On a previous thread, my counter-intuitive claim that verbal definitions are superfluous to science survived objections. I have been wondering if some further unconventional notions would survive a Troppodile attack.
Because natural science is effective, I suggested that we should do social science the same way. For example science does not define things but works out relationships between two or more objects (or concepts). That is, the only way we know if a concept (mass, time, heat, pulse, blood group…) exists is if we know its relationship to something else. Science also measures rather than counts and of course—of course!—scientific knowledge is falsifiable.
Social science is very far from this. Mel complained, with reference to the concept of power: “nothing like a consensus on anything ever emerges and … abstract and incomprehensible theory 1.”
That sums up the position throughout social science (except economics). In addition, almost nothing has ever been shown to be false—or could be. After a century (or two and a half millennia) of such failure would it not make sense to adopt the methods that have proved so efficacious in natural science? Here is an illustration of that efficacy.
First, take power. Clarify initially by excluding “power-to.” This comes from technology, is always increasing, and is not social science per se. Our problem concerns “power-over”—power over people. This is social (and zero-sum). Consider the first chapter of David Knoke, 1990, Political networks: the structural perspective. Knoke nominates two kinds of power-over he calls influence and domination. Many have made a similar distinction but Knoke suggests a relationship. He points out that two propositions give four combinations (“truth values” as the logicians say): 1: only influence, 2: both influence and domination, 3: neither influence nor domination, 4: only domination. From these Knoke deduces the four kinds of social environments where each power combination would be found. I will paraphrase his results as 1: free market, 2: hierarchy, 3: egalitarianism, 4: fatalism. This is standard natural science deduction from a hypothesised relationship. Is it valid? Set “power” aside for the moment.
Second, consider page 86 of economist Sam Bowles, 1998, “Endogenous preferences: the cultural consequences of markets and other economic institutions.” Journal of Economic Literature XXXVI:75-111. Bowles reckons there are two kinds of human interaction: impersonal and durable. He sets out the four truth values, namely: the first only, both, the second only, and neither. He deduces the kind of social interaction which follows from each. These he calls 1: ideal market, 2: bureaucracy, 3: community, 4: ascribed market—which I call fatalism. Different starting hypothesis but the same four types as Knoke. Now set Bowles aside.
Third, consider a famous paper by organisation theorist William Ouchi, 1980, “Markets, bureaucracies and clans.” Administrative Science Quarterly 25(1):129-141. He sees two kinds of economic exchange conditions which he calls “goal incongruence” and “performance ambiguity” whereby the former refers to everyone having their own different goals and agendas, and the latter is where it is difficult to distinguish individual contribution such as with complex teamwork. He sets out three of the four truth values: the first only, both, and the second only, and he deduces 1: markets, 2: bureaucracies, 3: clans. He doesn’t mention the fourth value (i.e., neither of his two conditions) but it is readily seen to be fatalism.
Three people hypothesising three different pairs of issues all getting the same four types. I have found a total of eight theorists who have taken two social relationship concepts (mostly quite idiosyncratic and not confined to modern society) and deduced four types. Save for a couple of reasoning errors they all get the same four.
A relevant digression: these four types are not only theoretical. Unlike personality types or any other psychology classification, we can actually see examples. Politically, we might characterise them as: 1: Hayekian dries, 2: Burkean wets, 3: the Rousseauian left, and 4: Homer Simpsonesque populism. Sociologically they’d be: individualism, hierarchy, egalitarianism, fatalism. They are visible everywhere people are free to interact.
Apparently, the four are the only coherent social worldviews and social arrangements (i.e., ideologies) that are possible. Apparently, the four types are “natural kinds” which is to say they exist, like the objects of natural science, independent of human perception. They are patterns of perceptions yet they exist independent of my, or any theorist’s, perceptions.
The same applies to the relational pairs. If all pairs only ever yield these four types, then the pairs of issues are inflexibly related to each other. Thus the relationship of every rational social issue to every other rational, social issue is specified. For example, because they yield the same division of the four types, Knoke’s “influence” is in a sense the same as Bowles’s “impersonal” and the same as Ouchi’s “goal incongruence.” So those concepts are also natural kinds. “Same” means a logical person must hold the same belief on all—agree with them all or reject all.
Evidently, a social science theory can be constructed by adhering to the natural science paradigm. Falsification criteria are obvious: just find a pair of rational, social items that don’t fit with those four types or which deductively yield some other types. Potential falsifications are countless. It’s classic Popper: any number of confirmations cannot prove the theory but a single case—a single pair of items—could falsify it.
Since the theory encompasses every social view and social structure, the four types are specified in considerable detail so empirical falsification is fairly straightforward: find types in the real world which conflict with them. Alternatively, find organisations or individuals whose beliefs or preference sets are coherent but contradict the theory.
So—a solution to social science’s ancient problem. Achieved by doing as natural science does, namely the theoretical deduction of relationships and avoidance of definitions. Notice, incidentally, that as in natural science, observation plays no role.
- increases[↩]
A few random comments:
1) your idea of what constitutes good and bad is far too restrictive, and some of your ideas from Popper are more complicated than you suggest. For example, whilst falsification is a good idea, one problem is that many models have many parts, and so you can only falsify one part of the model. It’s also the case that falsification assumes models produce essentially dichotomous alternatives that can be evaluated perfectly, but that isn’t the case either, and never will be for many interesting problems.
2) Just using one paradigm is a bad idea. If you look at the history of science, then what you what you find is generally a tight coupling between theory and data driven observations. So simple observation may well be useful.
For example, take Paul’s last post on racism. Let’s say we didn’t know anything about it. Even in the worst case, having some sort of taxonomy would be useful (“blacks will find it hard to get into the housing market; whites will find it hard to get mathematical jobs etc.). Of course, once you have lots of data, it might give you ideas about theory you didn’t think of before (or indeed unexpected findings, like bus drivers don’t dislike Asians), and so you can start coming up with different theories for the emergence of the data you found. It may also constrain those theories.
2) Many theories in social science are often awful to start with. But they often drive interesting question. For example, Freud is crazy. But the idea of processing things outside of awareness is a good idea.
3) Cherry picking beautiful examples from natural science gives a false impression of what’s going on. Try looking up topics like “computational chemistry” and so on, and you’ll find they’re basically differential equation soup areas which are no better than social science in terms of the precision of things.
