- Prices in the marketplace are determined by supply and demand
- Capitalism leads to faster economic growth than the alternatives
- Unless you have very strange priorities, free trade is a good idea
- Minimum wages increase unemployment

Those are all debatable or "yes, but". Prices being set by supply and demand applies to highly competitive markets, not ones where game theory applies. Capitalism, maybe, but not laissez faire capitalism; all the big success stories had a lot of state involvement. Security and redundancy aren't that strange and could be reasons for inhibiting free trade. Simple theory says minimum wages should increase unemployment, but empirical studies didn't find good evidence of that.

I do think there's unappreciated science merit in economics, but those weren't the most solid examples. Or I know too much economics for your point. :)

- Personality is about 50% biologically determined
...and the other 50% isn't determined by the parents, is the really surprising bit.

- Animals react to reward and punishment in extremely predictable, almost mathematical ways

Behaviorism? I think that's problematic, though I couldn't say much about it. But one of my profs noted that the Skinner box took quite a lot of trial and error to develop and be adapted to the rat, and lab animals are typically in highly unnatural and stressed conditions. (Hungry, isolated, off their natural circadian rhythms...)

I think Scott actually picked good examples, so I guess I'll dip my toes in.

1) Game theory always applies. It's just a specific modeling tool. I can model perfect competition in game-theoretic terms; it's very easy. That doesn't imply the prices suddenly aren't set by supply and demand. What, exactly, do you have in mind? Bertrand and Cournot competition? They're just models of what determines supply that are slightly more complicated than the traditional formula.

2) I agree this one requires more qualification. Many argue even laissez-faire requires significant state involvement, and of course it does. I think a better, uncontroversial (among economists) way to phrase it is this: complete collectivization of capital is almost universally bad for growth, and collectivization of agriculture is pretty much a collective death sentence.

3) "Strange" for economists can just mean "nationalist." We don't particularly care about securing jobs for Americans over Chinese. This is the most common justification for protectionism. And of course we happily point out that American consumers suffer as well, so it's definitely "strange" to prop up a few Americans at the expense of the rest in the name of helping Americans. Your moral calculus has to get very specific ("All Americans are more important than Chinese and steel workers are the most important Americans." Huh?)

Having security over goods--for instance, not relying on "foreign oil"--has been shown empirically to be a bad justification for protectionism. The two best predicting factors of foreign manipulation of prices to harm a country are war and protectionism. Protectionism sparks trade wars, as it leads other countries to institute similar tariffs. Touting volatile prices as a justification for a policy that causes volatile prices is definitely "strange." And in any case, subsidization of an industry is lower-cost.

4) EVERYBODY agrees that the minimum wage causes unemployment. You too. If we raised it to $1000/hr, most people would lose their jobs. You have to be radically opposed to all of the findings of economics to believe otherwise. In this sense, Scott is right, and there's nothing wrong with his example.

The controversial question is on what margin it causes noticeable amounts of unemployment. Some people contend that on margins of around an extra $1/hr across decently large periods of time (if you don't allow long periods of time between wage increases, you run into the $1000/hr problem), there is no noticeable effect on labor unemployment. In other words they claim that, under some common circumstances, low-wage labor operates in an effectively monopsonistic market. This is NOT equivalent to the proposition that minimum wages don't increase unemployment. Economists will tell you that minimum wages above MP=MC will still cause unemployment.

(As for the truth of that matter, I think the proposition that minimum wages cause unemployment is still obviously correct and the people who don't are misapplying Bayes' Theorem. I'd be happy to give an explanation to anyone who wants it, but it's basically irrelevant for this post.)

1) I don't know those names. But consider haggling between a monopoly and a monopsony.

3) I don't think it's been shown so thoroughly. Economists do tend to assume away war and hostile behavior, and to concentrate on efficiency as opposed to redundancy, and most don't think that much about long-term changes. Paying more for food to avoid the risk of famine can be quite rational. Specializing in commodity export can increase short-term wealth, but leave one more vulnerable to changes in commodity markets, vs. protectionism and nurturing of a diverse industrial base. Ignoring the reality of nation-states is itself 'strange'.

