I am creating thermal models involving ADI's LTC6952 devices. The datasheet for the component only provides the junction-case thermal resistivity so, due to the large exposed pad on the part, I made the assumption that heat transfer is biased towards the board (θ_jb << θ_jc). I'm hoping that someone will be able to confirm this assumption and, if possible, provide more realistic values for the junction-board thermal resistivity. A ratio of how much heat is expected to go towards the board vs the case (for example, 70/30 or 85/15) would be sufficient.
The answer is somewhat dynamic as the percentage of heat that is transferred out of the package with the ThetaJC will increase with increased airflow or if you have a heatsink that sits on top of the package. It also changes depending on how good your board was designed to let heat escape out of the bottom of the package (number of via's, GND planes...)
For a no airflow, no heatsink case, with the recommended number of GND via's to a solid GND planes, most of the heat will flow out of the bottom of the back (>90%)