Is Social Science an oxymoron?

As a social scientist (my future) and a person of biology and scientific research (my past), I can understand your criticism – it’s thrown around a lot in academic circles, this kind of attempt to discredit – however, as you note, the umbrella of social science covers a lot of areas that do utilize the ‘scientific method’ and all the ‘rigorous’ aspects that come with it. The early pioneers you mention understood that science, to them (and to me), didn’t just mean ‘get to facts using hypotheses, etc.’ because it was social science and wasn’t always having to do with questions that can be addressed that way. I like this part from wiki:

“The term may be used, however, in the specific context of referring to the original science of society established in 19th century sociology. Émile Durkheim, Karl Marx and Max Weber are typically cited as the principal architects of modern social science by this definition.[4] Positivist social scientists use methods resembling those of the natural sciences as tools for understanding society, and so define science in its stricter modern sense. Interpretivist social scientists, by contrast, may use social critique or symbolic interpretation rather than constructing empirically falsifiable theories, and thus treat science in its broader classical sense. In modern academic practice researchers are often eclectic, using multiple methodologies (for instance, by combining quantitative and qualitative techniques).”

I think that no social scientist should put forth conclusions and state that they arrived at that conclusion using the scientific method if that wasn’t the method they used. However, I don’t think many social scientists do that – just like ‘normal’ scientists, they put forth theories about humanity, about societal patterns and understand that ultimately much of it is conjecture but the importance of the contribution of social science to me is invaluable. As a sociologist, I firmly believe that ‘regular’ science was NEVER objective in its trajectories and to this very day carries with it all the flaws of humans – that’s why many scientific ‘truths’ can be questioned by sociologists because there are social forces shaping science, affecting scientists and each period of discovery needs to be assessed within its social context. For example, my future research will be grounded on the work of social scientists who work on addressing evidence out there claiming to show sex differences between men and women – as it so happens, it is the regular scientists, in this contested topic, that are using VERY shady science, straight down to lies, which then gets disseminated to people and affects policy and educational options, etc. It is the social scientists in this field that have taken the task of debunking their faulty methodology, of asking about correct scientific method and whether all the steps were followed. So I am glad to be part of a field that often keeps the ‘objective’ scientists in check. Though, I call myself a sociologist rather than a social scientist because the emphasis, at least for me, is on the social rather than the science – but you will never hear me proclaim otherwise.

I think it’s a science. They tell us a lot about people and societies based on case studies, experiments, research, observation studies… so on. Researchers develop ideas and theories based on what they already know about people; how their brains function and make connections, how we learn and grow, how people interact with others – each in this list from a different focus within the social sciences.
It’s not just people making stuff up and finding things to support their statements. A good researcher revises their theory when they come across data that doesn’t fit with what they thought, just like any other scientist would.

You’re not going to succeed in reclaiming the word science.
Human irrationality prevails again.

Most of the people that call themselves scientists are just lab technicians anyway.
Linguistics is a science despite the fact that very few people manage to talk perfectly, just like physics is a science despite the fact birdshit falls from the sky.

I would argue that social science is closer to philosophy, the first science.

Well maybe the name is an oxymoron, for the reasons you’ve listed, since I don’t think a traditional scientific approach is used to study these subjects, nor do we know nearly enough about them (But that may stand for more scientific things just as much, such as space travel, the medical field or animals.) to use methods which require at least a certain amount of confirmed knowledge, but it does remain that there’s something there to study. Things like psychology or behaviors can definitely borrow the same scrutiny that more established scientific matters do.
Hell I’m sure even philosophy can, since I’m of the mind that philosophy is what we called psychology before we knew what that was.

@Symbeline I certainly did not mean to imply that the subjects are worthless or should be abandoned. Art, literature, music, dance, and tons of other disciplines are not science, but are certainly very valuable studies.

behavioral and developmental psych seem to me to still be little more than pseudo-science

What?

@nikipedia “What?” Is no argument, question or proper response.

