Science is more than its methods (but social science currently isn’t)

Paul called my attention to this piece (behind a pay wall…of course), titled “The science in social science” and written by anthropologist H. Russell Bernard. When I was doing my graduate work at UCONN, we commonly referred to Bernard’s Research Methods in Anthropology as our “methods Bible,” so I went into his article with favorable expectations. Unfortunately, I think he engaged in some logical leaps that I just can’t make. From his abstract:

A recent poll showed that most people think of science as technology and engineering—life-saving drugs, computers, space exploration, and so on. This was, in fact, the promise of the founders of modern science in the 17th century. It is less commonly understood that social and behavioral sciences have also produced technologies and engineering that dominate our everyday lives. These include polling, marketing, management, insurance, and public health programs.

This is similar to what Gary King has done in his presentation on “Quantitative Social Science”, as I discussed in a recent post. Both King and Bernard have taken instances where researchers have collected and analyzed information on people and then derived insights based on that research that ended up helping people make money or avoid conflict or more efficiently allocate resources, and have claimed those cases as examples of the ways in which social science has been successful.

I don’t deny that people have done some pretty cool things with information about people’s behavior. I just have a problem calling those cool things “social science” just by virtue of the fact that people’s behavior was the thing being studied. This is an issue on which I’m still trying to form a clear and coherent opinion, so let me try to walk through my reasoning.

On an intuitive level, there seems to be a difference between a set of principles that embody clear expectations about how some aspect of the world should work, and the instantiation of those principles in technologies. Maybe we could call the set of principles “scientific knowledge” and, following Bernard’s lead, refer to the instantiation of those principles as “scientific application” or “engineering.” The applications tend to work – meaning they tend to do what they are designed to do – because their ability to function depends on outcomes (reactions? events?…not sure what the best word would be) that happen consistently under certain conditions, and the applications themselves either create those conditions or take advantage of those conditions when they occur naturally. For example, refrigerators work because certain organic compounds react in a certain way to differences in pressure and heat, and because (most) refrigerators themselves create conditions conducive to those compounds reacting in the way needed to cool stuff.

The fuzzy middle ground between scientific knowledge and scientific application seems to be the area of systematic observation. If you observe something long enough and consistently enough, you can develop expectations about how what will happen next. Humans do this intuitively, which is why we don’t constantly get surprised when people we know do all the things they normally do. When my three-year-old daughter pretends to not hear me when I tell her it’s time to brush her teeth, I don’t get surprised. I’ve seen her do that countless times before.

When it comes to research, I would lump both data collection and data analysis into this observation category, since many of the things researchers are interested in like trends or latent patterns by definition can’t be observed directly. A lot of what we call “analysis” seems to be just a way of observing things that we can’t see with our un-augmented senses, so in that sense a statistical analysis isn’t all that different from a microscope or any other piece of observational equipment used in the physical sciences. Most of the examples of “social science” Bernard gives in his article seem to be focused almost entirely on the observation part of the picture: people looked at stuff and noticed patterns and then acted on their observations.

True, acting on those observations certainly qualifies as “application”, but I’m not convinced we should call that science. I certainly don’t consider my parenting to be science, even though it involves a lot of pattern recognition. It seems to me that the criterion of replicability is the key here: if replicability is important, and I think it is, it’s because a replicable study can be repeated in lots of different situations, and from the standpoint of generating knowledge, the only reason I can think of that you would want or need to reproduce the same procedures in lots of different situations would be to discover patterns that occur generally. If a pattern occurs in one situation, it could be because that pattern pretty much always occurs under certain conditions and those conditions happened to be present in the original observational context. The only way to establish the generalizability of a pattern is to show that it actually occurs generally.

This generalizability of core principles is, to me, what differentiates science from plain-old pattern recognition, and it’s also what seems to be absent from Bernards’s (and King’s) examples of social science achievements.

