MN Senate: 12 key counties in projected 523-vote Franken recount gain

[ProgressiveEconomist]

The more seats Democrats gain in the Senate, the easier will be President Obama's task of bringing about dramatic changes on all the issues that inspired his victory.

The Minnesota US Senate contest is slated for automatic recount after November 18th.

As of Friday, Coleman was ahead by 221 votes.

Assuming Al Franken gets the same percentage of recounted undervotes that he got of machine-scannable votes in each county, how many votes would he gain or lose in the recount?

My estimate is: ***523***. He'll gain over 100 votes in Hennepin, in Ramsey, and in St. Louis counties (see column 1 of the table below for an estimate). But he'll lose more than 62 votes in the other 9 counties listed.

In each of the other 75 counties in Minnesota, Al would gain fewer than 37 votes or lose fewer than 40 votes.

WHAT'S YOUR OPINION?

Al will do better in a given county--
--The more undervotes there may be (Column 2 below); and
--The greater his percent margin over Coleman in machine-scannable votes (Column 3);

My measure of potential undervotes is the difference between the total vote for President and the thotal vote for Senate so far. This is an underestimate of the number of ballots with a vote for President but no vote for Senate. (The difference is the number of ballots with a vote for Senate but no vote for President).

It is assumed that 100 percent of voters for President actually intended also to vote for Senate; scale down each estimate if you believe this percentage is lower

For comparison with the estimated recount gain, the total Senate vote as of November 7th is listed in Column 4.

1--------- 2------------- 3------------ 4----------
FNETEST PSDIFF ALMARGIN SENTOT COUNTY

0897.437 __ 6430 __ 013.9570 __ 657296 __ 27:Hennepin
0588.955 __ 3280 __ 017.9559 __ 274104 __ 62:Ramsey
0184.025 __ 0830 __ 022.1717 __ 117952 __ 69:St._Louis

-062.957 __ 0297 __ -21.1978 __ 044698 __ 71:Sherburne
-081.061 __ 0369 __ -21.9678 __ 065209 __ 86:Wright
-089.082 __ 0691 __ -12.8918 __ 078065 __ 73:Stearns

-093.052 __ 0472 __ -19.7144 __ 066662 __ 70:Scott
-093.385 __ 0358 __ -26.0851 __ 049327 __ 10:Carver
-097.644 __ 1092 __ -08.9417 __ 075377 __ 55:Olmsted

-098.378 __ 1011 __ -09.7308 __ 136053 __ 82:Washington
-101.537 __ 1186 __ -08.5613 __ 181035 __ 02:Anoka
-125.206 __ 1610 __ -07.7768 __ 223859 __ 19:Dakota

Data were downloaded from the link at http://electionresults.sos.state.mn.us/ElecMenu.asp .

[democrattotheend]

1. Are we sure that all the undervotes have Senate votes on them?

Isn't it possible that a lot of them are people who chose not to vote for Senate?

I voted a straight Democratic ticket on my ballot in Virginia except for one office, which I chose to leave blank due to issues I have with the Democratic incumbent. Not everyone intends to vote for every race. Hopefully that's a small percentage though.

[ProgressiveEconomist]

4. Scale down column 1 by the percentage you think is appropriate--It IS assumed that all

ballots with a scanner-readable Presidential vote but no scanner-readable Senate vote will turn out to have a human-readable Senate vote.

With such a small margin for coleman in the tally so far, even if we scaled down the estimates by half, Franken still would win, wouldn't he?

[Johonny]

2. After 2000

It's amazing that every state in the Union didn't pass a law stating that once election were within the counting error of the vote count an automatic run off was needed. If Al wins on a recount it will be an ugly and unpopular way to enter the senate.

[ProgressiveEconomist]

6. Why an "automatic runoff" rather than the automatic recount current law provides

when the plarality margin is under half a percent (IIRC)?

Hand recounts cost 3 cents a ballot; how much would a runoff cost?

And would you exclude Barkley and write-ins from a runoff? Wouldn't that change the dynamics of the election entirely?

And what about turnout in a runoff? If turnoff drops dramatically from Election Day, why would the runoff be perceived as fairer than a hand recount of the original ballots?

[Johonny]

8. because if it's within the counting error

then it's within the error limits to which no accurate determination can ever be made. The best you can say is so and so is X percentage likely to have won the most votes. For whatever reason people assume you can accurately count every single vote.

If you count the votes up once and say A won and then count them again and say B won... people won't trust the vote count and the candidate enters congress with a certain taint. Al would be way better off with a run off that had the two highest vote totals. Why include a third party that didn't come close to winning? That makes no sense. Some states have built in run offs win no one wins a majority. That's an ok system. But it still fails the case where the vote falls within the counting error of the process. Humans have all this collective scientific knowledge of statistics and then we forget about them when we write laws for such things as the census and counting votes. It's amazing.

