Every night, around the time the sun sets, my local television station runs an evening news broadcast. It’s been this way my entire life – a setting sun means that the evening news will be coming on shortly.

But what if we said this: Since the sun always sets at about the same time as the evening news starts, the good folks broadcasting the news at CBS, NBC, and ABC must cause it to set.

Of course, that’s a ridiculous claim to make; Brian Williams and Diane Sawyer don’t make the sun set any more than the crew at Good Morning America causes it to rise. But this example highlights an error that many people make when they confuse correlation with causation.

An unfortunate example of this error appeared recently in a magazine called The American Banker. Two sociology professors from George Washington University (GWU) took to the opinion page to promote a recent study which claims an increase in the concentration of payday lending stores causes an increase in violent crime.

The big problem with such a broad claim is that the study in question finds correlation between payday lenders and violent crime, not causation. This kind of academic overreach is not only irresponsible – it’s also dangerous. Strong conclusions based on faulty logic could cause policymakers to take rash actions with harmful unintended consequences for the poorest among us.

Let’s take a step back and look broadly at what correlation means. There are all sorts of phenomena that may be related to, but not caused by, each other.

For instance, a study might find that an increase in diet soda consumption is associated with weight gain. Since diet soda doesn’t have any calories, it’s unlikely that it caused the weight gain. Rather, people who are already overweight may be more likely choose diet soda over regular to avoid the additional calories. Perhaps they’re still gaining weight, but the soda has nothing to do with it.

To actually prove whether one event caused another, scientists perform a “controlled experiment.” In these experiments, scientists observe two groups of people that are similar in almost all ways except one- the product or behavior whose effects they’re trying to test. Prescription drug companies often do this, testing an active drug on one set of individuals and giving a sugar pill (or “placebo”) to the other set.

Despite these time-tested principles of determining cause and effect, some researchers and professors still refuse to call a spade a spade – they might detect correlation, but their research interests or other biases lead them to overreach and call it causation.

This brings us back to our professors from GWU. They’ve studied low-income neighborhoods in the Seattle area, and found that violent crime tends to increase after payday lending establishments set up shop – hence, their claim about the link between crime and payday lending. However, many of these poor neighborhoods studied were already characterized by violent crime before the payday lenders moved in, and were likely to see more of it in the future.

Without setting up a controlled experiment, where two similar groups are tested, the professors cannot positively claim that an increase in payday lending establishments led to an increase in violent crime. Following the professors’ flawed logic, you could look at any business concentrated in a low-income neighborhood – liquor stores, bail bond establishments – and claim that they also led to an increase in violent crime. But the world isn’t that simple, and we do a disservice to real social problems in low-income neighborhoods by pretending that it is.

The danger of faulty conclusions shrouded in academic language is that policymakers may take them as gospel. Across the country, states have chosen to place restrictions on payday lending establishments. Yet these lenders are often the only source of funds available for consumers who lack access to more traditional lines of credit. Given that, it’s not surprising that recent economic research found “borrowers in states that permit more payday lending are less likely to be denied credit generally and have lower delinquencies.”

What’s the bottom line here? Be careful when using statistical relationships alone to generalize how the world works – you may miss the forest for the trees. Wolf Blitzer probably doesn’t make the sun set each night, and payday lenders probably aren’t the reason that low-income communities struggle with higher rates of violent crime.