"If in the control group at the 24th month, 5 rats would normally have gotten cancer, and 2 happened to get lucky, the paper largely becomes non-statistically significant."
Not quite. Looking at the numbers, it seems highly unlikely those results are statistically significant. If a couple of rats got lucky the paper would have shown no difference, or the opposite (that GMO is good for you!). Showing a difference and being statistically significant are not the same thing.
As far as I know there's only one GM crop being used that produces an insecticide, and that's one that's got a lot of evidence that it doesn't affect mammals at all. Someone posted elsewhere that it's approved for use on organic crops.
MOST GM crops that are being currently used, including the one under discussion, are modified to be resistant to externally applied insecticides and herbicides.
Um, Popper was a smart guy, but he's not an oracle of truth. Also, it's easy to quote him out of context to support an argument he didn't actually make.
It is most certainly possible (also advisable, and one of the basic tenants of science) to look at multiple studies to lend greater support to a particular hypothesis. In it's simplest form it's called "replication." It's also known as "meta-analysis" or "meta-studies," which can be fairly simple or can be fully quantitative and very rigorous.
Popper isn't exactly wrong, but he's talking about a particular type of philosophical argument. The upshot is that science (nor anything else) can ever "prove" anything useful. But it CAN provide probabilistic knowledge.
An experiment (i.e. an observation) can absolutely show that something doesn't cause significant harm (defined however you like) to such and such confidence. A proper meta-analysis of multiple studies can increase the confidence, decrease the threshold for the possible amount of harm or both.
Unless of course you're a follower of Hume and won't assume that past experience provides information about the future. In that case you might as well find a nice hole to squat in and smoke up.
A p-value that isn't "corrected" is a p-value for an experiment you didn't do. The p-value for the experiment you DID do IS enough. Taking only very slight liberties, it is the probability that a positive conclusion (there is a difference) is wrong.
Of course, if you do the wrong test, your p-value is invalid. Also if you don't "correct" it.
"But no, you don't withdraw published papers for bad science - you release another one proving the original was bad. (Unlike the original Lancet paper, which was discovered to be fraudulent which does demand removal)."
You absolutely do retract published papers for bad science. You don't retract them for incorrect conclusions, but you DO retract them for things like fraud, misrepresentation, unjustified conclusions, etc. I read the paper. It looks like these guys played some fancy analysis games to get some barely significant results regarding lots of things OTHER than tumors, then concluded that rats fed GMO corn got more tumors. That's bad science. I would never have accepted the original paper. If I were the EIC I'm not sure I would have forced it's retraction, but it's not an unreasonable move.
You had the most damning problem of all with your first paragraph. If it ain't got stats, it ain't science. The only science you can do (temporarily) without stats is theory, and that's only science if you're going to test it with experiment... using stats.
If they'd done proper stats it would have taken into account their plethora of experimental groups and ensured that they didn't get any positive results at all.
I didn't believe you that they hadn't done any stats so I looked up the paper. The only one I could find that fits is Food and Chemical Toxicology 50(11) pp. 4221-31. They DID do stats, but complicated unconventional ones that I find pretty suspicious (as in, a regular analysis didn't show anything so they kept trying until they found something that did). As far as I can tell, the analysis they actually did doesn't have much to do with the conclusion that GMO fed rats got more tumors.
You don't even have to do that if you drive. If you drive over they just ask some questions (you don't get out of the car) and scan your passport. Unless of course something goes wrong....
You've gone to great lengths to explain correlation != causation. Unfortunately, like so many people who go to great lengths to explain correlation != causation, you've completely missed the fact that there's no correlation in the first place.
Did I tell you about the time my friend and I got taken down in the customs line at gunpoint by every single customs agent at the crossing? It turns out they mistook my pasty Scottish self for some kind of international badass. A black international badass. They still had to ask the question though... "are you black?"
