I’m working with some collaborators to model crop yield variability response to soil organic matter contents–yield variability should decrease with more organic matter because organic matter retains water well and can buffer drought effects.
We have yield on soils, climate, and yield for all US counties for at least 20 years.
My colleagues and I would usually tackle this by calculating a statistic–like coefficient of variation. That would then reduce the data so you’d have only one value per county.
Do any of you more experienced Stan folks have thoughts on time-series (or even ODE-based) models that might help us model the data without having to aggregate up to a one-value-per-county statistic?