4) Even some of the most hated areas of social sciences (e.g., qualitative analysis), have some important things to say despite often being entirely theory free. If you want to convey the human experience to someone else (e.g., “How does racism feel to you”), just giving them some numbers to quantify the extent and some equations to model the learning of attributions isn’t exactly useful. So if you want your theory of racism to be understood by the average person (important in some areas of social science), or indeed to give some hard-to-quantify increase in the understanding of whats going on such methods might help.
5) There’s any number of interesting and useful findings that we wouldn’t know about if we stuck to your ideas. For example, what makes a good parent? That’s a good social science question, and something useful to know. But if we applied strict falsification type methodology, we’d get nowhere.
6) There are problems simply not amenable to falsification style logic (basically, many quantitative areas), since specifying the outcome is not simple. These occur in both social and natural sciences, even with things where we have some idea of the outcome. e.g., “How do I best manage the flora on this island when I don’t know what to expect or exactly what I want?”
“There’s any number of interesting and useful findings that we wouldn’t know about if we stuck to your ideas.”
Hear hear.
It’s funny. This stuff can be diverting, but I can’t see it being much use. And there’s a kind of intellectual authoritarianism in it. Popper’s inspiration was to design a demarkation between the club of science and the unscientific. If it had succeeded, well and good, but it didn’t succeed. Popper shared the intellectual authoritarian leanings of logical positivism, but whereas, like the Titanic it sank on its maiden voyage, Popper’s contribution was subtler and more durable. But at least as far as its influence in social science is concerned, it served the cause of scientism, not of science. Go have a look at the cliometric rubbish produced in the 1950s. A complete waste of time.
But of course this isn’t to say that I don’t believe in falsificationism. Just that it’s no great key to anything much – just a bit of the commonsense background as we go about our business of debating the way the world is and what might be made of it.
It’s funny, a lot of this kind of epistemology is done by scientists in their dotage. They try to come to some more general conclusions about what it is they’ve been up to. But while they might give some kind of elegiac pleasure they don’t contribute much to the practice of the discipline from which they are generalised. Meanwhile the one discipline that marched in lock-step with Popper’s platform is neoclassical economics – a modern scholasticism par excellence. A discipline in which the hard-won commonsense of decades gone can be lost in a blizzard of equations, in which people are taken seriously for developing elaborate theories which explain the Great Depression as a spontaneous holiday.
MP:
“So—a solution to social science’s ancient problem. Achieved by doing as natural science does, namely the theoretical deduction of relationships and avoidance of definitions. Notice, incidentally, that as in natural science, observation plays no role.”
What the heck? Observation plays no role in science? Telescopes? Microscopes ? Imaging devices? …
The comparison of old and contemporaneous aerial photographs is one of the main ways of doing historical ecology. We learn a great deal from it. See here for instance http://ianluntresearch.wordpress.com/2013/03/13/the-candles-of-dunkeld-2/
Seriously, this is one of the wackiest posts I’ve ever seen on this website.
Peter Singer pwns Karl Popper.
Mel,
Please entertain the possibility that there is a miscommunication rather than that the post is ‘wacky’. It’s a serious post by a serious person.
And I agree re the Peter Singer job on Popper to which I’ve linked before.
And this:
“Achieved by doing as natural science does, namely the theoretical deduction of relationships and avoidance of definitions.”
As I said on a previous thread, there is no commonly agreed upon definition of species. Charles Darwin wrote about this problem and it is as yet unresolved. In fact it will probably never be resolved.
Mike says we needn’t worry about definitions because it has no implications but it does. It has serious implications for species conservation programs, for example. Here is old Carl Zimmer article on the issue.
Once again I urge anyone with a genuine interest in the topics raised by Mike to read Chalmers’ introductory book on the philosophy of science, What is this thing called Science?. It is a slim volume and it is written in plain English. As far as I can tell, it is the generally accepted primer for undergrad philosophy of science students in the Anglosphere.
Apparently, my offering clear falsification criteria is a bad thing. Incredible.
Conrad—
1. Certainly falsification is often not straightforward. This is a well-worn discussion under the rubric “naive falsificationism.” So what? Am I at fault for offering clear falsification criteria? Surely there is no comparable offer in all social science. Is that not cause for celebration?
2. Observation. I spelt out how to do empirical falsification. How could you empirically falsify if you didn’t observe? My point was that observation plays no role in the theory—something you can confirm by perusing my description. Observation plays no role in any scientific theory. You theorise, you test by observing, you rethink your theory, you observe again and rethink, etc, but the observations do not enter the theory. If what you call “tight coupling between theory and observations” ( a very unscientific statement) ever allows observations to modify your theory you have left science behind and have floated off into that psychology/psychology fudge known as “developing a scale.”
3. Differential equations and computational chemistry? In social science we have not yet got to Copernicus. You mention “precision.” I didn’t claim precision. When forming four truth values, true or false (presence or absence, yes or no) is all there is. Rigid, yes, precise, no.
4. Blest if I see the relevance. Yes, we can know useful things without science. So therefore we shouldn’t do science? Humans knew plenty before science came along. I showed how to do social science as science. How is that bad?
5. “There’s any number of interesting and useful findings that we wouldn’t know about if we stuck to your ideas.” We have this conversation before. I say all such findings are factoids. Useful bureaucratically, worthless scientifically.
” For example, what makes a good parent? … if we applied strict falsification type methodology, we’d get nowhere.” How do you know? We get nowhere now. Anyway, it’s academic for there isn’t any theory of parenting to be falsified. For millions of years we knew how to be parents, then social science came along and suddenly we don’t. Real (falsifiable) science, such as medicine, has transformed children’s lives but child-raising remains a matter of fashion. Don’t you think a theory which specified the relationship of every social issue to every other social issue would have implications for parenting?
6. You manage the flora on that island by drawing upon your extensive botanical knowledge—hard, nitty-gritty, falsifiable science. To manage anything is not science but bureaucracy and without the science you will manage far less well. It is not a bad comparison to the situation in social science: got no hard, nitty gritty.
1. No, all you are doing is appealing to a case which doesn’t exist most of the time.
2. That’s wrong. If I observe unexpected data, then I must update my theory or think of something else.
3. Most questions don’t have yes or no answers. This is just not how the world works. If you think like that, you’ve missed 99.9% of what’s interesting about human behavior.
