly than just this statistics bit.
My impression is that, in certain sectors of the Humanities, something like this stance is more or less the default position.
Apparently I'll have to go for the Second Church of Beelzebub Rampant.
And "hot blonde" as some kind of counterweight?
Best seesaw ever.
85: It explains a lot of the disagreements we have, in that I am always trying to tell stories about individual people I know, and then someone shows up to tell me that "societal attitudes" are statistically totally different, so I am wrong. I have never figured out what the fuck that has to do with all the people I personally know.
I married a hot blond. There is something to be said for it.
then someone shows up to tell me that "societal attitudes" are statistically totally different
If they don't cite any specific piece, you can probably just ignore them.
83, 100. For example, when you make a joke that relies on lots of shared assumptions, you let other people know that you expect them to feel the same way. Since it's easy to notice these shared assumptions in familiar contexts (like a joke between one individual and another who know each other, have a worplace and ranks there in common), people keep doing this even when the context is unfamiliar or uncertain, like when something is published.
I'm curious, do you think that old letters can reveal social attitudes? I sympathize with papers like this being carelessly thought through, but it's not an insane thing to do, looking for unintentional revelations.
I married a hot blond. There is something to be said for it.
Particularly if you live in a cold climate.
103: At this moment, I'm working on a project related to "societal attitudes" about obscenity, but my point is to show how each writer in this ongoing conversation has a wildly different assumption about how the average person thinks and responds to texts. Yes, I think texts can be used to show the author's assumptions about "societal attitudes," but that's a construct of the text, not evidence that real live people all thought the same thing.
88,95 Based on societal attitudes towards black T-shirts with words on them, I bet that Cyrus' friend wears these shirts more often than Mr Blandings. p=0.002
105. Does it make a difference to you if the texts are written to be published, against cases where the writer expected a small audience or none?
I married a hot blond. There is something to be said for it.
Don Draper cannot agree.
100: I once had an anthropologist friendish colleague try and justify his (half-drunk) casually misogynistic comments to me later on the basis of the necessity for social scientists to pay attention to general truths over individual variation. Per AWB's point, part of my response was that us folks who don't fit into standard issue gender boxes were significant (conceptually and just genearally) whether we proved to be statistically significant or not. I also told him I thought that was a bullshit excuse for thinking you could legitimately talk sexist shit about strangers at a bar.
106.2: It always makes a difference what the intended audience is. But in the period I'm looking at, there is no coherent conception of the public sphere, so there are quite a few people participating in this argument who just seem completely baffled by one another's assumptions about what or how the "public" thinks. I don't want to get much further into it, as any more detail would be identifying, but it is funny to see these various people writing "Everyone knows this filth is damaging to the moral life of young people" and then others writing, "Everyone knows those other people are perverts who jerk off to the idea of damaging the moral life of young people."
90: Exactly. When we talked about this I was tempted to dig up materials by or about them to show that some religions really are very non-committal, in case he just didn't know, but didn't bother.
95
Would it bother you if he just said he didn't want his kids raised in a church at all?
Maybe a little bit, but not as much as his current attitude.
It's not like I agree with her about religion; I think the benefits of the church-as-a-good-environment-for-children (CAAGEFC) idea aren't worth the costs. I think she's wrong. I just don't think that her CAAGEFC idea should be a deal-breaker, especially considering that (a) she just has a mild version of the CAAGEFC idea and would probably be satisfied with Unitarianism rather than Southern Baptism (for example), (b) his stated philosophy is just agnosticism rather than atheism, and (c) making a big deal about it at this stage of their relationship seems like borrowing trouble (they're both moving hundreds of miles apart soon due to dream jobs in opposite directions). I think he seems to be too hung up on philosophical minutiae rather than enjoying the relationship while it lasts.
And "hot blonde" as some kind of counterweight?
Forgive the superficiality - mentioning her profession was superficial of me too - it all just seemed like examples of how she's generally a catch. He's commented on how they get along well together, and she seems nice enough the few times I've met her.
Speaking of doctors, Sen. Tom Coburn, M.D. gives his name just like when he spam emails you. Ron Paul leaves off the "M.D." whenever I've seen the name given but he hasn't spammed me so I don't know for sure.
Wikipedia says he's a deacon. I'm hoping he goes with Sen. Rev. Dr. Coburn.
To the OP, I'm not a quanitative historian, so can't speak to 1) or 2), but Gonerill's right that this is a complaint that historians generally have about theory building and application in the social sciences. In grad school, my cohort thought that it was odd that the Sociology department on campus (and elsewhere) were suddenly having a World Systems Theory revival, as nothing seemed to have changed from twenty years prior when we all decided it was a pretty limited concept and dumb in its more totalizing applications. But globalization had suddenly become hot, so Andre Gunder Frank got some extra attention towards the end of his life, before folks remember that he was a disagreeable crank.
The other general critique that I'm familiar with, particularly of historical sociology, is that they want to use history as an explanatory backstory, but without treating it as part of a contingent process. Part of this is disciplinary grumbling ("You're not doing it right/using my work the way it was intended!") and part is more serious methodological critique of cherry-picking/teleology.
but I wondered what sociologists would say about historians
I think I linked to this comic the last time around, but it still cracks me up. A running joke* in my grad cohort involved my quant-head friend mocking another friend's research: "What's your method? Reflection?"
*Maybe you had to be there.
"What's your method? Reflection?"
I thought everybody decided to call it "thick description."
115: The historian would be standing between those two, either trying to break up the fight or (more likely) egging both sides on. "Hey, we use data, but we're not so full of ourselves to think that that makes us scientists!"
I have only skimmed the thread. There are some fundamental debates about whether Fisherian hypothesis testing is a legitimate enterprise at all. There's an argument from improbability at its heart: these data are improbable given the hypothesis, therefore, the hypothesis likely to be false. Some (many?) think that that argument is a bad one. Of course, Fisherian hypothesis testing is hardly all of statistics.
If studying historical literature is not, in some broad sense, helping us to understand "societal attitudes"* I have absolutely no idea why it is being taught in universities as an academic discipline.
*Understanding that this has to be done in a smart way, not in a stupid undergrad way.
119: If you would like some information about what English literature studies is, I'd happy to tell you. Hint: it's not Historical Sociology Lite.
Ditto the appreciation of 4.
49: Can you expand on this? I recently reviewed an article and asked the authors to change from *0.05, **0.005, ***0.001 to just *
"What's your method? Reflection?"
I went to a research presentation event thing - not a conference, but kind of conference like - a while ago and thought that if I'd submitted something and had it approved, I'd put up a slide that said:
Methods
1. Read stuff
2. Think about stuff I read
3. Write some stuff down
The research was pretty much all social science, though arguably some fell into the category of "not even...".
Whoops. Forgot the less than sign doesn't work. Imagine that it says "less than 0.05" after the *.
120 was bitchy. But seriously, that's not what we do. We teach the history of English literature in order to study the history of meaning-making in English, through forms and genres, tropes and rhetorical structures, narrative and poetic conventions, etc.--all of which require us to learn a little genuine history, theology, philosophy, etc. We don't do it to teach undergrads how to make sweeping generalizations about what "really" happened in real life.
And also that I asked if I was being weird about that.
My very small field is very anti-different degrees of significance but I don't know if it's widespread.
I didn't mean that to sound as dismissive as it came across. I've been around English departments all my life -- I just mean that there has to be some broader connection to social reality or else the entire enterprise collapses as pointless. Obviously English people are good at teasing out the ways in which texts reflect that reality in complicated ways -- thus making your undergrads conclusions stupid. But ultimately, if we're not just reading literature for the fun of it, there has to be some reason we care about the deep study of historical literature, and that has to be tied to reaching conclusions about social reality. "The presentation of sexuality in 17th century British court drama was really ambiguous and complicated in the following ways" is a conclusion about societal attitudes, just not a naive and overgeneralizing one.
We usually give the 95% confidence interval and not the p value.
126: Are conclusions about how language has worked to create meaning not in themselves a meaningful goal of research? I'd say that concerns about what actual people thought in their real selves are not at all the goal of my work, or of my students' work. But I am an extremist in this sense. We are studying literature and language, not real people, and studying language is as important as studying math.
Anyway, I was once subject to people with very strong opinions about putting in the exact p value, so I always put it in the raw tables and somebody else usually takes it out at some point.
I always thought it was weird when someone in a lit class I was in - ok, I only took two - had a paper topic on something like what Bleak House says about the working class in England (ok, probably not that for Bleak House, but something like that). Whenever I heard a topic like that, I'd figure maybe they've done some research outside the text that we all just read? Meanwhile, I wrote about spontaneous combustion as a literary device.
But I can't remember all my paper topics, so some of them might have failed on the same grounds.
meaning-making in English, through forms and genres, tropes and rhetorical structures, narrative and poetic conventions, etc.--all of which require us to learn a little genuine history, theology, philosophy, etc. We don't do it to teach undergrads how to make sweeping generalizations about what "really" happened in real life.
Right, sweeping generalizations are dumb, but studying "meaning making" in texts is ultimately a means of studying a particular way in which social meaning is generated and reaching conclusions about it. I think we probably agree in reality but are having a definitional problem with the not very clarifying phrase "societal attitudes."
So far I rate this thread as significantly less dumb than the previous one (p < .05), but it's early yet.
116: In this particular case, my friend's method is more like that in 122, which is why we all found the joke funny.
133: Way to take the time to get the 'less than' symbol to appear.
Are conclusions about how language has worked to create meaning not in themselves a meaningful goal of research
Absolutely those conclusions are a meaningful goal of research, but is not ultimately figuring out the ways in which language has worked to create meaning a particular means of figuring out how "societal attitudes" are created and reflected through a particular kind of art? I mean, ultimately, you are dealing with a real text that was in the real world and read by real people.
131: Unless they've done some serious historicist research, a paper like that won't make it past my desk. I'll just hand it back and tell them to write an English paper instead of whatever this is.
132: Sure, insofar as language is a social construct. But the assumption that meanings are somehow stable throughout a national or international public is idiotic. If you talk about texts as negotiations between the implied author of the text (not a real person, but the narrow projection of the understanding and worldview of the person who wrote the text extrapolated directly from the content of the text) and the implied reader of the text (again, not a real person, but the narrow projection of the expectations the text has for the understanding and worldview of the intended reader or readers), then you could make some very measured claims about what the text seems to assume about the way readers think. But you can't say that reading literature is a window into history, which is what students seem to want from it.