Something that recently helped me clarify my thinking on this subject was this blog post, provocatively titled “The surprisingly weak case for global warming.” The writer analyzed some climate data, using a random walk process as proxy for the null hypothesis, which might be stated roughly as “increases in the earths’ temperature do not represent a long-term trend.” If you read through the comments, and get past all the posturing and outrage that seems to typify climate research discussions, several people point out that the main problem with the writer’s findings is that global temperatures are not a random walk. Heat transfer happens a certain way based on the source of the heat, the receptivity of the heat’s target, filtering that happens in between, etc. When it comes to the sun and the earth, that transfer does not follow a random walk.

So the writer made a lot of systematic observations, but he interpreted them in terms of a model whose dynamics didn’t accurately reflect the dynamics of the system the model was supposed to represent. That seems to justify the assertion of some of the commenters that the methods didn’t have any obvious technical errors – the observation part of the writers activities was accurate – but the science was still wrong.

So, as much as I hate to say it, theory is what seems to differentiate science from just-plain systematic observation. I hate to admit that because theory is so abused in the social sciences (and has often been treated quite poorly in the physical sciences as well). In social research publications, theory seems to refer, in many cases, to just the vague notion of having ideas about what things are important or what things influence what else. But really rigorous theory – scientific theory – specifies what the parts of the system are, defines which parts interact with which others, and lays out the mechanisms by which those interactions take place. It’s something that can be diagramed out as blocks connected by labeled arrows.

Once you have rigorous theory – and I think a decent rule of thumb for “rigor” is to ask whether you have defined the theory enough that you can specify it mathematically (or at least simulate it computationally) – that theory can inform how you decide what to observe, and how to decide to observe it. Some kinds of observations (for example, seeing how well a pattern approximates a random walk) aren’t appropriate given certain theories (for example, a system of heat transfer that isn’t accurately approximated by a random walk).

For the most part, disciplines traditionally considered part of social science haven’t really produced rigorous theory. If you don’t have a rigorous theoretical foundation to prescribe an appropriate course of action in your research, then it seems the appropriate course of action is to observe whatever you can whenever you can with as little regard to theory as possible. Then look for consistent patterns. Then employ a minimalist theory (say, access to resources and information, outcomes from past behavior, routinization, etc. – the kind of stuff for which there is already ample evidence for behavioral influence from the cognitive sciences) to try to interpret as much of as many of the patterns as you can. If that minimalist theory provides a plausible explanation for all of the observations, then you might cautiously try to get more rigorous observations of those theoretical categories along with the behavior to see if they really are as good a set of predictors as the original data mining seemed to indicate. If the minimalist theory leaves some observations unexplained, only then should you make an attempt to expand the theory, but only as much as is necessary to account for the observations.

I remember reading a paper a while ago about animals modifying their environments. The authors did a simulation that showed that if an organism had the ability to observe the outcomes of its own actions, and also the ability to modify the environment to flag situations in which certain outcomes took place, that organism was able to develop a useful internal representation (principles) of the environment where previously it had lacked that. I think the same thing could be true of theory in the social sciences. I don’t think we need theory in order to get better theory. In fact, for the social sciences that are so stuffed with bad philosophizing and punditry masquerading as theory, we might actually need less theory. We can develop it from systematic observation alone, even though that observation all by itself doesn’t deserve the designation of theory, or science.

The problem I have with Bernard’s and King’s characterizations of social science is that they show that systematic observations have taken place, then show that the results of that observation have been turned into different sorts of applications, and then show that those applications did some useful things, and then declare that social science made those useful things possible. They never make the connection to any core principles of scientific knowledge. They never show that the applications were useful because they operated on more accurate or reliable expectations about how the world works. That last ingredient is important, given the difficulty social scientists have had demonstrating that they actually know what they’re talking about (discussion, discussion, example, examples with discussion).

I am all for systematic observation and analysis. And I’m very much in favor of making decisions based on observation and analysis. I think all the examples Bernard and King give are great illustrations of the ways we can do good and useful things based on systematic research. But the only way to attribute those successes to social science is to show that social science embodies a set of principles – not just analytic practices – that consistently improve our ability to maintain realistic expectations about the world.


16 thoughts on “Science is more than its methods (but social science currently isn’t)

  1. Using your logic, all of the historical sciences (e.g., evolutionary biology) are not science. There is not one formula of science that everything must follow.