@Zyx: OMG! I thought it was such a cleverly constructed argument! And I guess I am confused about what a question is! As for its propriety… are you now responsible for the Fluther guidelines? I am so glad you’re here to police the quality of my answers. Thanks! <3 I really appreciated this helpful exchange!

@nikipedia I suppose since the statement you challenged was mine, I should take up its defense, and not leave it to @Zyx . I stand by the statement. While I recognize that those branches of psychology now include elements established by the scientific method, those that practice them seem to mix in much that is not backed by experimental proofs so much as metaphysical belief. Is your experience that this is not true? If so, please point me to references that would enlighten me on the recent progress that psychology has made in being able to truly understand and engineer human behavior.

@ETpro: Can I kill two birds with one stone? (What an awful expression.) I am currently taking a class in the department of Psychology and Social Behavior, which I believe qualifies it as a discipline you’ve labeled “unscientific.” And I am supposed to be doing some reading for it right now. So let’s open up the reading and decide if it’s scientific or not.

Data point 1: DeRivera, J. (2000) Understanding persons who repudiate memories recovered in therapy. Professional Psychology, 31, 378–386.

Here’s the abstract: How can clinicians avoid participating in the development of what may be false memories? Persons who had repudiated “memories” of sexual abuse recovered in the course of psychotherapy were offered three possible explanations for what had occurred. Although a majority endorsed a mind-control model, a substantial minority endorsed self-narrative or role-enactment models. The processes endorsed were consistent with retrospective descriptions of therapy. A knowledge of these different models may help clinicians understand how clients develop false memories and avoid social influences that contribute to unproductive therapy.

While the authors don’t explicitly state an a priori hypothesis, I could easily extrapolate one (clinicians may influence false memory development through mind-control, self-narratives, or role-enactment). And they do an experiment (not described in the abstract) in which research participants filled out questionnaires designed to test this hypothesis, and then inferential statistical tests were performed to test specific hypotheses about the relationship between certain kinds of therapy and the models endorsed by the participants. Researchers were blind to the identity of the participants.

Seems pretty scientific to me.

Data point 2: McHugh et al (2004) Family relationships after an accusation. J. Nerv Ment Dis, 192, 525–531. (link opens pdf).

In this article, the authors do explicitly state their research questions, although again they fail to state a hypothesis: “Did retractors, the ultimate recon- cilers, always go through the returner phase? Was the process always unidirectional, or did returners ever change back into refusers or retractors into refusers? Was it crucial to reach the retractor phase for other forms of family life to proceed, or did the elephant in the living room aspect of the returner status interfere significantly with family harmony? What actions of the accused encouraged the accusers to move to returner or retractor status?”

In this study, researchers again used blinded survey questionnaire data to investigate their research questions. Again, they used inferential statistical tests to find the relationship between their independent and dependent variables. They use conventional statistical tests (chi-square) and thresholds for significance (p < 0.05).

So I’m pretty convinced. What else do you need to be convinced that psychology (in particular, behavioral psychology) is indeed a scientific field?

@nikipedia Granted some solid methods are being applied in the two above studies. I am not sure I am understanding either study from the details gioven, but what I am getting is that both have to do with interpersonal relations between people after charges or at least memories of sexcual abuse. What seems missing is a definitive proof the alleged abuse actually occurred. My guess is it likely did in some instances and dod not in others. I am unclear on how much you can discern about human behavior if you don’t bring that knowledge into the rquation. You may be observing two individuals doing the same thing, but for almost entirely different reasons.

@ETpro: That’s completely orthogonal to whether or not the scientific method is being employed.

@nikipedia I don’t think so. Questioning whether an experimental result is telling you what you claim it to be telling you is as straight-path as science can get. In any case, I did not intent the OP to suggest that these disciplines completely shun the scientific method. The question I meant to ask was whether they employ it as the foundation for their conclusions and practice in sufficient quantity to be rightly called sciences.

