When working with areal data in spatial regression, missing responses and predictors can be a major hurdle. This is especially true when the missingness is not random, like in cases where small population sizes in certain units lead to incomplete data.
The literature often suggests complex hierarchical models or elaborate ad hoc imputation methods to tackle this issue. However, these approaches can be overwhelming and computationally intensive.
So, what are some practical and relatively simple ways to handle missing data in spatial statistics? I’d love to explore some alternatives that don’t require a Ph.D. in statistics or a supercomputer.
Perhaps we can discuss some straightforward imputation methods or clever ways to work around the missing data. Any insights or experiences would be greatly appreciated!