  2. DWD, I’m not sure I follow. Evolutionary biologists haven’t posited mechanisms and models that generally hold within their relevant domains? I would have thought that evolutionary theory in biology was a great example of providing a set of principles “that consistently improve our ability to maintain realistic expectations about the world”. That’s what Darwin’s “On the Origin of Species” was all about … and I only refer to him particularly because his work occurred so long ago and therefore can’t really be considered an “exception”. I think I don’t understand what you’re trying to say.

  3. Dark Wing,

    I’m happy to revisit my logic, but your comment is a little too vague – I’m not sure what part of my logic necessarily leads to the conclusion that all historical sciences are not science. Can you elaborate? And while you’re at it, can you define “historical sciences”? I just want to make sure we aren’t talking past each other. My understanding of historical sciences is roughly in line with, but I don’t see how anything in my post declared any of those activities to be “not science.”

  4. Doc,

    Your post reflects basic scientific research and its helical process of discovery. I agree hypotheses leading to marketing or commercial applications is far from what Leedy and Ormond refer to as ‘theory’ or ‘an organized body of concepts and principles intended to explain a particular phenomenon.’ Nor does such ‘scientific application, engineering, or knowledge’ emody the word ‘science’ as defined by Webster as a ‘systematically organized body of knowledge about a particular subject’; the National Undersea Research Program (NURP) as ‘a process of discovery that allows us to link isolated facts into coherent and comprehensive understandings of the natural world’; or from NASA, ‘knowledge and understanding in a multi-disciplinary approach… [of] processes and interactions (cycles)… [and] use of physical and chemical laws with mathematics to describe… complex interactions [that are] accurate, predictive models.’ I’m not sure I follow Dark Wing’s comments above as well?!

  5. Schaun,

    The philosopher of science, Ian Hacking, writes about observation, theory, and experiments, etc. in his defense of experimental scientific realism. Here is a section from one of his books that is relevant to your blog post.

    “One can say that most experiments don’t work most of the time. To ignore this fact is to forget what experimentation is doing. To experiment is to create, produce, refine and stabilize phenomena…Folklore says that experiments must be repeated…It is clear that a variety of experiments is more compelling than repetitions of the same event…Roughly speaking, no one ever repeats an experiment. Typically serious repetitions of an experiment are attempts to do the same thing better–to produce a more stable, less noisy version of the phenomenon. A repetition of an experiment usually uses different kinds of equipment…It might seem as if there is one domain in which experiments must be repeated. That is when we are trying to make precise measurements, of, for example, constants of nature such as the velocity of light. We ought, it may seem, to make many determinations and average them out…But even in this domain what is called for is a better experiment, not repetitions of less good trials on less good equipment” (pp. 229-32).

    He agrees with you, Schaun, that good experiments can be conducted without a heavy load of prior knowledge of theories. However, he acknowledges that a minimum of ideas are required to use equipment and plan an experiment, but he distinguishes this as a weak requirement when compared to the strong requirement which insists that no experiment can be conducted with any sense unless it is first guided by developed theory. According to Hacking, Karl Popper can be considered a paradigm case of one who prioritizes theory over observation. Popper in 1934, argued that all observation statements are interpretations in light of theories. Before Popper, there was Justus von Liebig in 1863 , who said, “Experiment is only an aid to thought, like a calculation: the thought must always and necessarily precede it if it is to have any meaning.”

    Hacking provides a counter example in the eighteenth century chemist Humphry Davy. Davy experimented on gases produced by algae. Davy had no prior theory about gases produced by algae. He simply noticed that algae did in fact produce gas under some conditions and not others. Then he introduced a flame to the gas to determine its combustibility. These findings led to further experiments. So, Hacking maintains that Davy engaged in experiments inductively, contrary to Liebig’s insistence that science proceed deductively.