And I’m trying to provide evidence that at least the disciplines I noted do. Maybe historically they weren’t held to such high standards but my experience with contemporary psychology is that it’s every bit as rigorous as the life sciences. (I can’t speak for natural sciences.)

@nikipedia I shouldn’t be keeping you from your studies. But I do see flaws in what I think the experiments you mentioned were menat to deal with. I do not think those methods would give reliable data.

Whether the specific techniques employed will effectively address the question is, again, orthogonal to whether or not the studies are using the scientific method. I’ve been avoiding addressing the specific issue you raised because there’s a tremendous body of literature that I think satisfactorily addresses that question. I can send you links if you want.

@nikipedia No need. I can Google it and find them. If there is something you feel is particularly strong at proving your case, that I would appreciate a link to.

@ETpro I’m sorry but I can’t believe that you don’t regularly then see flaws in ‘real’ scientific studies – because if I remember correcly during my pre-medicine and graduate studies, there was no shortage of current scientific articles to tear down because they were atrocious examples of science, yet published in the bestest coolest like smartest international journals. I’ve been published for my research numerous times and I know what it takes to put forth evidence and develop correct experiments and I’ve taken enough advanced statistics courses to know that what gets put out there by scientists and then picked up by the media is so often utter crap.

No, as long as scientific method and empirical data is involved.

The main problem with social science is, in many ways, the same problem with medical science: people.

Science can give us demonstrable evidence of something as a fact the less subjectivity is involved on either side of the experiment. As examples, we attempt to control this in experiments on the side of the person performing it through double-blind process, and on the side of the subjects of the experiments through appropriate sampling.

The problem with science is, of course, that rarely is subjectivity eliminated…and the more it is a necessary part of the experiment, hypothesis, or data the less it should be used to provide an answer, and more to provide ways we can try treatment, see how the knowledge can be used in policy development knowing that it’s not firm, etc.

Math, for instance, is science. Everything else falls below math. The hard sciences require certain assumptions on our part (subjective assessments of what’s probably the case based on what we know) and therefore puts forward theories as truth until proven otherwise.

Moving below that, where people are a necessary part of the equation, then we are in the realm of “this might be the case” in terms of most of the answers we get. But those are still useful. As @nikipedia points out, there are many experiments that are repeatable and carefully controlled. However, in dealing with psychological reaction, the variables are so numerous it is impossible to control for all the variables. Additionally, generalization is often a problem because we can often find problems in the sample – either through selection, size, etc. The problems of the experiment aren’t do to them being preformed without scientific standards, but rather because these problems are necessary and unavoidable. In medical science this is less the case, but because human subjects can’t be used in most cases, and because in treatment people often lie about potential causes, the clinician or physician moves forward based on the best information he or she has at the time.

Therefore, the problem is not in how we view whether or not the areas of knowledge are scientific, but exactly the nature of the results we get from the experiments performed in the area. Because the essence of science is that it constantly attempts to disprove itself, then a “social science” can be considered more honest that the answers are more stepping stones to better answers. Much of the problem, in fact, in the scientific community’s communication with the general community is what @Simone_De_Beauvoir refers to above – something is stated as a “fact” when scientists mean “everything that we’ve predicted has come true so far,” whereas a common understanding of fact is that a fact is a thing that is real, and true, and proven beyond a doubt. It is a fact that I just took a breath.

Social science as much as hard science allows us a better understanding of the world as we know it. However, neither should be understood to have provided an answer – just our best answer so far.

Ironically, in terms of general conception, it appears that if it is an oxymoron, social science is, in fact, more intellectually honest than science as a term, standing alone.

@Simone_De_Beauvoir In fact I do.

@ETpro So, if you do, then you should criticize ‘regular’ science for being as flawed as social science. Neither are perfect.

