Statistics and social sciences

Why don't social science departments (particularly psychology) just employ full-time statisticians to assist with research design and data analysis on projects? This would greatly improve the quality of the research, and reduce the need for endless future replications with slight variations improving some aspect of the methodology and/or data analysis.

OP again, I'll try to explain why I think this.

I've just completed my MSc in Psychology, and I averaged 98% in my statistics modules this year, and had similarly near-perfect scores during my BSc. So I'm some sort of stats genius right? Wrong, I'm nowhere near. Trying to deal with real data, instead of the nice datasets they give you in tests, complicates things hugely. I still have no idea what I'm doing when I try to analyse the data for my dissertation. I'm sure that I'm making shitloads of mistakes, and I keep having to try and Google answers because what I'm doing extends far beyond my course material.

That's why psychologists need to stop pretending we're capable of doing the stats ourselves (unless we study stats full-time, separately), and get proper statisticians to assist us instead. Whilst I've been studying I had to fit my stats modules in around many more psych modules, so I only have a very basic level compared to statisticians. Still, I'm technically able to be published in scientific journals (actually, data from my BSc dissertation is about to be published in EJSP). These journals are awash with research which have errors which will be obvious to statisticians, so why not just employ them full-time to improve the quality of the research? Every psychology department would benefit from this.

>So I'm some sort of stats genius right?
Only if you studied in Europe and not grade inflated US memeschools.

I'm in the UK, and did my BSc and MSc at two different Russell Group unis. Not quite Oxbridge, but Top 10. To be honest I thought that students in the US might actually be more adept at stats seeing as their courses last longer and typically have a broader focus, but that's a complete guess.

I've talked to some social science students about this. Here are their reasons (I shit you not):

>It's unethical to reduce people into commodities
>Statistics is "notoriously unreliable"
>The math is too hard to follow
>Correlation does not imply causation

So in sum it's because the people that study social sciences are retards

>The math is too hard to follow

That's sort of my point. You cannot reasonably expect social science students to understand what are (to me, at least) fairly difficult mathematical concepts when they only study statistics for something like 1/8th of their degree. It requires full-time study, starting at the basics and working from there. In my stats lessons I quickly learned how to do a battery of different statistical tests on SPSS, and didn't really have time to learn what any of the stats jargon actually meant, or what corrections were appropriate when your data isn't perfect, and so on. Also, my stats lecturers would often say "This is the correct way to do it, but it's frequently ignored in the literature", suggesting the problem is pretty endemic.

So it makes sense to me to just hire some full-time statisticians to oversee the whole research process. The social science specialists can still learn the basics so they can follow the rationale for certain things, but the statisticians have the final say. Creates a whole new industry for statisticians to just do stats all day, and greatly improves the quality of research (much of which, at the moment, is worthless).

I have some experience in psychology's research department and i share your concerns. I would frecuently see phd candidates having a hard time with data analysis (and neuroimaging analysis too). It is obvious that the basic formation doesn't prepare you for this stuff.

Maybe this is just my personal experience but where i stayed, phd researchers were just expected to figure out everything by themselves, until it's good enough to pass the peer-review phase.
Hiring a stadistics professional or a neuroimaging technician to assist every project would be too expensive. Besides, why would you invest in that if you have plenty amateurs that can do a "just ok" job?

Statisticians probably have better things to do than correct worthless social science research to be honest.

>why would you invest in that if you have plenty amateurs that can do a "just ok" job?

Because the research turns out to be relatively shit-quality, and you're not really any closer to proving your hypotheses. It's not the problem of any one deparment though, it's more to do with ineffective peer review and journals which publish low-quality research.

I doubt it to be honest. We don't need the very top statisticians, who will no doubt be doing their own research - just anyone who's capable of driving up the quality of the research. The type of graduate who would go into market research or some other bullshit like that otherwise.

You don´t get it. The entire field of psychology would fall apart if your suggestion were to be implemented. Psychology would lose its last remnants of scientific credibility. Profssional psychologists are naturally not very eager to enable their own own demise. There is no mystery there as to why real statistics is not applied to the social sciences.

I don't think this is true - there are a select few departments (e.g. Oxford, Cambridge, Stanford etc) who consistently produce high-quality research, all I'm suggesting is that other departments catch up and journals tighten up their quality standards. Realistically, it would probably be easiest to do that for universities if they just employ a statistician to oversee the research process.

it's not a good idea to outsource your "core business". social sciences are very statistical at their core, and hence why grad schools for social sciences are increasingly quantitative

>SPSS
Kys yourself

Well in my experience, psych grad school still isn't quantitative enough to make qualified experimental psychologists competent enough to complete high quality research. I'm in the UK so it may be different across the pond, but I think it holds true there too. The replicability crisis is at least partly a consequence of this.