    When you say that theory is what distinguishes science from systematic observation, are you implying that in cases like Davy’s, no science took place until the theories arrived on the scene post hoc? Perhaps you do not want to call an experimental chemist a scientist until he also has some theory to explain his experiments, but I don’t see any good reasons for doing so. Hacking also points out that the facts about the conduction of heat and electricity in metal, i.e., the experimental results and the laws that described them were established in stages between 1853 and 1911. However, the coordinating theory that explained these results and laws did not arrive on the scene until 1957 with the introduction of a quantum mechanical understanding of superconductivity. Hacking writes, “Let us not pretend that the various phenomenological laws of solid state physics require a theory–any theory–before they were known. Experimentation has many lives of its own” (pp. 164-165).

    The source for all of this is Ian Hacking’s 1983 book, Representing and Intervening: Introductory Topics in the Philosophy of Natural Science.

  6. Jeremy,

    I think I was looking less at science-as-practice or science-as-findings and more at science-as-discipline. Bernard and King both try to take specific historical outcomes that have turned out to be practically beneficial and attribute that benefit to “social science.” I’m arguing that, in order for attribution to be warranted, we have to establish that the target of that attribution (the field of study) has some characteristic without which the object of the attribution (the outcome) would not have been possible…or, at the very least, would have been much less probable. The practice of systematic observation cannot be that crucial characteristic unless we want to jettison all of our intuition about scientific inquiry and define, say, parenting as a science. The findings themselves cannot be the characteristic because the validity of the findings can only be evaluated in light of the outcomes of acting upon those findings, and those outcomes are the very thing we are trying to attribute to some cause. Saying the outcomes can be attributed to the findings just begs the question. So, it seems to me, the best candidate for an appropriate target of the attribution is the principals (framework, theory, etc) that identified what things ought to be studied and constrained the ways the researching findings could be reasonably interpreted.

    So, if pressed to take a stand on the issue, I would call what Davy did observation rather than science, but I don’t think I feel very strongly about that particular definitional issue. I’m concerned about attribution, because that has consequences for what gets recognized, studied, funded, and otherwise supported going into the future. If we’re going to take some desirable (or undesirable) outcome and try to attribute it to something, we have the responsibility to justify our implicit assertion that that something is coherent enough to be a reasonable target of attribution. I don’t think social science has any coherent attributed beyond (within a particular scope of research activities) the practice of observation, and because observation can be inherent in a lot of activities that we don’t normally call science, I don’t see a justification for attributing realized outcomes to the discipline itself.

    But, as I said in the post, I’m still trying to figure out what I think on this issue. The only conclusion I feel really strongly about is that many social scientists try to take credit for stuff that they really shouldn’t :-)

  7. Schaun,

    If the practices of science (observation, experiment, and theory construction) do not distinguish science from practices such as parenting (casual observation, casual, experiment, and casual theory construction), then perhaps “aim” can be a distinguishing feature. I am assuming, contrary to you, that theory is not the distinguishing feature. I can argue that parents frequently develop casual theories in order to explain and coordinate their observations and to produce predictions.

    As an aside, scentia, the Latin term used in ancient and medieval times, referred to logical demonstrations of general or necessary truths, usually done mathematically or geometrically, although scientia could be gained in other disciplines. After the seventeenth century, science began to refer to observation and experiment (Godfrey-Smith: 2003).

    Hacking’s book is about distinguishing scientific realism from scientific anti-realism. One of his interlocutors, Bas van Fraassen, writes to do the same thing. In making their distinctions, they highlight the aim of science. For a scientific realist (Hacking), the aim of science is to provide a true story or picture about what the real world is like, ‘real’ being defined as the world existing independently of our minds and thoughts, and ‘true’ being defined as theories that reference the real world the way the world is. For the scientific anti-realist (van Fraassen), the aim of science need not make any appeal to the world as existing independently of our minds. But this distinction is not the point of your blog post.

    My point is that if you are looking to demarcate science from parenting, one way you can do so is to say that science aims to give a kind of story (true or just instrumental) about how the world works, while parenting does not aim to provide such a story. Parenting is just about surviving each day while keeping the kids alive. The same practices can occur within both science and parenting–observation, experiment, and theory construction. Science remains unique because of its aims, and perhaps because its practices, although also found in parenting, are unique in their constraints, applications, and generalizability.

