Last week in this space, Naomi Schaefer Riley laid out the experience of the Seneca Nation of Indians with a program very similar to a “Universal Basic Income”—a policy idea that has gained currency among both liberals and libertarians lately. Like many Native American tribes, the Seneca Nation distributes profits from its gambling operations to its members on a per-capita basis. Here is how the program works, per Riley:

The current annuity for an adult between the ages of 18 and 60 on the reservation is about $8,000, disbursed in quarterly payments. Elders get a larger amount, in a kind of American Indian social security program. Half the money for children under 18 is given to their parents, and the other half is put into a trust. When a Seneca youth turns 18 and can show that he or she has graduated from high school or earned a GED, he or she receives a lump sum of $30,000. Those who don’t get a high-school degree have to wait until they’re 21 to receive the money. With each passing year, the annuities get larger, because the tribe invests its earnings.

The results have not been obviously positive. Riley’s reporting reveals that, upon receiving these payments, “the best thing most young adults do … is to buy a new truck. These are kids who have never had very much before; so when someone hands them a huge check, they clearly don’t know what to do. Store owners report that young people will come in to buy candy, handing $50 or $100 without expecting any change. These young people seem to have no concept of saving or investing.”

Unfortunately, however, it is rather difficult to look at the Seneca Nation through a more statistical lens, comparing the outcomes of its members with nationwide trends. The tribe has about 8,000 registered members, fewer than one-quarter of whom live on its two reservations, so its coverage in annual Census surveys is rather scant. The poverty rates Riley cites for the reservations have wide margins of error despite being based on five years’ worth of Census data (an incredible ±30.7 points for the smaller one), and the Cattaraugus County poverty rate, which she also cites, is of limited usefulness because the county is just 3 percent Native American despite the presence of the reservations.

The program has hardly eliminated poverty, in other words, but the data are far too noisy to say much more than that. I would also note that the official poverty measure is highly flawed and that poverty is not the only measure of whether a social program is working. It may be better to look elsewhere for insights into the Universal Basic Income.

Three key sources are data on lottery winners, “income maintenance” experiments conducted in and around the 1970s, and other research on Native American tribes with per-capita payments. They teach three main lessons: 1) Large, unconditional payments—possibly including the $8,000 annual sums the Seneca pays out—can undeniably be destructive; 2) even smaller payments may reduce the work effort of those who receive them; but 3) there is also evidence that these programs improve the well-being of poor families and the children in them.

Lottery winners are a group Riley mentions, and they have been much discussed elsewhere, so I won’t belabor the point. But research shows that lottery winners are at a substantially elevated risk of going bankrupt. Simply put, enormous amounts of money do little good for people who don’t know how to manage such sums. Winners also become unmoored from the routines they were accustomed to; some commit suicide or fall into drug addiction.

What about the effect of more down-to-earth sums? That’s where evidence from a handful of 1970s-era experiments from the U.S. and Canada comes in. Over and over, these experiments revealed that giving people money reduced the effort they put into work, though they rarely withdrew from the workforce entirely.

Indeed, some liberal supporters of the UBI have touted this feature of the idea, saying it would force employers to provide jobs good enough to compete with a life on the dole. Other supporters, taking a longer view, argue that the UBI might be needed in a future world where technology does most of the work and there aren’t enough jobs for everyone. Personally, I’m far too conservative to take the first point seriously and think the second one is premature at this point.

But here’s the complicating factor: there was also some (murky) evidence from those experiments that giving money to families, especially very poor ones, could improve child outcomes, such as educational attainment. And newer studies, including some of Native American gambling revenue programs (spotlighted by Shawn Fremstad and Erik Stegman over at TalkPoverty in 2015), show similar findings.

One study is particularly comparable to the situation Riley writes about—with the important difference that the program paid only about half as much. The study treats the introduction of the program as a “quasi-experiment,” with families that didn’t receive the extra money as the control group. When their families got the payments, the poorest children got an extra year of school by age 21 and became 22 percent less likely to have committed minor crimes by the time they were 16–17 years old. (Interestingly, unlike the earlier studies, this one did not find a reduction in working time.)

In sum, a Universal Basic Income could be profoundly harmful if it is too big, and it could reduce work effort to some extent no matter its size. But the evidence also suggests it could help children, especially those in the poorest families.

Robert VerBruggen is managing editor of The American Conservative.