Email (record): Estimating the Causal Effect of Randomization Versus Treatment Preference in a Doubly Randomized Preference Trial