4) There's more than possible monopsony. There's the possibility of increasing the compensation of a large chunk of the economy increasing demand of the very goods and services they produce and *increasing* employment.

Which gets into advances our host didn't mention, like circular flow theory and how countries aren't like businesses. Or the paradox of thrift, or externalities. (Is the Prisoner's Dilemma social science or pure game theory math? Tragedy of the Commons?)

I'm also sort of skeptical of so-called "qualitative research"; while I think it's important, I don't know to what degree it deserves to be let into the Secret Science Clubhouse.

To answer this one, you need a definition of science. Is it a process (hypothesis testing?) A domain (psychology)? An approach (wanting to find out what is really going on)? Specifically, you need to decide (a) whether you will allow any theory generation into your definition, and (b) what are the boundaries on the kind of theory generation you will allow. Do you require that any scientific theory be deduced directly from previous scientific results, or are you willing to allow other methods?

My answer is "it varies". I would struggle to call qualitative research science when it's based on an interpretivist ontology because it seems to me that it's a different game. I'm mostly happy for realist or critical realist qualitative studies to be part of an overall body of scientific research, because almost none of what you list above is proven by one study, or even a few. They are jigsaw pieces, and although we have to quality-assure the jigsaw pieces, we can only truly advance our knowledge by looking at the picture they make. This does depend, however, on other factors including research methods, context and the quality of specific research.

It is worth noting that if you are going to exclude qualitative research, you lose Charles Darwin.

(Full disclosure: qualitative researcher.)

I think that, if nothing else, qualitative research makes an excellent starting point for quantitative research. If you don't even know what to measure, then getting out there, talking to a bunch of people, and forming some hypotheses gives you some good ideas.

I hadn't thought of it in terms of theory generation before. It seems sort of sketchy for theory testing, which is what I most commonly associate with science, but you make a good point.

In psychology, which is my discipline, a lot of quantitative research is also sketchy for theory testing, because of the complexity of the problems being studied. This isn't true for all types of psychological investigation but it is true for a great deal. Me, I like looking at quantitative and qualitative research together and seeing how the pictures that they draw can be synthesised into something that is more than the sum of the parts.

This seems tangentially applicable:
http://www.smbc-comics.com/index.php?db=comics&id=2703

I'd say that about ten of those statements are wrong.

This is so obviously that ancient trick where the wise old man says something like "a few of those very important facts are wrong, but I won't tell you which!" with intention of getting other people to double check all of them because old people are obviously lazy.

At best. At worst, it's lying without making it easy, or sometimes possible, to refute.

« - Languages form large families like Indo-European that can be used to trace the history and development of different peoples
- People have an almost-miraculous language instinct that can for example turns a pidgin into a creole in the second generation »


My first reaction to that : This isn't anthropology, this is linguistics.
The definition of anthropology given by Wikipedia is quite confusing to me, but I'm pretty sure that if you put sociology in a separate category, then you ought to put linguistics in one as well.

Anyway, little children have an almost miraculous language instinct that leads them to imitate and assimilate their parents language, and that of other persons close to them, but sadly this is not this easy for people above the age of ten.

Typhon

I'm here via ciphergoth. Most of this is excellent, but "Minimum wages increase unemployment" is not consensus among economists. I'm interested to know, however, why so many people believe it is.

"Until the 1990s, economists generally agreed that raising the minimum wage reduced employment. This consensus was weakened when some well-publicized empirical studies showed the opposite, although others confirmed the original view. Today's consensus, if one exists, is that increasing the minimum wage has, at worst, minor negative effects.[77]

According to a 1978 article in the American Economic Review, 90 percent of the economists surveyed agreed that the minimum wage increases unemployment among low-skilled workers.[78]

A 1992 survey by published in the same journal revealed 79% of economists in agreement that a minimum wage increases unemployment among young and unskilled workers."