@Simone_De_Beauvoir I have taken on other sciences in other threads. That just wasn’t the subject here. Also, while I see cosmologists go off on tangents at times (theorizing dark matter and dark energy to make the standard model hold up against new evidence it is flawed, for instance) I think the natural sciences are far more amenable to the scientific method and have far fewer variables to control for, and so errors are much more common in studies dealing with the behaviors of mankind and the social institutions he builds. That was why I picked on the poor social scientists. But that was certainly not meant to suggest we should abandon our efforts to understand these things. They are likely of far more importance to chart than a Universe that is expanding faster than the speed of light and will someday be invisible from here anyway.

@ETpro Ok. To me, it was an implied subject because when you say oxymoron, I read it as social science is an oxymoron because it’s not science – if that’s not what you meant, forget what I said then.

@Simone_De_Beauvoir I don’t think some of the social sciences, psychology, sociology, political science and the like are far enough along in understanding yet to be true sciences. They seem to me to be in the dawn of their understanding.

and have far fewer variables to control for

Unless you have reason to believe that one of the uncontrolled variables is systematically biased to be present more in one group than another, the only risk you run by including this uncontrolled variance is that of a false negative rather than a false positive.

Let me justify this with an example. I study human memory. Suppose I wanted to know if some factor, e.g., biological sex, influences memory. The independent variable I am manipulating is sex. And then there are literally thousands of other variables I am not controlling for: what town did each subject grow up? What color is each person’s hair? Does everyone shop at the same grocery store? Do people in each group jerk off at equivalent rates?

But it completely doesn’t matter. As long as that variance is spread equally between the groups AND as long as that variance has no influence on my dependent variable (i.e., memory), I can introduce all the variance in the world. The point of parametric inferential statistics is to measure whether the difference between my two groups—men and women—is greater than the difference within each group, which tells me whether or not my manipulation had an effect.

I can accidentally introduce so much variance within each group that my study needs huge quantities of people to have sufficient power to reveal a real effect. So I could accidentally conclude that biological sex has no effect when, in fact, it does, but I was unable to detect it owing to the huge amount of variance within each group. But this is a very different problem from a false positive—i.e., concluding that the independent variable has an effect when in fact it does not.

Sorry for the rant.

@nikipedia You are the expert, and I’ll be the first to admit my weakness in statistics and the fact I haven’t used what I did learn for many years. But I fail to follow the logic above. In your example above, you are trying to determine if gender has an effect on memory. Let’s say your data show that it does. That’s a positive. But what if any one of those other uncontrolled variables turned out to be what was actually causing that positive, and your sample happened to be skewed such than more of the men than the women happened to have that trait? Without a very large sample size, I do not see how you can rule out false negatives OR false positives.

@ETpro: Let me make sure I understand before answering. Which of these situations are you proposing?

(1) I find that statistically, men outperform women on a memory test, and I conclude that biological sex has an effect on memory. But what really caused the memory advantage was another variable, frequency of masturbation. It just so happens that both in my sample and in the population as a whole, males tend to masturbate more. Therefore, I erroneously concluded that males perform better on the memory task when really I should have concluded that masturbation frequency predicts success on the memory task.

(2) I find that statistically, men outperform women on a memory test, and I conclude that biological sex has an effect on memory. But what really caused the memory advantage was another variable, being born in the state of California. By completely random chance, I had more males born in California than females, therefore I erroneously concluded that being male gave people an advantage on this task rather than being born in California.

Which scenario are you suggesting? Or if you want my answers to both, I can do that too.

@nikipedia – Personally, I think that there is a mythology associated with statistical revelations in the social sciences, as well as the “hard” sciences, that just needs to be cleared up.

Causation is difficult to prove the more and more people are a necessary part of the study or experiment. It’s impossible to know all of the variables that come into play because people are actively deceptive, unlike pure studies of matter and interactions between material things. An atom won’t “lie” to you – but people will lie, leave things out, be unclear, remember things wrong, and all of this interferes with conclusions from experiments with them.

I blame the media and education, however, regarding these issues rather than the scientific areas of studies themselves…but I would like to see scientists communicate better with laymen, however.