But the assumption that meanings are somehow stable throughout a national or international public is idiotic.
Comity!
But you can't say that reading literature is a window into history, which is what students seem to want from it.
A dark, murky window that doesn't reflect things accurately unless you know a lot from other sources already, but still a window, surely. E.g. ancient and medieval history rely heavily on literary texts.
OK, I'm out of my depth, I should go back to seizing the assets of illicit music downloaders (not really).
real world and read by real people
I am not a mind-reader, not even of people I know who are alive and in the same room with me. I do not presume to read the minds of entire "societies" 300 years ago. What do "people" think about sex? Or women? Or labor? I don't fucking know. But I can tell you what a text implies about readers' expectations about sex, or women, or labor. I can't say that has much to do with what actual human beings who lived 300 years ago thought in their private thoughts about sex or women or labor. It might sound like a stupid distinction from outside literary scholarship, but it is probably the most important methodological distinction between literary history and other kinds of histories.
49, 121: It's not incoherent if you're using the Fisherian approach to hypothesis testing, which allows treating a p-value as evidence (I mean, some would argue that that approach is incoherent, but it is how the p-value is seen). It is incoherent if you're using the Neyman-Pearson approach.
The N-P approach is less vulnerable to accusations about its poor philosophical foundations, but I think most people, if they realized what Neyman-Pearson approach really gave them (i.e. a decision rule about when to act as if a hypothesis were true, but literally no comment on whether the hypothesis is, in fact, true) would find it an unsatisfying approach to trying to decide what data tell us.
Really, as soon as you get out of school, that stuff doesn't come up all that often.
it is probably the most important methodological distinction between literary history and other kinds of histories
I think this is really the key issue here. Both literary scholars and historians use literary texts, but they use them in different ways and for different purposes.
141: To 140, Moby? What does that mean? I mean, sure, philosophical foundations of statistics don't come up very often among people in general, or people who are employed doing data analysis, but I don't think that's relevant to their utility or interest.
I do not presume to read the minds of entire "societies" 300 years ago. What do "people" think about sex? Or women? Or labor? I don't fucking know. But I can tell you what a text implies about readers' expectations about sex, or women, or labor. I can't say that has much to do with what actual human beings who lived 300 years ago thought in their private thoughts about sex or women or labor.
I just don't get the jump from
(1) You can't naively read off a society's attitudes about X from literary presentations of X (including literary presentations of attitudes about x).
to
(2) You can't know/learn anything about a society's attitudes about X from literary presentations of X.
I'm not saying it isn't interesting. Just that most of this stuff is considered "settled enough to be going on with" and people go on with standard hypothesis testing as output from standard statistical software.
142: And I think the confusion about what literary scholars do is further complicated because, unlike a lot of other scholarly subjects, people read the same stuff I teach for the purpose of entertainment.
I had a really interesting independent study this year with a student who is a fan of the author he wanted to discuss. He reads these books for fun, not for his major, which is in the social sciences. I realized very early on that negotiating an English methodology with him on a set of texts that he reads for pleasure would be enormously difficult, and that I would have to be extremely clear about my expectations for methodological rigor in our conversation. Luckily, he's a super-bright guy and got on board, and ended up writing a beautiful final project, but it was tough, and it made me clarify to myself exactly what the study of literature is and what it's for. I'm glad I got a chance to think through it, but it was a bit nerve-wracking for a while.
144: My bias against (2) is that I work in a period in which there really isn't such a thing as "society." I have no idea what that term would be meant to include or disinclude in the 17-18th centuries, and why you wouldn't just go ahead and specify which community you're talking about. What did titled London male Tories in the 1720's think about female sexuality? Even then I'd be hard-pressed to make any generalizations, and I wouldn't ever do so based on literary texts alone.
But I can tell you what a text implies about readers' expectations about sex, or women, or labor. I can't say that has much to do with what actual human beings who lived 300 years ago thought in their private thoughts about sex or women or labor. It might sound like a stupid distinction from outside literary scholarship, but it is probably the most important methodological distinction between literary history and other kinds of histories.
Sure. It's important to have humility about the reach of what a deep analysis of a famous novel (for example) tells us about people's real lives. But stating what a text implies about readers' expectations on a lot of issues is no small beans -- that's revealing something interesting. And, ultimately, the deep study of the novel's treatment of various issues is interesting because it's revealing something about how meaning was constructed at a particular moment in time and in a particular manner (which may have been a very complicated process). If it's a subject completely divorced from the real world, (a) the analysis is likely to be poor and misleading and (b) who cares? Which is why most English people I know are super smart and super careful about knowing a lot about their period, etc.
Question asked at a conference I was at recently, when a discussion of systematic uncertainties got kind of contentious: "did you say 95% confidence interval, or 95% competence interval?"
AWB, thanks for clarifying; as always comments about methodology are interesting.
I had a really interesting independent study this year with a student who is a fan of the author he wanted to discuss. He reads these books for fun, not for his major, ...
That does sound like it would require a lot of time unpacking assumptions. It's often more work to go back over material that you believe you know, and correct your beliefs, than it is to learn something new.
149: What if we thought about literary scholarship as having more in common with art history than with the social sciences? That is, no one says art history is pointless just because real people in the 12th century are not accurately depicted by icon paintings. Or that the main reason to look at 12th-century icon paintings is to learn what life was like for people in the 12th century. We look at art to see how a particular work attempts to create some kind of meaning or experience from a perspective we can extrapolate from the work and for an audience whose assumptions are projected by the work, and even still, those extrapolations might have little or nothing to do with what the real painter (who may have been painting on commission or at swordpoint or God knows what) really felt about the content of the work, and the projections might have nothing to do with what real people (who maybe had to poop really bad that day when they were at church and saw it) felt about it.
Right, but no one would (or should, I know that there have been attempts to do this) say that art history reveals nothing about society. You're analyzing a particular form of social meaning, and if you divorce art history entirely from its context you're just not doing very good art history, and it's not very interesting.
Ach, I wish I had time to catch up on this thread, because I am quite sure we can make some rough generalizations about societies and their attitudes, and that we actually need to do this in order to function. For instance, "American society frowns on polygamy, while many African societies do not."
Certainly we need to be able to talk about societal attitudes to make sense of discrimination (on the basis of race, gender, religion, sexuality, etc.) which is something we are all concerned about.
What does it mean to say that there was no society in the 18th century? Did people live alone like orangutans?
||
South East of UK people, are we all still on for tomorrow night? With the lubbock-strandees?
Can someone remind me which pub it is in Pimlico?
>
My bias against (2) is that I work in a period in which there really isn't such a thing as "society."
AWB is Margaret Thatcher?
155: I wish someone would just tell me what "society" means.
Just that most of this stuff is considered "settled enough to be going on with"
By most people, true, but actually in my field (psychology, in a subcommunity at the most neuroscience-y end of it) I'm seeing the beginnings of a movement toward a Bayesian approach to inference. Why the likelihood approaches get no love, I don't really understand, but I think it's for, erm, historically/societally contingent reasons, not principle. I'm also not sure that most of this stuff is *rightly* considered settled enough to be going on with, and doesn't deserve a lot of examination.
There's a lot of reason to think that science that rely on statistical inference are really fucked up. Maybe I'm just imagining that it's more of a problem in sciences that rely on statistical inference -- sometimes when I chat with physicists they're all, let me tell you about unreplicatable results in *my* field. Anyway, I'm far from knowing the answer, but it's an interesting question -- in fact, a question for the sociology of science -- whether the choice of one versus another theoretical justification for statistical practices, and all the methodological variation that stem from it, leads the sciences that rely on statistical inference to be more fucked up than they would be otherwise.
144: I don't either, but I thought at first AWB was more taking the line that a single text from time and place Y relating to X is almost completely useless in building up an understanding of X in time and place Y.
My first exposure to the idea that "society" might not have the meaning I thought it did was in the book Debt!
[page 66] What makes the concept of society so deceptive is that we assume the world is organized into a series of compact, modular units called "societies," and that all people know which one they're in. Historically, this is very rarely the case. Imagine I am a Christian Armenian merchant living under the reign of Genghis Kahn. What is "society" for me? Is it the city where I grew up, the society of international merchants..., other speakers of Armenian, Christendom, or the inhabitants of the Mongol empire itself..?
...
[page 69] Let me return again to that word, "society." The reason that it seems like such a simple, self-evident concept is because we mostly use it as a synonym for "nation." After all, when Americans speak of paying their debt to society, they are not thinking of their responsibilities to people who live in Sweden.
The web of social relations in which individual humans are embodied? A literary text is just as enmeshed in a particular, historically contingent set of social relationships as all the rest of us are.
Can someone remind me which pub it is in Pimlico?
distinction between literary history and other kinds of histories
There's also a distinction in this thread between "society's attitudes" and "attitudes held within a society (or among members of that society)".
I wish someone would just tell me what "society" means.
Would it be fair to characterize comments about "social attitudes" as meaning two things (a) it is likely that a person with characteristics [X] will have, as a first approximation, opinion [Y] about subject [Z] and (b) that people are likely to expect that people with characteristics [x] . . .
That is, two statements, both about probabilities, the first about the probability that people actually believe something the second about the expectation of the probability that people believe something.
"What's your method? Reflection?"
Epoché!
163: So does society include servants? Merchants? Men and women? Straight and queer people? Children? Disabled people? The clergy? Teenaged girls? Illiterate people? Academics? I'm just trying to get a read on how broad my speculations about "society's" opinions has to be. 155 suggests that we could make speculations like "Eighteenth-century British people did not generally eat or enjoy Thai curry." Great! But if you're speculating that broadly, what's the point?
I think the disaster zone that is medical research suggests that the way that statistics is handled should not be considered settled. There's some real problems in medical science coming from the intersection of statistics with the current structure of funding, publishing, and research.
Methods
1. Read stuff
2. Think about stuff I read
3. Write some stuff down
When this sort of thing is said in literary studies, I pretty much immediately dismiss the speaker. Oh, you "just read" some stuff, okay! You don't "do theory" or any of that crap! Except that most of theory is about the ways that we read stuff, about examining and making more conscious the ways we approach texts and make meaning out of them. It's not like by refusing to deal with theory you don't take a particular approach, you just don't make it explicit to yourself.
159: I think part of it might have to do with the availability (or not) of the double-blind randomized trial in that field. The whole point of that type of experiment is to make sure that your prior conceptions don't influence your interpretation. At the very least, I'd be concerned that using an informative prior in certain ways is likely to allow the investigator's preconceptions to have too much rein.
