In order to truly calculate a fair popular vote total, you need to adjust upwards proportionally the number of caucus votes by the average ratio of caucus goers to primary voters.
I am going to try and prove this with some numbers from a simple scenario. Say there is one caucus goer for every 10 primary voters. Say there are 3 states with these results:
Primary 1 Clinton Wins 60-40%
Population 1000 - 300 Clinton Votes, 200 Obama Votes
Primary 2 50-50% Tie
Population 500 - 125 Clinton Votes, 125 Obama Votes
Caucus 1 Obama Wins 60-40%
Population 1000 - 20 Clinton Caucus Votes, 30 Obama Caucus Votes
If you add up the total votes you get: 445 Clinton to 355 Obama, a 55.6 to 44.4% Clinton win. You have two large states, with the same population, going for opposite candidates by the same percentage, which should cancel out. The other state is a tie, so the result should be a tie. But by just counting VOTES, this gives one candidate an unfair win. The smaller state primary vote is skewing the results in Clinton's favor in this example.
The fair way to create an apples to apples popular vote count is below the fold.
If you multiply the caucus vote by the primary voter to caucus voter ratio, 10, and then add them all together, Caucus 1 becomes:
Caucus 1 Obama Wins 60-40%
Population 1000 - 200 Clinton Caucus Proportional Votes,
30 Obama Caucus Proportional Votes
and the totals are 625 Clinton, 625 Obama, an even, fair tie.
So in order to have a TRULY fair popular vote total, the number of caucus votes in each state need to be multiplied by the average primary voter to caucus goer ratio (However that is figured out - probably from historical data) before they are added to the vote totals.
If my math is wrong here, or anyone thinks my logic is wrong, please discuss nicely below!