Each year your election department in your county in coordination with the WA OSOS ("Office of Secretary of State") attempts to update your database. You can imagine why! Voters leave the county, move inside the county, die, return, register, re-register, update their address, etc. Unlike an update in your mail address, the incentive for updating your voter registration is not quite so pressing. Indeed, some polling research indicates an unfortunate percentage of the electorate nationwide think that by changing their postal address their voter registration is automatically updated.
County voter databases eventually become collections of registrants that no longer vote in your county. To remedy this, WA State has implemented ERIC (Electronic Registration Information Center) in conjunction with the PEW Center for the States. Seven other states have also implemented ERIC and other VRM (Voting Registration Modernization) measures. ERIC uses multiple nationwide databases (DOL, birth/death certificates, 'motor-voter', other states and county voter registration rolls) to provide a window into 'Ambiguous', 'In-County', 'Cross-County', and/or 'Mailing Address' changes. However, via a public disclosure request ('PDR'), I have reviewed the ERIC driven updates for Whatcom county and find the numbers a smaller part of the "churn" volumes I describe below. WA OSOS online voter registration, political party registration operations, and voter self-maintenance together probably contributed as much if not more than ERIC to the "churn" rate in the last ten months.
A Churn Rate of 34% for this General Election?
I was surprised when querying two separate Whatcom County Voter histories (from 12/04/12 and 10/07/13) to find so much "churn" in the voter rolls. The last ten months reveal a small net increase in registered voters (429). However, this net increase hides a "churn" of over 23,000 voters almost evenly divided between those RegistrationNumber(s) ('registrants') that left voter the database (7106) and those that joined (7535). Another component of "churn" is seen when I try to approximate those voters that simply changed precincts (8634). This figures gives:Total Churn Approximation (RAW) =
(7106 + 7535) + 8634 = 23275 registrants either deleted, modified or added
Some idea of the significance of this "churn" could be crudely approximated by dividing total vote return in any general election by total churn for the last year's voter database. For the last presidential election (105K) and this upcoming election (guessing 68K):
((7106 + 7535) + 8634) /105000
0.2216667
((7106 + 7535) + 8634) /68000
0.3422794
or 22 and 34 percent respectively.
We could assume voters carry their predilections with them when they physically relocate (Big Assumption!). If true, then my "churn rate" computations would be less dramatic (but still dramatic enough!):
(7106 + 7535) /105000
0.1394381
(7106 + 7535) /68000
0.2153088
or 14 and 21 percent respectively.
I find this amount of change in the voter database in such a short period of time (10 months) significant. The R Code and further explanation for this post is here.
We could assume voters carry their predilections with them when they physically relocate (Big Assumption!). If true, then my "churn rate" computations would be less dramatic (but still dramatic enough!):
(7106 + 7535) /105000
0.1394381
(7106 + 7535) /68000
0.2153088
or 14 and 21 percent respectively.
I find this amount of change in the voter database in such a short period of time (10 months) significant. The R Code and further explanation for this post is here.