4. This is your restricted definition of science. If something you don’t like gives some insight into human behavior, you don’t like it. Sorry, but this is still science, just not how you like it. You are too concerned with the method and not the outcome.
5. Again, this is your restricted view of the world. There go infinite interesting questions. As for your suggestion about parenting, why not tell that DOCS. If they had better models, less children would be beaten by their parents and fewer children would be taken away when they didn’t need to be. This is why these questions are interesting. The fact they can’t be falsified based on some dichotomous criteria simply shows that the avenue which you think science should be pursued in is too limited.
6. You could do nitty gritty experiments, but you don’t because you don’t have infinite time and money. This is also why, for example, we don’t do all the nitty gritty experiments we’d like to do in schools either. So you must think of other ways.
Nicholas—
Thanks for defending me, Nick—I confess that I was grinning to myself when I wrote that statement about observations. I reckoned I’d get a rise out of Mel and/or Desipis. I hope I have now clarified the matter.
You think we wouldn’t know useful things if we stuck to my ideas. There is little danger. But I didn’t suggest ignoring everything else. I just showed a way—as no one ever did before—to do social science like natural science. Surely that’s good?
“…at least as far as its influence in social science is concerned, [falsificationism] served the cause of scientism, not of science.” I shouldn’t be surprised; no one in social science ever tried it (except in economics). I offered falsification criteria to give the maximum opportunity to find fault with my theory. I delivered myself up, saying what would prove me wrong. Scientism? It’s standard science as far as I know.
“This stuff can be diverting, but I can’t see it being much use.” What stuff? You are saying someone done this before but I don’t know of anyone who has. What are you thinking of? You don’t see any use in knowing the relationship of every social, rational thing to every other social, rational thing?
“…there’s a kind of intellectual authoritarianism in it.” Interesting idea. So let me say, with all the authority of a dilettante blogger: to do science you may not depend on definitions and you must state a relationship. Is that authoritarian? Or is it just facing the facts, just advising the investigator of mother nature’s secrets of the discipline which is prerequisite to understanding her?
I looked at cliometrics on Wikipedia. I cannot imagine how history could be construed as science and I didn’t see that being claimed. Whether or not it is rubbish, if you go around measuring things you think you see, you aren’t doing science. The things have to be out there, genuinely existing, independent of what anyone thinks. Nature calls the shots and all the scientist can do is discover.
“…a lot of this kind of epistemology is done by scientists in their dotage.” I didn’t do any epistemology. (Did I?) I just outlined a social science theory following natural science rules. I thought (still think) I did something no one, dotard or otherwise, ever did before.
Mike Pepperday:
” I just showed a way—as no one ever did before—to do social science like natural science. Surely that’s good?”
Actually I’ve read your post four times now and it still doesn’t make any sense to me. You’ve showed me nothing at all. I feel like I’ve just read the blurb for a pyramid sales scheme.
I don’t want to be unkind but please, for the third time, read the applicable literature. Secondly, learn to write in plain intelligible English so we can actually follow you. My suggestion is try reading the Singer article linked to above for an example of how to write about abstract ideas in an intelligible way.
My apologies for being blunt but I think its best that I don’t mislead into thinking you’ve had a Eureka moment. All the best.
Mel—
Are you clear about the role of observations now (see my response to Conrad above)? Be assured that I have heard of telescopes. I have, in fact, extensive professional experience in their use. I know about air photos too, having spent years in the business of taking them. Not long ago I mentioned I had smuggled something across the border at Konstanz. Here is another titbit: the contraband was a roll of aerial film.
“As I said on a previous thread, there is no commonly agreed upon definition of species.”
Hmm. That is what I say—said many times—and is what everyone says, including that Van Deemter who was quoted at length and who, I assume, is an expert. We are all entirely in agreement. Nothing in science can ever depend on a definition. Mother nature doesn’t give a rats about human definitions.
“It has serious implications for species conservation programs.”
Of course it does. Definitions are essential for bureaucracy. Nothing special about “species” here. I went on about that before, too. No matter what you try to manage you have to define it and boundary problems are common. Paul was saying you can’t even define male and female. I elaborated on the difficulty of defining a geological fault. It is therefore pretty obvious that a scientific explanation of nature simply cannot depend on any definition. Yet in social science this is attempted all the time—and, of course, it fails to make any progress. You mentioned power being “influence” and “coercion” and above I show how to cope with that—without definitions.
I endorse your admonition to read Chalmers. I have possessed the book for decades and read it through again only a year or two ago. He is the only philosopher I’ve read who talks about definitions (I am not knowledgeable on philosophy but I have done some hunting) and he agrees with me that science cannot depend on definitions. But you knew that, right?
“I’ve read your post four times now and it still doesn’t make any sense to me.”
I admit it’s fairly terse. I am not sure where you get lost but briefly:
If you have two propositions P and Q there are four possible combinations:
both P and Q,
neither P nor Q,
P without Q,
Q without P.
Knoke had two propositions about power and from their four combinations he deduced types to which I gave four general names. Not that hard. Too briefly: if there is influence without coercion the environment might be one of the free market. If you have both influence and coercion that would characterise authority and a hierarchy. If you had neither there is no power – roughly a hippy colony, a Greens party or the Amish or some such egalitarian set-up. And if there is only coercion then there is no choice about anything and people blow with the wind—fatalism.
Bowles had two propositions about economic interaction and deduced four types and lo, the four are the same as Knoke’s four.
Ouchi had two propositions which are pretty convoluted but the point is he deduces types from three of their four combinations and lo, they are the same as Knoke’s and Bowles’s. That the fourth type is fatalism is easy enough to see.
Everyone I can find who has deduced the consequences of two social propositions finds the same four types. So I say: you cannot name a pair of relational issues that yield anything else but these four. By relational issues I mean matters which we must deal with in order to live socially. Like apathy, competition, cooperation, crime, freedom, heroism, human nature, justice, leadership, religion, revenge, ritual, sin and many more. I think it would not include the personal and internal, such as depression or happiness, or emotional attachments to family or football club.
Does that help?
