OMNILIBRIUM
  Rational Discussion of Controversial Topics


GO TO THE MAIN THREAD Sort By:


melian 10 June 2015 03:33 PM
71%

Causality direction is indeed debatable. But while correlation does not imply causation, but the absence of correlation does imply the absence of causation.

I think, as a general rule, the burden of proof should be on those who want to create a new legal restriction rather than on those who oppose it. So if minimum wage laws are not correlated with any real positive effect, there is no reasonable justification for them.


stars0
Reply


pianoforte611 1 July 2015 10:13 AM
65%

but the absence of correlation does imply the absence of causation.


This is incorrect, there are causative relationships with zero correlation for many reasons (confounders, non-linearity, and a few others)

stars0
Reply


melian 1 July 2015 10:56 AM
70%

Can you give an example? I can see how confounding variables could make a correlation very weak, but zero?

stars0
Reply


pianoforte611 1 July 2015 07:27 PM
67%

An artificial example would be Z = X/Y where Corr(X,Y)=1

This correlation could be spurious or only true of some samples. Of course in reality confounding tends only result in this sort of problem when combined with non-linearity, lack of statistical power or imprecision in measurement. Lack of correlation is certainly evidence for lack of causation, but it does not imply it.

Regarding non-linearity, there are many examples:

http://math.stackexchange.com/questions/444408/zero-correlation-does-not-imply-independence

stars0
Reply


melian 2 July 2015 07:27 AM
72%

Thanks, that was interesting.

Btw, I think in most cases with non-linear dependence you would generally find a non-zero correlation between X^2 and Y^2 even when Corr(X,Y)=0.


stars0
Reply


ChristianKl 11 June 2015 04:00 AM
62%

Causality direction is indeed debatable. But while correlation does not imply causation, but the absence of correlation does imply the absence of causation.
In this case the data set you refer to is not strong enough for such a claim.
Countries don't randomly adopt minimum wage laws but do so due to political factors.

So if minimum wage laws are not correlated with any real positive effect, there is no reasonable justification for them.

The main positive effect that people want to create with minimum wage laws isn't increased GDP. It's that certain people are payed more money.


stars0
Reply


DanielLC 10 June 2015 10:41 PM
60%

but the absence of correlation does imply the absence of causation.


Didn't you say there was a correlation?

By contrast, countries with the highest minimum wages (relative to the GDP per capita) are among the world poorest.


It's unlikely to get no correlation when there's a causal structure involved unless it involves something correcting itself, but you can still get a negative correlation if there would otherwise be an even stronger negative correlation.

I think, as a general rule, the burden of proof should be on those who want to create a new legal restriction rather than on those who oppose it.


According to the Department of Labor:

A review of 64 studies on minimum wage increases found no discernable effect on employment. Additionally, more than 600 economists, seven of them Nobel Prize winners in economics, have signed onto a letter in support of raising the minimum wage to $10.10 by 2016.


Honestly, I'm skeptical of this too, but it's one thing to say they have the burden of finding evidence. It's another to say they have the burden of changing your mind.

stars0
Reply


melian 11 June 2015 12:14 PM
70%

Didn't you say there was a correlation?

No, just that there was no obvious positive correlation and many counter-examples.

it's one thing to say they have the burden of finding evidence. It's another to say they have the burden of changing your mind.

I would not put much weight on a petition by 600 left-wing economists. It would be easy to find 600 right-wing economists who believe the opposite (though it would definitely be a lot harder to make them sign a petition in support of lowering the minimum wage).

I would also take with a grain of salt any claims from the website that belongs to the Department of Labor (whose employees might lose their jobs if they contradict the official line of the White House).

Given the political biases in this question, the only evidence I would seriously consider is simple and easily verifiable statistics (something like in this example). Is it really an excessive requirement?
you can still get a negative correlation if there would otherwise be an even stronger negative correlation.

Sure. For example, taking a certain drug can be negatively correlated with being healthy because it is taken only by the sick people. But, in medicine, the drugs are tested by comparing two groups - sick people who get it and sick people who don't. Similarly, we might compare the effect of minimum wage laws on otherwise similar countries. If minimum wage laws work, we should see more poor people in countries without them.


stars0
Reply


DanielLC 11 June 2015 11:03 PM
60%

Given the political biases in this question, the only evidence I would seriously consider is simple and easily verifiable statistics (something like in this example). Is it really an excessive requirement?


If all you need is a correlation, then I'm sure there are plenty of people who can provide you with one that shows that minimum wage is great.

But, in medicine, the drugs are tested by comparing two groups - sick people who get it and sick people who don't.


It's by testing two randomly selected groups. If you just look at people who choose to take it or not take it, you don't get very useful results. You end up finding out things like that power lines cause cancer.

stars0
Reply


melian 12 June 2015 03:12 PM
70%

If all you need is a correlation, then I'm sure there are plenty of people who can provide you with one that shows that minimum wage is great.

Have you actually seen a study like that which made sense to you?
It's by testing two randomly selected groups.

