Conditional smooths y-axis scale (brms bernoulli splines model)


fit = brm(binary_Y_variable ~ s(age), data = data, family = bernoulli())

Conditional_smooths plot

What is the scale of y-axis? Odds ratios? Possible to show them as probabilities?

conditional_smooths(fit, effects="age")

Conditional effects plot

Shows probabilities, but the plot differs considerably (wide credible intervals). Is it wrong to show splines with conditional_effects() function?

conditional_effects(fit, effects="age")

Sorry for taking so long to respond to this post!

I’ve not worked with splines much, but @paul.buerkner will surely be able to explain this to you. I checked the help pages for these two functions and I can’t find a description of the y-axis per se, but in conditonal_effects() Paul write that it provides a data.frame containing: ‘predicted values of the response.’

conditional_smooths shows thinks on the link scale, that is, logits in this case. No way to change that right now. conditional_effects also adds the intercept, hence the difference uncertainty.