What's wrong with SPSS? It's the standard tool for psychologists, as far as I'm aware.

Evolutionary psychology PhD here.

OP, would we do occasionally our own stats, we typically bring in statisticians and methodology specialists when designing large scale studies and analysing data.

What exactly did you learn during your time studying?

Was your degree largely based on evolutionary thinking?

I'm curious, as there are a lot of humanities based meme degrees out there.

Also, if you would, what is your perspective on/understanding of SSRIs?

I prefer R, because I'm a pedantic autist.

>would we do

while we do*

So you're saying you do typically bring in specialists for the bigger studies? I've never heard of that happening before, but I'm glad that it does.

I've been at two of the top 10 unis here in the UK (or maybe top 15, depending on what table you use). So not quite top-tier, but not too far off. I did a general psychology BSc, then a MSc which was a sort of mix between performance, positive, and health psychology. I've had pretty limited evolutionary psychology if I'm being honest, obviously we learned about things like sexual and natural selection, but I've never actually studied a purely evolutionary psych module (which is a shame, as I find it interesting).

I'm generally opposed to SSRIs because I think all antidepressants are overprescribed (especially in the States) and psychotherapy would be more effective for the vast majority of less serious cases, but I don't know how they compare to other antidepressants. It's been a little while since I studied clinical psych unfortunately.

I've never heard of R to be honest, only SPSS, Minitab, and Matlab.

Yes, absolutely as it would be ridiculous not to.

So many things can be overlooked when it comes to developing study/trial structure, hence we bring in trial designers and stats professionals.

Well, all psychology is biological in nature and therefore evolutionary; the apparent divide is resultant of cultural interference with psychology as a science over the past century or so.

When I say biological, I mean based in biology, not that all aspects of psychology are genetically determined.

Psychology is essentially a sub-discipline of ethology, and technically speaking primatology.

However, these seemingly obvious connections are lost on many.

I agree with your comments on SSRIs and so does that data; it's good to hear someone talking sense.

>that data

the data*

Autocorrect is the bane of my life.

Also, I posted about using R.

It's open source and free to download, however it is less user friendly.

Although, it's far more variable and diverse than other stats software.

Yeah I tend to support the evolutionary psychology line of thinking, although most social psychologists seem to have the opposite opinion. I'm sure it's incredibly entry-level for someone studying for a PhD in the subject, but Pinker's Blank Slate was pretty inspirational in shaping the way I think about the brain, and about human nature in general. My department for undergrad was far more social psych in nature, and had (in my opinion) a good amount of low quality studies which, surprise surprise, favoured a 'progressive' way of thinking.

On the subject of SSRIs and other antidepressants, my clinical psych professors were very much in favour of psychotherapy over pharmacology, although maybe that's because they're not legally allowed to prescribe drugs. But either way, I agree with them - doctors are way too quick to prescribe antidepressants and then just assume the person will be miraculously healed with no psychotherapy. On the other hand, I understand that the demand for psychotherapy far outweighs the supply, so this might just be a consequence of the mental health system receiving inadequate funding.

>Yeah I tend to support the evolutionary psychology line of thinking, although most social psychologists seem to have the opposite opinion.

This is the insane thing about psychology: the majority of practitioners and academics seem to think that it is a choice whether one accepts the evolutionary basis of human behaviour or not.

Don’t get me wrong, I understand that you most likely do not feel this way, as you have made clear, however it’s a striking phenomenon, especially considering that we’re talking about a field of biological science.

>I'm sure it's incredibly entry-level for someone studying for a PhD in the subject, but Pinker's Blank Slate was pretty inspirational in shaping the way I think about the brain, and about human nature in general.

Not at all; it is a great book and a much needed one given the current state of psychology.

We need to raise the consciousness of layman and academics alike, regarding the nature of our evolutionary history.

>My department for undergrad was far more social psych in nature, and had (in my opinion) a good amount of low quality studies which, surprise surprise, favoured a 'progressive' way of thinking.

Unfortunately, it’s not too surprising as the ‘standard social sciences model’, as it has become known has roots seemingly everywhere.

