One pattern, which is perhaps not so Earth-shattering, is that Ken Buck did better than Jane Norton in more conservative counties during the March caucuses. McCain's presidential vote in 2008 is a reliable predictor of Buck's successes. This result is statistically significant (p ≤ .05) and holds even with all the control variables I could throw at it.
Interestingly, county ideology played no role in the Democratic contest. Liberal counties were just as likely to go for Romanoff as for Bennet.