[ProgressiveEconomist]

9. The state of MN disagrees, and so do I. See their recount documents at LINK

http://www.sos.state.mn.us/home/index.asp?page=1047 .

[Johonny]

11. I looked them over

Didn't see anything to convince me I'm wrong. The best the state can determine is which candidate had the highest probability of being the winner. With vote totals this low that's all they can do. That's basic statistics.

[ProgressiveEconomist]

15. What would YOU say is the margin of error for hand recounts of optical-scan-rejected ballots?

Minnesota election law recognizes a margin of error for optical scan equipment of at most 0.5 percent (50 per ten thousand). This is a very conservative estimate; for example, optical-scan counties in the Florida Y2K Presidential election had read-error rates of 0.03 percent (3 per ten thousand). Automatic hand recounts occur when the 0.5 percent threshold is breached.

A very thorough set of rules for carrying out hand recounts is given by Minnesota Statute 204C.21 and summarized in half a page at http://www.sos.state.mn.us/docs/pilingrecountmaterialspartisan.pdf . Rules for determining voter intent from a ballot are given by Minnesota Statute 204C.22 and summarized in 3 pages, consisting mainly of ballot diagrams, at http://www.sos.state.mn.us/docs/voterintentmaterials.pdf .

Teams of county auditors and city election officals (designated at http://www.sos.state.mn.us/home/index.asp?page=1041 ) have to agree on the arrangement of all questionable ballots in piles of various categories. In each pile, bundles of 25 ballots are criss-crossed to facilitate counting.

When a county's recount is finished, the resulting piles of ballots and summary forms are packaged and taken to the Secretary of State for checking and consolidation.

I'd say these procedures are virtually foolproof. Do you see any holes in them?

[ProgressiveEconomist]

7. .

[wellstone dem]

3. The difference now is 206 votes.

[ProgressiveEconomist]

5. 15 votes would not change this aanalysis much, given the two factors it is based upon.

I might re-run this before the vote is certified on November 18th, and certainly will after that.

Another improvement would be to use precinct data rather than county data.

[NewJeffCT]

10. And, if he is certified as behind, Coleman will stand by his word and concede

and, if you believe that one, I have a bridge in Brooklyn for you to buy...

[ProgressiveEconomist]

12. I'd bet Coleman's campaign people already have written the statement he'd issue

should Franken take the lead by the November 18th certification deadline.

The statement would explain why at that time insisting the recount proceed does NOT contradict previous statements by Coleman. That's EASY for any lawyer or PR flack, unfortunately.

[Occam Bandage]

13. " It is assumed that 100 percent of voters for President actually intended also to vote for Senate"

Bad assumption. Franken and Coleman both were kind of unpopular, and both ran negative campaigns. It's possible that many or most of those undervotes are blank.

[ProgressiveEconomist]

14. See post #4 above. What proportion would you say were INTENTIONAL undervotes

for Senate? What proportion of voters in each county who intended to vote for President did not intend to vote in the US Senate contest? Franken needs only a bit more than 200 added votes to win. 523 assumes zero intentional undervotes, but suppose 57 percent were intentional undervotes. 0.43 x 523 = 224, still more than Franken needs.

Anything more than 57 percent intentional undervotes is likely to swing the election back to Coleman, given these county-level figures.

[Occam Bandage]

17. With due respect,

I'm not going to even bother looking at the "analysis" of someone who didn't even bother looking up past undervote proportions. It's a bit like reading the economic forecasts of someone who starts off by saying, "Now, let's assume that everyone on Earth makes $20,000 per year...."

[ProgressiveEconomist]

16. TOTALS and complete results for all 87 MN counties, in 3 groups of counties

There are 3 groups of counties, depending on the range of estimated net recount gains by Al Franken.

In 17 "IMPORTANT" counties, Franken nets at least 12 votes or loses more than 40 votes. 12 of these counties already have been discussed in the OP. The split at 40 was inspired by the fact that there are no losses between just under 40 and well over 62. The average gain is over 55 votes, for a total of more than 940 votes.

This gain of 940 votes in the most important 17 counties is partially offset by losses in the other two groups of counties, for a total net gain over all 87 counties of 523 votes.

In each of 55 "minor-impact" counties, Franken's net gains and net losses are below 12 votes in absolute value. Franken averages a loss of 2.73 votes per county. But, since there are so many of them, the total loss from all 55 is more than 150 votes.

Finally, in 15 "small loss" counties, the loss averages 17.82 votes per county, for a total loss of 267.33 votes. The amount of loss for Franken in each of the 15 counties ranges between 12.14 and 39.59.