It would be better if we could make a controlled experiment on a large group of randomly selected countries. But a non-random selection from a limited set is still better than purely theoretical speculations.


stars0
Reply


DanielLC 13 June 2015 01:25 AM
62%

Have you actually seen a study like that which made sense to you?


I haven't actually looked at any studies on the matter. I just really don't trust correlation. I suppose this is the part in the argument where I'm supposed to not be lazy and actually hunt down studies to this effect.

But a non-random selection from a limited set is still better than purely theoretical speculations.


I disagree. If theory went one way and correlation went the other, I'd trust the theory. If I seriously looked for confounding variables and couldn't find any, then I'd start considering the study, but I don't think this is the case here.


stars0
Reply


melian 18 June 2015 04:50 AM
72%

I agree that correlation is not proof, but I would also be wary about trusting any theory that does not have supporting empirical evidence. In principle, it may be possible to arrive at the right answer by logical reasoning alone, but in practice this almost never happens. With may be one exception I donít recall any correct scientific theories that were based on pure logical reasoning, rather than on the experimental results pointing out errors in the previous theory. In politics, where our emotional biases are far stronger, pure logic is even less reliable.

stars0
Reply


DanielLC 19 June 2015 12:50 AM
64%

but I would also be wary about trusting any theory that does not have supporting empirical evidence.


If there's no empirical evidence of any real use in either direction, do you trust the theory that minimum wage causes unemployment, or do you trust the theory that it does not?

stars0
Reply


melian 20 June 2015 06:48 AM
72%

In the absence of a direct evidence, Iíd lean towards the theory that minimum wage does cause unemployment. But my belief would still be largely based on empirical data Ė almost in every other case raising prices results in reduced demand.

stars0
Reply


Fwiffo 17 July 2015 08:00 AM
56%

But is it not possible that other position is also supported by general principles that are also backed by empirical data? One could for example argue that usually political decisions are made for the public good and require extensive agreement to pass so that usage of minimum wage is likely to be effective or otherwise it would have fallen out of favour way sooner. That is you say it is an issue of "price lowering" but one could argue that it is an issue of "political consensus". If we regard both "price lowering" and "political consensus" to be backed up by empirical results both options are still emprically backed up (even if it is not direct). You would have to argue why the way you phrase the question is the more appropriate way of framing it.

It is more easy to be aware what indirect empirical evidence one has for ones own world view but that doesn't mean that opposing world views lack those too. Assuming that things that contradict with your world view must lack that basis if the relevant evidence is not presented in the same go means there is a presumption of arbitrariness until proven founded when dealing with the beliefs of others that is not in effect when dealing with your own thoughts.

stars0
Reply


melian 17 July 2015 09:27 AM
72%

political decisions are made for the public good

To use the popularity of the minimum wage as evidence of its effectiveness, you would need first to show that the above premise is correct in a very large majority of cases (as with price-demand relationship). Iím not convinced that itís true even in 50% of the cases.


stars0
Reply


Fwiffo 18 July 2015 07:42 AM
56%

That would be direct evidence. It is also somehwat odd of a stance that most political decisions are harmful. Wouldn't that pretty easily lead into conclusion that we ought to stop doing decisions as it has expected negative impact while doing nothing has no impact.

The indirect case would be that for example the method of taxing things that are problematic has been working. Tobacco troublesome? Tax it. Too many liqour related deaths? Tax it. Usually increasing the related tax makes people do less of the activity but usually you end up increasing the tax return ie the people don't decrease their activities in the same ratio as it is taxed. A minimum wage is pretty similar to making a tax for low wage employemnt and making a benefit program for low wage earners.

I thought that you were not implicitly critising the alternative as coming from fields that do not base their principles on empirics. I guess you might have been aware of this facet and even meant it. That is if you are allowed the benefit of indirect empricial evidence so should the other sides too. I thought that the difference on the relative strength of the indrect empirical stregnth was more even.

stars0
Reply


melian 18 July 2015 01:17 PM
70%

Wouldn't that pretty easily lead into conclusion that we ought to stop doing decisions as it has expected negative impact while doing nothing has no impact.

Having zero laws and regulations may be worse than having a million of them. But that does not mean that 600 thousands laws out of that million cannot be harmful.
I thought that the difference on the relative strength of the indrect empirical stregnth was more even.

Letís do a quantitative comparison. Iím aware of only two exceptions to the principle that raising prices decreases demand. One is when most buyers have no information about the product quality and the seller dupes them into thinking that higher price indicates higher quality. The second exception arose during very peculiar circumstances at the time of the Great Irish famine. Neither exception I believe is really relevant to the MW.

I understand you believe that most political decisions are beneficial, but do you also think that exceptions to this rule can be counted in single digits?
The indirect case would be that for example the method of taxing things that are problematic has been working.

Can't the same argument be made for taxing bad weather or people with cancer?
A minimum wage is pretty similar to making a tax for low wage employment

It is indeed similar and is likely to help eliminate low wage employment. The problem is that it might do so by making many low wage employees unemployed.


stars0
Reply


Fwiffo 13 June 2015 10:42 AM
54%

You are arguing that tampered evidence is better than lack of evidence.

stars0
Reply