Those can work, if you have clear alternative candidate models. But it’s not clear how you would have done that here, in, say the second problem. The model you would have mixed with is something like “lithium-7 is actually reactive on relevant timescales here”; that’s not really a model, barely a coherent assumption.
Isn’t this more or less what mixture models were made for?
Those can work, if you have clear alternative candidate models. But it’s not clear how you would have done that here, in, say the second problem. The model you would have mixed with is something like “lithium-7 is actually reactive on relevant timescales here”; that’s not really a model, barely a coherent assumption.