No it doesn’t help. You just keep digging yourself into a deeper hole. Let’s again take your claim that definitions are unimportant and then consider this statement of yours:
” And if there is only coercion then there is no choice about anything and people blow with the wind—fatalism.”
If you’d bother to look up the definition of fatalism you would know it doesn’t mean anything like a power relationship that involves only coercion. Fatalism is a philosphical term and it has different meanings in different traditions. Here is the relevant Stanford Uni Encyclopedia entry.
I also found this statement and example illustrative of a rarely seen level of intellectual laziness and general cluelessness:
“If you have both influence and coercion that would characterise authority and a hierarchy. If you had neither there is no power – roughly a hippy colony, a Greens party or the Amish or some such egalitarian set-up.”
The Greens have rules. Members can be expelled.
The Amish uses shunning among other things as a method of rule enforcement. This doesn’t fall neatly into coercion or influence but it is an exercise of power.
Hippy colonies have rules and enforcement of one type or another just like every other community. Look ’em up. Google is your friend. Try the Christiana community near Copenhagen for instance.
You also bizarrely claim you haven’t defined power yet you have very clearly defined it as influence and or coercion.
Don’t give up you day job, Mike.
I don’t think trying to understand complex social concepts identified using vague language, such as ‘power’, ‘interactions’ or ‘motivation’ by framing them as logical propositions is going to lead anywhere useful.
Mel—
“If you’d bother to look up the definition of fatalism you would know it doesn’t mean anything like a power relationship that involves only coercion. Fatalism is a philosphical term and it has different meanings in different traditions.”
Further evidence that definitions are a waste of time. You are displaying the same in-depth knowledge of fatalism as you displayed of telescopes and air photos. Ditto re Greens, hippies and Amish. Eight theorists started out with eight very different pairs of propositions and they all wound up deducing the same four types. One of which I have labelled fatalism. It’s only a label. Earlier I used Homer Simpsonesque populism. Got a definition of that?
Mel, I don’t think you want to understand.
Desipis,
You don’t think specifying the relationship of every social thing to every other social would lead anywhere useful? Okay, so why not show me? The means to prove me wrong are readily available. I set them out.
Do you think the usual social science muddle will lead anywhere?
Do you think there’s any worth in the aspects of society that we don’t understand scientifically? Has there been any improvement in society (other than scientific progress) in the last century, or is everything the same as the 19th century but with cooler toys? Did civilization improve much between when the apes climbed down from the trees and the development of the scientific method during the 17th century? Surely you can see that exploration of social issues, even outside a scientific framework, has contributed greatly to improving society over time.
A series of rhetorical questions—not questions but opinions. Also basically off-topic.
I say social science fails in that knowledge does not accumulate. (Nothing is ever disproved. There are virtually no actual theories that can be tested.) If social science adopted standard science methods this might be rectified. I have illustrated how this might be done.
But if we take your questions as genuine questions (i.e., ones where the answer is not taken as known) then it would be pertinent to ask how we would distinguish social progress that was caused directly or indirectly by science from progress that just happened. After all, when the “dark ages” gave way to the “enlightenment” that was also the beginnings of science. To what extent was that a coincidence?
Interesting question. Can you find a way to falsify the hypothesis that social science provides benefits to society?
Mike Pepperday:
“Eight theorists started out with eight very different pairs of propositions and they all wound up deducing the same four types.”
And based on this you have developed a theory for how the social sciences should work. In other words you’ve gone from a small set of observations to a general theory. This is the exact definition of inductive reasoning. You do know this, don’t you?
Do you remember what Chalmers said in his book about Naive Inductivism?
You also make the pompous claim that your half-baked theory somehow captures the reason for the success of the natural sciences!
Sorry Mike but this is seriously lazy thinking.
Conrad
1. I don’t know what this “case” is you say doesn’t exist most of the time. I would have to disagree since I cover the gamut of social relationships. However, no matter how restrictive you think it, has anyone offered comparable falsification criteria? No? Then celebrate! Falsification in social science at last! Frightfully restrictive but at least a start.
2. Yes, if the data don’t fit, by all means rethink the theory. We do not need to disagree here. The point is the data are not in the theory—your new thoughts are. If you dream and wake in the night with a new insight, rethink the theory. The dream isn’t in the theory—your new thoughts are. If your colleague points out a flaw in the logic, rethink the theory. Your colleague is not in the theory—your revised thoughts are.
A lot of things might influence your thinking and it is obvious that data would be significant. The bad thing is to think that data drive theory. Bad because (I think I’m doing epistemology now, Nicholas) it leads social science up the garden path—e.g., to developing all those dopey “scales.” From Comte to Popper, every philosopher who has looked at the matter says: theory precedes observation. But it is more than that: theory is despite the data. That’s why new scientific knowledge is so often a surprise. If science came from data it wouldn’t have taken 10000 years of civilisation for science to come along. If science came from data the birds and the bees would be doing science.
Galileo said that the period of a pendulum is independent of the amplitude. He did not observe this. His friend and rival, Guidibaldo del Monte, experimented and said it was wrong. Galileo insisted. His pendulum, he said, had no friction, its string had no weight, and its weight had no volume. Del Monte mocked him, saying that is not how pendulums work. I have read that young people given pendulums in science classes never discover Galileo’s theory. Who turns out to have been right? There’s a big, big lesson here.
“Data mining” is currently the in-thing. Very useful. But some think it is science which is delusional.
3. Your assessment of what I might be missing is ad hominem. “Most questions don’t have yes or no answers. This is just not how the world works.” That’s channelling Guidibaldo del Monte. Your assessment of how the world works would be in accord with virtually every social scientist who ever lived. And where has it got us? As Mel said, “nothing like a consensus on anything ever emerges and … abstract and incomprehensible theory [increases].”
Newton’s first law says an object moves at constant velocity in a straight line forever. Since there is not a single example in the whole universe, you might say it is not how the universe works. Yet this law is essential to understanding the motion of everything from raindrops to galaxies. It is usual for theory to be, to the undisciplined eye, “not how the world works.”
4. Getting personal again, you tell me: “You are too concerned with the method and not the outcome.” No way. It is the failure of a century of intensive social science to produce an outcome (other than abstract and incomprehensible theory) which motivates me. What works is what works. Science (real science, hard science, hard-core science, my “restricted vision of science”—whatever you want to call it) works. Therefore we should try to adopt its methods.
