Drug Testing
The other day the Minotaur had a pre-employment drug test. These tests have become a standard for all sorts of employment. The fact that such an invasive process has become a standard in the US is appaling. This is all the more true because they are of dubious value. The doubtful value of drug testing stems from two distinct problems with the drug testing system.
The first problem is the basis for testing in the first place: that somehow drug use has an adverse impact on job performance such that employers have a right to know about usage history prior to employment.
First off, let's eliminate the idea that a history of illegal drug use is an indicator for disrespect for the law that an employer might wish to consider. The testing does nothing to show illegal usage of drugs, only that drugs which are currently illegal in the US have been used. The difference is that the test totally ignores the idea of use of drugs in a legal setting outside of the US which if done inside the US would be illegal: that is, it says nothing about a man's respect for the legality of drug "X" in the US if he uses drug "X" while in Mexico. While it could be argued that an educated person should know of the applicability of drug laws to US persons outside the US, such laws of misfeasance are sufficiently esoteric that we cannot conclude disrespect for the law based on ignorance of it. By way of illustration, how many people know whether or not it is legal to use, while on vacation in Canada, an over-the-counter cough medication purchased in Canada and which contains large amounts of codeine? This isn't the kind of knowledge we can attribute to the reasonable layperson.
Given that, there is the real issue of whether illegal drug use actually impairs performance (Some groups have suggested that some illegal drug use actually improves performance under some conditions. http://www.druglibrary.org/schaffer/lsd/weil1.htm ). If it does lead to detectable impairment of performance, then the performance impairment is detectable in and of itself, and there is no need to have drug use act as an superfluous and imperfect test for something we can observe. On the other hand, if the drug use does not lead to detectable differences in performance, then drug use does not correlate to poor performance and the test is pointless. I would add here that there are some jobs where the performance is sufficiently critical that, by the time poor performance is detected, irreparable harm is done: airline pilots, bus drivers, etc. For these individuals, where malpractice is "life or death", I can see finding proxies like drug use history to predict performance. On the other hand, in the vast majority of jobs, service industry, sales, many kinds of engineering, performance which is impaired is easily detectable prior to any significant harm being done to anyone. In this latter case, why bother testing? Just look at performance: if it is good the first 90 days, keep the employee, if not dismiss them… As such, since the vast majority of jobs allow direct detection in a timely manner, widespread drug use history testing is a pointless, expensive, and invasive exercise which should be discontinued.
A second, and arguably far more important, problem with drug use history testing is the "false positive paradox". If it is true that illegal drug usage is an asocial aberration, then it should follow that it is rare. Unfortunately, testing for rare conditions, even with very accurate tests, leads to very unreliable "positives". To illustrate, imagine an employer with 1000 employees who decides to put them through a drug test. Our employer selects a very good test which is accurate 99% of the time. That is, 99% of the negatives are shown as negative with 1% erroneously positive and 99% of positives are shown as positive with 1% erroneously negative. Reflecting the expected rare nature of illegal drug usage, we will assume that only 1% of the employees are drug users: 10 use drugs. What results should we expect if we neglect statistical variation? We should find that, of the 990 non-drug users, 980 will test negative correctly and 10 will test positive incorrectly. Of the ten drug users, we expect all ten to come back positive. Thus, we have 20 positives as a result and ONLY HALF are correct!! This is NOT a very far-fetched scenario. It illustrates the problem with testing for a rare condition: the positives are VERY often incorrect. The correct way to address this problem is to always retest the positives: two positives in a row would be very reliable indicator. Unfortunately, that isn't how the test is applied in employment testing: if you fail once you're done- you don't get the job. As such, many qualified and clean people are being turned away to the detriment of the employers and the employees.
Drug testing is not only pointless, expensive, and invasive. It is also being applied incorrectly with inefficient results. We would all be better off if the majority of employers began recognizing the low cost-benefit provided by this hack system of drug testing and move away from it.
The first problem is the basis for testing in the first place: that somehow drug use has an adverse impact on job performance such that employers have a right to know about usage history prior to employment.
First off, let's eliminate the idea that a history of illegal drug use is an indicator for disrespect for the law that an employer might wish to consider. The testing does nothing to show illegal usage of drugs, only that drugs which are currently illegal in the US have been used. The difference is that the test totally ignores the idea of use of drugs in a legal setting outside of the US which if done inside the US would be illegal: that is, it says nothing about a man's respect for the legality of drug "X" in the US if he uses drug "X" while in Mexico. While it could be argued that an educated person should know of the applicability of drug laws to US persons outside the US, such laws of misfeasance are sufficiently esoteric that we cannot conclude disrespect for the law based on ignorance of it. By way of illustration, how many people know whether or not it is legal to use, while on vacation in Canada, an over-the-counter cough medication purchased in Canada and which contains large amounts of codeine? This isn't the kind of knowledge we can attribute to the reasonable layperson.
Given that, there is the real issue of whether illegal drug use actually impairs performance (Some groups have suggested that some illegal drug use actually improves performance under some conditions. http://www.druglibrary.org/schaffer/lsd/weil1.htm ). If it does lead to detectable impairment of performance, then the performance impairment is detectable in and of itself, and there is no need to have drug use act as an superfluous and imperfect test for something we can observe. On the other hand, if the drug use does not lead to detectable differences in performance, then drug use does not correlate to poor performance and the test is pointless. I would add here that there are some jobs where the performance is sufficiently critical that, by the time poor performance is detected, irreparable harm is done: airline pilots, bus drivers, etc. For these individuals, where malpractice is "life or death", I can see finding proxies like drug use history to predict performance. On the other hand, in the vast majority of jobs, service industry, sales, many kinds of engineering, performance which is impaired is easily detectable prior to any significant harm being done to anyone. In this latter case, why bother testing? Just look at performance: if it is good the first 90 days, keep the employee, if not dismiss them… As such, since the vast majority of jobs allow direct detection in a timely manner, widespread drug use history testing is a pointless, expensive, and invasive exercise which should be discontinued.
A second, and arguably far more important, problem with drug use history testing is the "false positive paradox". If it is true that illegal drug usage is an asocial aberration, then it should follow that it is rare. Unfortunately, testing for rare conditions, even with very accurate tests, leads to very unreliable "positives". To illustrate, imagine an employer with 1000 employees who decides to put them through a drug test. Our employer selects a very good test which is accurate 99% of the time. That is, 99% of the negatives are shown as negative with 1% erroneously positive and 99% of positives are shown as positive with 1% erroneously negative. Reflecting the expected rare nature of illegal drug usage, we will assume that only 1% of the employees are drug users: 10 use drugs. What results should we expect if we neglect statistical variation? We should find that, of the 990 non-drug users, 980 will test negative correctly and 10 will test positive incorrectly. Of the ten drug users, we expect all ten to come back positive. Thus, we have 20 positives as a result and ONLY HALF are correct!! This is NOT a very far-fetched scenario. It illustrates the problem with testing for a rare condition: the positives are VERY often incorrect. The correct way to address this problem is to always retest the positives: two positives in a row would be very reliable indicator. Unfortunately, that isn't how the test is applied in employment testing: if you fail once you're done- you don't get the job. As such, many qualified and clean people are being turned away to the detriment of the employers and the employees.
Drug testing is not only pointless, expensive, and invasive. It is also being applied incorrectly with inefficient results. We would all be better off if the majority of employers began recognizing the low cost-benefit provided by this hack system of drug testing and move away from it.
posted by Minotaur at 2:32 PM
0 Comments:
Post a Comment
<< Home