Earlier quoted context omitted.
the point is there's not really a way for us to tell which is the "substance". It's all just correlations and it's all just a matter of what aspect you choose to highlight
You don’t think you can isolate variables and analyze effects? For example, you can’t study the opinions of people with high education who live in rural areas?
Group B: College Educated + Rural
Group C: No post-secondary + Urban
Group D: No post-secondary + Rural
Fairly simple analysis to do, and I suspect (as I think you do) that the trend will align with a split between A/D vs B/C not A/B vs C/D.