I'm not entirely sure in what tangible ways using Bayesian inference and a prior based on the principle of indifference is different from using regular stats.
Also, my lawn and the kids on it.
While we're on the subject of English as a discipline, I was wondering the other day what the really awesome results are in the field. Of course most good scholarship in all fields is detailed technical and not of very broad interest, but occasionally you get real breakthroughs. What's a nice example or two from say the last 50 years of major breakthroughs in English scholarship? (I.e. something interesting, genuinely new, and universally acknowledged as true by experts, and accessible enough to explain why it's interesting to the generally educated public.) Most fields I know well enough to come up with a few examples, but I don't actually know English well enough to do so.
172: if you choose not to choose you still have made a choice
That lyric pretty much explains everything.
170: In what way is it a disaster?
172: It was (or would have been) a joke.
173: In the long run what you want is some centralized systems updating their priors based on all properly conducted research (including negative results). There'd still be some initial bias, but over time you'd expect those to go away, and at any rate the biases of the individual experimenters would not come in.
I mean, it's certainly producing fewer deadly-disease cures per dollar than back in the day, but those were easier diseases and ethics cost money.
I have a pile of books on the historical foundations of statistics. It's very interesting. I am also embarrassed that I've forgotten everything I read in them that would meaningfully contribute. Still, let me show off that I have some books!
174: I heard that they've found conclusive proof that William Shakespeare was actually William Shakespeare.
176: Companies run multiple studies and bury the results that don't come back "significant" and then publish the ones that do. The placebo effect is growing over time which means we're doing something horribly horribly wrong.
The placebo effect is growing over time which means we're doing something horribly horribly wrong.
Really? I thought it just meant we've done a great job of convincing people that drugs really work.
If we were sure that's what's going on that's fine, but it still results in huge problems that if you did studies early enough your drugs are ok, but now they wouldn't be good enough to beat a placebo!
178, 182: I agree that making sure negative results are reported is important, but there are real differences between studies that look at different populations. And conducting an actual trial costs millions to tens of millions. No one is deliberately trying to retest the same compounds on the same types of people with the same conditions to make money.
I mean, they'll try it for a new labeling indication of something that has already proved a big seller, but that's slightly different and mostly a problem caused by IP law.
174: Queer theory is a good example, in that it has given us terms for articulating social, erotic, and sexual relationships that have always been in the texts we study, but have been to some degree rendered invisible by the enforcement of heteronormativity in the late 19th and in the 20th centuries. A lot of good useful theory works this way, in that the texts we read explicitly address ways of thinking about race, sexuality, gender, class, labor, disability, globalization, whatever, but the standard 20th-c interpretations have been blind to these issues in texts. Gulliver's Travels is a great example of a book that is full to the brim of explicit commentary on every single one of those issues, but has been read by professors of, say, *my* professors' generation, as a clever satire on specific political situations in the early 18th and little more.
Selfishly, I'd say that the return of serious engagements with the history of rhetorical and aesthetic theory have been profoundly transformative to English methodologies. Real interdisciplinary scholarship (not "I said 'money' so now I'm an economist; I said 'brains' so now I'm a neuroscientist"), especially in theology and philosophy, have made huge changes in my field.
My impression, from reading historians online, is that historians think that every fact is a precious, unique snowflake, and the act of putting two facts together into a "generalization" is an act of appalling violence. Anthropologists seem to a somewhat similar attitude.
In contrast, most social scientists would regress apo's penis length on their mother's cat, if they could get a paper out of it.
At least one of those seems like it would be a constant and you need two variables to regress.
188: I knew someone would make that correction. I'm glad it was you, Moby.
186 is all methods. As for truly groundbreaking applications of those methods to particular texts, I could give examples, but the best ones were so successful that they now seem banal. Like, of course we read Dracula as an expression of anxieties about Eastern European immigration, venereal disease, and female and queer sexualities. Duh. That seems completely obvious to even the dumbest undergrad now. Thirty years ago it would not have been.
We look at art to see how a particular work attempts to create some kind of meaning or experience from a perspective we can extrapolate from the work and for an audience whose assumptions are projected by the work, and even still, those extrapolations might have little or nothing to do with what the real painter (who may have been painting on commission or at swordpoint or God knows what) really felt about the content of the work, and the projections might have nothing to do with what real people (who maybe had to poop really bad that day when they were at church and saw it) felt about it.
This is actually something of a bummer for me. A lot of art history is much, much more theoretically interpolated than I would like, and the very few examples I've found that talk about, for example, the opportunities for painting on commission or the contractual structures of commission-painting over time, or the way that guilds protected or shared information about the aging process of varnishes, tend either to be maddeningly vague or insanely expensive. I'm really curious about the physical and material history of art production, much more about that than about the evolution of discourses or tropes in art depiction. It's a lot easier to do scholarship on the latter than the former--which isn't to say that the tropes and discourse stuff isn't also interesting!--but I'm bummed not to be able to find much material art history.
186, 190 have now completely destroyed my understanding of your position, AWB. Aren't these arguments almost exactly the kind of arguments that you rejected up-thread? These arguments only make sense if they are part of historically-contingent societal attitudes.
"Social attitudes" in social policy research are investigated using humungous quantitative surveys and generally reported in rather dry statistical write-ups.
Outside that somewhat specialised field my experience is that it's a term used loosely and misleadingly to suggest the dominant ideology in a given time and place, usually by people who are nervous about the idea of ideologies at all and scared of the idea of dominant ones.
190 is the sort of thing I was looking for.
A literary text is just as enmeshed in a particular, historically contingent set of social relationships as all the rest of us are.
Sure it is, but how much of those relationships can you get at from the texts themselves?
I taught Darnton's The Great Cat Massacre this semester, which is him trying to get at the culture of 18th century France through a set of textual viewpoints, but in order for it to have any meaning he has to divide French societies into a number of different overlapping classes and orders and he's drawing on a great deal of background materials to enable his interpretations. The book is 35 years old at this point, and there are many areas where it overreaches, but I think it supports AWB's point that you have to know a lot more about author and audience than a literary work itself can tell you in order to say anything meaningful about the cultural world that produced it.
Even if we treat society as some kind of heterogeneous but coherent unit, ala early Bourdieu, it's very difficult to get a read for norms or cultural presumptions from literary sources alone, because they are self-consciously constructed representations of society for some end. You can maybe tell which ideas the author is endorsing, which attitudes or behaviors they think should be dominant, but it can be difficult to tell whether those ideas reflect or challenge the social order as they understand it.
It's even harder to get at the doxa of a culture from literary worlds, at those ideas or attitudes that are so uncontested as to be assumed. There's a whole body of historical/cultural theory that holds that there is no concept of individual liberty prior to the Enlightenment/French Revolution. At this point, however, I think it's impossible to distinguish between the absence of such a concept and its grounding in a different set of surrounding concepts.
I'm really curious about the physical and material history of art production, much more about that than about the evolution of discourses or tropes in art depiction. It's a lot easier to do scholarship on the latter than the former--which isn't to say that the tropes and discourse stuff isn't also interesting!--but I'm bummed not to be able to find much material art history.
It sounds like you might be looking in the wrong discipline. Art history may not look at that sort of stuff much, but I would think history might (especially in subfields like economic history and history of technology). For certain times and places even archaeology may be useful for this.
I feel like people in this thread keep using the nonsense phrase "societal attitudes" just to make me lose my fucking mind.
I feel like people in this thread keep using the nonsense phrase "societal attitudes" just to make me lose my fucking mind.
I feel like they are instead responding to your use of it in the first place.
I'm really curious about the physical and material history of art production, much more about that than about the evolution of discourses or tropes in art depiction. It's a lot easier to do scholarship on the latter than the former--which isn't to say that the tropes and discourse stuff isn't also interesting!--but I'm bummed not to be able to find much material art history.
My memory of undergraduate art history classes is that Michael Baxandall is good on this.
198: How about "intersubjectively developed and maintained localized peer-group attitudes among the (specify)"?
198, 199: I checked. teofilo is right. You started it, AWB!
I would guess that the proportion of material art history within economic history or history of technology is not much higher than you find in art history.
It's probably like book history/history of the book: you can find it in a number of fields, but you still have to do a lot of looking in any of the fields where you'll find it. Although book history seems to have gotten more popular as the book has (allegedly) begun to die.
190 doesn't talk about "societal attitudes." It's about the means by which the implied author of Dracula develops the vampire in particular ways in order to exploit what he perceives as these anxieties in his implied readers for a set of complicated and even conflicting literary purposes, about which literary scholars still argue. You can't just read Dracula and say it's proof that "society" in the 1890s thought this or that.
202: I started it in order to complain about idiotic undergrad methodologies that are banned in my classroom!
"I learned it from watching you!"
I would guess that the proportion of material art history within economic history or history of technology is not much higher than you find in art history.
Probably true, yeah.
I feel like people in this thread keep using the nonsense phrase "societal attitudes" just to make me lose my fucking mind.
It's an abstraction, like "science" or "natural." Used in certain contexts, it's empty or not very informative. Used in a sentence like,
Societal attitudes towards sex-selective abortion are different in contemporary China and the United States.
it's meaningful, if still needing a lot of qualification depending on what point you want to make.
205: We're a bunch of crazy rebels. You tell us something is banned and we just have to do it.
See, for example, analogies.
Sorry, I got called away. Thanks, Criminally Bulgar--I may have been trolling for recommendations there.
I would guess that the proportion of material art history within economic history or history of technology is not much higher than you find in art history.
Yes, I suspect that it's very much a function of ideological school (materialist, Marxian) and period than of departmental affiliation.
AWB, you are arguing against a strawman position that no one here is advocating, or is even sympathetic with. No one is saying that you can just naively assume that a literary text provides unambiguous evidence from which you can make a naive generalization about the viewpoint of an entire nation at any given point in time. But each and every one of the examples in 186 and 190 is an example of using understandings about the social contexts in which a literary text is embedded to help explain that literary text, and of using the literary text to help explain that social context. Otherwise, what on earth does it mean to say that Dracula is "an expression of anxieties about Eastern European immigration, venereal disease, and female and queer sexualities."
210: I think Chandra Mukerji's From Graven Images might have a bit of what you're looking for. But I only read a chapter of it and that was a long time ago.
No one is saying that you can just naively assume that a literary text provides unambiguous evidence from which you can make a naive generalization about the viewpoint of an entire nation at any given point in time.
