In the midst of an unprecedented epidemic of opioid overdose deaths in a country at nearly-unprecedented levels of economic inequality, what’s the best way to spend scarce taxpayer funds to fight addiction?

1. Put more money into evidence-based treatment that has been proven to cut the death rate in half.
2. Fund programs that provide naloxone—the antidote to overdose—to people who may be present when there is an overdose, and that have evidence suggesting reduced mortality.
3. Pay to drug test people who apply for food stamps—and to defend against inevitable lawsuits because this has previously been found to be unconstitutional.

If you chose 3, congratulations! You are actually Wisconsin Gov. Scott Walker! His plan, announced this month, involves screening single, non-parent food stamp applicants via questionnaires that essentially ask them to list the drugs they take—and then urine testing those who answer honestly. Wisconsinites would be better off burning their money than spending it on such failed approaches to addiction. Programs to drug-test poor people are not an effective way of diagnosing addiction, they certainly don’t prioritize those most in need of immediate care, and they don’t even save money.

To start, Florida’s program to drug test all welfare recipients not only failed to cut costs but didn’t pass constitutional muster: A federal appeals court ruled that it violated the Fourth Amendment because it tested people at random and the state couldn’t prove there was a “substantial need” to test this particular population. Four months of testing on 4,086 applicants yielded 108 positive tests—finding a prevalence of drug use that is far lower than found in the general population at a cost of over $100,000, plus an unknown and likely higher amount in legal fees.

Second, to try to get around that pesky constitutional requirement for evidence of at least suspicion of drug use or safety concerns, other states have chosen to use self-incriminating screening questionnaires, as Wisconsin now plans to do. These have resulted, unsurprisingly, in the detection of even fewer drug users at an even greater cost.

In six states that tested over 74,000 welfare applicants in 2014 at a cost of nearly $1 million, each had positive test results of less than 1 percent and most had rates under .01 percent. A seventh state, Arizona, which tested only 42 people out of 142,424 applicants, had a positive rate of 16 percent, but was clearly much more selective in choosing whom to test.

Marijuana is not only the most commonly used illegal drug but also the one most likely to show up in urine tests for technical reasons. Not surprisingly, most of those who tested positive were pot smokers, users of one of the least harmful substances.

Due to a quirk of physiology, people who smoke marijuana regularly can test positive for it when they are actually not at all impaired and it can stay in the system for at least a month, not hours or days like other drugs. Practically, this means that a positive marijuana test doesn’t mean someone is either a safety risk or addicted. In other words, it doesn’t provide any useful information.

Instead, what marijuana testing tends to do is drive people who want to smoke pot to take substitutes like synthetic cannabinoids— so-called “legal highs” like K2 or spice—that do not show up on tests for marijuana. These substances, which can be bought online or in “head shops” and even some convenience stores, are far more dangerous and impairing. A fatal overdose of marijuana has never been reported—but these synthetics can and have killed.

While some have been made illegal, synthetic cannabinoid products are frequently replaced with different and not-yet-banned chemicals, which haven’t even been tested on animals, let alone FDA approved for human use. Obviously, this is not going to help food stamp recipients: If they take these drugs to avoid positive marijuana tests, they pose greater risks to themselves and to potential employers.

So, basically, Walker’s plan is an expensive intelligence test for drug users. Many of the smartest will simply lie on the questionnaires. Those who are compulsively honest—often, people already following recommendations of recovery programs and therefore unlikely to need further help—will get sent for tests that show, unsurprisingly, they are already drug-free. Those who are both unwilling to lie and unwilling to give up drugs may use more dangerous substances, either synthetic cannabinoids or drugs like cocaine or opioids that will only show up if they’ve been used within a few days.

In the context of a situation where people who are desperate for treatment often cannot get it, where people who are addicted to opioids who do not get effective treatment are at extremely high risk of dying, spending money this way is not only stigmatizing and cruel, it is counterproductive. Spend the money to make more treatment available to those who are seeking it—and will therefore actually be the most likely to benefit. Spend the money on naloxone, which will save lives so that people can survive long enough to seek treatment. Spend the money on food stamps, so that hungry people can eat.

Or, if we’re really so desperate to drug test people who get government benefits, why not require testing for those who take the mortgage interest deduction? Alternatively, we could test those who want to avoid the estate tax—surely stoned or otherwise impaired rich people are a greater risk to the economy and they can also afford their own treatment, so that would save money too? And, how about testing members of Congress or state politicians like governors, whose compulsive return to bad policies despite negative consequences is clearly suggestive of addiction?

Of course, we won’t do this because the reason to drug test poor people is not to help reduce addiction or even to save money. The idea is to humiliate them into avoiding seeking the benefits to which they are entitled. It is the United States that should be ashamed, not those who need aid.