And someone's far more likely to have taken Econ 101 and its analysis of price floors, or even to just grasp intuitively that higher prices mean buying less of something, than to have heard of the empirical studies in question.

Thanks for that, saved me looking those surveys up again.

OTOH (from the wikipedia article):

A 2000 survey by Dan Fuller and Doris Geide-Stevenson reports that of a sample of 308 economists surveyed by the American Economic Association, 45.6% fully agreed with the statement, "a minimum wage increases unemployment among young and unskilled workers", 27.9% agreed with provisos, and 26.5% disagreed. The authors of this study also reweighted data from a 1990 sample to show that at that time 62.4% of academic economists agreed with the statement above, while 19.5% agreed with provisos and 17.5% disagreed. They state that the reduction on consensus on this question is "likely" due to the Card and Krueger research and subsequent debate.[80]

A similar survey in 2006 by Robert Whaples polled PhD members of the American Economic Association. Whaples found that 37.7% of respondents supported an increase in the minimum wage, 14.3% wanted it kept at the current level, 1.3% wanted it decreased, and 46.8% wanted it completely eliminated.[81]

Surveys of labor economists have found a sharp split on the minimum wage. Fuchs et al. (1998) polled labor economists at the top 40 research universities in the United States on a variety of questions in the summer of 1996. Their 65 respondents were nearly evenly divided when asked if the minimum wage should be increased. They argued that the different policy views were not related to views on whether raising the minimum wage would reduce teen employment (the median economist said there would be a reduction of 1%), but on value differences such as income redistribution.[82] Daniel B. Klein and Stewart Dompe conclude, on the basis of previous surveys, "the average level of support for the minimum wage is somewhat higher among labor economists than among AEA members."[83]

In any case, economists who disagee that "minimum wages increase unemployment" do not have similar status to creationists!

Wait, epidemiology is a social science now?

A lot of people consider anything that relies on statistical crunching of data to "prove" a hypothesis to be considered a "social science."

[I say "prove" because it may just be a unrelated correlation rather than a direct cause and effect. There is a lot of art involved in seeing such trends in data. I know at least one researcher who used exactly the same set of epidemiology data to prove both sides of an argument (depending on which side was paying him).]

Of course, to properly evaluate an experiment in epidemiology it would be necessary to infect a target population with a known pathogen and observe the spread. Unfortunately there are a few ethical considerations that make this approach unlikely to be used, so most observations are performed in hindsight. Nothing wrong with this - it's much the same approach as accident investigation attempts to work out what went wrong with an engineering or aviation disaster, so that what is learned can be applied to prevent (or curb) future outbreaks/accidents.

The thing is there is no one type of scientific research, and if you work in cross-disciplinary fields you will soon discover this. For example, I was a physicist working in medicine - and medical grant boards are trained to look at research in termes of drug trials (I was once asked what the "control arm" of my physics research was at one grant interview). From my point of view a ridiculous comment, but it's the only way to test the variations in drug trials. Similarly science in social science means something yet again, mainly because the aforementioned ethical considerations make it difficult to set up an actual experiment using people.

http://www.quotationspage.com/quotes/John_Kenneth_Galbraith

"The only function of economic forecasting is to make astrology look respectable."

And geologists can't figure out exactly when the next earthquake will happen, and meteorologists can't figure out whether it will rain a week from next Tuesday, but it doesn't completely discredit those sciences.

I'd run geology against astrology any day.

Some of your reasoning is circular. The fact that economists agree that the gold standard is a bad idea is not necessarily proof that it is. Consider that flat earth society members all agree that the earth is flat.

It doesn't prove it, but it's pretty strong evidence!

Whilst it's not looking likely we can do science with economists, we can do science *on* economists.


Hypothesis : Economists know and can achieve less than they think they can.

Method : Find a group of economists, give them an explicit target for an economic variable and allow them the means to control that variable, let them do so and see if they succeed.

