JASP graphs for accuracy data goes above 1 when CI are added

Mila_Marinova

Hello everyone,
I was wondering about a feature in the JASP graphs. I want to plot accuracy data (hence the max is 1), but when I add the 95% CI, the y-axis goes above 1 (see below). I have received some comments that I need to correct these graphs, but I do seem to find where this is coming from. Any suggestions?
Thank you in advance.

EJ

Hi Mila,

If you have accuracy data, then for a single individual you can use the binomial, for which the CI respects the bound. However, I assume that you have multiple individuals, and you analyze the proportion correct across subjects. You can use an ANOVA for this as an approximation, but the CI is based on a Gaussian model and will not respect the bound. The standard statistical model is a hierarchical binomial model, and here the group level parameter will not exceed the bound. But this is a more complicated analysis, one that is not implemented in JASP yet. You could use the R package brms for this, but that won't give you a Bayes factor. You could also stick to the Bayes factor for your model inference, but for your parameter CI report something else (like the CI from brms or from a bootstrap). Not ideal, obviously, but it is unlikely to matter much. Of course your analysis is far from the only one with this problem -- usually people do not report the CI and the anomaly remains hidden...

Cheers,
E.J.

Mila_Marinova

Hi E.J.,

Many thanks for the clear explanation. Definitely comes in handy!

Best,
Mila