2016 Postmortem
Showing Original Post only (View all)Fun with California ballot numbers [View all]
I've been fiddling with numbers from CA to examine scenarios that could result is a Sanders win. I'm reluctant to post this because I know I'll get flamed, but I think it will be interesting to some people, so here it is.
_________________________________________________________________________
In California, the ballots that remain to be processed are mainly from people who sent their mail-in ballot very late. Older people tend to vote early, so they are over-represented in the counted votes. Younger people tend to do things at the last minute. Their votes will be over-represented in unprocessed ballots.
One scenario that would result in a Sanders' win would be if about 67% of the remaining ballots were from voters age 18 to 34 (or another demographic that skews as heavily to Sanders).
Perhaps that's outside the realm of possibility, but "just for fun," I've provided background information and calculations below.
According to the Capitol Weekly / Open California survey, voters age 18-34 broke for Sanders 78% to 22%.
The split for 35 and up can't be calculated based on the charts in the survey, but looking at the breakdown for 35-44 (Clinton 45%), 45-54 (Clinton 55%), 55-64 (Clinton 57%) and 65 and up (Clinton 65%), a 40/60 Sanders/Clinton split for voters age 35 and up wouldn't be too far off. That's what's used in this calculation.
According to the AP approximately 8.9 million voted. According to the CA SOS unprocessed ballot report, there are 2,423,607 unprocessed ballots. To estimate the number of ballots processed, subtract the unprocessed ballots from 8.9 million. The result is 6,476,393. Based on results published on the CA SOS site the number of ballots counted so far (all party primaries) is 5,629,170. If 6,476,393 were processed, and 5,629,170 counted, the rejection rate is 13%. For purposes of this calculation, a rejection rate of 15% is used (assuming the rate for unprocessed will be higher than processed).
The unprocessed ballot report includes the last time each county updated it's report. The last update for many were on 6/7, 6/8, or 6/9. This goes for some big counties. For example, the last report for San Diego, Orange, Alameda, Riverside, and Sacramento was 6/8. So, the unprocessed ballot total of 2,423,607 doesn't include mail-in ballots received between report date and deadline for receipt (6/10).
Assuming approx 3% of voters sent their ballots on 6/6 or 6/7 (to arrive 6/9 or 6/10) there would be about 250,000 late arrivals that have yet to be reported.
As reported here, 3,817,713 of the ballots counted so far are Democratic primary ballots. The total counted in all primaries is 5,629,170 (http://vote.sos.ca.gov). Therefore, 67.8% of all ballots counted were Democratic primary ballots. In the calculations below it's assumed that the same percentage of unprocessed ballots will be Democratic primary ballots.
Here are the numbers that go into the calculation. For anyone who feels like driving themselves nuts fiddling with numbers too, I've assigned variables and provided a "step-by-step" guide to make it easy to recalculate using updated reports or adjustments to the assumptions.
If you identify errors or neglected factors, let me know.
The Numbers:
C = Counted for Clinton
C = 2,128,194 a/o 6/10 report
S = Counted for Sanders
S = 1,653,416 a/o 6/10 report
99% of the Democratic primary ballots counted are allocated to Clinton and Sanders (the other 1% is distributed among other candidates)
U = Estimated unprocessed (2,423,607 a/o 6/10) plus an estimated 250,000 that arrived between last county reports and 6/10 deadline.
U = 2,673,607
M = How many votes you want Sanders to win by
M = 10,000
The Steps
Step 1. Estimate how many unprocessed are Dem primary ballots (P)
P = .687U
P = 1,836,768
Step 2. Estimate how many unprocessed Dem primary ballots will be valid (V)
V = P - .15P
V = 1,561,253
Step 3. Estimate how many valid Dem primary ballots will be allocated to Clinton and Sanders (X)
X = .99V
X = 1,545,640
Step 4. Calculate Sanders target counted ballot total for win by M votes (T)
T = (C + S + X)/2 + M/2
Using M = 10,000
T = (2,128,194 + 1,653,416 + 1,545,640)2 + 5000
T = 2,668,625
Step 5. Number of votes Sanders needs to reach target (Y)
Y = T - S
Y = 2,668,625 - 1,653,416
Y = 1,015,209
Step 6. Calculate number of valid Dem primary unprocessed that would need to come from 18 - 34 year olds (W) to give Sanders the win.
Y = .78W + .4(X-W)
1,015,209 = .76W + .4(1,545,640) - .4W
1,015,209 - 618,256 = .38W
W = 1,044,613
Step 7. Determine what percent W represents in relation to V.
W/V = Percent of valid Dem ballots that need to come from 18-34 year olds for Sanders win.
1,044,613/1,561,253 x 100 = 67%