Results : In 2003 we set the Bank of England (a group of economists) the target of maintaining a long term average of 2% CPI inflation in the UK. Jan 2003 - June 2012 has an average rate of 2.75% CPI, missing the target by 37%.

Conclusion : Economists know much less than they think.

(1) Don't you need to measure how much they think they can achieve as well as how much they actually do achieve? (2) Don't you need a similar belief vs fact test for other sciences, to be sure it's specifically an economics problem?

Well the modern world is full of things built to very precise tolerances that come out of the hard sciences, e.g. we use physics, chemistry and engineering to predict the amount of fuel that needs to be put in an aeroplane before it goes on a transatlantic flight, it's pretty rare for an aircraft to run out of fuel 2/3rds of the way across.

Aren't you comparing a single problem specifically selected for failure to a single problem specifically selected for success? One could just as easily say "Physicists still can't create fusion power, but an economist can balance her checkbook just fine."

Yeah, really we'd need to do a meta-study only it's very hard to find examples where economists say they will achieve measurable outcome X and then give us a sensible chance to measure if they did.

??? He didn't assert it's a specifically an economics problem, and so does not need to test that. One can assert, and test, whether sparrows fly separately from the question of whether birds in general fly.

If you're right about his intent, I'll concede the point right away. The evidence is that economists do indeed have a very serious overconfidence problem, as does all or nearly all of humanity.

Bad example.
http://www.bankofengland.co.uk/education/Pages/targettwopointzero/mpframework/currentinflationtarget.aspx

CPI is subject to factors beyond central bank control (droughts, oil shortages), and accounting for them would have big costs (stagflation). I'm surprised the Bank uses a CPI target, actually, rather than core inflation, which is more something they can control. Note oil prices have gone up a lot in that period.

Also, AIUI the Bank has tried for loose monetary policy to offset the stupid austerity policies of Cameron and Clegg.

So this is more a case of the economists not aiming for the target they're aiming at. The Bank certainly *could* bring the inflation rate down, at the cost of even higher unemployment.

Yes, I read that before posting. To quote the bit in bold in the middle which is repeated multiple times,

monetary policy is aiming to ensure that the inflation rate is 2.0% on average over time

over the last 8 years they've spectacularly failed to hit this target as the average inflation rate is 2.75%. Which means,

1: they can't control the average inflation rate

or

2: they're deliberately lying when they say it's their target and they've missed it intentionally


So I'll revise my conclusion, they're either idiots or scumbags.

Neither.

For one thing, they don't set the inflation target, the government does. So while I think "CPI of 2.0%" is an idiotic goal, the idiots are outside the Bank. Blame George Osborne, perhaps.

For another, the Bank has a second aim of meeting government targets for growth and employment. Having two goals implies tradeoffs, especially when the government is actively sabotaging employment.

As for being scumbags, what's more important? Removing 0.75% of extra inflation, or having a lower unemployment rate?

The inflation target is primary, the others are secondary. To quote George Osbornes letter to the BoE,

The monetary policy objectives of the Bank of England are to maintain price stability and subject to that, to support the Government's economic policy, including its objectives for growth and employment.

So to answer your final question, the inflation is more important than the unemployment. The job of the BoE is to implement the policy, not make it.

The Banks is *also* supposed to be independent. And a survey of news articles clearly shows that it is trying for monetary stimulus to the economy, though not enough IMO.

So the claim that the Bank's "failure" to keep CPI at 2.0% refutes economics is debunked.

Bank of England annual report 2012, page 35

Core Purpose 1 — Monetary Stability
Strategic Priority 1 Keep inflation on track to meet the
Government’s 2% target.


