[EL] Gerrymandering, the wasted votes / efficiency method, and geographic concentration
Daniel A. Smith
dasmith at ufl.edu
Sat Jun 17 14:44:04 PDT 2017
I've always been perplexed by the logic of the efficiency gap...
Nick and Eric provide an example of how the efficiency gap works in
their Chicago Law Review article.
They write (p. 851-852) (notes and italics omitted):
The efficiency gap, then, is simply the difference between the
parties’ respective wasted votes, divided by the total number of
votes cast in the election. Wasted votes include both “lost” votes
(those cast for a losing candidate) and “surplus votes” (those cast
for a winning candidate but in excess of what she needed to
prevail). Each party’s wasted votes are totaled, one sum is
subtracted from the other, and then, for the sake of comparability
across systems, this difference is divided by the total number of
votes cast. Figure 1 below shows how this calculation is carried out
for the hypothetical district plan discussed in the Introduction.
The bottom line is that there are 200 fewer wasted votes for Party A
than for Party B (out of 1000 total votes), resulting in an
efficiency gap of 20% in Party A’s favor.
Here's their Figure 1:
Let's leave aside for a moment some of the assumptions that go into
calculating the efficiency gap and the creation of their table (such as
no minor party candidates contesting the 10 seats, or that each seat
actually has two candidates on the ballot (as opposed to uncontested
seats that have to have imputed result totals)).
Let's instead fill in the table with what on the face seem to be very
competitive districts: Every seat is decided by just 2 percentage
points, but Party A wins each of the 10 districts, 51% to 49%.
Sure, Party B loses all the seats, but they all are competitive; each of
the seats could potentially flip to Party B in subsequent elections.
If I were Party B, or a candidate or supporter of Party B, I'd take
these districts any day of the week. Even if Party A strategically
gerrymandered them, each district is winnable given a wave election, the
right candidates, better messaging, higher (or lower) turnout, etc.
Yet, the efficiency gap in my hypothetical scenario is 48%, more than
twice that of Nick and Eric's example. Despite the overall
competitiveness of my example, it would be a clear target for litigation
according to the efficiency gap logic.
I guess my question to Eric and Nick and the broader listserv is, how
does the efficiency gap serve as a useful step in identifying a partisan
gerrymander given this obvious flaw?
dan
--
daniel a. smith, ph.d.
professor & university of florida research foundation professor
political science internship program coordinator
department of political science
303 anderson hall | phone: 352-273-2346
po box 117325 | fax: 352-392-8127
university of florida | email: dasmith at ufl.edu
gainesville, fl 32611-7325 | http://people.clas.ufl.edu/dasmith/dasmith/
https://twitter.com/electionsmith
On 6/16/2017 7:21 PM, Eric McGhee wrote:
>
> I invented the efficiency gap, yet I would still agree with Michael
> that it seems unlikely a ruling for the plaintiffs in Whitford would
> rely exclusively on that metric. (I would take issue with Michael’s
> claim that the efficiency gap is a “simple transformation” of existing
> measures under all conditions, but that’s another story.) It’s also
> worth noting that the legal standard Nick Stephanopoulos and I
> proposed (see http://chicagounbound.uchicago.edu/uclrev/vol82/iss2/4/)
> would still treat a plan’s observed efficiency gap as only one step in
> a process that would also allow a state to defend its plan as
> unavoidable for other reasons.
>
> Fortunately for the Whitford plaintiffs, every conceivable measure of
> bias or intent suggests the WI plan is a strong outlier. This includes
> a variety of tests that account for the state’s underlying political
> geography. To my mind, this makes the Whitford case less a test of a
> particular standard or measure, and more a test of whether the court
> really wants to get involved in partisan gerrymandering in the first
> place.
>
> Eric McGhee
>
> *From:*law-election-bounces at department-lists.uci.edu
> [mailto:law-election-bounces at department-lists.uci.edu] *On Behalf Of
> *Michael McDonald
> *Sent:* Wednesday, June 14, 2017 4:12 PM
> *To:* Election Law Listserv <law-election at uci.edu>
> *Subject:* Re: [EL] Gerrymandering, the wasted votes / efficiency
> method, and geographic concentration
>
> The efficiency gap is a simple transformation of prior seats to votes
> partisan bias (gerrymandering detection) measures. If one party wins
> more than 50% of the seats with 50% of the vote, all measures will
> indicate that the party is favored in a redistricting plan. For this
> reason, the efficiency gap is no more or less challenged than other
> partisan bias measures and I suspect Justice Kennedy will find it
> wanting for a bright line that identifies when a constitutional
> violation has occurred, as he has done with every other measure
> presented before him.