No one here is, but it sounds like that's exactly what her students tend to say (or at least imply), which is what she was complaining about in the first place.
Which is to say that I don't think it's really a strawman, but also that it's not clear that there's actually a real dispute over this going on in this thread.
I mean, it's certainly producing fewer deadly-disease cures per dollar than back in the day, but those were easier diseases and ethics cost money.
That, and apparently there's not an endless supply of efficacious molecules. Genuinely "new and improved" lately is really difficult to pull off.
I take AWB to be saying that Dracula is not an expression of "society's" "anxieties about Eastern European immigration, venereal disease, and female and queer sexualities" because you can't take Stoker or Dracula's readers as stand-ins for some general "society."
the enforcement of heteronormativity in the late 19th and in the 20th centuries
By whom?
I think we can all agree that AWB's students would be better off if they sat down and read some Bakhtin. Or possibly rolled some Bakhtin into cigarettes and smoked them during the siege of Stalingrad.
217 - The shivering Chaplain robed in white, / The Sheriff stern with gloom, / And the Governor all in shiny black, / With the yellow face of Doom?
Right, but I don't think anyone here is saying that the undergrad students were right on. We're saying that rejecting the dumb undergrad methodology complained about is not at all the same thing as rejecting the idea that literary texts are socially embedded, and that context matters for explaining the text and that the text can help you explain the social context.
I don't think in practice this is at all controversial; AWB doesn't just sit in her office, read an 18th century novel, and then tell us her thoughts about it; she knows a lot about the 18th century, the background of the author, the social and literary-historical culture in which it was produced, etc., and the ultimate product of the research is designed to tell us something about the ways in which 18th C authors as a group thought about these issues. I think what's actually going on is that she knows too much about the 18th C context, not too little, and thus is afraid to generalize. But being cautious about generalization doesn't require, or even imply, going full throttle with "there is no society" and "literature tells you nothing about society."
Halford, just tell me what you mean when you say "society" and I'll tell you whether I agree with you. Who is included in your understanding of 18c "society"? That's all I want to know.
24.3 is sinister Bayesian propaganda. "Population mean" is a completely clear atheoretical notion that does not require a prior to make sense of.
I think the danger of statistical methodology is the idea that there actually can be any kind of consistent statistical methodology that can be mechanically applied to produce "truth". All we have is some data, and some tools that either summarize the data, or give us some ability to apply skepticism to what the summary is telling us. The place to look for methodology is not within statistics, but in how disciplines carry out their long-run research programs. Statistics can only ever be a small part of that.
172:
I sympathize with the point that there is no neutral methodological standpoint. But neither is it clear to me that lots of talking about methodology ends up making for better readings of texts. Would you be happier if someone said: "I pretty much proceed on an ad hoc basis throughout, but am open to questions about methodological presuppositions I might be making, whether knowingly or not"?
223.2: well sure. Until our automatic inference algorithms are perfect, that is.
222 further: You're right that not making generalizations is a product of knowing enough about 18c London to differentiate between coffeehouses, rather than even saying coffeehouse society. I just don't think you get to a mature understanding of a period by making generalizations that ignore complexity and specificity. I'd rather see undergrads do very deep, extremely careful research and analysis of a specific problem in a specific text than try to do the undergraduate version of a sweeping survey of society's attitudes toward women or whatever. If you want to talk about the representation of women by male authors of the 1740s as a dissertation project that you research over several years, I'm willing to buy that it is an English research project. But when I've read undergraduate theses that attempt to say how "society" felt about gays in the 90's or how "society" felt about abortion in the 50's, on the basis of some aleatory similarities between a couple of novels, it is almost always ignorant sexist/homophobic shit that ignores that women and gays actually existed and had their own ideas and might even be considered members of "society," if we weren't so busy thinking up generalizations.
222: If the book in question is Dracula, then the answer is "readers of Dracula in that era".
225: Shhhh! Don't let Moby know they're actively working on making him umemployed.
227: s/b "implies readers of Dracula in that era," no? I have no idea what every real reader of Dracula got out of it. I can make some educated, informed assumptions about what the text implies that readers want or will get from it.
But that's what makes your work, interesting, right? In some sense, the mental states of people in the past is completely inaccessible to us, since we have such indirect reflections of those mental states. So your job is to provide the education, informed guesses.
Saying part of what drives Dracula is "anxieties about Eastern European immigration" is interesting in a way that saying that what drives it "anxiety about nuclear war," and that's not just that amount of textual evidence you can find to support the first one is greater than the second. If we could somehow find out that there was no anxiety about Eastern European immigration, and that Stoker wasn't personally anxious about it, then it makes the first reading much less interesting.
undergrads...extremely careful
Those sound like different undergrads than the ones I knew.
I'm really curious about the physical and material history of art production, much more about that than about the evolution of discourses or tropes in art depiction.
Michael North's Art and Commerce in the Dutch Golden Age is really good. He goes into a lot of detail about how many paintings there were, who paid for them. Even families that didn't have much often owned paintings.
I'm very interested in learning more about the economics of Chinese porcelain, but even the physical locations of some kilns are not known.
More broadly to Upetgi's question, as a layman I'm not sure that this is a useful way of thinking about the field. The hard sciences, by and large, deal with facts: "HD 189733 has spectroscope reading X" and "one possible cause of spectroscopic value X is the presence of water vapor on a planet". English, like most of the arts and many of the social sciences, deals much more with interpretation. Interpretations are going to be much less likely to be universally adopted among professional scholars of a field than facts are*, and while I could imagine that some promising researcher could make her bones with, say, tangible evidence that William Sly was actor who played the Third Assassin, that sort of thing seems to more the domain of historians.
* Look at the muddle economics has gotten into as what seemed to be universally accepted interpretations became political expedient for people to ignore. Or climate science, although that's a slightly different case.
I think people keep inventing new statistics to make me feel old and dated.
All I've been arguing all along is that literary study should not see doing history or sociology as the endgame of its arguments. Our work should be fundamentally about literature and language and how they have worked. Sure, that requires being extremely well-educated about the historical phenomena that the content of texts addresses. But the purpose of thinking about literature and language isn't to tell us what everyone everywhere once thought about something, any more than one studies historical theology to get an accurate representation of how every single religious person in an era thought about and experienced God.
Admittedly, the semester I spent studying School for Scandal was along your lines; I (tried to) analyze it in the context of the food-culture of the time and ended up discovering great things about the invention of takeaway, which informed some elements in the eventual production. Perhaps that's more normal than it seemed in the context where I did it (the tiny theater department of an engineering university).
Honestly if there aren't at least a bunch of things that are close to universally accepted among experts in a field, then I don't think that it's scholarship worth doing. But I see no reason to assume that other people's fields are not making progress or figuring things out.
224: I haven't done a study according to any scientific method, but my impression/experience is that - in history, at least - people who have actually taken the trouble to be reflective about their methodology tend to produce more interesting work, even if that work does not explicitly incorporate all that much from the theory/models/methods that they engaged with.
(To a certain kind of social scientist, it still all just looks like reading and writing, which is what my joke would have been aimed at. I mean, to the extent that I'm still doing history, I don't actually think of myself as "just reading".)
236: See? That's some interesting historicist research! I have a student who is spending the summer doing some extremely deep international research on mathematical theories and education in the late 18c and its relationship to [specific identifying stuff]. But it's still an English project because he's looking for instances in which autobiographical narratives use the language of mathematics in order to dog-whistle the author's gentleman status to other gentlemen. It's about rhetoric, even if he has to learn a great deal about how mathematical study seems to have signified a certain kind of class status to other members of that community.
Let me take a different example, as a test case for your view. HP Lovecraft's early fiction is clearly driven by anxiety about immigration -- it's not subtext, it's text. This early fiction barely registered on the world when it was published, and would be completely forgotten if it wasn't by a famous author.
In his later stuff, you can still see where the anxiety about immigration peeks through, which suggests that this anxiety is an important reason why Lovecraft wrote the stories that he did. But the fact that his later stuff -- where the anxiety is considerably more submerged -- is what made his reputation suggests that the anxiety is less important for the effect on the reader.
In that sense I would say we can provide evidence for how a particular work is actually affecting readers.
238: Within specific fields in English, you'd find that it's the same with us. Here is a particular theoretical approach to the history of the novel that dominates the discourse and what goes on a syllabus in our field for fifty years. Then it changes when there is a new dominant approach that emerges and the former understanding is discarded.
I'm not saying that undergraduate overgeneralizations are defensible, not even a little. One can assume, though, can one not, that when someone speaks of 'society' they mean to exclude those persons/groups that one would properly characterize as 'marginalized' -- along whatever axes are relevant. Society thus defined is a completely fallacious description of a population, but may be a useful concept when looking at how opinions of the non-marginalized affect other non-marginalized people.
230 - Not to step on AWB's line, but there's rarely a unified readership to consider; I'm not sure there's any reason to think that Bram Stoker, a financially successful Irish theatrical manager and a friend of Walt Whitman's, felt personally threatened by Eastern Europe, and Dracula wasn't initially a huge commercial success. But weird feelings about Catholicism are clearly the water in which Stoker swam.
241: OK, but do you say that Lovecraft merely breathed in a universal xenophobia and breathed out his fiction? Hell no; that motherfucker is CRAZY. The way he expressed his xenophobia is really special and weird and should be analyzed, sure, in the context of an audience he assumed would share his xenophobia to some degree, but also to whom he is advertising xenophobia in an extremely specific and unique way.
I teach Blake all the time, and he is also completely nuts and wonderful, and nothing makes me sadder than when a student says, "Oh, yeah, 'Auguries of Innocence,' it's just like what everybody says about like karma, how what goes around comes around." Yes, it's JUST LIKE your fucking stupid cliché. What are we doing here? Let's just translate all of poetry into clichés and stereotypes then go home.
246.last: Comity! King Lear is about how shit happens, and then you die.
237: In 1993, I took my first stats class. Nobody mentioned Bayes and I felt young. Now I sometimes hear about Bayes and feel old.
243: Yes, this is what I want to argue against--this idea that "society" exclusively means "oppressors." Women don't get to be part of "society." Gay people don't get to be part of society. Black people aren't society. When students say "society," they mean to specify rich straight white men, so why not just say that? They think it's so liberal to talk about how society hates women, and I'm like fuck you, man, there were all kinds of women in society and your refusal to acknowledge that they were there and that their ideas also matter is really fucking sexist.