Bank of England annual report 2012, page 36

The Bank of England Act 1998 delegates to the Bank’s Monetary Policy
Committee (MPC) the task of setting the Bank’s policy rate —
and other monetary instruments, such as asset purchases — in
order to:
• maintain price stability, and
• subject to that, to support the economic policy of Her
Majesty’s Government, including its objectives for growth
and employment.
The exact definition of price stability is laid out annually by
the Chancellor of the Exchequer in the annual Budget
statement; presently it is to achieve a CPI inflation rate of 2%.


There are two mentions of unemployment in the report, one on page 9 where high unemployment threatens to push the inflation target below 2% justifying quantitative easing to force inflation, the second on page 20 again to justify quantitative easing to force inflation.


The monetary stimulus is justified repeatedly in their annual report as a method of preventing deflation / low inflation in order to hit the 2% target. A target they're pretty shit at hitting.

Selection bias; conclusion is pitifully weak.

Original paper appears to have been written for humorous intent rather than as a serious contribution to the field?

What the fuck are you referring to? How did it not occur to you that you should reveal the thing you're talking about?

Sorry, I was criticising my original posting, not yours (assuming you are the same anonymous as the other one). I agree entirely with the first anonymous' selection bias comment.

Here's my breakdown:

1. There are descriptive sciences. These are like geology, zoology, observational astronomy, archaeology, physical anthropology, and maybe parts of history. These clearly deal with facts, and not so much with theory, but they don't really need theory. If the questions you're answering are "What happened? What kinds of things are there?" then you can just go answer them, without as much mathematical apparatus as physics uses. These are obviously sciences, and successful ones.

2. There are undeveloped sciences. This is where theory would be *awesome* but the problems are hard and we haven't developed the theory yet. I think huge swaths of biology and medicine go here, as well as psychology, and maybe linguistics. Physicists are kind of wrong to sneer at these fields as "soft" (hey, we can't help it if living things are complicated!) but the real problem with having an undeveloped field is that you can get away with publishing stuff to lower standards of rigor, because we're all in the dark (compared to the "hard" sciences.) So the best biologists and psychologists are incredible pioneers, but the average paper published in those fields is...not so great. It's not that there hasn't been progress; there's been real progress, and there are also a shit-ton of "HERE THERE BE DRAGONS" spots.

3. There are "sciences" that are impossible to separate from politics. I'd put economics and sociology here. Economics? You'd think a laissez-faire lady like me would be shouting the loudest that econ is a science! I was almost an econ major! And...more or less, yeah, I believe economic theory. But the more I learned about it, the more questions I had about how much it could really predict. And academic economists are almost never separate from the world of offering prescriptive advice (whether to government or industry.)

Think of it this way: when you talk about a "liberal physicist" or "conservative physicist," you're talking about his private life. If you talk about a "liberal economist" or "conservative economist," you're talking about his work. It's a good test in general. "Liberal archaeologist" sounds like someone who voted for Obama. "Liberal sociologist" sounds like someone who promoted liberal ideas in his work.

I don't know much sociology, but it seems to start with the presumption that behavior is deeply affected by social institutions. Which is a communitarian point of view, right off the bat. Some people will stop you right there and disagree with you. And we all know the Haidt study about sociologists leaning liberal.

I think this is probably what folks are thinking of when they think "social sciences." They see a basically political, normative activity. And they think, "Hey, if you're just going to do politics, then I have the same right to participate in politics as any academic."

4. It never occurred to me that epidemiology was a social science. It's empirical AND it has theory, and it's very good at making predictions! It's just plain science.

The liberal and conservative macroeconomists agree more than you think. Your perception is probably from the politically tinted news you get rather than journals.

While the social sciences are astoundingly successful (I'll just go with your standard), they're more prone towards making, accepting, and promulgating specious conclusions -- this is the point I think you missed in the (admittedly hyperbolic) quoted comments. Gwern maintains a good huge cache of evidence in support of this position on his dnb faq.

The bottom line is that we should be skeptical of everything coming out of the social sciences, and more so than fields like physics.

Also, I appreciate that you're trying to recalibrate a community towards a more nuanced and reasoned position re the social sciences. LWers are often blinded by contrarian wiles until someone comes along and slaps some sense into them.

