Note: For some of the definitions, turn to the Minimal Set.
When looking at elections, we always hear about turnout numbers, but we rarely question what the turnout or the turnout rate really is. We also assume that this measure is always the same, hence we call it turnout.
Let’s take two republican US states, Texas and Wisconsin, to see if this is true:
Besides calling the valid votes the turnout, Texas uses the following formula:
valid / registered * 100 or else as valid / voting_age * 100
Wisconsin defines Turnout as:
ballots_cast / voting_age * 100
Ballots cast is the most probable explanation for the number under General Election Votes. In comparison: valid + scattering = 2,976,150 (99.07%); valid without scattering = 2,953,386 (98.31%).
Probably due to the recount, the full data is still missing (8 municipalities in Wisconsin haven’t yet published their final official results in regard to ballots cast and registered voters).
Conclusion:
These two turnouts are not comparable because they use a different equation or a different understanding of the word turnout.
Let’s first start with a definition what turnout is and how it is defined in most national elections and how we have seen it used in all US county results of last year’s presidential election so far.
Turnout
The definition of turnout for the US is:
turnout = ballots_cast / registered * 100
This formula gives you the turnout as percentage of ballots cast in respect to registered voters. Important: ballots cast are all the votes, may they be valid or not (= residual vote).
This formula works very well for the United States where all voters have to register prior to vote.
A representative example for the turnout is here:
247,836 / 328,294 * 100 = 75.49 ; The turnout is 75.49%
In most of Europe the formula would be:
turnout_eligible = ballots_cast / eligible * 100
Michael McDonald from the @ElectProject also uses the Voting Age Population (VAP) and compares it to the eligible, or as he calls it, the Voting-eligible Population (VEP).
turnout_voting_age = ballots_cast / voting_age * 100
So there are now three possible turnouts. We know that the number of people in the voting age group is bigger than the eligible group as some people are not allowed to vote. The difference can be very small as in the Netherlands or quite high as in Florida (around 1.6 Mio).
We also know that not all people register to vote, so the eligible population is always bigger than the registered population.
Otherwise said:
voting_age > eligible > registered
And if ballots_cast are constant, the turnout will be inversely proportional:
turnout_registered > turnout_eligible > turnout_voting_age
In conclusion we can say that the turnout should always be a measurement of ballots cast, which is the valid and the residual vote combined, divided by the relevant population that can vote, which is either eligible or registered. If the voting age population is used, always call it the turnout of the voting age population.
And if there are different populations, call the turnout accordingly, e.g.:
turnout_voting_age = ballots_cast / voting_age * 100
Outturn
The outturn (our creation) is a more radical measurement than the voting age turnout. The formula is:
outturn = valid / population * 100
It combines two ideas:
- It only uses the valid vote, as in the Texas example on top. The reason is that ballots_cast hides the huge variation of the residual vote.
The residual vote in the Florida election jumped for example from 0.75% in 2012 to 1.67% in 2016. Compare that to the even lower number of 0.45% in the Netherlands. - It uses the whole population and disregards the voting age, eligibility and registered voters entirely. The reason is that this gives the full measure of representativeness.
A turnout of 80% can equally be 640'000 people casting their ballot with 800'000 registered voters in a population of 1 Mio, as it can be 640 people out of 800 registered voters for the same population.
This is not possible with the Outturn. For the first Outturn with assumed 620'000 valid votes, we would have 62%, for the second Outturn with 620 / 1 Mio, we get 0.062%.
If you think, that this doesn’t make sense, remember how congressional districts in the US are allocated. They are population- and not registered voters-based (although some politicians are eager to change that).
In the future, we might also have countries that lower the voting age to 16 (in the 26th amendment to the US constitution the voting age had been lowered from 21 to 18 in 1971), as suggested by the Economist, so it will be easier to compare populations as a whole than different VAP’s.
Let’s calculate now the Outturn for Texas and Wisconsin:
Texas: 8,969,226 / 28,004,748 * 100 => Outturn: 32.03%
Wisconsin: 2,953,386 / 5,781,337 * 100 => Outturn: 51.08%
The valid vote in Wisconsin is without scattering (22,764 votes), as only registered write-ins are considered valid since 2013.
Texas Population was 27,862,596 on July 1, 2016, we extrapolated till November 8, there must be roughly 28,004,748 people. In Wisconsin, we started with 5,778,708 on July 1, 2016 and got to 5,781,337 on November 8.
References:
http://www.electproject.org/2016g
TX Results: http://elections.sos.state.tx.us/elchist319_race62.htm
TX Turnout: http://www.sos.state.tx.us/elections/historical/70-92.shtml
Valid Write-ins in Texas: http://kxan.com/2016/10/18/texas-will-only-count-13-write-in-candidates-for-president/
WI Turnout: http://elections.wi.gov/elections-voting/statistics/turnout
Registered and Ballots Cast: http://elections.wi.gov/node/4952 (8 municipalities are still missing)
Valid & Scattering (after Recount): http://elections.wi.gov/sites/default/files/County%20by%20County%20Report%20President%20of%20the%20United%20States%20Recount.pdf
Scattering: http://elections.wi.gov/node/3283
Valid Write-ins: http://www.nbc15.com/content/news/8-presidential-write-in-candidates-in-Wisconsin-398922821.html
Texas, Wisconsin on July 1, 2016: https://www.census.gov/quickfacts/table/PST045215/48,55,00
California lists eligible and registered voters and provides both turnouts:
http://elections.cdn.sos.ca.gov/sov/2016-general/sov/2016-complete-sov.pdf