Wouldn't that be nice. The various US border agencies seem to operate under the very loosest of laws and most of the time seem to be making it up as they go along. Also, "it's the only government in the world I'd trust to obey the law?" Seriously? After all the treaty breaking, Geneva convention flaunting, populace monitoring and prisoner torturing? I guess if by "obeying the law" you mean abusing any loophole possible, redefining words to create loopholes where convenient ones didn't exist and secretly writing new laws if all else fails, then sure.
Um, no. Hospitals are generally government institutions. Each hospital has an administration, which is responsible to either a regional health authority or, in some provinces, they skip the region and have a provincial health authority. The provincial health authorities are responsible, via the minister of health, to the provincial government. Provinces are in charge of health care, but the federal government collects most of the taxes that pay for it, so if a province wants to do something health-wise the feds don't like, they hold up the money. That lets the federal government exercise general control over the kind of health care that's provided across the country. Special equalization payments also exist so that poor provinces can maintain the same standards as the rest of the country.
Clinics and non-hospital doctors' offices may or may not be privately owned.
Pharmacists are supposed to be more than pill counters. They're highly educated drug experts, far more so than physicians. The pharmacist is supposed to check what the physician prescribed, make sure it's correct, a reasonable dosage, and doesn't conflict with anything else the patient may have or be taking. It's very useful for the pharmacist to know the diagnosis, in order to do his job.
That's silly. Those supercomputers are general purpose machines with fast interconnects. If you need to do a hard computation, they're what you want. For doing something trivial like mining bitcoins you can use a much simpler system, which is consequently much faster and cheaper.
You don't think the NSA uses PCs to brute force RSA keys or analyze every phone call made by everyone everywhere, do you? If they seriously wanted to mine bitcoins they'd just set up a nice run of half a billion ASICS and laugh at your "bajilionty petaflops!!11."
Why they'd want to is far beyond me. Bitcoins are probably much easier for them to steal than to mine. And I'm sure they enjoy everyone broadcasting their transactions for easy capture and analysis. Kind of like credit card transactions without that pesky cash economy.
That's part of the problem. People are putting real value into the bitcoin market, value that the holders of old coins can come and claim whenever they feel like it. If any significant amount of old coins are cashed in, it's going to cause the mother of all crashes.
AFTER that happens, bitcoin might be a reasonably stable currency to do things with. It depends on whether the cashing in kills it or not.
Higher frequencies don't work very well through obstructions (these guys specify line-of-sight). Low frequencies though, go through walls better. Presumably you could record and compress audio, then retransmit it using lower frequency sound. The problem is, creating low frequency sound waves requires large speakers, and we hear fairly well down to quite low frequencies.
Modulating the frequency of a conventional light source is pretty difficult. You could use an LED and slightly manipulate the colour mix though.
Sorry, I'm not an extreme capitalist (nor an American, where lots of the extreme capitalists live). Business exists to serve the people, and the people have both the right and the responsibility to regulate it, particularly when, in the absence of that regulation, the behaviour of business becomes detrimental. Particularly when that business becomes large enough that it exerts significant control on the market.
Switzerland is on the right track.
And by the way, I own stock in several companies, and I vote it.
Outlook has to be the crappiest native mail client, but it's still a native app.
On the other hand, it could just be because Yahoo mail manages to be even worse. A guy at work today was just swearing at his yahoo mail giving another "server not available" error this morning. I'd ditch it too.
Which is a strong argument in favour of externally mandated checks and balances. The chief executives cannot be board members, the board must be elected by direct shareholder vote, executive salary caps, uniform company severance packages, etc.
When corporations were fairly small and their success or failure wasn't much more important than a spectator sport freewheeling corporate crap was fine. With companies that are so important that they are too big to fail, making or breaking entire national (or international) economies, that's not acceptable.
Monarchs are selected for their ability to take and hold power. What other criteria would you use? The ability to make the average person happy? That's called democracy.
Slashdot Mirror
"If in the control group at the 24th month, 5 rats would normally have gotten cancer, and 2 happened to get lucky, the paper largely becomes non-statistically significant."
Not quite. Looking at the numbers, it seems highly unlikely those results are statistically significant. If a couple of rats got lucky the paper would have shown no difference, or the opposite (that GMO is good for you!). Showing a difference and being statistically significant are not the same thing.
