The Case For Range Voting: Democratic Primaries

Ian Kerr, November 20, 2016

In a recent post, I mentioned the notion of Range Voting in Democratic Primaries.

For full details on the mechanics of it, see rangevoting.org. The process in brief: it's the same process by which they judge figure skating or gymnastics in the Olympics. Each judge (voter) gives each contestant (candidate) a numeric score (not just the contestant/candidate they like the most); highest average score is the winner.

First, I'll provide an example that starts with real Primary Election results and theorizes how Range Voting might have changed the outcome.

Hypothetical Example - Arizona Democratic Primary 2012

By the time Arizona held its Primary, there were only two active candidates in the race, Hillary Rodham Clinton and Bernie Sanders. The actual vote totals had Clinton winning by 15%:

Candidate	Votes	Pct.
CLINTON	262,459	56.3%
SANDERS	192,962	41.4%
OTHERS	10,600	2.3%
	466,021

What the raw vote totals do NOT tell us is the level of affinity for each candidate among voters for whom that candidate was NOT their first preference.

Suppose ...

A) these voters could have cast their ballot by Range Voting -- allowing them to rate each candidate on a scale of 0-to-9 (9 best).

B) each vote from the actual Primary (above) were worth the maximum 9 points to the candidate for whom the vote was cast. That would leave the following number of voters the opportunity to further rate each candidate:

Candidate	Additonal Ratings*
CLINTON	203,562
SANDERS	273,059
OTHERS	455,421

*Total ballots cast minus number of actual votes received.

C) Clinton supporters had a generally more-favorable view of Sanders than Sanders supporters had of Clinton... such that the table below describes the frequency with which each candidate received each possible additional rating:

Additional		Candidate
Rating	CLINTON	SANDERS	OTHERS
9**	5.0%	7.0%	5.0%
8	5.0%	15.0%	13.0%
7	5.0%	15.0%	13.0%
6	5.0%	15.0%	13.0%
5	15.0%	11.0%	13.0%
4	15.0%	11.0%	13.0%
3	15.0%	11.0%	13.0%
2	15.0%	11.0%	13.0%
1	10.0%	2.0%	2.0%
0	10.0%	2.0%	2.0%

**there would be no rule against giving two or more candidates the same rating, including the highest rating.

D) THEN the final result would give Sanders the win:

RATING	CLINTON	SANDERS	OTHERS
9	272,637	212,076	33,371
8	10,178	40,959	59,205
7	10,178	40,959	59,205
6	10,178	40,959	59,205
5	30,534	30,036	59,205
4	30,534	30,036	59,205
3	30,534	30,036	59,205
2	30,534	30,036	59,205
1	20,356	5,461	9,108
0	20,356	5,461	9,108
AVERAGE	6.68	6.86	5.11

Also relevant... the total number of "enthusiastic" supporters (voters who gave each candidate a 7, 8, or 9) and "detractors" (voters who gave each candidate a 1 or 0).

Candidate	Enthusiastic	Detractors
CLINTON	292,993	40,712
SANDERS	293,994	10,922
OTHERS	151,781	18,216

What does this illustrate?

1) Someone with as much as a 15-point margin in a "check one box" vote might NOT be the most popular candidate in the race. Only Range Voting could illustrate this.

2) The most popular candidate almost always has to be the one with the most "enthusiastic" votes. 

3) The candidate with the most "detractors" almost always will NOT win. 

#2 & #3 are especially important in a Primary election... because there's a General Election to be won thereafter. The enthusiastic voters are the ones who will knock doors, make phone calls, put up yard signs, give people rides to the polls, talk up the candidate with friends... they're the foot soldiers in a successful ground game. The detractors will entertain the thought of voting for a different candidate in the General Election... and some will actually do so.

Again the breakdowns above are not based on any scientific study of Arizona's Democratic Primary voters... but I don't think it's unrealistic. This example raises the possibility of 40,000+ Clinton detractors having voted in this Primary (more than 1 in 12 Primary votes). Third-party candidates garnered more than 140,000 Arizona votes in the General Election. It's totally possible that 3-in-10 of the 3rd party votes came from Clinton detractors who voted in the Democratic Primary.

What do you think?

Circular Firing Squad: Cease Fire!

Ian Kerr, November 20, 2016

A dozen days following an election that stunned the 48% of America (at last count) that voted for Hillary Clinton and many of the 5.3% who voted for someone else. Social media remains fertile ground for the blame game.

Instead of wasting keystrokes and LIKE buttons on blaming each other, can we come to agreement on the following things?

1. The Democratic National Committee needs to be led by a new chairman who 
• will be able to work the job full-time
• has not held high public office in the last two years
• will not endorse or oppose any candidate vying to be the party's Presidential nominee in 2020 until the conclusion of the primaries and caucuses.
• will work for the Party to be more of a reflection of the aspirations of its voters rather than the wants of its donors or other highly-influential individuals.
2. The DNC should scrap the concept of Superdelegates.
3. The concerns of working-class and middle-class whites hold much in common with those working-class and middle-class minorities. The Democratic Party has not stressed that enough.
4. When turnout is high in-general, it favors Democrats. The advantages Democrats have with respect to the voting tendencies of diverse voters do not accrue to Democrats if diverse voters do not turn out.
5. If it would be hard very soon to abolish the Electoral College, we should work in the various states to pass state laws that would require the Electoral Votes of that state to accrue to the winner of the national popular vote.
6. Polling. Trump won states despite the polls saying he was behind in those states by more than a normal polling error. Clinton made campaign decisions based on those poll numbers. Either get better polling or treat "5 points ahead" as "tied."

