Just The Numbers: Behind Sanders-Wiggins 2010: Monday: Projected Vote Totals

Monday, November 8, 2010

Monday: Projected Vote Totals

Nineteen counties will report updated vote totals, with the bulk of the new numbers coming betwee 4:00-5:00pm (Whatcom reports at 5:30, Spokane at 6:00). And yes, I'm including Grant County.

Before we get to projecting new totals for today, let's start with some basic assumptions:

1. Most of the nineteen counties will report the same number of net new ballots today as they did on Friday. However, when that would put a county near or over the amount of EBOH -- or if the county did not report on Friday -- we will assume that the county will only count 80% of it's remaining ballots on hand. Yes, this is a guess, but it's based on the concept that there will always be a pile of ballots that require hand counting, visual inspection, re-completion (if voter intent is discernable), or some other manual handling. Also, there will undoubtedly be ballots that will be shelved for signature mis-matches, voter cast two ballots (duplication), or a host of other reasons. Because these ballots are set aside -- but are possibly still factored into the EBOH count, depending on each county's policies -- this count dwindles downward very slowly while the ready-to-count ballots are quickly zero'd out. Think of it as a curve approaching zero but taking a really long time to get there.

2. The overall county average thus far for SMOV and VCR will be used for each county UNLESS the county reported data on Friday or Saturday AND the Friday/Saturday totals contained at least 500 ballots and at least 10% of the total ballots cast in the race. These qualifiers are added to ensure we are using numbers of a statististically significant sample size. [Note: When I use overall SMOV/VCR percentages, I will indicate it with a (T); when I use Friday or Saturday SMOV/VCR data, I will indicate it with a (F) or (S)]

On to the numbers.

Rather than try and type this all out, I took a screenshot of my projections for today and I'm attaching it to this post. That way, you all can see the numbers I used (and challenge any assumptions I made accordingly).

[EDIT: I accidentally left Clallam county off the original chart above (sorry Clallam!) I've updated the graph and changed the numbers in the below paragraph to reflect the +191 net votes that Clallam is expected to provide to Wiggins today.]
It's very hard to predict how many ballots will actually be counted today, but using the estimates above, I'm predicting that we'll see another 175,052 ballots counted and 128,316 votes cast in the race. I expect Wiggins to pick up a net of 8,470 votes, reducing Sanders's lead to 4,781 votes total.

Stay tuned. I'll be posting frequently as individual counties start releasing updates at 4pm.

2 comments:

MichaelpNovember 8, 2010 at 2:02 PM
Your chart is so superior to mine. I'm a little embarrassed.

BUT - My models project a Sanders lead at the end of the day of 4-6k, so we are, so far, in a similar position.

Baited breath, my worthy opponent, baited breath.
ReplyDelete
Replies
MichaelpNovember 8, 2010 at 4:39 PM
KC numbers created a drastic hit.

Sanders needed to be about 6k up after KC drop (then 8k at the end of the day) to pull it off.

There is now no statistical model that gives Sanders a win.
ReplyDelete
Replies

Add comment

Terminology

The Terminology post is quickly working its way down the page, so here it is, reproduced at the top, for all time. A quick introduction to the terminology I use on this blog:

SMOV: Sanders Margin of Victory (%). Within a defined universe of ballots (say, one particular county's ballots on one particular day), it refers to the percentage of ballots cast for Sanders. [Ed. Note: I'm not expressing an ideological preference: as of this posting, Justice Sanders is leading the race; and using the leading candidate makes the math easier.] Accordingly, a SMOV of 51% means that if 100 ballots were cast, Sanders received 51 and Wiggins received 49.

VCR: Votes Cast in Race (# or %). Not every ballot contains a vote for Sanders or Wiggins. Indeed, on average so far, just under 77% of all ballots cast have contained a vote for either candidate. Some voters may have chosen to write in a candidate -- it's even possible that they may have chosen to write in Sanders or Wiggins, which creates some interesting possibilities (which may be explored in a later blog post). But largely, it's because people chose not to vote in this particular race. They may have been single-issue voters (i.e. the U.S. Senate race), or they may have only cared about the initiatives; I have several friends who only cared about I-1100 and I-1105. Or they may have just stopped at the first page and decided to go do something else. Either way, the VCR% plays a big role in our analysis.

EBOH: Estimated Ballots on Hand (#). This refers to the number of ballots that each county is estimating still remains to be counted. EBOH x VCR% would yield the total number of votes that still remain to be allocated between the two candidates. Note that EBOH is not a static number. Each day in which votes are counted should result in a net decreate in the EBOH total, as votes from that category are counted and allocated to the candidates. However, some ballots may be disqualified for any number of reasons; further, disqualified ballots may be later added back to the EBOH pile if they are later verified by affidavit. Furthermore, ballots may still be arriving, which would increase the EBOH total. Although most of these ballots will be disqualified for being postmarked after election day, some may be from overseas military voters or may be correctly postmarked and will be counted.

SAND or WIGG: I sometimes refer to the candidates as SAND (Sanders) or WIGG (Wiggins) for short. I intend no disrespect.

KPS: A reference to King, Pierce, and Snohomish counties, the three largest counties in Washington in terms of registered voters. There are 3,601,152 registered voters (as of Election Day 2010) in Washington; King has 1,069,791 (29.71%), Pierce has 410,081 (11.39%); and Snohomish has 377,739 (10.49%), for a total of 51.58% of all registered voters in Washington.

KPSS: Same as KPS, but add Spokane county; it has 261, 250 (7.25%). KPSS together has 58.84% of the total registered voters in Washington.

KPSSCTKW, or THE BIG EIGHT: Yes, this probably seems a little ridiculous, but stay with me. Same as KPSS, but now add the counties of Clark (219,616, 6.10%), Thurston (149,024, 4.14%), Kitsap (143,796, 3.99%), and Whatcom (116,576, 3.24%). The Big Eight are the counties with over 100,000 registered voters each, and together, they comprise 76.31% of the total registered voters in Washington. [Ed. Note: Yakima is very close to joining this club at 99,568; sorry Yakimanians, I had to draw the line somewhere. Go sign up 432 more voters, and you're in! (next time)]