5. I didn’t suggest the problems of parenting were not interesting. I just said parenting is fashion, not science. As I asked before: wouldn’t a theory that set out the relationship of every social issue to every other social issue be relevant to parenting? Surely it has to be. It’s limited? I don’t know your reason for saying this but I suppose no theory is unlimited. However limited it may be, at least it’s falsifiable theory, not fashion.
6. You could do nitty gritty experiments? No no. The nitty gritty comes from theory and you haven’t got any theory so you can’t experiment—not meaningfully anyway. Social science has been doing “experiments” for a hundred years. To no result.
In the whole of social science (except economics), I think the only scientific theories would be a couple of precepts from “learning theory” and the claim that “democracies never war against each other.” Are there any other falsifiable theories? That is some outcome to show for a century of academic effort.
Relation 1: Source of moral authority – Human (H) or Abstract idea (A)?
Relation 2: System for establishing moral standards – Traditional (T) or Radical consensus (R)?
Combination 1: (HT) Monarchy
Combination 2: (HR) Democracy
Combination 3: (AT) Theocracy
Combination 4: (AR) Fascism/Communism
QED.
Desipis
Thank you! That is the sort of thing I need to hear—a view from a quite different angle. (Mind you, I certainly also needed to hear that offering falsification criteria was such a sin; I’m still coming to terms with that.)
But it’s a pair of issues. You have named four issues. It’s two issues from which to form the four “truth values.”
Dichotomies have to be genuine for a universal theory. An example of a dichotomy would be: human authority and not human authority. (H and ~H in logician’s jargon.) Assuming that dichotomising is okay (and those eight authors do so assume) then H and ~H includes everyone and no one can be in both categories.
The pair also have to be a useful pair—i.e., they should deductively deliver four clear results. (The theory is not refuted by an unclear result.) In practice that seems to mean that they should have sufficient contrast that they are independent of definition. For example, “influence” and “domination” in the context of power require only a brief indication of their meanings (an example or two) and then, because the two have some contrast, the plausible variations in their definitions have no effect upon the four truth values deduced from them.
You could try H and A as a pair and see where the truth values, H&A, ~H~A, H~A, and A~H, lead. I will ponder it. The other pair—moral systems—don’t seem to be contrasting and I don’t know that I can make much of “radical consensus”—it sounds as if it needs a long definition to convey its meaning which would rule it out. What would contrast with “traditional”?
Thanks for having a go.
Hi Mike,
I think it is great you want to continue to debate this and, like Nick, just because I disagree with you does not mean I don’t recognise that you mean what you say and are trying to debate it seriously.
Having said that, I am afraid I am fully agreeing with Conrad and Nick: you elevate a set of ideas that have been helpful in the pursuit of science to an absolute dictum.
Rather than rehash the particular points we have gone over again and again in various blogs, let me offer, primarily for your amusement, an analogy I like to use when thinking of how I do social science. I will probably put it up on a blog of its own at some later point :-)
I see science very much as finding oneself initially on the Northpole wanting to eventually get as close to the equator as possible. You pick a ride on the biggest iceberg going that direction. You have no realistic hope of making it, and you start out very cold and isolated as the iceberg gradually breaks up into thinner sheets as you get closer. It makes sense once a while to stand on a the biggest iceberg in the convoy and ride the ocean currents with it that get you closer to your goal, but no matter how big and solid looking the iceberg you choose, it melts and breaks up way before you get to your destination and you drown.
So the person truly determined to get as close as possible must sometimes go along with a big iceberg and not move on it, but he must always be prepared to tag along another iceberg that comes along or breaks from the existing one. And his last moments before drowning will always be spent dancing on sheets of thin ice.
Of course, one can always decide never to try to get to equator in the first place and simply declare to others that they will get to it by taking the next iceberg to float in that direction.
So if you see someone standing still on an iceberg going in the right direction, I would say he is not very close to his destination at all. Pity the poor schmucks forced to dance in their final moments as they struggle to get another meter closer.
oh, and in case the analogy was not clear, I think of you as the guy on the pole saying to those around him that they should find a big iceberg to go down south with rather than already start dancing on the thin sheets that also flow in that direction. Good advise, up to a point.
“you elevate a set of ideas that have been helpful in the pursuit of science to an absolute dictum.”
Yep, that’ll do. What gets me is that adopting the tried and true should be considered radical. It just shows how adrift social science is.
Your analogy is unclear. If you are talking about persuading others of one’s opinions then it does not particularly apply to science. More likely you are saying I am on an iceberg that will sink. I am baffled. Was Galileo on an iceberg? Did it sink? You have lost me.
I thought the analogy was rather clear. An iceberg is a fundamentally inadequate vehicle for trying to get to the equator. The point is that scientific method is a fundamentally inadequate vehicle for understanding or solving many social issues. They both might be able to make incremental progress, but there is an overriding issue (heat/complexity) that will none the less prevent them from getting anywhere near completing their journey. Maybe once hell freezes over they’ll both be a success.
The problem with taking a falsifiability stance is that social theories are inherently difficult to falsify. When working with natural science there aren’t that many fundamental forces in the universe. This make it possible to construct experiments that isolate the hypothesis from all other factors.
In social science there are thousands of intersecting hypothesis and millions of potential factors that can influence outcomes. It’s simply not practical to construct experiments that can replicate situations with sufficient precision to isolate the impact of a hypothesized phenomenon. This means there will always be an explanation in defense of a theory (‘the outcome was caused by some other factor’) and hence endless possibilities for debate even after ‘falsifying’ experiments have been performed.
One of the hidden metaphors in this, I think is of science as a kind of ineluctable engine of truth – or the elimination of error. But when the world is complex enough, in fact as Poincare discovered, it doesn’t have to be very complex at all, prediction becomes pretty much impossible over any decent horizon.
This is true of weather forecasting which doesn’t get any better at predicting the weather 4 months out and probably never will. Indeed one might be able to use science to come close to proving that it won’t. So that is then the limit of our predictive knowledge.
That’s analogous to the situation we’re in with social sciences, only not only is society vastly more complex than the weather, but it contains nodes of consciousness and free will, such that if you predict something, that prediction itself can change the behaviour of the objects of one’s study.