And can we consider the following:

7. Democrats should select their nominees for public office via Range Voting.
   For a technical explanation, see http://rangevoting.org/
   I'll offer a strategic explanation in a later blog post. 

How To Watch Coverage of the 2016 Presidential Election Returns

Ian Kerr
November 6, 2016 (Updated 7am MST - November 8)

Before you turn on your TV set Tuesday, you need to ask yourself three things:

1) Are you an eligible voter or can you become one via same-day registration?
2) Have you yet to vote?
3) Are the polls open in your jurisdiction?

If the answer to all three of those questions is YES, then get off this website and go vote.
-----
The following is meant to help Election-Night (or, if you’re watching from the Eastern Hemisphere, Election-Morning) viewers interpret the Electoral Vote scoreboard as it unfolds. If you’re not familiar with the workings of the U.S. Electoral College system, consider reading https://en.wikipedia.org/wiki/Electoral_College_(United_States).

The table below reflects an approximation of the Electoral Vote scoreboard at various times of the night (and early morning, with reference to U.S. Eastern Standard Time).

TABLE 1 – Approximate Electoral Vote Count at various Time Points

My data – 

A) Poll-closing times for each state

B) The state’s electoral vote counts, and

C) The odds for given candidates to carry each states (treating parts of Maine and Nebraska as “states” in their own right) per http://fivethirtyeight.com as of 6am Eastern Standard Time, Monday November 8. Consult that website for its methodology of developing their odds.

The following table indicates which candidates are most likely to be declared projected winners and the various states and approximately when.

TABLE 2 – Approximate Times At Which States Would Be Projected For Various Candidates

Note – Maine {ME} and Nebraska {NE} award two EVs for their states At Large (AL), then one for each Congressional District; ME has 2, NE has 3. Numbers in parenthesis that follow these state codes represent the number of the Congressional District(s) encompassed in a particular projection; "(AL)" means "At Large".

My methodology –

1) Whichever candidate the odds indicated to have the best chance of winning each state, I eventually credit that candidate with the requisite number of electoral votes.

2) In the Clinton and Trump columns of the table above, you will see postal codes of certain states appearing BEFORE the forward slash {/} and some appearing AFTER.

3) If the odds data indicated any candidate to have better than an 86% chance of carrying a particular state, I believe the networks will declare that candidate to be the projected winner of that state within 5 minutes of the poll-closing time of that state. These are the states in the Trump and Clinton columns with postal codes placed BEFORE the forward slash.

4) To the extent the odds are not as good as 86% for a candidate, I believe the networks may wait for more data to declare projected winners there (they may not wait so long; I’m being conservative). Those states initially appear, at their poll-closing times, in the Too Close column.

5) At the approximate time when the media will make projections for these initially-Too-Close states, the postal code for that state will appear in two places: a) in the Clinton or Trump column, but AFTER the forward slash AND b) in the appropriate “Shift: 'Too Close' to _____” column.

6) I have chosen to vary the time interval between the poll-closing time and the approximate time the projection would occur according to these variables: a) how close the odds are {closer = longer}, b) recent history of long queues of voters in line at poll-closing time, especially in urban areas {i.e. Ohio}, c) varying degrees of difficulty in vote-counting {i.e. Virginia has 95 counties and 38 cities doing vote-counting}, and d) past experience {i.e. 2012} of how long it takes projections to occur when there is a close result (I used this article, which described some of the 2012 timeline of when networks projected winners of various states: http://www.hollywoodreporter.com/news/nbc-news-declares-obama-winner-387191).

How I plan to use this data

Again, this is an approximation of the Electoral Vote counts for each candidate at various times of the night based on 1) a respected oddsmaker’s calculation of the probable outcomes and 2) my not-terribly-scientific view of how those outcomes will manifest themselves during Election Night coverage.

When I watch election coverage Tuesday, this table should help frame the answers to questions like:

A) “Who’s ahead or who’s behind right now?”
The answer isn’t just the raw electoral vote total – it’s the total relative to what might be expected at a point in time. If Trump is ahead by a count of 160 EV to 120 EV for Clinton at 9:05 pm EST on Tuesday, it may be a mistake to say Trump is “ahead.” It's more truthful to say he’s “performing as the oddsmakers expect.”


B) “What are the surprises? And what do the surprises mean?”
The vote count will differ from the tables above for two general reasons – 1) one candidate bucked the odds and won a state unexpectedly or 2) the media projected a winner in a state earlier or later than the approximation called for. The first surprise is more consequential than the second.

NATURE OF 11/7 UPDATES -- Due to changes in odds:
Moved FL, NC and NV to Clinton from Trump
Changed projection times for AK, CO, IA, ME(AL), ME(2), NE(2) & WI.
NATURE OF 11/8 UPDATES -- Due to changes in odds:
Moved ME(2) to Clinton from Trump
Changed projection times for AZ, GA, ME(AL), MN, NE(2) & VA.