(cross posted at VoteForAmerica.net: Part 1, Part 2, Part 3)
NBC is now reporting Hillary Clinton will be named SOS after thanksgiving.
----------------------------------------------------------------------------------------------------
--------------------
Recount Regression: Day 2
----------------------------------------------------------------------------------------------------
--------------------
Original Recount
Coleman: 534475 - 534687 = -212
Franken: 494804 - 494930 = -126
Franken Net Today: 43 (exactly the same as yesterday)
Total Franken Net: 86
Franken Deficit: 129
Day 2 Complete. If these graphs make no sense, read the audit results (below) and then day 1 (further below) for additional context.
The dotted lines represent the regression result from Day 1:
The dotted lines represent the regression result from Day 1:
Precincts with 0-3494 Votes:
Coleman Gain: -514.680649876274
Franken Gain: -295.310532175405
Franken Net: 219.37011770087
All Precincts Extrapolation Results:
Coleman Gain: -539.655665129103
Franken Gain: -311.596023329002
Franken Net: 228.059641800101
----------------------------------------------------------------------------------------------------
--------------------
Audit Regression
----------------------------------------------------------------------------------------------------
--------------------
Before I get into the recount, I want to provide some additional information regarding the audit. If you don't care about the audit, you can skip to the second portion of the article for just recount analysis.
The following analysis uses data provided by Prof. Jeffrey B. Lewis of UCLA who coauthored Residual Votes in the 2008 Minnesota Senate Race. My analysis is entirely independent and not in anyway shape or form associated with the aforementioned individuals or institutions. The data contains an exhaustive compilation of the revised Post Election Audit Review published by the Office of the Secretary of State in Minnesota.
I parsed the data and created the graphs shown above. The three graphs immediately preceding this paragraph depict the number of votes gained or lost with respect to the total votes recorded for each precinct included in the audit. The dots represent the change between the machine counted votes and hand recorded votes surveyed in the audit. A dot located at (443, 2) represents a gain of 2 votes by the given candidate in a precinct with 443 total votes. Each approximation line uses a least-squares nonlinear regression for arbitrary fit functions as provided by Mathcad's genfit() function.
You'll notice that the graphs range from 0 to 2,645 votes; this is because the largest precinct surveyed in the audit contained just 2,645 votes.
The following functions correspond to the resulting regression's interpolation. The input x represents the number of votes in a given precinct. Although it appears as though the first two terms cancel, the truncated three decimal numbers are in actuality much more precise:
The following graph illustrates the best fit regression functions, developed above, extrapolated out to include the vote totals of all 4,130 precincts statewide. As of right now (12:55 AM CT 11/20) the Otstego precinct of Wright County currently leads the state in total votes with 6,621.
I plan on rehashing this analysis with the current results of the recount within the immediate future. Today Franken gained 43 votes as a result of today's recount over the certified returns published on November 18th:
Nov. 18 Recount Change
Coleman (R) 195,708 195,638 -70
Franken (D) 180,950 180,923 -27
Franken Net Gain: 43
Franken made up much of his ground in St. Louis County, as I have previously suggested, by gaining 40 votes and losing 16. St. Louis County has recounted just 77 of 232 precincts (33.18%) and has contributed to a 23 vote decrease in Coleman's lead. Overal, Franken gained votes in 50 precincts, and lost votes in 79 while Coleman gained votes in 40 precincts and lost in 92.
While today's result appears to be good news for Team Franken, the result of a Ramsey County lawsuit may be even better. The Franken Campaign sued the St. Paul county for the names of voters whose absentee ballots had been rejected. The presiding judge issued this statement: "With each passing hour, the Franken campaign is irreparably harmed in its efforts to ensure that each valid vote is properly counted." Franken's attorneys said the ruling strengthens their case in other counties and with the canvassing board over the ongoing issue of "wrongfully" invalidated absentee ballots.
----------------------------------------------------------------------------------------------------
--------------------
Recount Regression: Day 1
----------------------------------------------------------------------------------------------------
--------------------
Using the pretense of information contained above I've incorporated the initial recount returns into the regression mechanism.
For a more complete explanation of the process you can read the preceding text, if you have not already done so. The following results will still make sense, but if you're interested in the math, I tried to explain it in greater detail above.
The following two graphs depict the number of votes gained or lost with respect to the total votes recorded for each precinct that has completed their recount. Each dot represents the change between the originally certified result and the post recount tally in a given precinct with x number of total votes:
And the current functions used in conjunction with the regression:
Now again using those functions, the following two graphs further illustrate the regression interpolation. The graph below illustrates the regression on precincts whose vote totals are less than 2,668; the largest precinct that has completed a recount.
The graph below goes further and fits the regression onto all 4,130 precincts statewide; this forces 214,813 additional votes, that lie in precincts that surpass the 2,668 vote threshold, into the predetermined trend.
Using the previous two graphs, and the functions they represent, a projection can be made for the cases covered. The first case simply includes precincts with 2,668 or less, while precincts above and beyond that figure are entirely ignored. For each precinct, the total vote total is taken and applied to the listed function for each candidate. The result is then added to that candidates sum, and the next precinct is calculated. This process is done using precinct results from the final certification.
Coleman Gain: -407.600629299817
Franken Gain: -194.977755790799
Franken Net: 212.622873509018
Franken nearly makes up the pre-recount deficit of 215 votes, while 214,813 votes are entirely ignored. If those additional votes are applied to the process, Franken takes a decisive lead:
Coleman Gain: -540.336882632741
Franken Gain: -237.764374574729
Franken Net: 302.572508058013
As of right now (4:16 AM CT 11/20), after 15.49% of all votes have been recounted, Franken stands poised to become Minnesota's next Senator. Later today, after the second day of recounting is over, and the officially unofficial results are posted at the SOS site, I'll post another article with the new data included in the analysis. As each day goes by, the graphs, and projections should begin to converge. If Franken gains another 40 votes throughout the upcoming day, Team Coleman is definitely in trouble. The only way to find out is by counting the votes.
I plan on updating these graphs at the end of each recount day; SOS posts results at 8 PM CT. I will post my results until the conclusion of the process. Also if you have any questions ask them in the tipjar, and I'll try and answer them in the comments.
Update: w0ot! Rec-List. Keep it here and I'll do an update with the day 2 recount results at 8 PM CT after the SOS releases today's data.