[Jeff Chandler]

Likelihood to Vote Calculation
.
Each respondent is given a rank based on a calculation of their likelihood to vote. That calculation is based on two factors: their demographic profile, such as race, sex and age, using historic demographic voting patterns released by the Census Bureau after general elections; as well as their self-reported likelihood to vote. To improve the accuracy of the calculation, Reuters/Ipsos also conducted an exercise in the spring of 2016 matching respondents with actual voter registrations and voting history. All respondents are then ranked according to our calculation of their likelihood to vote.

You can throw all that out the window. There is a large crossover of democrats this year, but the big story is the numbers of people who haven't voted in a long time. Those people won't be included in the "likely voters".

[GilGil]

That was my whole point in posting this. In just a few weeks we will be talking about president elect Trump. This race is over!