    Perhaps ‘aim’ is not adequate to satisfy your desire to characterize science as a discipline instead of as a set of practices. However, I am wondering why you would want to define science as a thing apart from it’s practices. Hacking argues that he is a special kind of scientific realist. He claims to be an experimental realist. For him, that means what is real is whatever we can manipulate in experiments, especially if what we we manipulate can also cause changes in other things. For example, use an emitter to spray some positrons at another molecule and observe effects in that molecule. For Hacking, this means that positrons are real–because through experiments (practice) we manipulated them and used them to manipulate other things. In that moment, he could care less about any theory that may explain what just occurred in the experiment. The theory can come later. For him, science occurred during that experiment, even if the experiment was a failure by conventional scientific standards. Surely, we do not limit science as a label only for the success parts of science. On the other hand, van Fraassen maintains that theory is always relevant and always present.

    I know you said you wanted to define science as more than its practices in order to facilitate attributing things to science. But I don’t know exactly what it is, then, you would be attributing things to. Perhaps it is a category mistake to think that things can be attributed to science in the same way things can be attributed to practices of science. You cannot show up at my campus, see all the dorms, admin buildings, class rooms, and athletic facilities, and then ask me to show you the university. The university is not a thing in the same category as the things that make up the university. To see all those facilities is to see the university.

    Perhaps you can say that science is any “set of practices involving observation, experiment, and theory construction” that also purports to or aims to provide a story about how the world works. Eliminating social science from this set still does not provide a clear target for attributions. Within the remaining set of “science” you still have distinctions between realist, anti-realist, positivist, logical positivists, logical empiricists, instrumentalists, and on and on and on.

    So, perhaps no traditional designation for a discipline will be coherent enough to warrant attributes to itself. And if you make theory the criteria to distinguish between science and non-science, then you rule out all those cases where experiments without theory have achieved significant results within the scientific community.

    In the end, we will necessarily have to fund opposing sets of scientific practice because there is no a priori way to know which ones will produce effects that satisfy our interests. Once you talk about funding and interests, you are talking about politics.

  8. Jeremy,

    Hmm…I don’t think I meant to turn this discussion to the question of what is or is not science. In reality, I think I’m most comfortable going with your summary that “no traditional designation for a discipline will be coherent enough to warrant attributes to itself.” I think the most accurate way to characterize the situation is to attribute findings to the individual people who made them and to attribute the good or bad consequences of acting on those findings to the finders as well as to the actors (but probably mostly to the actors).

    I don’t think parenting is just about surviving each day. I don’t think I’m the only parent to come up with stories about how the world of my family works. We do the same thing with our co-workers, our neighbors, our classmates, and all of the other social circles we run in. We come up with generalizations about what to expect in certain situations. So the storytelling aspect is not a characteristic of science that differentiates it from all the things that we normally don’t consider to be science.

    I don’t think it’s useful to differentiate between science and politics – at least, those distinctions aren’t any more useful than most disciplinary distinctions within social science – and that’s why I think disciplinary attribution can be useful even if it’s not entirely accurate. Attributing a body of findings, or the consequences of acting on those findings, to a discipline allows for a sort of shorthand for differentiating theories and practices that improve expectations from those that don’t. Questions of funding and interests and support are at least partially related to the extent to which people see value in a particular endeavor – the extent to which they see that endeavor leading to improved expectations – and therefore operate as a proxy measure for the success of the endeavor itself.

    My main concerns here are (1) people often claim similarities among individual researchers’ projects and findings and (2) people often attributed good or bad outcomes to those perceived similarities. Group definitions based on those similarities will always be inaccurate to some extent, but I think it’s fair to assume that the similarities themselves do exist and that the disciplinary distinctions are one imperfect measure of those similarities. So the question then becomes: under what circumstances are we justified in attributed good or bad consequences to the shared characteristics that a discipline embodies? I don’t like using subject matter or methods as the criteria because that definitely leads to miscategorization: no discipline, especially from among the disciplines whose researchers have at various times claimed membership in the social sciences, has a subject matter or set of methods substantially different from all other disciplines. So an alternative criterion (and I’m sure there are many other possibilities) is to use theory.

