Not sure if anybody has done one of these yet, but here we go. I decided to do a simple average between Gov/Sen contests where applicable. It worked well in around two-thirds of the states or more, but there are a lot of exceptions and stipulations listed below:
If there was only one contest, then that is the only one used. I also combined results - even if there was a large difference between contests - in all states where exit polling was available AND Democrats were on the ballot in each applicable race.
As mentioned on the graphic (and above), some of the averages may skew the results. One example is Oklahoma, where Dorman came close to breaking 40% among whites. In the two Senate contests, however, whites were closer to 26% Democratic. For fairness' sake here, I averaged the two Senate contests together and counted them as one rather than averaging both Senate results against the Governor's race, which would essentially weigh the Senatorial white support double that of Gubernatorial white support. Ditto for SC.
Other notable examples of skewing will be in places like Maine, Iowa, Rhode Island and Hawaii. Examples of bad candidates where there was only one race (OH & NV) make things a little messy, too. NV deserves its own category; I made a special formula for calculating it as mentioned on the graphic.
I also opted not to average four other states: TN, AL, AK & KS. Tennessee's situation was similar to Nevada's in terms of whether or not there was actually a Democratic candidate running for Governor. AL had no contested Senate contest. AK & KS had independents running for Governor and Senate, respectively.