Adam Kucharski

And ultimately, it comes down to this issue that for a lot of problems, we can't necessarily intervene and randomise.

but there might be a situation that's done it to some extent for us so the classic example is the Vietnam draft where it was kind of random birthdays were drawn out of lottery and so there's been a lot of studies subsequently about the effect of serving in the military on different you know subsequent lifetime outcomes because broadly those people have been randomized you know it was for a different reason but you've got that element of randomization driving that and so again you know with um

some of the recent singles data and other studies, you might have a situation, for example, where, you know, there's been an intervention that's somewhat arbitrary in terms of time, like, you know, it's a cutoff on a birth date, for example.

And under certain assumptions, you could think, well, actually, there's no real reason for the person on this day and this day to be fundamentally different.

I mean, perhaps there might be, you know, effects of kind of cohorts if it's school years or this sort of thing.

But generally, this isn't the same as having people who are very, very different ages and very different characteristics.

It's just,

Nature, or in this case, just a policy intervention for a different reason, has given you that randomization, which allows you, or kind of pseudo-randomization, which allows you to then look at something about the effects of an intervention that you wouldn't as reliably if you were just kind of digging into the data of kind of yes, no, who's received a vaccine.

Yeah, I think this is an issue that I think a lot of people get drilled into in their training.

You know, just because there's a correlation between things doesn't mean that that thing causes this thing.

But it really struck me as I talked to people researching the book, how in practice in research, there's actually a bit more to it and how it's played out.

So first of all, if there's a correlation between things, it doesn't tell you much generally that's useful for intervention.

You know, if two things are correlated, it doesn't mean that changing that thing is going to have an effect on that thing because there might be something that's influencing both of them, you know.

If you have more ice cream sales, it will lead to more heat stroke cases.

It doesn't mean that changing ice cream sales is going to have that effect.

But it does allow you to make predictions, potentially.

Because if you can identify consistent patterns, you can say, OK, if this thing's going up, I'm going to make a prediction that this thing's going up.

So one thing I found quite striking actually talking to research in different fields is how many fields choose to focus on prediction because it kind of avoids having to deal with this cause and effect problem.

And even in fields like psychology, it was kind of interesting that

There's a lot of focus on predicting things like relationship outcomes.