Suppose someone believes something wrong ( imagine, for example, that person believes in learning styles ). You explain why it is incorrect. What happens next? Will that person adjust, or will they still hold on to the mistaken idea?

A large study from 2023 attempted to give an average answer to that. Their conclusion? “On average, attempts to correct misinformation have little effect.” A little discouraging, right?

But in a new blog post on Datacolada , Joe Simmons explains in detail why that conclusion is actually entirely off the mark. You can’t and shouldn’t compare apples and oranges.

What turned out? The researchers threw two very different things together:

1. How many people changed their minds after a correction
2. How much the wrong belief stuck

It’s like taking the average of how fast people slow down in a corner and how hard they accelerate afterwards. You might get zero, but that doesn’t mean nothing happened.

Thanks to open data, which is a good thing, to be clear, Simmons was able to get started himself. What did he think?

• People did adapt after a correction (d = 0.48, which is a strong effect)
• At the same time, some of the misinformation remained (d = -0.39)
• But if you simply add the two together, you get a weak average effect (d = 0.11)

And that’s exactly what the original study did.

So in summary:

1. Averages can be misleading. Especially when you combine things that don’t belong together.
2. Corrections do work. People learn, just not perfectly.
3. Open data makes these kinds of insights possible. Without that transparency, no one would have been able to expose this fallacy.

When someone claims that correcting misinformation doesn’t work “on average,” it’s good to ask: what exactly are you talking about? Because an average that combines opposite effects sometimes doesn’t tell you anything at all. ( or sometimes people even combine things that have nothing to do with each other ).

Or as Simmons nicely sums it up: “It would be crazy if both corrections and misinformation had no effect whatsoever – and that is exactly what the average shows.”