Marginal_effects() vs marginal_smooths() y axis

When I use marginal_effects() vs marginal_smooths() on the same bernoulli model the y axes are different. What are the values being reported on the y axis? In neither case does it resemble the model intercept or the proportions I’d expect based on the raw model data.

Family: bernoulli
Links: mu = logit
Formula: result ~ s(week)
Data: data (Number of observations: 9199)
Samples: 4 chains, each with iter = 2000; warmup = 1000; thin = 1;
total post-warmup samples = 4000

Smooth Terms:
Estimate Est.Error l-95% CI u-95% CI Eff.Sample Rhat
sds(sweek_1) 2.56 1.15 0.92 5.28 931 1.00

Population-Level Effects:
Estimate Est.Error l-95% CI u-95% CI Eff.Sample Rhat
Intercept -1.34 0.03 -1.39 -1.29 3759 1.00
sweek_1 -1.26 0.48 -2.31 -0.41 1313 1.00

marginal_smooths shows the raw spline on whatever scale it is computed. In your case on the logit scale.

Ok, thanks - and which scale is the marginal effects plot on? My plan is to convert the Y axis to probability.

On the scale of your response unless you change that.