Random Drug-testing in Schools

RedJoker

cavedave wrote:

These tests could be used for some proper investigation but then you are not really relying on the random drug tests.

This is true. However, it would be foolish to solely rely on one test no matter how accurate.

It will mean that instead of investigating the entire school going population, you could just investigate a percentage of them. For example, students that tested positive could take a further test, this would have to be completely independent of the other one, any overlap could mean falsely identifying the same thing.

cavedave wrote:

My argument is with the Prosecutor's fallacy that random tests make likely. This fallacy is an important one as you could end up as the defendent at the wrong end of it and with a Jury who are innumerate.

This is obviously a big concern. Conditional probability and Bayes theorem aren't taught at leaving cert level. I doubt that all testers, principals, teachers and parents would have a lot of experience with them.

Also different areas will have different percentages of drug users which will change the percentage of false positives, the smaller the number of drug users, the more likely you are to get false positives. Rural vs. Urban may produce different percentages of false positives.

2Scoops

RedJoker wrote:

Dave,

There are two types of possible errors in testing, type I and type II errors. The test could produce a positive result when there was no drug use or could produce a negative result when there was drug use. Either error can be labeled type I or type II, however usually the more serious error would be labeled type I.

For example, when testing for a disease, the type I error would be giving a negative result when the disease was present.

Your values of 50% and 91% that you cited earlier in the thread are meaningless, the actual percentage is entirely dependent on the probability of type I errors, type II errors and the percentage of the sample tested that are taking drugs.

If you have statistics for these than it is possible to work it out using the following formula:

P(D|P) = P(P|D)P(D)/P(P)

Where P(D|P) is the probability of being on drugs given you tested positive, P(P|D) is the probability of testing positive given that you are on drugs, P(D) is the percentage of people who are on drugs and P(P) is the probability that a random person will test positive.

This is when we want to know what proportion will be falsely accused. We could also find out what proportion will be falsely cleared.

These tests could possibly be used as a screen for further investigation.

Speaking of errors...

A type I error is when a significant result is found when none really exists. In this case, it would be a false positive test result. A type II error is when a non-significant result is found when there really is one. In this case, a false negative. They are not interchangeable based on the perceived 'seriousness' of the error, only on the test design. There is also the rare type III error in directional hypothesis testing - fascinating stuff!

An interesting aside, type I is commonly considered to be more serious due to final stage medical research on drug efficacy. So saying some new drug works when it doesn't and marketing it it to patients would be a financial and ethical disaster. Saying a drug doesn't work when it really does and not marketing it is only an ethical disaster:)

Anyway, back on topic: what drugs do they intend to test for and what are the respective sensitivities and accuracies of the tests? If anyone can answer these then we can really get stuck into the mathematical argument.

cavedave

Anyway, back on topic: what drugs do they intend to test for and what are the respective sensitivities and accuracies of the tests? If anyone can answer these then we can really get stuck into the mathematical argument.

Good point unless we know this information no false positve rate can be established. The fact that none of these facts are mentioned does not inspire belief that the whole issue will be handled rationally though.

Executive Steve

Judt wrote:

People who take drugs buy drugs, paying money to drug dealers, who are murdering criminals. If criminals paid tax on their ill-gotten gains, would it be acceptable to you?

they are murdering criminals because they trade in commodities that are incredibly lucrative sources of income when all goes well, and are liable to get them a decade in prison (more than a child rapist) if things don't go well.

cocaine is ridiculously cheap to manufacture; the reason it costs so much is that to make it economical to import you have to handle quantities that will get you put away for a very long time.

given the fact that it's perishable, and that the normal legal saftety net that would apply to "legal" commodity traders, small businessmen, SME's etc don't really apply in the drug market the only way you're going to safeguard yourself from competition is aggressive protection.

granted, it certainly doesn't help that the sort of person who has a few kilos of coke stashed away and a sawnoff shotgun to defend it is probably dipping into his own supply, utterly paranoid at the thought of the cops, his competition, and god knows what else, but believe you me the stuff you read in the papers is pretty much the tip of the iceberg...

and lets face it, oppressive prohibition has been the cause of all of this.

if you are going to punished equally harshly whether you're dealing coke by weight or smoke by weight, you're going to trade where the money is...

it always astounds me that the politicians who claim to be free market economists and to have rational brains can't see this...

Judt

Ahem. The legalisation of drugs issue is a different one - as I said, I'd be open to considering it for marijuana, but harder drugs like cocaine? Well, you're pissing in other peoples pots there - Columbia and Afghanistan being two good example of countries that produce these drugs. Somewhere along the line, somebody gets murdered. But so long as it's "over there" I guess it's alright...

Some drugs are just too dangerous, to the person and the people in the way of the supply lines, to be made legal.

Executive Steve

why are the people over there getting murdered then?

noone gets murdered for exporting cotton, raisins, bananas...

Executive Steve

please note - i do think heroin and cocaine use are serious public health problems; i just believe that there would be an awful lot of money saved (as well as lives) by dealing with them as public health problems.