No offense, but you completely misread my point (I will let others judge about theirs), and, because you quoted my comment, I will have to answer and clarify.
 
I do not hate social sciences anymore than I hate chemistry. It is just that social sciences have not, to my knowledge, reached the past-Newtonian stage yet. Their way of deriving reliable conclusions from organized data is not at the stage where basic principles are developed from interpreting. It is quilt, not cloth: so far some correlations are spotted and that's it. In fact, I am talking exactly about what you are talking in the "hindsight bias" paragraph: in some bodies of knowledge we are just not there yet. We will be, God willing., but right now the exact reason criminologists don't have a good consensus about the sentences is that no one yet can do the experiment with replicatable results and find out. What is considered publishable in our world (as a criminal defense attorney I happen to read some articles about penitentiary issues) is nowhere near publishable by the standards of, say, molecular medicine. And in regard to how criminologists can advise police in regard to their ways: maybe in some better world they are supposed to, and maybe if you have a sufficient amount of kids some of their descendants will survive to enter a society where it works that way. 

 Also, no offense, I find your selection of examples rather random. I get psychology, I kinda, though not really, get economy, but how is epidemiology a social science? Isn't it an offshoot if medicine? You name anthropology but not history? If you define social science as "branch of knowledge that has to do with people's behavior vs. physical objects" than linguistics is a social science too, and you might as well ask why we, who think that psychology is currently in process of developing basic laws, do not believe in the usage of ablative in Latin. That's as cognitive a bias as it gets, and it is plain unfair.

I'm not sure what exactly you mean by "post-Newtonian". Who was the Newton of cell biology? What about the Newton of oceanography? If Newton was the Newton of physics, how come most physics has nothing to do with gravity?

It seems that most fields are made up of piecemeal facts and independent explanations for those piecemeal facts. Social science has had its great tiers-together - Adam Smith, B.F. Skinner, etc - but most of what it does is small incremental progress on specific problems, just like everything else.

I agree that my selection of what to call a "social science" was kind of random, though it was partly motivated by Wikipedia and partly by what sciences I hear people attacking.

I apologize if I misconstrued your comment, but the comparison to astrology (which wasn't a science that lacked unifying principles, but a non-science that was completely wrong) was kind of misleading.

Who was the Newton of cell biology? What about the Newton of oceanography?
Great question, but I am a wrong person to ask. A cell biologist of an oceanographer may be able to tell you. Reason I mentioned him at all was that Newton is not so much about gravity as he is about general principles. Scientists discovered situations where Newtonian physics is not applicable as such (similarly to how Euclidean geometry is not applicable to a spherical surface) but no physicist after him discovered yet that some part of Newton was wrong in the sense Ptholemean celestial spheres concept is wrong.

And I made no parallels with astrology. "Saturn in the wrong house" was a hundred percent alchemical reference (since I was using an alchemical analogy to being with), and a serious consideration for an alchemist (which btw Newton was too). But Alchemy is ok, chemistry grew out of it, and thought it is not a science, it is a science fetus in a manner of speaking. Paracelsus, using alchemical and astrological references, successfully treated syphilis with quicksilver. But he failed in some other endeavors, and alchemy does not work anyway, which was the bottom line of my point: sporadic empirical successes and pieces of knowledge successfully stumbled upon does not make a body of knowledge into science. You need an understanding of some general principles. With some of the social sciences that you mentioned we are not there yet. With some others we are, and for that reason I am surprised they made your list: I thought them to be parts of some actual science and therefore, well, science.

Consider me (the author of the second quote) chastened. I keep seeing social scientists draw strong conclusions from weak data, but this perception is based on anecdata too weak to draw strong conclusions about the field.

Part of the definition of "social science" is a science that has to be done via complicated correlational and observational methods.

This explains a lot. Social science is like philosophy: its successes grow up and become their own fields and aren't social any more.

Well, that settles that.