Minnesota and Instant Runoff Voting

by Leigh Pomeroy

Last week Minneapolis voters overwhelming approved instant runoff voting for future city elections. Instant runoff voting, or IRV, allows voters to rank their candidate preferences on their ballots. The idea is that if no single candidate receives a majority of 1st choice votes, then the candidate receiving the least number of votes is eliminated and the 2nd choice of those who voted for him or her is then allocated to the remaining candidates.

Here's a theoretical example. Let's say Clint Bunsen, David Ingqvist and Clint's sister-in-law Arlene Bunsen are running for mayor of Lake Wobegon, Minnesota. The vote is tallied, and the results are as follows:

Candidate:	A. Bunsen	C. Bunsen	D. Ingqvist	Total
1st choice votes	43	35	55	133
1st vote percent	32.3%	26.3%	41.4%	100.0%
Those voting for A. Bunsen voted their 2nd choice for…	-	5	36	41
Those voting for C. Bunsen voted their 2nd choice for…	30	-	2	32
Those voting for D. Ingqvist voted their 2nd choice for…	40	10	-	50
Those voting for A. Bunsen voted their 3rd choice for…	-	40	0	40
Those voting for C. Bunsen voted their 3rd choice for…	28	-	3	31
Those voting for D. Ingqvist voted their 3rd choice for…	9	40	-	49

Because no one candidate received a majority of the vote, the candidate with the least number of 1st choice votes is eliminated. That's Clint Bunsen. Now the 2nd choice votes of those who preferred him over the other two candidates are allocated to the remaining candidates. (NOTE: Not every voter cast 2nd and 3rd choice votes, either because they were confused by the new system or because they didn't care or, in the case of Wally from the Sidetrack Tavern, they were protesting the change from the old system.)

Thus, the instant runoff results look like this:

Candidate:	A. Bunsen	C. Bunsen	D. Ingqvist	Total
1st choice votes	43	35	55	133
1st choice percent	32.3%	26.3%	41.4%	100.0%
2nd choice votes of those who preferred C. Bunsen allocated to remaining candidates	30	-	2	32
Instant Runoff Total	73	-	57	130
Instant Runoff Percent	56.2%	-	43.8%	100.0%

Although Arlene Bunsen finished 2nd in the 1st balloting, she wins the election with 56.2% of the runoff vote because she was the overwhelming 2nd choice of voters whose initial choice was Clint Bunsen.

So this is instant runoff voting. Lest is sound strange to Americans, it is not to citizens of Australia, Ireland, New Zealand and Fiji, all of which use IRV or a similar system in many of their municipal, state, territory or federal elections. In the U.S., IRV has been adopted so far in:

• Alameda County, Berkeley, Davis, Oakland, San Leandro, and San Francisco City and County; California
• Takoma Park, Maryland
• Ferndale, Michigan
• Minneapolis, Minnesota
• Burlington, Vermont
• Pierce County, Washington

IRV still has a long way to go, and implementation in the cities and counties where it has been approved can take time and incur expenses. For example, electronic voting machines are not yet set up to process the system, and in Minneapolis, implementation is estimated to cost $1.2 to $1.4 million.

Yet the advantages in the long run should outweigh the costs and confusion inherent in making the transition. In some cities and counties it eliminates costly primaries and special elections. But the main advantage of the system is that it yields a clear winner — a candidate who is the definite 1st or 2nd choice of the voters. For example, below is a possible scenario of how the recent results for Governor of Minnesota might have turned out with IRV:

Candidate:	Pawlenty-Molnau	Hatch-Dutcher	Hutchinson-Reed	Pentel-Provencher	Brown-Nelson	Davis-Soderberg	TOTALS
Vote %	46.7%	45.7%	6.4%	0.5%	0.4%	0.2%	99.9%
Votes	1,028,568	1,007,461	141,735	10,850	9,649	3,776	2,202,039
Pawlenty-Molnau 2nd choice pro forma %		5.0%	90.0%	0.0%	4.0%	1.0%	100.0%
Pawlenty-Molnau 2nd choice pro forma votes		51,428	925,711	0	41,143	10,286	1,028,568
Hatch-Dutcher 2nd choice pro forma %	5.0%		85.0%	10.0%	0.0%	0.0%	100.0%
Hatch-Dutcher 2nd choice pro forma votes	50,373		856,342	100,746	0	0	1,007,461
Hutchinson-Reed 2nd choice pro forma %	30.0%	55.0%		10.0%	4.0%	1.0%	100.0%
Hutchinson-Reed 2nd choice pro forma votes	42,521	77,954		14,174	5,669	1,417	141,735
Pentel-Provencher 2nd choice pro forma %	5.0%	20.0%	75.0%		0.0%	0.0%	100.0%
Pentel-Provencher 2nd choice pro forma votes	543	2,170	8,138		0	0	10,850
Brown-Nelson 2nd choice pro forma %	90.0%	0.0%	5.0%	0.0%		5.0%	100.0%
Brown-Nelson 2nd choice pro forma votes	8,684	0	482	0		482	9,649
Davis-Soderberg 2nd choice pro forma %	90.0%	0.0%	5.0%	0.0%	5.0%		100.0%
Davis-Soderberg 2nd choice pro forma votes	3,398	0	189	0	189	0	3,776

My pro forma allocations of voters' 2nd choices are purely speculative. Obviously, anyone wanting to play with the math can do so. (Please email me if you would like a copy of the spreadsheet.) Given the above assumptions, the 1st two elimination rounds would take out Davis-Soderberg and Brown-Nelson, but the result would still yield no candidate with a clear majority.

Here is how the 1st two elimination rounds would look:

Candidate:	Pawlenty-Molnau	Hatch-Dutcher	Hutchinson-Reed	Pentel-Provencher	Brown-Nelson	Davis-Soderberg	TOTALS
Vote %	46.7%	45.7%	6.4%	0.5%	0.4%	0.2%	99.9%
Votes	1,028,568	1,007,461	141,735	10,850	9,649	3,776	2,202,039
Davis-Soderberg 2nd choice pro forma %	90.0%	0.0%	5.0%	0.0%	5.0%		100.0%
Davis-Soderberg 2nd choice to remaining candidates	3,398	0	189	0	189		3,776
1st elimination net %	46.9%	45.8%	6.4%	0.5%	0.4%		100.0%
1st elimination net vote	1,031,966	1,007,461	141,924	10,850	9,838		2,202,039
Brown-Nelson 2nd choice pro forma %	90.0%	0.0%	5.0%	0.0%			95.0%
Brown-Nelson 2nd choice to remaining candidates	8,684	0	482	0			9,167
2nd elimination net %	47.3%	45.8%	6.5%	0.5%			100.0%
2nd elimination net vote	1,040,651	1,007,461	142,406	10,850			2,201,368

After eliminating Pentel-Provencher, there would still be no winner. Only after reallocating the Hutchinson-Reed vote would one candidate emerge with over 50% of the IRV calculated tally.

The result? Hatch-Dutcher by a mere 3,871 votes out of an adjusted total of 2,171,299 votes cast, for a margin of less than 2/10ths of 1%.

Here is how the final elimination rounds would work:

Candidate:	Pawlenty-Molnau	Hatch-Dutcher	Hutchinson-Reed	Pentel-Provencher	Brown-Nelson	Davis-Soderberg	TOTALS
Pentel-Provencher 2nd choice pro forma %	5.0%	20.0%	75.0%				100.0%
Pentel-Provencher 2nd choice to remaining candidates	543	2,170	8,138				10,850
3rd elimination net %	47.3%	45.9%	6.8%				100.0%
3rd elimination net vote	1,041,193	1,009,631	150,544				2,201,368
Hutchinson-Reed 2nd choice pro forma %	30.0%	55.0%					85.0%
Hutchinson-Reed 2nd choice to remaining candidates	42,521	77,954					120,475
4th elimination net %	49.9%	50.1%					100.0%
4th elimination net vote	1,083,714	1,087,585					2,171,299

Not being a statistical guru, I confess there may yet be errors in the formulations. Further, IRV formulations vary from one implementation to another. Nevertheless,this is approximately how the system might work if adopted for statewide in Minnesota. The only problem with an election this close, however, is that these results would trigger a recount, which is a whole separate can of worms.