I'm doing a project for my intro class where I am analyzing each US congressional district by compactness using area and perimeter (in order to quantify gerrymandering).
My question is which projection to use for this analysis. Presumably, each state draws their congressional districts using their own state projection, so would this mean that a reliable analysis could only be performed if I perform 50 different analyses, one for each state?
Would using a country-wide analysis not provide effective analysis?
For the area calculations, you'll want to use an equal area projection like Albers (look under projected coordinate systems, continental). When we did some research on it, the biggest differences in area calculations between different "equal area" projections were due to how many vertices there were in the features.
Distance/length calculations are harder because no projection preserves all distances. You might break it down into UTM or state plane zones or state-wide systems when available to try to reduce the distortions in the distances due to the projection choice.
Melita