Last week, Third Way released a report stating that people who attend public four-year universities graduate at higher rates if they have a goldilocks level of debt—some, but not too much. At best, that finding is flawed and unhelpful. This and other studies that use loan debt as their unit of analysis are dangerous because policymakers can latch on and argue that “some debt is good, too much debt is bad,” when that claim is simplistic and completely unprovable with the available data.

First, the study’s findings. The author, Rachel Dwyer, a professor at Ohio State, found a relationship between graduation rates for students at four-year public colleges with some debt, versus those with none or a lot. It’s a inverted-U shaped curve, which Dwyer uses as evidence to suggest that students taking on some debt at public four-year universities graduate at a higher rate, but too much debt can be a bad thing. She also found that for those who attend private four-year institutions, the more debt they take on the more likely they are to graduate. Dwyer used data from the National Longitudinal Survey of Youth, which, until recently, annually tracked one cohort of students who were 12-16 years old in 1997, when the survey started.

The problem is that Dwyer starts with a lot of assumptions before asking her research question. She assumes that the amount of debt a student has is one of the primary factors determining whether a student completes. That assumption is hardly settled–there is very little research to suggest that debt is a primary driver of completion. Because she starts with a flawed question-are students more likely to graduate as they take on more debt?- she’s going to get a meaningless answer.

To Dwyer’s credit, she’s doesn’t make causal claims, and was careful to avoid any strong conclusions from the data. Third Way’s Senior Vice President for Policy Jim Kessler was not so careful, telling the Wall Street Journal that having some debt was an incentive for students to graduate.

Maybe. Maybe students with some debt work harder at school because they have “skin in the game,” as Kessler called it. But 18-year-olds graduate (or don’t) from high school with no debt, and no one seems to be arguing that those students would be more likely to graduate if they had some debt.

Far more likely, students who are motivated to graduate will do whatever it takes, which increasingly includes taking on educational debt. That means there is a relationship between debt and graduation, but we really have no idea what the exact nature of that relationship is, and it may not even be that important. For example, if I had data on how many times students brush their teeth per day and their graduation rate, I might conclude that students who brush their teeth more are more likely to graduate, but it’s unlikely anyone would take those findings seriously.

Too many people with debt fail to graduate. Over 20 percent of borrowers default, and of those defaulting, 97 percent never earned anything higher than a certificate degree. Many of those people aren’t even in the dataset that Dwyer used, because it only measured four-year institutions and many of defaulters never attend that type of institution. If we’re worried about defaults, it’s the wrong dataset. If we’re worried about graduation rates, it’s the wrong variable.

If students are going to take on debt, they need to graduate. But whatever other issues one has with debt, it is likely neither the inherent problem, nor the solution, to students dropping out of college.