  9. Schaun,

    Let me first try to summarize what I think your main points are.

    1. Theory differentiates science from ordinary systematic observation and analysis.
    2. Good scientific theory explains observations in terms of the world as a mechanical system. Such explanations should specify:
    a. entities
    b. relationships between entities
    c. mechanisms that determine those relationships (rules? laws? conditions? or some combo of conditions and laws?)
    3. Rigorous scientific theory specifies a, b, and c either mathematically or computationally.

    I will make one more attempt to remove theory as the demarcator between science and other forms of investigation. I will argue that how one uses models to represent the world is a better demarcator. All of my argument comes from Ronald Giere’s paper, “How Models are Used to Represent Reality,

    It’s fair to say that theories are intended to represent aspects of the real world. There is a relationship between the theory and the real world, and that relationship is defined by how well the elements of the theory hook up to what is really going on in the world.

    Your concern for justifying what we attribute to scientific disciplines, given your contention that theories are the primary distinguishers, is a concern about how well the theories do their job, how well the theories represent what really goes on in the world.

    Giere provides a way of talking about the activity of representing, without having to use ambiguous terms such as theory and law. Without using those two words, he defines the activity of representing in terms of principles, specific conditions, models, hypotheses, and generalizations. He even includes a space to account for a person’s intentions or purposes when they try to represent the world.

    “We should begin with the activity of representing, which, if thought of as a relationship at all, should have several more places. One place, of course, goes to the agents, the scientists who do the representing. Since scientists are intentional agents with goals and purposes, I propose explicitly to provide a space for purposes in my understanding of representational practices in science. So we are looking at a relationship with roughly the following form: S uses X to represent W for purposes P.”

    S uses X to represent W for purposes P. Here, S is the investigator and X is the model.

    “It is models that are the primary (though by no means the only) representational tools in the sciences.”

    A scientist constructs a model by using principles and specific conditions. Applying the model to the real world “generates hypotheses about the fit of the specific models to particular things in the world, hypotheses that may be generalized across previously designated classes of objects.”

    “It is not the model that is doing the representing; it is the scientist using the model who is doing the representing. One way scientists do this is by picking out some specific features of the model that are then claimed to be similar to features of the designated real system to some (perhaps fairly loosely indicated) degree of fit. It is the existence of the specified similarities that makes possible the use of the model to represent the real system in this way.”

    If I understand correctly, your problem with Bernard and King is that they claim certain benefits for social science, i.e., social science caused some good things to happen, but that they do not establish “social science” as a necessary element in the causal chain that led up those benefits. The reason they don’t, or cannot, defend “social science” as a necessary element in the causal chain is because social science, at least the bits pointed to by Bernard and King, does not have rigorous scientific theory or, in a different phrase you used, “any core principles of scientific knowledge.”

    However, your mixing of the terms ‘theory’ and ‘core principles’ may be creating some confusions.

    In the traditional sense of theories we think of empirical laws “as generalizations that are both universal and true.” Giere, on p. 475, explains why this concept of laws forces them to ultimately be false or only vacuously true. Instead it is better to think of them as principles and to figure out “how the principles function in representational practice.” They function as templates, along with specific conditions, in the construction of models.

    If we use Giere’s model-based understanding of the practice of representing, then you can get rid of the terms theory and law. You can focus on the use of core principles within and to construct useful models.

    “The required qualifications, then, concern only the range of application of the model. One need only indicate, tacitly or explicitly, where it applies or not, and to what degree of exactness. One cannot directly test principles by empirical means. One can only test the fit to the world of particular models that incorporate the principles. Thus, empirically, there is no basis for preferring one set of these principles over the other.”

    So, back to your original concern about if theory is the best indicator of what demarcates science from other forms of systematic observation and analysis.

