I’ll go with the exit polling data, not your system.
Norm the data to the actual results.
If you reject “my system,” you reject all current exit data.
And, just so I understand...
Gallup says 13% of non-white voters voted for Nixon in 1972.
NYT says Blacks voted 18% for Nixon, and Hispanics voted 35% for Nixon.
And, neither Gallup or NYT published voter turn out by ethnicity, so we have no idea what base they were using, although it can be estimated.
Which of those data sets is correct?
Or, maybe both data sets are incorrect?