>On the subject of SSRIs and other antidepressants…

I agree and to be perfectly honest the largest driving forces behind SSRIs and other antidepressants are industry funded studies, conferences, continued education programs, drug reps and a sea of publication bias, methodically flawed studies and dodgy statistics.

Psychotherapy can work wonders for individuals suffering from disorders involving identifiable triggers and associated irrational thoughts, which may also act as triggers (for example, in the case of health anxiety and certain cases of depression).

In other words, if the over-primed neural networks associated with anxiety/depression correlate with observable external triggers, then it is possible to prevent these networks from firing by reprogramming how one responds to triggers on a case by case basis, which can be achieved with cognitive behavioural therapy and personalised preventative procedures.
Not to mention simple talk therapy, as this may provide individuals a platform on which to explore their experiences and personal grievances.
Heck, sometimes just having somebody to talk to can make a world of difference, after all we are social animals living in an increasingly antisocial world.
Listen OP, I’m currently drunk on cheap champagne however I’d love to talk more with you about these issues, so I’ll return when I’m in a more able cognitive state or perhaps make a new thread on evolutionary psychology.

>On the subject of SSRIs and other antidepressants…

I agree and to be perfectly honest the largest driving forces behind SSRIs and other antidepressants are industry funded studies, conferences, continued education programs, drug reps and a sea of publication bias, methodically flawed studies and dodgy statistics.

Psychotherapy can work wonders for individuals suffering from disorders involving identifiable triggers and associated irrational thoughts, which may also act as triggers (for example, in the case of health anxiety and certain cases of depression).

In other words, if the over-primed neural networks associated with anxiety/depression correlate with observable external triggers, then it is possible to prevent these networks from firing by reprogramming how one responds to triggers on a case by case basis, which can be achieved with cognitive behavioural therapy and personalised preventative procedures.

Not to mention simple talk therapy, as this may provide individuals a platform on which to explore their experiences and personal grievances.

Heck, sometimes just having somebody to talk to can make a world of difference, after all we are social animals living in an increasingly antisocial world.

Listen OP, I’m currently drunk on cheap champagne however I’d love to talk more with you about these issues, so I’ll return when I’m in a more able cognitive state or perhaps make a new thread on evolutionary psychology.

Are you unaware of the existence of Econometrics (and thus Econometricians)? Statistics and maths are a huge component of Economics -- one of the social sciences -- that everyone is rigorously taught (at my university at least) and if you really need to dive into some hardcore methodology there are specialists for it.

Cool that you got 98%, but it's a BSc. BSc means absolutely nothing. I'm guessing you didn't know of anything past OLS by the end, right? What kind of rubbish degree makes you work with made up data instead of actual datasets?

People need to stop being whiny bitches and just learn the maths and statistics to do proper research. Many Economics degrees seem to fuck this up (not at my uni, fortunately), but boy it sure seems like it's infinitely worse for fields like Psychology. Completely undermines an otherwise useful and interesting field of study.

In fairness I think most social psychologists would accept at least some degree of human nature, the question is just about how much of our behaviour is shaped by our evolutionary history - and both sides of the debate have a certain amount of merit. But I'd agree that most departments tend to ignore human nature, and that's for political rather than scientific reasons.

I still think antidepressants are useful in cases of severe depression/anxiety - people just won't commit to or respond to therapy otherwise. But yeah, as you say, it is far more useful to reverse these unhealthy neuroplastic changes which are associated with depression. There seems to be a narrative associated with depression that it develops by 'chance' and is purely physiological, like a physical disease. I understand that they want to stop people with depression blaming themselves, but it's bullshit all the same and harms the healing process because people think they just need pills to get better.

98% was in the MSc, and the datasets were real but specifically chosen so they'd work nicely with each test. Actually I'd never heard of OLS and had to look it up on Wikipedia, which is sort of my point. I'd come across some of the terms before and have a vague idea of what they mean, but not really. In psychology we just learn the process for different parametric and non-parametric tests and when to use each one, without really learning the theory or jargon behind it.

That's not necessarily a criticism of the way we're taught though - with the amount of other modules we have from the various different disciplines within psychology, there's really no more room for extra statistics modules. Between 1-4 tests were taught each lecture, I don't think they could have reasonably taught it in any greater depth. It might be different in economics, I don't know. But for psychology, either the course needs to be extended and statistics needs to take a much greater part, or specialist statisticians need to be brought in. I'd personally be happy with the latter, there's nothing wrong with sharing expertise in my opinion.