    Instead of using theory as the demarcator that can be used to criticize social science, we can say of certain social scientists:
    – that they misrepresented the world by applying poorly constructed models to the world, given whatever the scientist’s purposes were
    – that because their model is a misrepresentation, then what they claim as similarities between their model and the world are not similarities
    – that these false similarities, then, cannot be the grounds for the scientist’s findings
    – and finally, then, that any public attributions of cause or of success, are not warranted for this instantiation of this particular model given whatever the scientists purposes were

    In the end, it may turn out that there are no principles sufficient enough in regularity, to be considered principles of behavior, or principles of culture, or principles of society (assuming that such principles would have to be something other than mechanical principles). If that is the case, then no models constructed with such principles can be used by a scientist to adequately represent aspects of the real world. This, to me, raises even more interesting questions.

  10. “Most of the examples of ‘social science’ Bernard gives in his article seem to be focused almost entirely on the observation part of the picture: people looked at stuff and noticed patterns and then acted on their observations.

    “True, acting on those observations certainly qualifies as ‘application’, but I’m not convinced we should call that science.”


    Huh? Isn’t looking at stuff and noticing patterns the very essence of science? Take, for example, when Gregor Mendel looked at his pea plants over several generations, noticed a pattern in their trait inheritance, and developed genetic theory. Was this somehow not science?

    — Ashkuff | | How to use anthropology, in business and ADVENTURE!!!!

  11. Jeremy,

    I don’t see much of a difference between models as you (and Giere) describe them and theory as I described it. In both cases, it seems we’re talking about concepts used to organize and prioritize information. I don’t think statements about concepts (theory, models, core principles etc.) being “universal” and/or “true” are very useful. It seems more appropriate to talk about consistency or reliability. Does employing a particular concept consistently (across time and across cases) lead to more realistic expectations about the world than not employing that concept? If so, then that concept is universal and true enough for practical purposes.

    What I was trying to say in my original post is that, for some good or bad outcome to be justifiably attributed to some sort of collective (a field, discipline, etc.) instead of to the individual people who were the proximate causes of that outcome, the outcome needs to be attributable to the use of concepts generally accepted by members of that collective and not just to observational or analytic methods commonly employed by members of that collective. So I guess this goes toward addressing Ashkuff’s comment – looking at stuff an noticing patterns is a necessary component of science, but it’s not sufficient. For purposes of attribution – as I said before, I’m not trying to define what is and is not science in general – the concepts are a better differentiating characteristic than the methods are.

    The reason I dispute Bernard’s and King’s attribution is because they haven’t demonstrated that most of the good outcomes they cite are the result of concepts that reliably lead to more realistic expectations about human behavior in general. Not all of their examples suffered from that problem. Bernard’s desensitization example, for example, seems to be a pretty good illustration of beneficial outcomes (helping people overcome phobias) that can pretty clearly be attributed to behavioral/social science. In most cases, however, Bernard and King only demonstrate that some people relied on observations that allowed for more realistic expectations about human behavior in one-off situations. It doesn’t seem right to credit those outcomes to social science.

  12. I think that there is an overly rigorous definition of “Science” being applied here.

    I think that the first thing we need to consider is how much younger social sciences are (e.g. Psychology) when compared to our paragon sciences. Systematic, observational research is a cornerstone of all science and I would argue that it is, alone, science. It is, however, incomplete science when experimentation is not included. Think back to the basic descriptions of the scientific method, the first step is always observation. Theories, however, definitely exist within the realm of social science. Take for example the concept of framing discovered by Kahneman and Tversky. This has done more than contributed to some marketing and money schemes–it has fundamentally defined how humans view the world. Consider the Milgram studies and the effects of authority on people. Both of these phenomena are well researched, are supported by fairly robust theory, and have been replicated through scientific rigor.

    Granted, there is a lot of bad science, pseudo-science, and non-science being conducted in the name of “social science” but to dismissively sweep the entirety of that broad domain under the rug is really a disservice to scientists everywhere. Consider the task at hand for social scientists. Every single behavior of a living creature that a social scientist examines subsumes dozes (or perhaps hundreds or millions) of processes. Each of those processes exists at a different level of analysis–perhaps it is a more fine-grained social science or basic biology or basic chemistry or maybe even physics. As far as I’m concerned, the ability to reproduce a behavior in humans even a third of the time or half of the time is a pretty sizable feat.