Somehow I don’t think being dragooned into your four quadrants is going to solve that problem. Still, I could be wrong. But given the implausibility of your claim to virtually all those on this blog, the obvious way forward – and perhaps the only one for me (and I’m guessing others here) is for you to stop bossing us about, and instead demonstrate some surprisingly useful results from this framework. Then I’ll be all ears and so will Conrad and Paul – promise!
Mel will probably just continue to be rude to you – but perhaps I’m being unfair ;)
Nick,
I’m not trying to be rude but I am teasing a little;)
If Mike really believes in his “four quadrants”, as you so appropriately put, it should submit a paper to journal. Here is the journal of the philosophy of science association webpage.
If Mike gets his paper published I’ll eat my shorts :)
Desipis—
“scientific method is a fundamentally inadequate vehicle for understanding or solving many social issues.”
Is that what the analogy means? Then the claim is baseless for how would anyone know? The scientific method has never been tried. It works so well for natural science—why wouldn’t it work for social science? But to say it won’t work when it has never been tried is bad enough; to keep on saying it when there is an example of it at the top of this thread—words fail me.
“The problem with taking a falsifiability stance is that social theories are inherently difficult to falsify.”
1. What social science theories? There aren’t any social science theories! Billions of words published every year in learned journals and not a theory to be seen. Correct me if I am wrong but economic theories have been falsified because there actually are some theories. Consequently, knowledge in economics is greater today than two centuries ago. There exists a body of economic knowledge that can be debated. You can’t say that for psychology or sociology.
2. Inherently difficult to falsify? I have put together a social science theory with quite straightforward falsification criteria. I didn’t even invent it—I just said relationships are needed and took other theorists’ truth values from two dichotomised issues . Bingo—falsifiable theory. By the way, the elementary logic that two propositions give four truth values is hundreds, maybe thousands, of years old.
“In social science there are thousands of intersecting hypotheses and millions of potential factors…”
Excuses, excuses. That is no less true of natural science. Indeed, social science may be simpler. The theory I outlined covers the rational and social (which is not everything but is still vast) and it says all social human worldviews and arrangements come in one of four types and the factors that make these types number perhaps a couple of hundred, all of which have a fixed place. Not very complicated.
“It’s simply not practical to construct experiments that can replicate situations with sufficient precision to isolate the impact of a hypothesized phenomenon.”
You are confirming my basic assertion: a century of effort for nix. The language is revealing: “phenomenon.” The phenomenon will be some figment of some academic’s fevered brain, something that he or she has, God help us, defined. The notion of testing someone’s definition is ridiculous yet that is, in effect, what social science “experiments” do.
The so-called experiments are not testing a relationship—a relationship between two or more things that do not need defining. If knowledge is to accumulate, I cannot see that there isn’t any option on this. Science theory (real science, the kind that works) consists of relationships between objects that do not depend on definitions. Why would anyone expect procedures that deviate from this to get results? And it doesn’t get results.
How many academic “experiments” are done annually in social science? 10000? 100000? If you exclude the survey stuff it must still be thousands of publications every year. Utterly barren except with respect to the careers of the perpetrators. If there is a hypothesis (like as not invented after the testing) most tests are done with a view to confirming it, not refuting it. So bad is this, that if the hypothesis is not confirmed, it isn’t published. Thousands upon thousands of true hypotheses! How come we still know nothing?
Because it isn’t science. And in most of these cases, perhaps all, if there is a hypothesised theoretical relationship, it plays no role because no test is made of it; instead a statistical model is adopted. The hypothesis, whatever it was, is thus made irrelevant to the test which consists exclusively of mathematics put together by Mr SPSS and his siblings. This maths is applied in all cases, irrespective of any hypothesis. Theory-free testing!
You read some game theory result and you say “Oh, that’s interesting. Fancy that!” and then two days later you can’t remember what it was because it was a context-free factoid.
You suggest there will always be ways to excuse experiments that do falsify. Oh yes. That occurs in natural science—Kuhn says it is common there. But it does not excuse science from producing falsifiable theories. The theory I outlined above is falsifiable theoretically and it is fairly straightforward—as, I should think, theoretical falsification of economics would be. Empirical falsification is a lot more difficult—everywhere. Empirical falsification of economic theory is quite another matter. It is apparently hopeless. It seems the only economic theories that are empirically false are those of one’s opponents.
“The scientific method has never been tried”
Yes it has. It’s been tried any number of times (including in my work, see e.g., here). It works quite well for somethings, but it’s not good for everything.
What you are saying is basically what the behaviorists were saying a century ago — That their scientific method was the best for everything, and that the only decent solutions to problems came from applying their scientific logic. But it turned out to be wrong for a number of reasons. Perhaps the most fundamental is that it seems pretty implausible that we’ve discovered the type of scientific methodology that is not only general enough to solve all problems, that it provides the best solution to all problems, and will for all eternity. This is why you need different approaches — For all I know, some of the great scientific discoveries will turn out to be dead ends in 100 years.
Mel—
Induction? Fine. The fact that, to date, successful science theory (apparently always) consists of relationships between two or more undefined objects, does not prove that this will always work.
That the sun has risen every day in the past does not prove that it will always rise. Nevertheless, we do not plan our lives on the expectation that the sun will not rise tomorrow.
Induction is how we learn. It is how the pigeons in the Skinner box learn, how the Thanksgiving turkey knows it will be fed. Who watched Q&A last night? AC Grayling said Bertrand Russell had a clean mind for he was always changing it. Miaow!
Desipis
“Interesting question. Can you find a way to falsify the hypothesis that social science provides benefits to society?”
No. Nor could you falsify its inverse. Its unfalsifiability means it is not, in the strict scientific sense, an “interesting” question.
The benefit or otherwise of anything to society has to be just about the most political thing you can ask.
What any politically aware person can do is take a shrewd guess at predicting which people will endorse or disapprove of which programs. That is a bit different, I know, but is beginning to escape the definition of “benefit,” transforming it into a question of who has what views of “benefit.” It is people’s views that my theory is dealing with. So instead of arguing about what “benefit” would mean, I would sort the world into four possible meanings of benefit.
If you read what I said of the four types (they are pretty plain) you can make your own guesses. I won’t because if I tried it would be off the cuff and it isn’t necessarily easy.
