When making a claim that something exists, it's enough to find just one example of it. I think there are enough anecdotes to say that moral licensing definitely exists.
That said, I agree the research on it is low-quality and that it's probably not so widespread that it's something to worry about when you're doing good.
Can you add some description to help me interpret the funnel plot? It's not obvious to me what's happening or why that means what it sounds like you're saying it means
Imagine a bunch of people run studies on moral licensing with a wide-range of sample sizes. Studies with smaller samples sizes will have higher standard errors and so be less precise. But all studies should still be clustered around the real effect size. In an ideal world, the correlation between effect and sample size should be unbiased.
If people start testing Advil, studies will cluster around its real effect on pain reduction. But smaller studies will be further off in both directions. If small-n studies exist only in one direction that implies that something weird is happening: authors choose not to submit those studies, journals choose not to publish those studies, or worse-case small-n studies are manipulated to get results that are positive.
When making a claim that something exists, it's enough to find just one example of it. I think there are enough anecdotes to say that moral licensing definitely exists.
That said, I agree the research on it is low-quality and that it's probably not so widespread that it's something to worry about when you're doing good.
Can you add some description to help me interpret the funnel plot? It's not obvious to me what's happening or why that means what it sounds like you're saying it means
Imagine a bunch of people run studies on moral licensing with a wide-range of sample sizes. Studies with smaller samples sizes will have higher standard errors and so be less precise. But all studies should still be clustered around the real effect size. In an ideal world, the correlation between effect and sample size should be unbiased.
If people start testing Advil, studies will cluster around its real effect on pain reduction. But smaller studies will be further off in both directions. If small-n studies exist only in one direction that implies that something weird is happening: authors choose not to submit those studies, journals choose not to publish those studies, or worse-case small-n studies are manipulated to get results that are positive.
Ah okay great, thanks! That helps. Nice article. I like you say 'first of all this is fake because it's social psychology' hahah