The point of Moll Flanders, as I have been told by like 500 undergrads, is that "A girl's got to do what a girl's got to do." I have started explaining even before they pick up the book that I most certainly do not want to hear that "A girl's got to do what a girl's got to do." And the conversation is quiet as they vibrate silently in their seats and then someone says, "I just really don't see how this book is not about how a girl's got to do what a girl's got to do."
I teach Blake all the time, and he is also completely nuts and wonderful, and nothing makes me sadder than when a student says, "Oh, yeah, 'Auguries of Innocence,' it's just like what everybody says about like karma, how what goes around comes around." Yes, it's JUST LIKE your fucking stupid cliché. What are we doing here? Let's just translate all of poetry into clichés and stereotypes then go home.
One of my favorite professors was really good about forcing people to continue to reference the text during class discussions, rather than wandering off into personal associations. It made for good, very intense classes.
It did, perhaps, also keep the class closer to his personal vision of the text, as well as the text itself, but I do think it can be very good to be constantly asking, "can you find a specific passage in the text that would illustrate what you're talking about?"
We should put agronomists in charge of left-wing politics
This would imply a more or less direct handing of the torch from JK Galbraith to John Quiggin (both of whom started as agricultural economists), so yes.
I take it that the problem isn't just saying "society says x", but saying that without knowing what the hell they're talking about, instead going off of vague feelings about what the past was like.
I think the deepest problem with stats is that it encourages you to assume reality follows the ergodic axiom. Which sometimes seem to be right but in many / most fields of study (especially the social and human sciences) is wrong. Related, it encourages ignoring theoretical assumptions -- there are usually lots of unexamined auxiliary hypotheses involved in converting a statistical correlation into a causal statement, which is really what people are interested in. Even with all the attention to correlation/causation I think that's still true.
by the way, I just saved the world (or at least, convinced a Reuters journalist not to run a story that would quite possibly have caused a catastrophic bank run in Greece, a few days earlier than they were due one anyway), on Twitter. I will accept the gratitude of the Greek people at any time between now and never (do I hear "seller at Never"?). Twitter is ace.
This early fiction barely registered on the world when it was published, and would be completely forgotten if it wasn't by a famous author.
I am enjoying the tension between the two clauses of this sentence.
Go order yourself a homeopathic cocktail, young Davies!
253: I didn't know that. I just like the word agronomy.
I also punched someone at the weekend.
I punched someone on Monday [gloves, etc], but I expect yours was more al fresco and less good-natured? I assume the person had it coming?
This is the kind of spirit I want to see more of at the London meetup.
Maybe "the weekend" is the name of a bar.
ttaM, you have to recall dsquared's previous "fights." He probably waved at a guy thinking they were acquainted but actually weren't.
this idea that "society" exclusively means "oppressors." Women don't get to be part of "society." Gay people don't get to be part of society. Black people aren't society. When students say "society," they mean to specify rich straight white men, so why not just say that? They think it's so liberal to talk about how society hates women, and I'm like fuck you, man, there were all kinds of women in society and your refusal to acknowledge that they were there and that their ideas also matter is really fucking sexist.
Yes, indeed, and studying literature gives you an insight into why this is true! But I don't think sociologists go around (when they talk about society at all, except as a shorthand) thinking that "society" just means "oppressors."
If they meant "oppressors", they could just say "The Man".
Everybody's punching for the weekend.
Everybody's Society's punching for the weekend.
Thanks for saving the world, dsquared. I'm Greek-American for what it's worth.
272: It's worth about 134€ (or 23,000 drachma).
I live 25 miles from a town called Frankfort. Surely that's what he meant.
Stupid Germans spelling things funny.
Speaking of things related to German cities, I heard that Danzig was going to switch to polka music and call itself Gdańsk.
129: We are studying literature and language, not real people, and studying language is as important as studying math.
Caveat: I haven't followed this thread much, or followed its tone.
Small point, though: I thought you studied literature, not language use. I agree that studying literature is not the same as studying real people.
I study rhetoric, which is in between, but there are people in English who study literature, and others who study language. I don't exclusively study literature (about half of what I do is on literature), because I work on the forms of argument across genres of writing.
The average high in August in Gdańsk is 68F and the average low in January is 26F? That sounds pretty nice, now that you don't need to deal with communism. For Pittsburgh the numbers are 82F and -6F and Pittsburgh has a very moderate climate compared to the Midwest.
And I will read (am reading) the rest of the thread.
I read the Pittsburgh numbers wrong. It's 82 for the average August high and 21 for the average January low. Still, a wider range.
282: Thanks. I wasn't completely aware of those distinctions. I'm interested, and have yet to catch up to the thread -- have to have some dinner now. Maybe I can contact you off-blog sometime if this is reviving something that did, after all, come up here 150 comments ago.
83
WHAT IS A SOCIETAL ATTITUDE? This is making me nuts.
The average opinion about some subject.
If the distribution has more than one or is seriously not normal for another reason, the average is a completely misleading and pointless measure of central tendency.
287: OK, but what does that mean in a time when there isn't any way of knowing what "average" means because the public sphere is dominated by people who aren't in any way average, and are just imagining what average folks must think? We can learn a lot from literature about what very smart, interesting people think about average folks, but why not leave it at that?
92
As an example of how screwy statistics are in practice, note that the average error in election polls from the last 3 weeks prior to an election is larger than the typical reported margin of error. ...
This is likely because the reported margin of error is just the statistical error based on the sample size. It does not include other possible errors like a biased sample, invented data, poorly worded questions, people lying, people changing their minds in the last three weeks etc.
Sorta seems like a bank run in Greece has already started.
291: Right, which exactly illustrates the point that overemphasis on formal statistical tests leads to reporting misleading results.
293: I'm not going to google this stuff tonight, what with Greece having a bank run and it being late, but I'm fairly certain that polls specifically constructed to predict an election and run by people who predict elections for a living are not off by more than the margin of error very often. At three weeks out, you can't expect 95% accuracy in the margin of error, but that change isn't why the polls are usually off. The stuff in the media usually has no estimate of who is actually going to vote and this is why the general polls you see reported are so often wrong. Of course, the media often has no interest in predicting an election until after the votes are cast and once that is done they don't seem to have much trouble calling elections on the basis of polls (some of them exit polls, but still).
The survey groups often differ in how they determine who is likely to vote, but the results are always very different than when you don't correct for that. Not biasing the sample and having properly worded questions are fairly standard skills that are widely taught in graduate schools and that someone conducting repeated polls can account for. There will be mistakes, but they are rare among the professionals.
289
OK, but what does that mean in a time when there isn't any way of knowing what "average" means because the public sphere is dominated by people who aren't in any way average, and are just imagining what average folks must think? We can learn a lot from literature about what very smart, interesting people think about average folks, but why not leave it at that?
You have to specify what group you are averaging over and whether each member is equally weighted. This generally involves giving more weight to the opinions of the powerful. Which makes sense if you are interested in how society treats average folks. And common people will often internalize the rules society imposes on them so their opinions won't be that different.
294: Nate Silver says:
For a variety of reasons, the magnitude of error associated with elections outcomes is higher than what pollsters usually report. For instance, in polls of Senate elections since 1998 conducted in the final three weeks of the campaign, the average error in predicting the margin between the two candidates has been about 5 points, which would translate into a roughly 6-point margin of error. This may be twice as high as the 3- or 4-percent margins of error that pollsters typically report, which reflects only sample variance, but not other ambiguities inherent to polling. Combining polls together may diminish this margin of error, but their errors are sometimes correlated, and they are nevertheless not as accurate as their margins-of-error would imply.
I thought he also had a post specifically on whether newspapers when reporting on polls should include estimates of all errors, rather than just formal sampling errors, but I can't seem to find it.
294
The survey groups often differ in how they determine who is likely to vote, but the results are always very different than when you don't correct for that. Not biasing the sample and having properly worded questions are fairly standard skills that are widely taught in graduate schools and that someone conducting repeated polls can account for. There will be mistakes, but they are rare among the professionals.
I think this is too optimistic. If an organization conducts many polls you can plot the errors. The curve will sort of look normal but the width will be a bit wider than predicted (additional error) and the mean will often be offset (bias, so called house effects in which some organizations consistently favor Republicans or Democrats).
292
Sorta seems like a bank run in Greece has already started
It's a bit of a puzzle to me why there is still any money in Greek banks that can be pulled.
296: Is the average error of polls of Senate elections since 1998 significantly different from the average error of election prediction polls?
(Just kidding, I'm going to take Silver's word for it though I note that he's not being very specific about whether certain types of polls might be better than others.)
And I'll also note that Silver is certainly not arguing for less emphasis on formal statistical tests. He's arguing for a different formality that seems like what medicine calls meta-analysis.
Oh hai, statistics question: suppose I have two similar probability distributions and I want some way to quantify how many points, on average, I have to draw from the distributions before I can decide which one describes my data. What's the right way to do this? (I have a couple of things in mind that I could handwave as giving me roughly what I want, something Kolmogorov-Smirnov-y and something Kullback-Leibler-y, but I'm a little unclear on whether there's an obvious canonically "right" thing to do.)
301: well, a power analysis, I guess. If you have some sense of how far apart the means are.
Depending on the details of the question.
I'd go with Kolmogorov-Smirnov since Russians are generally good a math-y stuff.
I don't think means are relevant; the distributions aren't remotely normal.
Also, SAS and Stata's online docs have much, much more about Kolmogorov-Smirnov.
I feel like in my classes we mostly used K-L divergence for model comparison. So that might be good, too. Just do what you feel.
Very convincing mathematical arguments, Moby.
I feel like more people have heard of Kolmogorov-Smirnov so going with Kullback-Leibler might make me sound smarter, or at least like I know what I'm talking about.
Kullback-Liebler is neat but then I remember being fed notes of caution about the non-symmetricness of it. Notes of caution that I now totally don't remember, and my Bishop book is at the lab.
So... good luck!
We should totally have a regular feature where Sifu and Moby answer stats questions.
Become a Bayesian and make up your answer.
Good fucking thing you're not in literature, then! All your K-S and K-L would be for naught in the first place, silly. Philosophy would also have to show you the door.
[I am fooling around. I have no idea what K-S and K-L are. Good thing I don't do that.]
I've never heard of Kullback-Leibler, which doesn't mean much. The other one is pretty common.Is this kind of thing what you need?