Now I will see if I can answer Nicholas.
Conrad—
As you say: “their” scientific method. I don’t actually agree with your description of behaviourism but that doesn’t matter because in spirit you are right. I have on several occasions read authors saying the scientific method has been tried in social science and doesn’t work. (The only name that comes to mind is Bent Flyvbjerg.) To which I say: bollocks. They have never tried it.
For example who has woken up to the matter of definition independence? I think no one except Chalmers. I had a read of that 1974 Singer article. He discusses whether all swans are white and of course he is perfectly aware that it is an argument over a definition. But he is oblivious to the fact that that rules it out as science. If you see the verb to be (is, are) in a purportedly scientific statement, it is a signal that it is about a definition. Into the WPB with it! It’s funny. Kurt Lewin in the 1930s insisted that science does not deal with intrinsic properties (e.g., being white) which is much the same thing. No one took any notice and no one took any notice of Chalmers a generation later.
Referring to my theory, Nicholas said “implausible” and now so have you. Surely this is a mark in its favour. It doesn’t prove anything of course but historically, isn’t implausibility a prerequisite for a genuinely new theory? Given a century of intense but vain investigation by a lot of very brainy people, how likely is it that the True Theory, when it turns up, will be immediately plausible? It has just got to be a surprise.
Other approaches. It is really immaterial how many other approaches there are. I just say that the only approach that leads to cumulative knowledge, to causal understanding, is hard-core science.
I am trying to write a reply to Nicholas’s last. It’s a problem for I did leave out some things in the original post. Got to strike a balance between brevity and clarity.
Mike now says:
“If you see the verb to be (is, are) in a purportedly scientific statement, it is a signal that it is about a definition.”
Paul Frijter’s bus paper has 679 is/are statements. I guess it ain’t science.
To what extent can you be confident that social phenomenon will persist long enough to be comprehensively studied within some cumulative knowledge framework?
Desipis
The theory I have outlined is eternal and universal. It applies to the aliens from ALpha Centauri as long as they discuss how they should live together.
Nicholas —
“demonstrate some surprisingly useful results from this framework. Then I’ll be all ears and so will Conrad and Paul – promise!”
Given the historic reception of new ideas, I am not that sanguine. Conrad would have me burnt at the stake. Who was it said that science only progresses with the funerals of the old guard?
It’s a big ask. A theory is to explain how things work. One may hope there is a use for it. If there is a use, can it be explained in a few paras? Not sure this is the right place but I’ll give it a go.
The original post, long as it is, left out a fair bit. I’m still leaving a lot out. After some writing and discarding, I have decided to feed the Troppodiles some big lumps of meat and see if they can be digested. It might take some chewing.
Concomitance of issues.
This is the big omission from the original post.
There are four types, 1,2,3,4. You can divide them three ways: (a) 2+3 v. 1+4; (b): 1+2 v. 3+4; (c): 1+3 v. 2+4. Knoke’s “influence” divides them as (b) and his “domination” divides as (c). Bowles’s “durable” is (a) and his “impersonal” is the division (b). You would have to look back to the post to check that I have got that right. Complicated? Don’t worry; though everyone says social science is inherently more complex than natural science, the complications here are trivial compared with physics.
I have written these three (a) (b) (c) such that the first of the division accepts and the second rejects. For example, the Types 2 and 3 accept “durable” and the Types 1 and 4 reject it.
This limitation to three divisions indicates that a logical person (or organisation) who prefers “influence”, also prefers “impersonal” because they both divide the types according to (b). (They will be Types 1 or 2.) Similarly, the person who rejects influence must also reject impersonal. (Types 3 and 4.)
Now if there are only four types then all “issues” (Is there a better name?) must conform. That is, every rational, social issue divides people in one of those three ways. There are just three fundamental internally consistent preference sets. You can probably write down a couple of dozen relevant matters and imagine there might be a hundred or two (e.g., self-identity, mother nature, risk, equality, resource management, shame, guilt, authority, honesty, trust, punishment, blame, responsibility). Each of those can be expressed as an issue or issues classifiable as (a) or (b) or (c). No issue is classified on its meaning—we’d become mired in the definition bog if we tried that—but on how it fits, when related to a contrasting issue, to the four types.
The result is that if you believe in, say “influence” you must believe in everything else that is (b). If you reject “influence” you reject all of (b). So (a), (b) and (c) are coherent preference sets, sort of super issues. Most things are (a) or (b). The other one, (c), which concerns coercion, is rather less populated.
Here is one more solid lump. There are four viable types constructed as follows. 1: accept (b) and reject (a) and (c). 2: accept all three. 3: Accept (a) and reject the others. 4: accept (c) and reject the other two. You can persuade yourself of that by checking the examples given in the post. Plain enough. But what of the other four permutations, namely of accepting two super issues and rejecting the third and of rejecting all three? These aren’t viable. No logic ever leads there. The attempt always leads to inconsistency. This turns out to be a huge restriction, tightening interrelationships and multiplying the falsification possibilities. In social science if there is a single falsification possibility it would be a marvel. Here there are thousands.
Use of the theory (Hirschman):
This should be the right forum to talk about AO Hirschman’s “exit, voice and loyalty.” (Btw he died a few months ago.) He postulated these three tactics (say, by the firm or by employees) when a firm is in trouble: depart, campaign to fix it, or put up with it. His thesis (1970) has generated a substantial literature which finds that his exit and voice make sense but his “loyalty” is confusion. The literature is right. Exit and voice fall into place on (b) and (a). (Can you work that out—or at least see that that allocation looks reasonable?)
1. It is fairly easy to see that “loyalty” is a misnomer and is Type 4—i.e., not an issue but a type. There does not seem to be any specialised word to go to (c) which would deliver the types. Plenty of non-specialised words like “coercion” or “authority” but nothing of particular contrast to exit and voice.
2. Hirschman explicitly said that exit and voice were not mutually exclusive yet he did not examine the result of both together. That combination yields the Type 2 bureaucrat (who exits some activities and agitates to reform others).
3. H. did think his theory had much wider implications (divorce, competition, black power) and other writers have applied it to all sorts of things which appear quite unrelated to firms in trouble. Anyway, it is now sorted out and connected to everything else in the social universe.