>
> Simulations are problematic for another reason. Redistricting is such
> a complex graph partitioning problem that enumeration of all feasible
> redistricting plans is impossible in finite time. As a consequence, we
> cannot know the properties of any simulation algorithm that does not
> generate all feasible plans with equal probability. The only algorithm
> that is guaranteed to create plans with equal probability is one that
> randomly assigns blocks to districts and rejects resultant plans than
> are not legal. (This is similar to random sampling from a survey
> perspective.) Unfortunately, this algorithm produces feasible plans
> with exceedingly low frequency as to make it impossible to use in
> finite time. All other proposed simulation algorithms have never been
> proven that they can randomly sample. Indeed, we have shown that two
> proposed algorithms fail random sampling on a small toy example (see:
> https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2583528). Failure
> to demonstrate the ability to randomly sample means that an algorithm
> could be biased in unknown ways, and might indicate the presence of a
> gerrymander when none exists, or fail to detect a gerrymander when one
> exists.
>
> The paper I link to describes the method of revealed preferences that
> the Florida and Wisconsin courts applied to determine a partisan
> gerrymander occurred. The method of revealed preferences simply looks
> at the adopted plans as they were developed and traces how partisan
> goals were elevated over traditional redistricting principles. In both
> cases, the courts waved legislative privilege to enable plaintiffs to
> see how the legislature incrementally traded off state constitutional
> or other traditional redistricting principles for partisan advantage
> through the course of generating test maps. In Florida, the efficiency
> gap was not put before the court in evidence, and simulations failed
> so spectacularly that the district court judge did not even bother to
> describe the evidence in his ruling. In Wisconsin, the court found a
> traditional partisan bias measure created by a consultant to the
> legislature was informative, as was the efficiency gap, in tracing out
> the legislature's preferences. I suspect that if Kennedy rules
> favorably for the Wisconsin plaintiffs, it will be on on the strength
> of the revealed preferences illuminated by applying both partisan bias
> measures to the test plans defendants produced in discovery, and not
> on the efficiency gap alone.
>
>
> ============
> Dr. Michael P. McDonald
> Associate Professor, University of Florida
> 352-273-2371
> www.electproject.org <http://www.electproject.org>
> @ElectProject
>
> On Wed, Jun 14, 2017 at 4:52 PM, Mark Scarberry
> <mark.scarberry at pepperdine.edu <mailto:mark.scarberry at pepperdine.edu>>
> wrote:
>
> My thanks to Bruce and to another list member who replied
> off-list. Has this point been made in any of the briefs?
>
> Mark
>
> Mark S. Scarberry
> Pepperdine University School of Law
>
> ------------------------------------------------------------------------
>
> *From:*Bruce E Cain <bcain at stanford.edu <mailto:bcain at stanford.edu>>
> *Sent:* Wednesday, June 14, 2017 1:43:49 PM
> *To:* Mark Scarberry; Election Law Listserv
> *Subject:* Re: [EL] Gerrymandering, the wasted votes / efficiency
> method, and geographic concentration
>
> Mark
>
> The efficiency gap is highly problematic for many reasons, and
> would be especially bogus for the reason you suggest if
> cross-sectional, over time data are used rather than simulations,
> as is the case in Whitford…see Wendy Tam Cho’s web page or email
> her at <wendycho at illinois.edu <mailto:wendycho at illinois.edu>> for
> several forthcoming publications on this topic…
>
> B
>
> *From: *Mark Scarberry <mark.scarberry at pepperdine.edu
> <mailto:mark.scarberry at pepperdine.edu>>
> *Date: *Wednesday, June 14, 2017 at 1:30 PM
> *To: *Election Law Listserv <law-election at uci.edu
> <mailto:law-election at uci.edu>>
> *Subject: *[EL] Gerrymandering, the wasted votes / efficiency
> method, and geographic concentration
>
> I'm sure someone must have made the following point (if it is right).
>
> The wasted votes / efficiency method for measuring partisan
> gerrymandering would seem to take "one person one vote" to the
> next level. It would benefit parties (currently the Democratic
> party) whose supporters are geographically concentrated.
>
> Is that observation correct? Cites would be appreciated.
>
> Mark
>
> Mark S. Scarberry
> Pepperdine University School of Law
>
>
> _______________________________________________
> Law-election mailing list
> Law-election at department-lists.uci.edu
> <mailto:Law-election at department-lists.uci.edu>
> http://department-lists.uci.edu/mailman/listinfo/law-election
>
>
>
> _______________________________________________
> Law-election mailing list
> Law-election at department-lists.uci.edu
> http://department-lists.uci.edu/mailman/listinfo/law-election
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://webshare.law.ucla.edu/Listservs/law-election/attachments/20170617/8bce6e8f/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: fekjibkodaaonkda.png
Type: image/png
Size: 60011 bytes
Desc: not available
URL: <http://webshare.law.ucla.edu/Listservs/law-election/attachments/20170617/8bce6e8f/attachment.png>
View list directory