312: Yeah, I'm a little confused about that. Also the use of it that I've seen before that seems to answer the question I want depends on some prior probability, and I don't want people to pick on me over that. Maybe I'll stick with K-S, which I think I kind of understand.
Stand outside Nate Silver's house with a boom box while wearing a trench coat and ask him.
K-S
trying something new or unfamiliar and failing leaves a bad taste for everyone, especially the fuck-up. Do a bunch of jackknifes with K-S to get an estimate of how many samples should work, increase that estimate by 2 or 10 depending on how many machines you can use. Is efficiency critical?
301
Oh hai, statistics question: suppose I have two similar probability distributions and I want some way to quantify how many points, on average, I have to draw from the distributions before I can decide which one describes my data. What's the right way to do this? (I have a couple of things in mind that I could handwave as giving me roughly what I want, something Kolmogorov-Smirnov-y and something Kullback-Leibler-y, but I'm a little unclear on whether there's an obvious canonically "right" thing to do.)
You might just try simulation. Run a few thousand random simulations with a 50-50 prior and your proposed decision rule and see what your error rate is as a function of the number of points. Make sure you are using an adequate random number generator however.
318: Don't be ridiculous. I would stand outside Cosma's house.
Someone would probably call the cops.
321 doesn't bode well for unfoggedydecahedon.
Anyway, bootstrapping, jackknifing, and other repeated sampling things are a good example of something where computational ease still matters unless you really do have a whole bunch of computers.
Some version of 320 is pretty clearly the obvious lowbrow physicist thing to do. When in doubt, simulate! I just thought there might be a more canonical answer. 319 would sound good if I knew what a jackknife was. Is a jackknife like a bootstrap? Wasn't there something about this in Orwell's "Politics and the English language"? No, wait, that was a jackboot in the melting pot, right?
Jackknifing is just bootstrapping half-assed.
324: For a simple one-dimensional distribution (maybe two, as a next step)? Is it really that bad?
Imagine a jackknife stabbing a bootstrap... forever.
Unless you have a whole bunch of computers.
279-280 Yes they do spell things funny, e.g. they call Słubice Frankfurt.
I've never had bootstrapping take that long but I've never messed with it with particularly big datasets.
You could try exact logistic regression if you want to see it go really slow but I've only seen that give problems when I was doing it on purpose to see if I could clog the computer. It wasn't data you'd normally use that for. Otherwise, I've never run anything that took more than an hour or two unless there was an error and that is rare.
I mean, it is rare that it takes an hour or two, not that I make an error.
Essear, have you looked at the Cramer-Rao or the Chapman-Robbins bound, if you're comparing a parametrized or a discrete set of models?
really regretting I don't have time to read all the comments, but here's a few things:
pithy memorable lies with large impact
the new yorker has a good reasonable profile of Clay Christensen. Funny thing on Christensen, everybody and their brother seems to have analyzed that hard disk data which is his only academic publication. A few (confession: friends) have found his story doesn't hold up so well. But the story is so compelling and he has so many good examples I find myself partly believing and using the story even though I know it's not consistent with at least some of the data he claims.
Context is just another name for covariates
No, now context is the source of "natural experiments". That way we don't have to worry about simultaneous causation.
We should put agronomists in charge of left-wing politics.
Reminds me of Jane Smiley's Moo, where the radical Chairman X of the horticulture department is leading his band of loyal associates and revolutionary grad students to gradually plant more of the campus than they're formally supposed to. He's a wonderful character.
"Historically, the fundamentals of statistics include some presumptions which I find totally sketchy."
There are lots of things to turn to if you're looking for things to criticize about statistical practice "in the wild", so it's tough to say what your friend had in mind. Bayesians, which seem to include most real statisticians, criticize the philosophical underpinnings of common frequentist approaches. Yet frequentists searching for the institutionalized yet fairly meaningless 5% significance level still dominate the actual practice of social science. In my field, which follows economics, more careful approaches to causality than pure regression analysis are coming to dominate. And yet if you take the time to tear down the really complex "cutting edge" approaches, which are cited by many, they are usually built on questionable assumption after questionable assumption. Unfortunately, complexity wins out over clear but limited conclusions.
BTW, that full new yorker story has wonderful material if you're looking to feed your contempt for business schools. Christensen was classmates with Jeff Skilling and remembers as a good man with a family and implies it was misplace priorities (too much "short term" focus on work) which led him astray. The story doesn't mention that he's a close friend of Romney who has been out attesting to Romney's good character and deep caring for other people.
335: I haven't. But maybe I should. I think the simpleminded K-S thing is roughly what I want, anyway, for now. A ballpark estimate that is comprehensible to most of my readers. Anything too fancy would be a bad idea anyway, given the size of the uncertainties involved in what I'm doing.
Unfogged has really gone down hill since all the lawyers were replaced by statisticians Ogged left.
A mathematician, another kind of mathematician, and a statistician.
338
I haven't. But maybe I should. I think the simpleminded K-S thing is roughly what I want, anyway, for now. A ballpark estimate that is comprehensible to most of my readers. ...
Simulation (if only as a sanity check) still sounds good to me.
It's weird to think that the euro might unravel over the next two weeks. I sort-of suspect almost everything you hear in the media is posturing by people leaking a hard line to reporters to improve their negotiating position. The danger, though, is that if panic sets in among the public that the euro is about to break up, the reaction could be enough to destroy the euro.
I don't see how the euro can survive a general bank run in the southern countries. The ECB could take steps to stop it, but this depends critically on whether they really are incompetent nutjobs, or are just pretending to be incompetent nutjobs. (Though I also now wonder how much of this is because of incompetent reporting/analysis. Last week some ECB or Bundesbank official was quoted as saying that the job of the ECB is to target the inflation rate for the euro-zone as a whole, and if that meant higher inflation in Germany, then that was Germany's problem to fix. If you're going to have a euro, then this clearly has to be true, but I read all kinds of weird analysis about what this statement meant.)
It might not be so much that the ECB et al are incompetent nutjobs but that they are operating within political and legal constraints [in terms of the various EU treaties] that make lots of choices impossible, or near impossible.
I laughed quite a lot as I was reading 270 onward. (I always skip Shearer's comments, of course.)
At the weekend?
It's like you've never even met British people. Protip: When they say "in hospital" they mean in the hospital.
WTF am I doing up when I have to lead two workshops tomorrow with an untested curriculum?
Why am I continuing to have a conversation with myself, Sifu style? It's not as though I have the same charm to pull it off.
344. All that is true, but you also have to allow for the possibility that the ECB et al. are incompetent nutjobs. This seems to be Krugman's take for example.
I don't think either incompetence or constraints are the issue. The ECB folks are very smart, and shockingly unconstrained, but they have different values than most of us, and a different mental model of how the economy works. They don't want Greece to leave the euro, no, but they prefer it to the steps that would be needed to stop it.
I would agree with 350. You don't have to be incompetent or insane to be wrong about the way the world works, or to have different values and goals. Lots of competent and sane people thought, for example, that scurvy was an infectious disease rather than a deficiency, and did so a century after the reverse had been shown to be true. Lots of competent and sane people thought that it was a good idea to preserve slavery.
You don't have to be incompetent or insane to be wrong about the way the world works
Agreed, up to a point. If you're sufficiently wrong then you get to be incompetent or insane; I would for example question whether a physician who continued to treat scurvy as an infection after it was conclusively shown to be a deficiency was competent, even though they might have been sane. The BMA doesn't think Andrew Wakefield is competent: they struck him off and quite right too.
The ECB was called into being to serve the interests of the EU. If they are clearly not doing so, their competence must be in question, even if "nutjob" is extreme.
If you're supposed to be an expert on something, and you have a wrong model of how that things works, and the fact that your model is wrong is well-documented, then what word is there to use other than "incompetent"? It doesn't matter how prestigious of an auto mechanic school you went to, if you think that automobile engines run on mayonnaise.
At this point, we're not even talking about being familiar with the Great Depression. We didn't know what the consequences of letting a money center bank go bankrupt, but in 2009 we learned that apparently they're bad. We don't know what happens in the modern fiat world if all bank depositors flee Greece, Spain, Portugal, and Italy. I'm comfortable that history teaches us we're happier not knowing.
I mean, read this Bloomberg article, and tell me the officials quoted here are not incompetent. (The optimistic interpretation is that they're lying as part of some sort of gamesmanship.)
The first two paragraphs convey the main message:
The European Central Bank is conducting a comprehensive review of all its policy tools and has no immediate plans to increase stimulus even as market tensions mount, two euro-area officials said.The review, mandated by the central bank's six-member Executive Board, intends to assess the effectiveness of its measures, including the bond-buying program and long-term refinancing operations, and is scheduled to be completed in June or July, said the officials, who spoke on condition of anonymity because the deliberations are private. A third official said the ECB may not consider taking any further policy action until July, and that the bank sees current market tensions as a way of focusing politicians' minds on reform efforts.
Are they high? The last sentence makes me hope that they're lying to send a message to politicians. But if they're serious, then Europe is fucked.
Fair point. I think I was unconsciously using "incompetent" when I should have used "stupid". But "scurvy is an infection" was pretty much medical consensus around 1900 despite Lind's work in the 18th century, so I am not sure that you can use that to say "every doctor in Edwardian Britain was incompetent".
Bill Mitchell post. MMT guru. God, he does go, but linked for this snippet:
I have previously written about the Eurozone armaments trade and how that is an intrinsic element in the imbalances that have arisen since the Eurozone was established. Please read my blog - The value of government - for more discussion on this point.The German magazine Der Spiegel carried a photo of a big U boat on November 11, 2010 carrying a Greek flag and said that:
Modernste U-Boote mit Brennstoffzellentechnik: Trotz Finanzkrise haben etwa die Griechen insgesamt sechs Boote der Klasse 214 bei den deutschen Howaldtswerken (HDW) geordert, die zu ThyssenKrupp Marine Systems gehören.
Meaning? These state-of-the-art U-boats (submarines) are equipped with fuel cell technology : Despite the financial crisis, the Greeks have ordered a total of six boats of the Class 214 from the German company HDW, which is owned by ThyssenKrupp Marine Systems.
The Greek government (when they had one) was being coerced at the time by the Germans into seriously cutting back their social spending but at the same time were still buying very expensive arms from the Germans.
The Euro madness is characterised by a strong Germany exporting to weaker nations a lot of military trash while at the same time demanding the same nations who buy their exports to cut government spending (but presumably not on goods that the Germans sell).
