(Revised November 2022)

While the presence of disparities in mental healthcare is well documented, the mechanisms of such disparities are less understood. In this paper, I develop and estimate a structural model of diagnosis for a prevalent child mental health condition, Attention Deficit Hyperactivity Disorder (ADHD). The model incorporates both patient and physician influences to highlight various mechanisms of mental health diagnosis and sources of disparities. Using electronic health record data and novel natural language processing techniques applied to doctor note text, I estimate gender-specific model parameters and decompose the male:female ADHD diagnostic difference of 2.5:1 observed in the data. Counterfactual simulations show that only 33% of this diagnostic difference can be explained by underlying symptom prevalence, with the remainder driven by differences in diagnostic thresholds. I find that physicians view missed diagnosis to be costlier than misdiagnosis, especially for their male patients, and I discuss reasons why this may be economically warranted..