Additional-Review-(marginalization-over-joint-population+EoS-uncertainty).md

@@ -60,7 +60,7 @@ and there is an extra factor of `p(data|Lambda)` included within the integrand.
 As is shown in the [technical note](#technical-note), we instead wish to compute
 
 ```math
-S = \int d\Lambda p(\Lambda) \int d\theta p(\theta|\mathrm{data},\Lambda) \Theta(\theta)
+S = \int d\Lambda d\theta p(\theta, \Lambda|\mathrm{data}) \Theta(\theta) = \int d\Lambda p(\Lambda|\mathrm{data}) \int d\theta p(\theta|\mathrm{data},\Lambda) \Theta(\theta)
 ```
 
 and to do so we must normalize the sum over `\theta_i` to remove the factor of `p(data|Lambda)`.