...
Translation: "Germany ranks third in the world behind the US and Russia in world sales of armaments. According to data available from the acknowledged Stockholm Peace Research Institute, the German share of the world market over 2005-2009 averaged eleven percent. The largest buyers of military equipment made in Germany were Turkey (14 percent of total sales), Greece (13 percent) and South Africa (twelve percent). In 2008, the federal government allowed military equipment worth almost six billion euros to be exported".
Just so you know what kind of economic system DD and Walt want to keep stable and going strong.
356: So this is going to be one of those times where, even though I am almost always polite to you, you're going to shit on me?
357:So what do you want in #353, Walt? Really, what is your plan and goal? "We must avoid catastrophe" means status quo or not?
Jodi Dean, who has been on the ground all along, declares the Occupy Movement dead and buried.
I won't link, but Michael Hudson tore Krugman apart over at Keen's place. Krugman was always a Rubinite.
4-5 years on, and here we are watching JP Morgan counting their gambling losses while strips of flesh are being torn off the people to bandaid this corrupt incompetent totally destructive financial system. How much longer do we pray that just one more small reform will make it all better? Better for fucking whom, and we know who whom is. Not you or me.
You can't control finance. Burn it, bury it, and drive a stake through it's heart.
349
All that is true, but you also have to allow for the possibility that the ECB et al. are incompetent nutjobs. This seems to be Krugman's take for example.
But isn't it also Krugman's take that the Euro was doomed (or at least a very bad idea) from the start? Which I suppose is consistent with its promoters being incompetent nutjobs but also may mean there is not a lot to be done at this point.
345
... (I always skip Shearer's comments, of course.)
Math is hard.
I haven't gone into the history of PK's ideas on the Euro in detail; I'm not advocating his views uncritically - hence the expression, "for example" - but as far as I understand it his view is not that the Euro was necessarily a bad idea, but that it was appallingly badly designed from the outset. And that seems to be self evident whether you look at it from the point of view of a German financier or a Greek worker.
I think talking about the average opinion of society on some issue is more often than not terribly misleading. Opinions can't be added up and averaged.
OK, maybe if the opinion is numerically expressed: "what do you think the top rate of tax should be?" But how on earth do you average out people's opinions on immigration, for example?
And how on earth do you average out people's thoughts about something like sex, that they may not even be fully aware of?
363
... But how on earth do you average out people's opinions on immigration, for example?
Perhaps by asking how many immigrants a year should be allowed? And I think it has meaning to average yes, no responses to get the fraction of the population that favors certain positions. On many issues opinion will be quite lopsided.
More generally you can try to construct a scale by asking a bunch of questions about immigration which will capture something about society's overall attitude (positive, negative or mixed) towards immigration.
Math is hard.
Isn't that what Barbie used to say before parents complained about how Ken shouldn't have a nickname for his penis?
365, please read 363, where I actually mentioned this!
364: indeed.
364
And how on earth do you average out people's thoughts about something like sex, that they may not even be fully aware of?
You can look at behavior (for example fraction of illegitimate children or fraction of cases where the putative father is not the real father) as well as stated attitudes.
Math is hard.
And you have to thelebrate it every day! More than oneth on Thunday! It's a hard life being a prietht.
And how on earth do you average out people's thoughts about something like sex, that they may not even be fully aware of?
Only urple knows.
But how on earth do you average out people's opinions on immigration, for example?
http://www.pollingreport.com/immigration.htm
364: Get a sufficiently large n of randomly selected people into a huge orgy, and observe the relative frequency of different behaviors?
Or, more seriously, you can at least describe people's public opinions, then survey them on private behaviors.
Except AWB is talking about people from a few hundred years ago. To learn about their sex lives, you can't do an opinion survey. All we have are written records and the fossils of people who were trapped in lava while having sex.
372. Yes you can do all that stuff, but it doesn't get you any closer to an average.
You can't really generate an average attitude towards sex, but you might be able to come close to a modal answer to any given question about sex. I think that when people talk about 'societal attitudes' it's something close to the mode that they mean. At least inasmuch as they actually mean anything particularly well defined.
374
Yes you can do all that stuff, but it doesn't get you any closer to an average.
The fraction of people believing (or doing) this or that thing is an average.
And in other news the Germans are now saying the Greeks should try a bit of stimulus. Construct a sentence using the words "up", "it", "couldn't", "you" and "make".
The fraction of people believing (or doing) this or that thing is an average.
The fraction of people believing [how strongly? on how much information? on what authority?] (or doing [how often? how willingly? how commitedly?] this or that thing is an average [a mean? a median? a mode?] And supposing we agree to accept the most facile possible interpretation of this, what does it usefully tell us about anything?
The ECB was called into being to serve the interests of the EU.
I can't really get into the details of this ECB argument, much as I'd like to, but this really isn't true. It was called into being to control inflation in the Eurozone, and nothing else, which is a very different kettle of fish. The Eurozone is not the EU, and controlling inflation is not the same things as serving interests in general.
378
... And supposing we agree to accept the most facile possible interpretation of this, what does it usefully tell us about anything?
It's the same as any other simplified model of reality, some are more useful than others. But I think for example something like the change in the fraction of people in the United States supporting gay marriage over time does convey some useful information.
A ballpark estimate that is comprehensible to most of my readers.
Huh. I assumed this was for your new hobby, not for your actual job.
190
Speaking of Bram Stoker and Dracula I was under the impression that Stoker was the kind of author English professors looked down on and would not consider a suitable object of study. How are opinions evolving in this regard? Does shifting focus just reflect the fact that it is hard to say something original about someone well studied like Shakespeare?
379: Right. Deposing the Prime Minister of Italy is right there in the job description.
Which I suppose is consistent with its promoters being incompetent nutjobs but also may mean there is not a lot to be done at this point.
There is a lot to be done, but it's politically impossible and arguably even with the political will couldn't be implemented in time because of the institutional constraints ttaM mentions. Pretty much the only thing that can save the euro (with all its current members) is a proper fiscal union involving both loss of national fiscal sovereignty (not going to happen) and massive transfers to southern Europe (not going to happen), combined with a change in the ECB's mandate to target employment as well as inflation (not going to happen).
383: I meant that as an existence proof, not that we are always able to take an average of people's attitudes about sex, but that it would mean something to do so.
There is a lot to be done, but it's politically impossible and arguably even with the political will couldn't be implemented in time because of the institutional constraints ttaM mentions.
The distinction between this and there not being a lot to be done is too nice for me. Greece will presumably leave the Euro over the summer; anybody else for sure?
A proper fiscal union is not necessary in the short run. All that's needed is the ECB to monetize the sovereign debt. This would involve ignoring their inflation mandate in the short run, but it's hard to see how "preserving the existence of the euro" doesn't trump "ensuring the euro doesn't inflate".
All that's needed is the ECB to monetize the sovereign debt. This would involve ignoring their inflation mandate in the short run, but it's hard to see how "preserving the existence of the euro" doesn't trump "ensuring the euro doesn't inflate".
a) Because it is explicitly forbidden by the EU treaty and by the ECB's founding statutes. The ECB cannot monetise sovereign debt without an EU-wide renegotiation of the treaty. Which has already happened, and it failed to produce anything remotely adequate.
b) Because, technically, the ECB doesn't care about the euro per se. It cares about prices. In practice, the ECB cares about the euro because without it, it can't control prices. But regardless, both legally and culturally, what the ECB cares about is inflation and nothing else.
Technically, a fiscal union isn't necessary in the short term. But the short term measures that are needed aren't institutionally or politically possible. There was a time when I thought we could have a short term fix as a bridge to the long term solution, but Germany has made it clear that's not going to happen, not to mention still being in denial about the depth and breadth of the crisis. And Greece's (and Spain's) deterioriating economic situation, governmental weakness and understandably angry citizens mean that things are being brought to a head.
Oh here
Paul Krugman's Economic Blinders
Michael Hudson at Steve Keen's place. Keen has been on a bit of a jeremiad against Krug for a while, currently getting a bunch of guest posters to help out.
Unfortunately, Mr. Krugman's failure to see today's economic problem as one of debt deflation reflects his failure (suffered by most economists, to be sure) to recognize the need for debt writedowns, for restructuring the banking and financial system, and for shifting taxes off labor back onto property, economic rent and asset-price ("capital") gains. The effect of his narrow set of recommendations is to defend the status quo - and for my money, despite his reputation as a liberal, that makes Mr. Krugman a conservative.
As far as i remember, Krug wants deficit spending (everywhere), in other words more gov't debt owned and controlled by banksters. Krugman does not want to rev up the printing presses and launch the helicopters.
2) Krugman, DeLong, and the other Neo-Keynesians are very adamant about being Keynesians only at the "Zero-Bound" and monetarist at all other times. He will fuck us, and kill the boom when wages start to rise.
Or, alternatively, Krugman is a fan of currency depreciation.
Old nasty Keynesianism, the trick of real wage deflation with nominal wage inflation. I am not a fan.
Krugman is just another neo-liberal on vacation at the ZLB.
Does "monitize the sovereign debt" mean issuing eurobonds backed by all Eurozone countries?
No, it means, effectively, the central bank directly financing government borrowing. As opposed to providing banks with liquidity collateralised with government debt. Or, as in the case of the SMP, buying government bonds in the secondary market, ostensibly to preserve the functioning of the policy transmission mechanism.
I should add, joint Eurozone bonds are in my opinion a necessary part of the long term solution, but, again, Germany is adamant this will not happen without a full-on fiscal union. And, to be fair, the German constitution does prohibit them without it.
388: a) Sure they can. They can buy it in the open market. b) What happens to price stability when the euro zone breaks up?
389: Krugman's solution is for the Fed to commit to a high inflation target. In theory helicopters could be involved, but he thinks the main channel is through expectations, so if the central bank is sufficiently credible then it will happen without firing up the helicopters.
The fact that a bunch of people on the left have decided that Krugman is the Worst Person Ever reminds me of, well, the entire history of the left from inception to the present day. I can't really blame Keen, since Krugman was publicly somewhat insulting towards him, but for everyone else it's just old habit.
Also, Krugman is a New Keynesian. Neo-Keynesianism is something that died out in the 70s. Nick Rowe is plausibly a neo-Keynesian.
Does "monitize the sovereign debt" mean issuing eurobonds backed by all Eurozone countries?
No, it means, effectively, the central bank directly financing government borrowing.
Borrowing?