1--------- 2------------- 3------------ 4----------
FNETEST PSDIFF ALMARGIN SENTOT COUNTY
_________________________________________________________________________

0523.119 __ 24817 __ --------- _ 2,885,438 __ TOTAL FOR ALL 87 MN COUNTIES
_________________________________________________________________________

"IMPORTANT" COUNTIES
0897.437 __ 6430 __ 013.9570 __ 657,296 __ 27:Hennepin_______01
0588.955 __ 3280 __ 017.9559 __ 274,104 __ 62:Ramsey________02
0184.025 __ 0830 __ 022.1717 __ 117,952 __ 69:St._Louis_______03

0036.933 __ 0699 __ 005.2836 __ 027,235 __ 85:Winona_________04
0021.439 __ 0115 __ 018.6429 __ 018,334 __ 09:Carlton__________05
0020.003 __ 0169 __ 011.8362 __ 019,018 __ 50:Mower__________06

0019.546 __ 0380 __ 005.1437 __ 031,417 __ 66:Rice ____________07
0014.484 __ 0154 __ 009.4052 __ 024,242 __ 31:Itasca___________08
-062.957 __ 0297 __ -21.1978 __ 044,698 __ 71:Sherburne________09

-081.061 __ 0369 __ -21.9678 __ 065,209 __ 86:Wright___________10
-089.082 __ 0691 __ -12.8918 __ 078,065 __ 73:Stearns___________11
-093.052 __ 0472 __ -19.7144 __ 066,662 __ 70:Scott_____________12

-093.385 __ 0358 __ -26.0851 __ 049,327 __ 10:Carver___________13
-097.644 __ 1092 __ -08.9417 __ 075,377 __ 55:Olmsted__________14
-098.378 __ 1011 __ -09.7308 __ 136,053 __ 82:Washington_______15

-101.537 __ 1186 __ -08.5613 __ 181,035 __ 02:Anoka__________16
-125.206 __ 1610 __ -07.7768 __ 223,859 __ 19:Dakota__________17

0940.520 _ 19143 __ --------- _ 2,089,883 __ Subtotal 17 "important" counties


"SMALL LOSS" COUNTIES
-012.142 __ 0064 __ -18.9718 __ 018,965 __ 43:McLeod_________01
-012.202 __ 0089 __ -13.7100 __ 010,124 __ 20:Dodge__________02
-012.229 __ 0092 __ -13.2926 __ 012,616 __ 42:Lyon __________03
-012.730 __ 0063 __ -20.2069 __ 004,736 __ 59:Pipestone______04
-013.637 __ 0132 __ -10.3312 __ 019,204 __ 05:Benton_________05

-013.745 __ 0413 __ -03.3280 __ 028,846 __ 14:Clay __________06
-014.424 __ 0078 __ -18.4920 __ 004,894 __ 67:Rock __________07
-014.988 __ 0058 __ -25.8411 __ 007,639 __ 68:Roseau_________08
-017.495 __ 0116 __ -15.0817 __ 019,938 __ 30:Isanti_________09
-018.734 __ 0198 __ -09.4615 __ 015,135 __ 60:Polk __________10

-019.191 __ 0124 __ -15.4764 __ 020,793 __ 21:Douglas________11
-021.034 __ 0156 __ -13.4834 __ 029,147 __ 13:Chisago________12
-021.995 __ 0154 __ -14.2823 __ 016,811 __ 03:Becker_________13
-023.195 __ 0263 __ -08.8192 __ 034,901 __ 18:Crow_Wing______14
-039.588 __ 0199 __ -19.8935 __ 032,488 __ 56:Otter_Tail_____15

-267.328 __ 02199 __ ----------- __ 0,276,237 __ Subtotal 15 "small loss" counties