    Social scientists don’t have the luxury of pure scientific control. They cannot isolate a single behavior from other behaviors as chemists can isolate a single ion from a molecular compound. Social scientists also don’t have the luxury of the groundwork that the hard sciences–the scientific method has only really helped to create the realm of social science in the last century or two. Social scientists have to try to distill patterns out of complex and poorly understood systems of systems of variables. Not only are they studying impossibly complex phenomena, but they are doing so with very new tools. The tools available to the social sciences are analogous to the magnifying glass in chemistry–they don’t have their electron microscope yet. They’re a young science that is trying to gain a better grasp on the nature of the world at a very macro level.

    Social Science as a very, very broad domain (I find the term about as descriptive as lumping mathematics and biology in the same group). Sure, they have a lot of kooks and snake-oil peddlers lurking about, but that is hardly unique to any single domain of inquiry at any level. If they’re still in the infancy of their field, roaming around and observing the world as best they can–you can hardly use that criteria alone to say they are not a science–they don’t have the luxury of something akin to the theory of evolution to unify their domain (yet). They do have a lot of fledgling theories that continue to emerge. As the major, “real” theories in social science gain and grow, the others will no-doubt be subsumed or discarded as is the common practice of all science.

  13. Jeff,

    Interesting. Why would we want to adopt a less-rigorous definition of science? What do we lose by being more selective about the activities to which we choose to ascribe that title?

    I don’t find the argument about social science’s relative youth to be very convincing in light of the fact that medical science is roughly the same age as social science, and has been able to accomplish a whole lot more, even in the areas of public health where lab-like environments aren’t really feasible. I don’t see how medical problems, especially those tackled at the population level instead of the individual-organism level, are any less complex than the problems tackled by social scientists. I also think you’re drastically understating the complexity chemists and other “hard” scientists face in their work.

    I agree that social scientists don’t have great measurement tools, but that fact would engender more sympathy if I saw a greater proportion of social scientists clamoring to try out the better tools that have been developed in the last decade or so (I think Paul wrote about that a while ago:

    As I said in one of my previous comments in this thread, I’m not writing off social science. I *am* a social scientist. I’ve even given a specific example or two of real-world outcomes that can rightfully be attributed to social/behavior science. The point of my original post was to argue that we shouldn’t attribute outcomes to social science just because the outcomes dealt with people’s behavior.

  14. Schaun,
    I love Bertrand Russell’s example of expectation, observation and science. I quote:
    “The man who has fed the chicken every day throughout its life at last wrings its neck instead, showing that more refined views as to the uniformity of nature would have been useful to the chicken.” (RUSSELL, Bertrand: “Problems of Philisophy”, Ch. IV).
    I agree that observation per se is not science: observation is a necessary but not sufficient condition for science. Observation can result in descriptions. Descriptions are also necessary but not a sufficient condition for science.Building patterns is also a necessary not sufficient condition for science. Build in these patterns in a encompassing theory on the basically accepted premisses is sufficiently Science.
    Bertrand Russell’s example shows one thing very clearly: the difference between the induction of the behavior of non-conscious entities and human behavior. I don’t want to say that human behavior is principally unpredictable, but it is potentially unpredictable (we are conscious and non-conscious at the same time). So, in a row of events or behaviors a1, a2, …, an, behavior an+1 can have an unexpected and significantly different quality. We can see this phenomenon in many complex systems, and human beings and their behavior are one of the most complex systems. For me this is enough reason to reject Bernard’s simplified methodology. In almost Bertrand Russell’s words: a more refined views as to the uniformity of nature would have been useful to the anthropologist.

  15. Schaun, your post reminded me the disdain I acquired for the vagueness of social science and the reluctance to go beyond something that is conceptually replicated. Good to know there people who feel the same way about this! Might be worth starting a blog.

  16. Pingback: on the role of theory in science and social science | EIGHTEEN CHAINS

Comments are closed.