Still not following, GY Have we forgotten what "monetizing" means?
How about crediting citizen accounts with say a trillion Euros, no debt or creditor involved?
I don't understand how the various routes to inflation can have the same distributional effects. Helicopter drops, overnight lending rates, and buying assets seem like they should inflate different sectors of the economy, and only even out over some hypothetical long run. Am I right, wrong, or (most likely) not even wrong?
Krugman's solution is for the Fed to commit to a high inflation target. In theory helicopters could be involved, but he thinks the main channel is through expectations, so if the central bank is sufficiently credible then it will happen without firing up the helicopters.
Most people not named Sumner or Rowe believe that at least initial huge QE will be required, in order to establish craziness. This is what Krug eventually decided wuld be needed in his Japan paper.
Sumner wants the Fed to but bonds, then equities. He is a shill for asset inflation, not wage inflation. And asset inflation, transfers to the rich in the hope of trickle down, is really what NGDP targeting will be in practice.
Print and spend, without intermediaries. If we can't have Jubilee.
396: This has got to be right. Unless you believe that people instantly sense that the central bank has increased the money supply by x%, so you adjust all prices in your head by x%, then the way the money enters circulation has got to make a difference.
You know, the "lefty" economists keep talking about the horrible distribution of wealth and income, and call for higher taxes.
And then I ask:then what? How do we change the distribution and consumer confidence without inflationary wage expectations for the lower 60%.
And people like Rowe go berserk. The 70s, the 70s, absolutely cannot have expectations of wage increases over PCI.
So what do we do with the income we taxed in order to bring down the rich's share, buy back the electric companies? If we provide free services, roads, electricity, health care, that increases disposable income and we have real wage/income inflation.
I can't see their plan.
Sweet. After getting approval from both Walt and bob, I'm going for the parlay. For over a decade now, the Fed has been keeping interest rates near zero, doing so because (I think) they weren't seeing signs of inflation. Some classes of assets, like housing, were shooting up in price, and massive amounts of money were sloshing around in financial services. Is this a sign that the actions of the Fed were only loosely coupled to the part of the economy from which standard measures of inflation are calculated? Correct, incorrect, or incoherent?
I don't understand how the various routes to inflation can have the same distributional effects. Helicopter drops, overnight lending rates, and buying assets seem like they should inflate different sectors of the economy, and only even out over some hypothetical long run. Am I right, wrong, or (most likely) not even wrong?
Right. In fact, there's a debate going on in the UK right now about so-called "credit easing", which is basically quantitative easing targeted at particularly sectors of the economy (or rather sectors of the financial markets). Mervyn King isn't a fan, but some MPC members are. But, at least in the West, this is all more or less uncharted territory. Quantitative easing (by central banks) was until the crisis thought of pretty much entirely in terms of buying government bonds.
Still not following, GY Have we forgotten what "monetizing" means? How about crediting citizen accounts with say a trillion Euros, no debt or creditor involved?
Well, this is also forbidden by the ECB's statutes and almost certainly the EU treaty. Individual citizens aren't eligible counterparties for the Eurosystem, nor do they have eligible collateral, and, besides, the ECB wouldn't do something so inflationary even if it could.
For the avoidance of doubt, I'm not making normative policy judgements here. I'm just describing the political and legal circumstances. The ECB simply does not have the ability to do what many people, including myself, would like it to do. But at the same time, it probably wouldn't if it could.
The fact that a bunch of people on the left have decided that Krugman is the Worst Person Ever reminds me of the possibility that the only thing they ever read other than each other is Slate.
399:And this why we leftys are down on Krugman.
Krugman wants deficit spending.
We have had thirty years of deficit spending, and all the growth has gone to the top, because finance has captured the transmission mechanism.
I could give Krug a break like Keen, and say his type of Keynesian simply skips over banking or finance, or I could simply distrust him, because the fucker should have read Michal Kalecki thirty years ago.
Off topic, I watched "Merry Christmas Mister Lawrence" last night for the first time in 20 years. It holds up pretty well.
I await Bob's analysis.
Things the ECB/Merkel could do that are politically plausible - insist that austerity be focused on targets that will have the least negative multiplier effect, rather than the highest - e.g. wealth taxes on the very rich and forcing countries with non-functional tax collection to start collecting the taxes they are due from the wealthy; having the ECB place a de facto ceiling on sovereign interest rates, push for rapid wage growth in Germany, rather than the reverse.
What happens to Greece if a really radical party takes charge, and they decide to reject the bailout plan, default on their debt, and go back to the drachma? Would anyone loan money to the new government? Would the new currency immediately inflate out of control because the government keeps printing more to pay its bills?
405:Been at least that long for me. Funny that Kitano was known in Japan as a pratfall facepie comic, when we get his dramas and art films.
Heard "Young Americans" a little differently the other night. Oh, oh I want a young American. Oh Dave.
I'll get to Mr Lawrence after I read the book on homosexuality in Japan.
default on their debt
You fuck with the IMF at your peril. As in, "You are doing bidness with our sworn arch-enemy Greece?"
They are sooooo fucked.
The hospitals are running out of life-sustaining drugs, have no money and no credit. But they have six submarines, and Naples is pretty close.
Greco-Roman War II. You heard it here first, folks!
406: There are certainly things that can plausibly be done to make things less shitty than current policy. I'm just saying they're not going to stop the euro from falling apart.
Might I suggest that nukes will fix the Fukushima Greece problem?
Discussing the competence or incompetence of economists is like discussing the competence of astrologers. Or perhaps a better example would asking whether a competent mathematician is qualified to fix your car. They may be competent but not at that. I think Krugman's point was always that the monetary union needed some kind of fiscal union to back it up and that was not created. I think the assumption actually was that if you created the monetary union the fiscal union would gradually follow because it had to, except it still had to actually be done.
What's been getting me about the recent science/methodology threads is that the disastrous state of economics is entirely due to the attempt to apply scientific methodologies and pure statistical modeling to the study of human society. It's simply unworkable. Political economy rightly construed obviously does involve the search for causal mechanisms and the examination of data, but you have to recognize that as scholars economists are essentially historians using mathematical tools, and in practice they are irreducibly political managers of complex social systems.
404: Well, Krugman has in fact argued that direct debt relief of homeowners would be a good thing, if we could do it. (I count seven references to the phrase "debt relief" in End This Depression Now!) But there are major questions about whether that is politically feasible in the US (esp. given that the trigger for the formation of the Tea Party stuff seems to have been fears that the "wrong" kind of people would get health care/debt relief). He has also pointed out that mild inflation would in fact help in two ways: by allowing homeowners/consumers to pay down their debts in cheaper dollars (assuming some component of wage inflation), and by allowing the Fed to use a negative real interest rate to stimulate the economy, which would create jobs that help homeowners/consumers pay down their debts.
Eg., from chapter 7 of ETDN!:
It would seemingly be beneficial to both sides to have a program that offers troubled borrowers some relief while sparing lenders the costs of foreclosure. There would be benefits to third parties as well: locally, empty foreclosed properties are a blight on the neighborhood, while nationally, debt relief would help the macroeconomic situation.
So everything would seem to call for a program of debt relief, and the Obama administration did in fact announce such a program in 2009. But the whole effort has turned into a sick joke: very few borrowers have gotten significant relief, and some have actually found themselves deeper in debt thanks to the program's Kafkaesque rules and functioning.
407: in order: All manner of horrible shit, definitely no, and maybe or maybe not but the trouble is they need hard currency to pay for fuel and food, both of which Greece is a net importer. There would have to be rationing and GDP would likely fall 20% (basically a bit more than the gap between Greece and Croatia).
As a plan for avoiding austerity, leaving the euro involves a hell of a lot of austerity. Default within the euro is a much better idea, as evidenced by the fact that Greece has already done it.
More passages on debt relief from ETDN!. From chapter 12:
So in keeping with the principle of Rooseveltian resolve, aka "If at first you don't succeed, try, try again," we should try debt relief again, this time based on the understanding that the economy badly needs such relief, and that this should trump concerns that some of the benefits of relief might flow to people who behaved irresponsibly in the past.
Whoever is sitting in the White House next year will best serve his own political interests by doing the right thing from an economic point of view, which means doing whatever it takes to end the depression we're in. If expansionary fiscal and monetary policies coupled with debt relief are the way to get this economy moving - and I hope I've convinced at least some readers that they are - then those policies will be politically smart as well as in the national interest.
416: "definitely no" in the short term, sure - it's not like defaulting once cuts you off from lending forever. I can't see a Greek government - any Greek government - imposing an effective rationing regime either, but price hikes and shortages should have the same effect with rather more widespread misery.
49, 140: I think that at least some p-values should influence our beliefs in the strength of a hypothesis. If you flip what you a priori believe to be a fair coin, and you get 44 heads in 54 trials, do you still believe it to be fair? (If the coin is fair, then the two-sided probability of over 43 heads or less than 11 heads is 0.000004: 4 chances in a million.) Barring some other context, such as this particular sequence of trials being cherry-picked out of lots of other runs, I'm not inclined to give that coin the benefit of the doubt.
That's not an arbitrary example. It's the result of one particular sign test reported in the Bill James 1987 Baseball Abstract as part of his rookie study. And it's one where a priori I would very much like the null hypothesis (of no difference between the groups) to be true, because of the deeper social implications of rejecting that particular null. And yet I think that particular p-value is extremely strong evidence against the null. It doesn't just reject the null hypothesis at any reasonable confidence level, it rejects it, grabs the ball out of the air, drives the length of the court, and slam-dunks it at the other end.
(James didn't actually publish the p-values - most of his argument was based on summary group statistics. But he did mention the 44 out of 54 thing, and I can compute the p-value from there. He stated that he had run that particular analysis expecting to see no significant difference between those groups, was shocked by the results, and spent the rest of his article reporting on how he reanalyzed the data several different ways looking for statistical artifacts, or other possible explanations for the results.)
I'm not sure that a Bayesian approach would be helpful here. If I give the null hypothesis a high prior, is that because I believe it to be likely a priori, or because I wish to believe it to be likely? That kind of analysis seem pretty subjective - someone with different political opinions might assign very different priors and draw different conclusions. While the Fisherian approach just says: dude, it's overwhelmingly likely that there is in fact a real difference between these groups, assuming you haven't screwed up the analysis. Deal with it.
I am so late to this thread. Did anybody protest Chebyshev's inequality ?
studying language is as important as studying math.
Nowadays there are some reasonable methods to study literature.