This paper proposes a new panel data sample selection model with 1) ordered discrete choices in the selection equation and 2) non-parametric unobserved heterogeneity in the equation of interest. This method is used to estimate life-cycle earnings profiles using high-quality administrative data. We compare conclusions regarding the existence and direction of selection into (part-time) work among men and women across different panel data sample selection techniques. The main conclusion is that our new approach is able to control for important unobserved heterogeneity from intensive labor supply choices with important consequences for the existence and direction of selection in (part-time) work.