As far as I know there's only one GM crop being used that produces an insecticide, and that's one that's got a lot of evidence that it doesn't affect mammals at all. Someone posted elsewhere that it's approved for use on organic crops.
MOST GM crops that are being currently used, including the one under discussion, are modified to be resistant to externally applied insecticides and herbicides.
Um, Popper was a smart guy, but he's not an oracle of truth. Also, it's easy to quote him out of context to support an argument he didn't actually make.
It is most certainly possible (also advisable, and one of the basic tenants of science) to look at multiple studies to lend greater support to a particular hypothesis. In it's simplest form it's called "replication." It's also known as "meta-analysis" or "meta-studies," which can be fairly simple or can be fully quantitative and very rigorous.
Popper isn't exactly wrong, but he's talking about a particular type of philosophical argument. The upshot is that science (nor anything else) can ever "prove" anything useful. But it CAN provide probabilistic knowledge.
An experiment (i.e. an observation) can absolutely show that something doesn't cause significant harm (defined however you like) to such and such confidence. A proper meta-analysis of multiple studies can increase the confidence, decrease the threshold for the possible amount of harm or both.
Unless of course you're a follower of Hume and won't assume that past experience provides information about the future. In that case you might as well find a nice hole to squat in and smoke up.
You're right. Most of them are on your payroll. They're publicly funded.
You're also right, it's not "science" in the original sense, when most scientists were funded privately by wealthy patrons.
A p-value that isn't "corrected" is a p-value for an experiment you didn't do. The p-value for the experiment you DID do IS enough. Taking only very slight liberties, it is the probability that a positive conclusion (there is a difference) is wrong.
Of course, if you do the wrong test, your p-value is invalid. Also if you don't "correct" it.
"But no, you don't withdraw published papers for bad science - you release another one proving the original was bad. (Unlike the original Lancet paper, which was discovered to be fraudulent which does demand removal)."
You absolutely do retract published papers for bad science. You don't retract them for incorrect conclusions, but you DO retract them for things like fraud, misrepresentation, unjustified conclusions, etc. I read the paper. It looks like these guys played some fancy analysis games to get some barely significant results regarding lots of things OTHER than tumors, then concluded that rats fed GMO corn got more tumors. That's bad science. I would never have accepted the original paper. If I were the EIC I'm not sure I would have forced it's retraction, but it's not an unreasonable move.
You had the most damning problem of all with your first paragraph. If it ain't got stats, it ain't science. The only science you can do (temporarily) without stats is theory, and that's only science if you're going to test it with experiment... using stats.
If they'd done proper stats it would have taken into account their plethora of experimental groups and ensured that they didn't get any positive results at all.
I didn't believe you that they hadn't done any stats so I looked up the paper. The only one I could find that fits is Food and Chemical Toxicology 50(11) pp. 4221-31. They DID do stats, but complicated unconventional ones that I find pretty suspicious (as in, a regular analysis didn't show anything so they kept trying until they found something that did). As far as I can tell, the analysis they actually did doesn't have much to do with the conclusion that GMO fed rats got more tumors.
I don't get this. Your country has fair elections. Your governments, so far, haven't departed from your constitution that much.
If you don't like the candidates, run yourself. Or start a party and do it seriously.
Your problems aren't due to lack of a constitution or elections or tyranny. They're do to the vast majority of you just not giving a shit.
You don't even have to do that if you drive. If you drive over they just ask some questions (you don't get out of the car) and scan your passport. Unless of course something goes wrong....
You've gone to great lengths to explain correlation != causation. Unfortunately, like so many people who go to great lengths to explain correlation != causation, you've completely missed the fact that there's no correlation in the first place.
I believe all those people also had a Y chromosome instead of the normal double X.
Did I tell you about the time my friend and I got taken down in the customs line at gunpoint by every single customs agent at the crossing? It turns out they mistook my pasty Scottish self for some kind of international badass. A black international badass. They still had to ask the question though... "are you black?"
They did let me in eventually.
Ah, but you can't, because they're secret.
