The problem in both economics and psychology is the desire to use the understanding part of science in order to attain an engineering result of controlling people's behavior. Atoms don't care if they're controlled by other humans - but most humans do care. Humans care and many change their behavior in order to counter the attempted control.
Similarly, if one tries to find some "law" or regularity about how the market works, which can then be used to invest so as to make a premium profit, the behavior of others in the market will change to invalidate that law. (I call this Grey's Law)
An example would be a December-January buy - sell opportunity. It was seen that many investors sell stocks in December, in order to optimize their personal taxes, especially capital gains taxes. Thus there was a pre-Christmas market dip, followed in January with a bump up, as the cash came back in. As this became known and discussed, buyers bought sooner (in December), so as to invalidate the prior regularity.
"Buy low, sell high" always works - but it's not so helpful to know if any particular investment opportunity is low or high.
The 2006-2008 housing bubble crash (2006) & then the derived financial crash (2008) was because of lots of sloppy investments because "housing always goes up", was a regularity that had been true since before the S&L crisis (starting 1986), and especially thru the Y2K & dot.com bubble busting. Tho in 2003 The Economist was predicting a housing bubble crash - which didn't happen for another few years.
Similarly, Greenspan with his Dec 1997 "Irrational Exuberance" speech was sort-of-right about the coming bust, but that dot.com bust happened years later, March 2000.
Social science is good for understanding - but fails at "social engineering". Because those darn Free Willed humans don't keep behaving the same way when "nudged" or ordered to.
This is all absolutely true in economics. VERY discouraging to a data drive, free market economist like myself. And we are seeing a great deal of 'thought-influence' from NGO's that publish garbage cost benefit analyses- people think these are science.
Are we supposed to forget about the elephant--or, rather, the donkey, in the room? The "replication crisis" has the same explanation as the question, "Why do political pollsters always seem to make mistakes in the same direction?" Should not take Gauss to reckon up this one......
Let’s pretend for a moment that some super smart people from say the RAND Corporation figured out how to manipulate the thinking of the unvaccinated to change their minds. This crack squad goes door to door for a few months programming this change in the anti-vacc people. Now the RAND team wants to scale up their work. They codify the lessons from their field work into a set of algorithms. They then recruit agents to learn the algorithms and they go back out into the neighbourhoods of America for phase 2. Now they publish the results from their study and it’s presented on Fox News and CNN. All the predictable responses from the talking heads play out. After publicising the work, I ask you, could the study be replicated? Or would the anti-vacc person at the threshold of the entry to their door now retort, “hey I heard about your work manipulating people. No thanks prick.” ? You can’t replicate something in a different world. As Gary Taub always asks about a study, “what’s your background?”
I suggest the problem in the case of social science is that experiments with humans, which involve many uncontrollable and even unrecognized variables, are used to contradict social policies that were developed by civil society trial and error over decades centuries or even millennia. To call this high risk (irresponsible high risk) is understating the problem. It is hubris writ large. The results have been dreadful for society. Yet the government treats social science dictums as gospel. I make the same claim about economics. Time to recognize that economists do not understand enough about the ramifications of their theories to be making policy. We are seeing this play out today, with inflation, perpetual and rapidly growing debt, and a Fed with no way out. And after less than 15 years after the past policy disaster. Social sciences have done far more damage than good. Time to admit it and treat both as in the primitive stage of experiments “science”.
I agree with Daniel Kaufman and also Yuri Bezmenov. I would only add that it's a *whole* lot easier for *politics* to sway the results in social sciences. Which makes me consider a lotta the results with a great deal of skepticism. But that's just me.
The replication crisis in psychology is limited to certain research areas, such as social psychology and clinical psychology, in part because these areas of research tend to rely on arbitrary metrics and self-report data. Ironically, you state at the beginning of your talk that behavioral psychology in particular has been singled out for replication failures. Nothing could be further from the truth. Basic studies of learning and cognition are among the most replicable findings in all of psychology. These studies do tend to be experimental, are conducted in controlled laboratory environments, and utilize more standardized measures. The basic learning processes of Pavlovian and operant learning can actually be demonstrated in a classroom environment with the right equipment. For decades, students have conducted their own experiments for classroom assignments, and the basic findings regarding behavioral acquisition, reinforcement, extinction, punishment, and stimulus control can be very effectively replicated within these assignments, much like classic experiments in physics and chemistry are carried out by students in academic venues around the world.
The problem in both economics and psychology is the desire to use the understanding part of science in order to attain an engineering result of controlling people's behavior. Atoms don't care if they're controlled by other humans - but most humans do care. Humans care and many change their behavior in order to counter the attempted control.