"SMALL IMPACT" COUNTIES
0010.833 __ 0067 __ 016.1692 __ 006,902 __ 38:Lake __________01
0008.535 __ 0363 __ 002.3511 __ 034,707 __ 07:Blue_Earth_____02
0008.330 __ 0165 __ 005.0482 __ 017,115 __ 24:Freeborn_______03
0006.179 __ 0236 __ 002.6183 __ 021,999 __ 04:Beltrami_______04
0002.854 __ 0078 __ 003.6585 __ 006,724 __ 36:Koochiching____05
0002.738 __ 0022 __ 012.4474 __ 003,326 __ 16:Cook __________06
0002.654 __ 0022 __ 012.0638 __ 002,321 __ 44:Mahnomen_______07
0001.703 __ 0024 __ 007.0968 __ 003,410 __ 54:Norman_________08
0001.599 __ 0054 __ 002.9607 __ 009,356 __ 01:Aitkin_________09
0001.502 __ 0019 __ 007.9043 __ 005,225 __ 76:Swift__________10
0001.333 __ 0164 __ 000.8130 __ 018,081 __ 52:Nicollet_______11
0001.049 __ 0046 __ 002.2808 __ 014,337 __ 58:Pine __________12
0000.834 __ 0025 __ 003.3349 __ 004,168 __ 37:Lac_Qui_Parle__13
0000.731 __ 0033 __ 002.2155 __ 006,319 __ 12:Chippewa_______14
0000.620 __ 0017 __ 003.6456 __ 002,551 __ 35:Kittson________15
0000.301 __ 0014 __ 002.1505 __ 002,976 __ 06:Big_Stone______16
0000.000 __ 0000 __ -05.8510 __ 005,623 __ 32:Jackson________17
-000.038 __ 0009 __ -00.4171 __ 003,596 __ 26:Grant__________18
-000.267 __ 0012 __ -02.2266 __ 002,021 __ 78:Traverse_______19
-000.277 __ 0011 __ -02.5189 __ 005,558 __ 87:YellowMedicine_20
-000.383 __ 0013 __ -02.9439 __ 006,522 __ 61:Pope __________21
-000.361 __ 0022 __ -01.6425 __ 005,236 __ 83:Watonwan_______22
-001.039 __ 0017 __ -06.1113 __ 003,109 __ 41:Lincoln________23
-001.174 __ 0026 __ -04.5135 __ 008,109 __ 65:Renville_______24
-001.235 __ 0025 __ -04.9394 __ 005,608 __ 75:Stevens________25
-001.340 __ 0105 __ -01.2759 __ 011,129 __ 23:Fillmore_______26
-001.761 __ 0018 __ -09.7810 __ 004,795 __ 51:Murray_________27
-002.039 __ 0029 __ -07.0294 __ 014,980 __ 40:Le_Sueur_______28
-002.224 __ 0032 __ -06.9509 __ 002,158 __ 63:Red_Lake_______29
-002.544 __ 0026 __ -09.7840 __ 009,863 __ 81:Waseca_________30
-002.523 __ 0024 __ -10.5123 __ 006,012 __ 17:Cottonwood_____31
-002.894 __ 0069 __ -04.1949 __ 011,824 __ 79:Wabasha________32
-003.476 __ 0020 __ -17.3823 __ 007,709 __ 72:Sibley_________33
-003.760 __ 0031 __ -12.1283 __ 004,708 __ 45:Marshall_______34
-003.886 __ 0020 __ -19.4320 __ 007,148 __ 80:Wadena_________35
-004.094 __ 0045 __ -09.0974 __ 008,453 __ 33:Kanabec________36
-004.258 __ 0050 __ -08.5154 __ 008,103 __ 22:Faribault______37
-005.991 __ 0041 __ -14.6125 __ 012,503 __ 47:Meeker_________38
-006.474 __ 0053 __ -12.2148 __ 004,208 __ 15:Clearwater_____39
-006.958 __ 0089 __ -07.8181 __ 013,456 __ 48:Mille_Lacs_____40
-007.170 __ 0095 __ -07.5476 __ 008,718 __ 53:Nobles_________41
-007.813 __ 0051 __ -15.3191 __ 007,755 __ 64:Redwood________42
-008.390 __ 0088 __ -09.5337 __ 006,734 __ 57:Pennington_____43
-008.329 __ 0038 __ -21.9194 __ 003,376 __ 84:Wilkin_________44
-008.459 __ 0209 __ -04.0472 __ 010,674 __ 28:Houston________45
-009.199 __ 0064 __ -14.3731 __ 016,677 __ 49:Morrison_______46
-009.242 __ 0123 __ -07.5140 __ 025,672 __ 25:Goodhue________47
-009.511 __ 0068 __ -13.9863 __ 013,549 __ 08:Brown__________48
-009.532 __ 0064 __ -14.8942 __ 011,575 __ 29:Hubbard________49
-009.611 __ 0050 __ -19.2223 __ 002,263 __ 39:LakeoftheWoods_50
-010.383 __ 0093 __ -11.1641 __ 012,164 __ 77:Todd __________51
-010.442 __ 0094 __ -11.1084 __ 016,213 __ 11:Cass __________52
-011.534 __ 0112 __ -10.2984 __ 021,780 __ 34:Kandiyohi______53
-011.558 __ 0108 __ -10.7019 __ 019,548 __ 74:Steele_________54
-011.702 __ 0082 __ -14.2710 __ 010,672 __ 46:Martin_________55

-150.075 __ 03475 __ ----------- __ 0,519,318 __ Subtotal 55 "small impact" counties