Use of the theory (criminal court):
Adversarial and inquisitorial go to (b) and (c). It should be straightforward to reason that through. The interesting thing is the Type 3 who rejects (b) and (c). What do they want? They want reintegration with the egalitarian group—repentance. A sincere confession. This is employed round the world as “restorative justice” mostly for young offenders and minor offences though there was recently a murder forgiven (by the parents of the victim) in the US. Restorative justice will go to (a) and if you ponder it, you should see that that will suit the preferences of the other three types.
So the possible means of holding criminals to account fits straight into the theory, and is connected to every other rational, social preference. Not only that but by connecting in repentance as a process of criminal justice, Medieval trials and the show trials of the USSR and other eastern block countries can be analysed and compared, too.
Question: Does anyone know whether Mao, who was heavily into repentance, had contact with Protestant missionaries as a youth? This was suggested to me once but I have not come across any evidence.
Finally
Any contrasting duality that has currency or respect—academic or colloquial—had better fit the theory or something is not right. Kant’s price and dignity, Linton’s achieved status and ascribed status, Berlin’s two concepts of liberty, Kissinger’s conqueror and prophet. You can’t always deduce from such things (or it is tedious) but it is never hard to drop them into the places where they fit. In all cases it expands the duality into four, sharpens the concepts and connects them with everything else.
The above is all very terse but if the theory is what I claim, it can hardly be exhaustively described, let alone argued, in a page or so. I hope it is at least partly comprehensible.
Thanks Mike, but your comment about Conrad wanting you burned at the stake made me giggle. If you come up with the goodies people can’t really argue can they? I promise you Conrad won’t want you burned at the stake if you come up with something to show us why your theory of how to do social science isn’t a waste of everyone’s time.
Indeed.
Suggesting that nobody has ever REALLY tried to apply the scientific method PROPERLY to the social scientists sounds a hell of a lot like the special pleading of a communist.
It also sounds like concern-trolling. I get the feeling your real aim is not to help the social sciences progress but to ridicule them as they stand. And yes, given the way you’ve set up the goalposts, that must seem like it’s been a tremendous success. But, y’know… bollocks to that.
Mel Shakespeare circa 1620
Too hard, I think.
If my theory has any part of the truth, it could scarcely be otherwise. How readily were natural science theories understood? It took centuries. And social relations, everyone insists, are even more complicated than natural science. How easy is it to grasp any theory of elementary physics or chemistry? Or an economic theory, for that matter.
If I am right, those four types are real, as real as natural science objects, as real as heat or gravity or breath or blood group. It is not possible for them to be “phlogiston” or “downwards tendency”—which is what plenty of social science concepts are.
Although no one should simply accept this (on the contrary, one should seek to refute and ultimately falsify it), that’s not the hard part. The hard part is the logic. Given four types it is a fact that they can be pair-wise divided three ways. This is merely a tautology—not my opinion, nothing particularly to do with social science.
The same goes for my setting out which divisions go to which of the four types and how there are four further theoretical possibilities of accepting and rejecting the three super-issues (the four that, in my opinion, cannot exist because always illogical)—all that is tautologically inevitable given the existence of four types that are formed from pairs of dichotomised items.
The difficult part is the logic—and those logical interrelationships are needed to apply the theory.
I was looking through the thread and was struck by this from Conrad: “…once you have lots of data, it might give you ideas about theory you didn’t think of before … and so you can start coming up with different theories for the emergence of the data you found. It may also constrain those theories.”
Just so. Just so. I suppose 99.9% of social scientists would subscribe to that. It is a fairy tale. Newton’s first law and Galileo’s pendulum did not come from data and were not constrained by data; they were in spite of the data. A scientific theory transcends data. It deals in relationships between, if you like, Platonic forms.
These “data” Conrad refers to are, in practice, apparent data, consisting of counts of verbally defined items, not measures of relationships. No relationship is tested. Because social scientists believe in this just-so story, mountains of this useless stuff are published—and never looked at again.
The idea that one is doing science by measuring something you think you see—something you define in words—is just plain wrong. It is a caricature of science. It amounts to an attempt to measure a definition. Compounding the delusion is that it isn’t measuring but counting. To define things then count them is bureaucracy, not science. It has been basic to administration and management for thousands of years. It has its uses but it yields no scientific knowledge, i.e., no generality, no universality, no causality, no cumulative knowledge.
We often hear promoters of science proclaim the importance of “pure research” and point to examples of purely abstract findings which became technologically fruitful decades later. We don’t hear praise of pure research in social science. I suggest that there will be no progress till it is done. A research hypothesis would have to express a relationship between idealised objects which do not rely on definitions. Hard maybe, but it is the character of science relationships. It is the character of economics relationships. To mimic the basic science that worked so well in the past, that approach would need to be followed.
Hard to do and hard to understand even when an example is explicitly set out.
a link I found very interesting – lots related to scientific methods being applied in different situations.
http://www.thenation.com/article/174221/adventures-neurohumanities?page=full#
Mike’s post was a great challenge – I had to think and think about the idea that once someone states that “such a such is ……” almost automatically means they don’t understand scientific method.Very much correct as science is never static and all understandings are in flux and ever changing.
Perhaps the place to start is applying the Bell curve to all and any populations under analysis?
Thanks Murph. I read the neuro article. Even if they track every last neuron and synapse, I can’t see that it helps social science—they still lack the connection between the biophysics and the psychology.
The bell curve is a statistical distribution. Its (or another’s) application to measurement of populations is just mathematics. The problem in social science is population of WHAT?
That neuro article is about literature and the humanities. Not really my bag but I do think that if there is such a thing as psychology and sociology then they must be epitomised and crystallised in the characters and the moral lessons of literature. Shakespeare, for example, should be a rich source for psychologists to draw upon. He is ignored, as is all fiction.
Our time-honoured proverbs and mottos would surely form a concentrated distillation of how we think. What has social science to say? It is oblivious. Music and song? Anthropologists record exotic forms but mainstream psychology and sociology do not recognise their existence let alone offer theories to integrate and compare them.
The social sciences spurn the ordinary, everyday culture of song and story but in truth they are helpless. A century of ivory-tower research has brought the social sciences no ability to analyse ordinary, everyday, popular culture.
They can map brains till the fMRI melts down. It is not going to help. They need a theory.