Figure 2

Multinomial logit estimates of occupation: people with BD vs. population. Note Multinomial logit estimates and 95 percent confidence intervals of the parameters β_j in equation Pr(Y_it = j) = exp(β_j BD_{i +} θ_jF_i)/ Σ_k exp(β_k BD_{i +} θ_kF_i), where Y_it is the occupation of person i in year t, BD_i equals one for people with BD, and F_i is an indicator for females. This figure shows the five largest and the five smallest estimates of β_j, along with estimates of β_j for creative occupations, defined by 3-digit ISCO08 codes. Standard errors are clustered at the individual level.