Wouldn't that be nice. The various US border agencies seem to operate under the very loosest of laws and most of the time seem to be making it up as they go along. Also, "it's the only government in the world I'd trust to obey the law?" Seriously? After all the treaty breaking, Geneva convention flaunting, populace monitoring and prisoner torturing? I guess if by "obeying the law" you mean abusing any loophole possible, redefining words to create loopholes where convenient ones didn't exist and secretly writing new laws if all else fails, then sure.
Um, no. Hospitals are generally government institutions. Each hospital has an administration, which is responsible to either a regional health authority or, in some provinces, they skip the region and have a provincial health authority. The provincial health authorities are responsible, via the minister of health, to the provincial government. Provinces are in charge of health care, but the federal government collects most of the taxes that pay for it, so if a province wants to do something health-wise the feds don't like, they hold up the money. That lets the federal government exercise general control over the kind of health care that's provided across the country. Special equalization payments also exist so that poor provinces can maintain the same standards as the rest of the country.
Clinics and non-hospital doctors' offices may or may not be privately owned.
Pharmacists are supposed to be more than pill counters. They're highly educated drug experts, far more so than physicians. The pharmacist is supposed to check what the physician prescribed, make sure it's correct, a reasonable dosage, and doesn't conflict with anything else the patient may have or be taking. It's very useful for the pharmacist to know the diagnosis, in order to do his job.
Put the correct way, that's not a correlation.
That's silly. Those supercomputers are general purpose machines with fast interconnects. If you need to do a hard computation, they're what you want. For doing something trivial like mining bitcoins you can use a much simpler system, which is consequently much faster and cheaper.
You don't think the NSA uses PCs to brute force RSA keys or analyze every phone call made by everyone everywhere, do you? If they seriously wanted to mine bitcoins they'd just set up a nice run of half a billion ASICS and laugh at your "bajilionty petaflops!!11."
Why they'd want to is far beyond me. Bitcoins are probably much easier for them to steal than to mine. And I'm sure they enjoy everyone broadcasting their transactions for easy capture and analysis. Kind of like credit card transactions without that pesky cash economy.
That's part of the problem. People are putting real value into the bitcoin market, value that the holders of old coins can come and claim whenever they feel like it. If any significant amount of old coins are cashed in, it's going to cause the mother of all crashes.
AFTER that happens, bitcoin might be a reasonably stable currency to do things with. It depends on whether the cashing in kills it or not.
Higher frequencies don't work very well through obstructions (these guys specify line-of-sight). Low frequencies though, go through walls better. Presumably you could record and compress audio, then retransmit it using lower frequency sound. The problem is, creating low frequency sound waves requires large speakers, and we hear fairly well down to quite low frequencies.
Modulating the frequency of a conventional light source is pretty difficult. You could use an LED and slightly manipulate the colour mix though.
Sorry, I'm not an extreme capitalist (nor an American, where lots of the extreme capitalists live). Business exists to serve the people, and the people have both the right and the responsibility to regulate it, particularly when, in the absence of that regulation, the behaviour of business becomes detrimental. Particularly when that business becomes large enough that it exerts significant control on the market.
Switzerland is on the right track.
And by the way, I own stock in several companies, and I vote it.
It's a native app instead of a shitty web app?
Outlook has to be the crappiest native mail client, but it's still a native app.
On the other hand, it could just be because Yahoo mail manages to be even worse. A guy at work today was just swearing at his yahoo mail giving another "server not available" error this morning. I'd ditch it too.
Skype works extremely poorly on an air gapped machine.
Which is a strong argument in favour of externally mandated checks and balances. The chief executives cannot be board members, the board must be elected by direct shareholder vote, executive salary caps, uniform company severance packages, etc.
When corporations were fairly small and their success or failure wasn't much more important than a spectator sport freewheeling corporate crap was fine. With companies that are so important that they are too big to fail, making or breaking entire national (or international) economies, that's not acceptable.
Monarchs are selected for their ability to take and hold power. What other criteria would you use? The ability to make the average person happy? That's called democracy.