Well the thing is, the evolutionary framework used to describe human nature is wholly interactionist by definition.

There's no 'nature vs nurture' and there never has been; it's a false dichotomy and this is obvious to anybody who understands evolutionary biology.

To briefly sum it up in a somewhat pedantic fashion:

We are animals whose behaviour is governed by genetic and environmental programming, stored in the form of particular nucleotide arrangements and neural networks which correspond to our genetic code and an array of psychological mechanisms crafted by the evolutionary processes of natural selection.

These psychological mechanisms govern an array of behavioural programs, which deal with the adaptive problems humans faced throughout their evolutionary history.

Cultures, societies and individual personalities are the manifestations of genetic programming interacting with a wide range of external environments over time.

Culture is essentially the biological equivalent of cloud storage.

From the roaming bands of prehistory, to the economically and technologically specialised civilisations of today, our societies are built upon the foundations of our evolutionary programming.

This of course does not imply genetic determinism, nor any form clearly discernible deterministic nature; there are an unimaginably large number of hidden and imperceptible variables that affect the development of the human mind.

However, largely indiscernible environmental determinism, bound to a genetic infrastructure, is implied

Also yes without a doubt antidepressants, including SSRIs, have shown to be somewhat effective in cases of severe depression and anxiety, as informed by the data.

However, the 'pop a pill', paternalistic and simplistic nature of modern medicine, combined with the profit incentive guiding big pharma has lead to SSRIs and other drugs being handed out like candy.

I think every undergraduate psychology degree should feature a detailed module on evolutionary biology and that many a psychology department is in desperate need of a evolutionary informed intellectual makeover.

Also yes without a doubt antidepressants, including SSRIs, have shown to be somewhat effective in cases of severe depression and anxiety, as informed by the data.

However, the 'pop a pill', paternalistic and simplistic nature of modern medicine, combined with the profit incentive guiding big pharma has lead to SSRIs and other drugs being handed out like candy.

I think every undergraduate psychology degree should feature a detailed module on evolutionary biology and that many a psychology department is in desperate need of an evolutionarily informed intellectual makeover.

Yeah I don't mean to say it's nature OR nurture - it's well-known that genetic code and environment are always interacting and having reciprocal effects on one another. My problem is with social psychologists who observe a particular environmental phenomenon and use it to make large and unsubstantiated claims about society (which influences policy) without even considering natural individual differences.

Yeah, of course.

I'm not directing these statements directly at you by the way, I’m merely using our conversation as a platform to expand on related issues.

I know you know how it be and not be wishin’ it be how it ain’t.

>My problem is with social psychologists who observe a particular environmental phenomenon and use it to make large and unsubstantiated claims about society (which influences policy) without even considering natural individual differences.

Exactly.

Yeah, of course.

I'm not directing these statements directly at you by the way, I’m merely using our conversation as a platform to expand on related issues.

I know you know how it be and not be wishin’ it be how it ain’t.

>My problem is with social psychologists who observe a particular environmental phenomenon and use it to make large and unsubstantiated claims about society (which influences policy) without even considering natural individual differences.

Exactly.

I come from the social sciences (as opposed to the "real" sciences) and I can provide a few reasons as to why social sciences tend to avoid the rigorous statistical analysis characteristic in the life sciences, chemistry, physics, etc.

Social structures are incredibly complicated. Attempting to test a hypothesis is both difficult and often morally ambiguous. Even using the raw data taken from disinterested observations without interpreting the data through a given theory poses the risk of coming to factually untrue conclusions. This does not mean that statistical analysis is unwelcome, however, but that it unfortunately is of little use in many circumstances.

Coming from a social science background, I am uncertain if you are aware of this, so please be patient with me -- social structures are much more subjected to evolutionary pressure than biological structures are. I agree with you that culture is a manifestation of genetics with the environment, but culture is not the biological equivalent of cloud storage.

Culture is a process of change, always reorienting itself according to stresses present in a previous historical moment. As material environments are in a constant state of flux, so is culture; but culture exists on a time-lag. In order to respond to a given stimulus, first it must be recognized and a new strategy developed. But by the time that has occurred, the moment is gone, and new circumstances emerge. While biological evolutionary endowment is, as you say, the foundation, it is both less flexible, and less responsive, than the social evolutionary endowment.

I also agree. Antidepressants, stimulants, mood- or state-altering drugs are overused. Perhaps in extreme cases there is value for them, but this is treating the symptom and not the disease.