Similarly, if one tries to find some "law" or regularity about how the market works, which can then be used to invest so as to make a premium profit, the behavior of others in the market will change to invalidate that law. (I call this Grey's Law)
An example would be a December-January buy - sell opportunity. It was seen that many investors sell stocks in December, in order to optimize their personal taxes, especially capital gains taxes. Thus there was a pre-Christmas market dip, followed in January with a bump up, as the cash came back in. As this became known and discussed, buyers bought sooner (in December), so as to invalidate the prior regularity.
"Buy low, sell high" always works - but it's not so helpful to know if any particular investment opportunity is low or high.
The 2006-2008 housing bubble crash (2006) & then the derived financial crash (2008) was because of lots of sloppy investments because "housing always goes up", was a regularity that had been true since before the S&L crisis (starting 1986), and especially thru the Y2K & dot.com bubble busting. Tho in 2003 The Economist was predicting a housing bubble crash - which didn't happen for another few years.
Similarly, Greenspan with his Dec 1997 "Irrational Exuberance" speech was sort-of-right about the coming bust, but that dot.com bust happened years later, March 2000.
Social science is good for understanding - but fails at "social engineering". Because those darn Free Willed humans don't keep behaving the same way when "nudged" or ordered to.
This is all absolutely true in economics. VERY discouraging to a data drive, free market economist like myself. And we are seeing a great deal of 'thought-influence' from NGO's that publish garbage cost benefit analyses- people think these are science.
Are we supposed to forget about the elephant--or, rather, the donkey, in the room? The "replication crisis" has the same explanation as the question, "Why do political pollsters always seem to make mistakes in the same direction?" Should not take Gauss to reckon up this one......
Let’s pretend for a moment that some super smart people from say the RAND Corporation figured out how to manipulate the thinking of the unvaccinated to change their minds. This crack squad goes door to door for a few months programming this change in the anti-vacc people. Now the RAND team wants to scale up their work. They codify the lessons from their field work into a set of algorithms. They then recruit agents to learn the algorithms and they go back out into the neighbourhoods of America for phase 2. Now they publish the results from their study and it’s presented on Fox News and CNN. All the predictable responses from the talking heads play out. After publicising the work, I ask you, could the study be replicated? Or would the anti-vacc person at the threshold of the entry to their door now retort, “hey I heard about your work manipulating people. No thanks prick.” ? You can’t replicate something in a different world. As Gary Taub always asks about a study, “what’s your background?”
I suggest the problem in the case of social science is that experiments with humans, which involve many uncontrollable and even unrecognized variables, are used to contradict social policies that were developed by civil society trial and error over decades centuries or even millennia. To call this high risk (irresponsible high risk) is understating the problem. It is hubris writ large. The results have been dreadful for society. Yet the government treats social science dictums as gospel. I make the same claim about economics. Time to recognize that economists do not understand enough about the ramifications of their theories to be making policy. We are seeing this play out today, with inflation, perpetual and rapidly growing debt, and a Fed with no way out. And after less than 15 years after the past policy disaster. Social sciences have done far more damage than good. Time to admit it and treat both as in the primitive stage of experiments “science”.
‘Leads Dan and me to’ not ‘Dan and I’
I agree with Daniel Kaufman and also Yuri Bezmenov. I would only add that it's a *whole* lot easier for *politics* to sway the results in social sciences. Which makes me consider a lotta the results with a great deal of skepticism. But that's just me.
The replication crisis in psychology is limited to certain research areas, such as social psychology and clinical psychology, in part because these areas of research tend to rely on arbitrary metrics and self-report data. Ironically, you state at the beginning of your talk that behavioral psychology in particular has been singled out for replication failures. Nothing could be further from the truth. Basic studies of learning and cognition are among the most replicable findings in all of psychology. These studies do tend to be experimental, are conducted in controlled laboratory environments, and utilize more standardized measures. The basic learning processes of Pavlovian and operant learning can actually be demonstrated in a classroom environment with the right equipment. For decades, students have conducted their own experiments for classroom assignments, and the basic findings regarding behavioral acquisition, reinforcement, extinction, punishment, and stimulus control can be very effectively replicated within these assignments, much like classic experiments in physics and chemistry are carried out by students in academic venues around the world.
This would be a stronger comment with a link to at least one such often replicated finding.
So the academy strains to promulgate immutable truths in the social sciences, a fools errand!
This May explain why the woke have taken over social sciences - they can set the narrative and use hard to replicate studies to support it.