What is needed is a "lab rat" protocol for a simple (ideally inexpensive) experiment that has a good signal-to-noise ratio, e.g., two or more orders of magnitude above the noise floor. With that, any group that seeks to reproduce the purported effect can do so, characterize its error, investigate its mechanism through repeated trials, and test and confirm or refute various mundane hypotheses.
A room-sized calorimeter is an interesting idea. But if such a device is required to establish a result, it won't be convincing to very many people because all sorts of possible artifact will need to be ruled out, including going back to the room-sized calorimeter.