On your last point, you’ll get no disagreement from me. I think it’s incredibly important for scientists to communicate effectively with the media and with laypeople and to make an active effort to do so. I suggested that at our next lab meeting we discuss some potential social/ethical ramifications of some stuff we’re coming out with and my adviser initially looked confused, and then self-righteously pronounced that we will never let politics dictate our science (which is hardly what I was suggesting, anyway).

I will also not disagree that causation is difficult, if not impossible, to prove. But I think we can get close enough in many cases, and statistics are really the best tool we have to get there.

I disagree that more people—> more confounds. See my post above. The statistical test we use most often is called an ANOVA, or analysis of variance. This test measures the variance within each group and compares it to the variance between the groups you’re sampling. So you can introduce all of the variance you want to within each group—as long as that variance is balanced between the two groups, the variance between the two groups (attributed to your independent variable) will still emerge.

So all of the variance you get from working with human subjects is a-okay. People can lie and misremember all they want, as long as they do it equally between groups. If you have a real effect, it will still show up in your ANOVA.

@nikipedia – I don’t think that we really disagree on the variance issues. The problem I see is that people don’t understand both (1) controlled variance means that individual differences don’t disprove any results, and the results should be considered in enacting policies, laws, etc., which may help resolve issues that are raised by the results (i.e., anecdotal evidence isn’t proof against the results), and (2) controlled variance means that there are exceptions, so that we shouldn’t treat the results as conclusive to every member of the group the results provide (i.e., anecdotal evidence is important in terms of how we apply those policies, laws, etc.).

It goes to the education side – people have a hard time understanding that findings don’t go to proving who’s right or wrong on a social issue, but rather that there are factors we should be considering, and here is evidence of those factors.

Basically, social science often gives us results that demonstrate complexity in our lives and suggests more questions than it answers (which I think is awesome), and the media and education often focus on an individual studies to demonstrate that one side of an issue is wrong or right (which I think is troubling).

@nikipedia I meant more scenario B in your examples here. Since in your hypotheticals, scenario A was rightly gender based, the gender based conclusion would have been supported even though the reason was perhaps a bit different than one might suppose.

BTW, since you are a researcher i this field, does masturbation improve memory?

You know, I’m not sure anyone’s actually done that study. Think I can propose it for my dissertation? :)

So let me address this issue of accidental sampling bias. You are correct in pointing out that despite our best efforts to randomly select populations, completely random chance can aggregate people with one trait in one group. It sucks, and there’s nothing you can do about it.

But the primary goal of inferential statistical tests is to address exactly this problem: how likely is it that your effect is due to some kind of random sampling error? Most fields set the threshold at 5%, which is called the “p-value” (p = 0.05). The p-value tells you the likelihood that the observed difference between groups was due to random sampling error.

What this means is that you can do your statistical test, get a p-value of < 0.05, and conclude that your effect is statistically significant. This means that the difference between the groups is much more likely to have been due to your independent variable rather than sampling error, but the possibility that it was sampling error always exists.

You are correct in pointing out that you can decrease your p-value by increasing your N (number of people in your group). But for most scientists, p < 0.05 and the expectation that your effect will need to be replicated before it’s really believed are enough. Especially because no N will give you p = 0.0.

I think if you did propose it for your disertatin, you would find no shortage of willing test subjects signing up for the project.

Thanks for the information about the statistical sampling. I can certainly see that if a number of other researchers in various locations indepently try to verify your results, and all obtain confiirming results with a P = 0.05 or less, then you can be pretty certain your independent variable is correct.

@ETpro – I feel like recently I’ve needed studies and/or longitudinal or large-sample surveys or overviews of general behavioral information and characteristic differences among various in-groups to try to combat references to specific examples of behavior as proof that certain groups are the ones to “look out for.”

When used properly, even percentages can be used in at least an attempt to combat irrational prejudices.

@iamthemob Ha! Yes, I was scanning that discussion. Interesting.

If you find terrifying interesting…I agree