Posts by BenWilson
Last ←Newer Page 1 2 3 4 5 Older→ First
-
Speaker: Inequality: Too big to ignore, in reply to Matthew Poole,
the hypothesis behind any policy change is “This will make things better for people in our electorate by delivering xyz outcome,” surely?
That's not really very specific. I wouldn't call that a hypothesis, just a sales pitch. But what I'm getting at is that it's never suggested that they will make a change and then rescind it if it doesn't work. It's got to be accompanied by ridiculous levels of confidence about something that seems to be incredibly uncertain in reality.
-
Speaker: Inequality: Too big to ignore, in reply to Bart Janssen,
If we have enough observations and put together a good model the next experiment is obvious.
Yes, that's pretty much what I was suggesting to Rob before. It's not like it's a crime to look at your observations and then predict a model that looks like the pattern you see there. It's probably not going to shed huge light, but maybe what you're seeking at that point isn't insight, but greater certainty. But I suppose you could technically say that such a hypothesis has no place in that experiment, but could be in the next one.
But really we actually do make hypotheses.
I'd hope so. What I'm interested in is the way in which they come about - ultimately the aim IS the hypothesis, rather than just an agglomeration of unanalyzed data.
-
Speaker: Inequality: Too big to ignore, in reply to Bart Janssen,
Each time a government makes changes to policy that affect both inequality and economic growth an experiment is being performed.
Is it really an experiment if no hypothesis is made beforehand, as Rob Stowell asks?
-
Speaker: Inequality: Too big to ignore, in reply to Bart Janssen,
No way any experiment is simple enough to eliminate any doubt.
Of course that's an impossibly high expectation. But it can be a lot better than having no idea at all well before then.
-
Speaker: Inequality: Too big to ignore, in reply to Rob Stowell,
How could you design an experiment without doing this?
I'm not asking about the theory of doing experiments, I'm asking about the actual practice. It's quite possible to not make a prediction about what will be observed, but to just design an experiment that pits some variable against another, make observations, and then guess the hypothesis at some point that isn't right at the beginning.
-
Speaker: Inequality: Too big to ignore, in reply to Bart Janssen,
The best you get is hypotheses that are consistent with repeated observations.
I’m really curious whether, in practice, you make your hypotheses in advance of your experiments.
ETA: Actually, to be a bit more general, I'm very interested in every bit of light you can shed about how hypotheses come to you.
-
Speaker: Inequality: Too big to ignore, in reply to Bart Janssen,
I get that you are playing devils’ advocate but at some point it just becomes denial.
Believe me, it's not denial. My aim is more to question how it is that we can't adopt an attitude in which methodical experimentation is acceptable. That's not the same as just collecting data. It makes a conscious and deliberate attempt to discover the nature of causes. I'm not blaming anyone for not doing this, because it's not like there's someone even responsible. We're all responsible, and only in the way that people are responsible for not knowing something until they try to discover it. A quite normal kind of responsible, but also a changeable one.
For some reason you are treating their statements as if they have done no analysis at all and have no experience and knowledge at all.
Not at all. I'm asking some question, is all. I was actually hoping for this discussion not to be adversarial at all, and that you might have some useful insight into how economics could be more scientific, rather than concluding that the best we can do is collect some data and trust a bunch of experts to tell a good story, and throw up our hands in despair at the stupidness of people and the evilness of our political system.
-
Speaker: Inequality: Too big to ignore, in reply to Matthew Poole,
as I understand it you are on really shaky ground to make a change to conditions for a single subject and then claim causation of any observed outcome.
I don't think you even need to be a scientist to see that as completely and utterly false. I could drop a single rock repeatedly to observe extremely reliably that when I open my hand, it falls to the ground. It would only apply to that rock, at that point, but that might be enough. I could very reliably see that it was the opening of the hand that caused the rock to fall to the ground, not the other way around, just from the fact that causation can't flow backwards in time.
Anyways, I think you might be misunderstanding the level of proof that I'm talking about. There's a whole world between "we have no idea" and "we are absolutely certain" and it can take a long period of time to move from one to the other, and it is during that period that we are conducting science, not after it.
you would have so many other potential variables that without a control you would really struggle to assess much of anything
Can I ask whether you are aware of any such experiment being conducted to actually make that claim with any level of confidence? It sounds to me like you're saying no experiment can or ever has been conducted in macroeconomics. If so, I have to question on what basis you think we understand anything about it?
-
Speaker: Inequality: Too big to ignore, in reply to Matthew Poole,
How would you set up the necessarily-controlled macroeconomic environments in which to conduct such experiments? You need a country-sized (or at least one the size of a large city) economy with which to experiment, and another one to act as the control.
I don't know if that's true. You can experiment on a single subject without a control. My doctors have been doing it to me for years. Try a drug, observe effect. Try raising dose, observe effect, try dropping dose, observe effect. There's hundreds of variables they can't control, but over time, a pattern emerges, whether the drug has the desired effect, and what side effects it has. From the drug trials conducted on a massive scale they do know what to expect, but they still really can't be sure without just trying to see, what the effect will be on a particular subject. I buy in to the experiment because I'm even more interested in the answers than them, because it directly affects my wellbeing.
What I'm suggesting is that a social attitude about whether such controlled economic experimentation could be allowed might actually allow the society to genuinely discover what works for it. Since we're all subjects of such an experiment anyway, but just without any kind of methodology or consistency, or even plan, it doesn't seem like some kind of evil path to me. It seems like the difference between trusting medical science or runecasting and witchcraft.
-
Speaker: Inequality: Too big to ignore, in reply to Matthew Poole,
Just quickly, because I have to pop out for some unproductive behaviour, I don't claim to know the answer, hence the question. Try popping the other hat on for a sec and tell me how it could be made more scientific...I'd be interested to see what you can come up with.
Last ←Newer Page 1 … 190 191 192 193 194 … 1066 Older→ First