Professional Documents
Culture Documents
SSRN Id3027578
SSRN Id3027578
RAY J. WALLIN
Ray J Wallin has been active in local politics in Saint Paul and Minneapolis, MN,
writing and providing research to local campaigns. He has an engineering degree and an
MBA, and initially became interested in the efficiency gap when reading an article
concerning the Wisconsin case in MinnPost, a local online newspaper. He noticed flaws
in the efficiency gap method that did not reconcile and could not help but pull at the
threads.
TABLE OF CONTENTS
INTRODUCTION .................................................................................................... 3
CALCULATING THE EFFICIENCY GAP USING TABLES ......................................... 4
MANIPULATING DISTRICTS ................................................................................. 6
THE CHARISMATIC POLITICIAN MODEL ........................................................... 9
THE EFFICIENCY GAP EQUATION ...................................................................... 11
AN EFFICIENCY GAP EQUATION EQUIVALENT TO THE TABLES ....................... 12
THE EQUAL DISTRICTS EFFICIENCY GAP .......................................................... 15
A MINI-STATE LIGHTS THE WAY ...................................................................... 16
CONCLUSION ..................................................................................................... 18
INTRODUCTION
In the fall of 2017, the US Supreme Court will hear arguments in the Gill v
Whitford case to decide whether parts of the Wisconsin States redistricting
plan of 2011 rise above a threshold to be considered gerrymandered. In the
opinion of the lower court, the term Efficiency Gap, EG, is referenced over
two-hundred fifty times1 as the term has framed many facets of the trial. The
purpose of the EG is to calculate whether an elections outcome is fair, a term
also referenced in the opinion.
The efficiency gap is found by summing a states wasted votes district by
district and then dividing by the total statewide votes. The gap measures the
distance between the actual number of seats won by a party and the fair
number of seats the party should have won given the statewide vote count.
One can also look at the EG as measuring how efficiently a party converts its
votes to elected seats, or how few wasted votes a party accumulates statewide.
This paper has two parts. The first half attempts to understand the EG, its
strengths, its limitations, and whether it performs as a bellwether of
gerrymandered districting plans. The analysis is done with the help of two
tables presented to the district court2 in a report by Professor Ken Mayer, an
expert testifying for the plaintiffs in the Gill v Whitford trial. Table 8 from his
report, reproduced in this paper as Table 1, calculates the EG for Wisconsins
Act 43, a redistricting plan enacted by the Wisconsin state legislature in 2011.
Act 43 has a high EG, which, according to the theory, indicates
gerrymandering. Mayer then alters Wisconsins district lines to establish a more
fair districting plan which he calls his demonstration plan. Mayers
demonstration plan is Table 7 in his report and is reproduced in this paper as
Table 2.
The second part of this paper looks at the EG equation, how it is derived, and
shows that wasted votes introduce bias and should not be used. Along the way,
two new EG equations are derived, including the equal districts efficiency gap
equation, the correct equation to use when calculating an efficiency gap.
Finally, a mini-state example brings the thoughts of this paper together.
The main takeaway from this paper should have less to do with the EG methods
than the assumptions made when deriving them. In the United States, each
citizen is guaranteed the right to vote and each vote is treated individually. But
a gerrymandering analysis is a different animal as individual districts (in this
1
Gill vs. Whitford Complaint and Exhibits: US District Court for the Western District
of Wisconsin, Case: 3:15-cv-00421-bbc Document #: 166 Filed: 11/21/16
2
Analysis of the efficiency gaps of Wisconsin's Current Legislative District Plan and
Plaintiffs' Demonstration Plan, contains Mayers Tables 7 and 8: (some redundant
columns in Mayers table have not been reproduced) Gill vs. Whitford Complaint and
Exhibits: US District Court for the Western District of Wisconsin, Case: 3:15-cv-
00421-bbc Document #: 1-2 Filed: 07/08/15
4 Equal Districts Efficiency Gap
case) make up a state. It is the district, not the individual vote that needs to stay
whole. Therefore, common statements such as Party X won 55% of the
statewide vote but only 45% of the seats, are close, but biased.
Table 1 calculates the EG for Wisconsins Act 433 by summing the wasted
votes in the 99 Wisconsin Assembly races of the 2012 election. In the table,
Democrats receive 51.138% of the statewide vote but elect Assembly members
to only 42, 42.424%, of the 99 seats. According to the EG, if a party wins over
50% of the statewide vote, fairness dictates that the party should win over
50% of the Assembly seats. This clearly did not happen under Act 43.
Democrats were inefficient at converting their votes to wins and wasted a
good share of their votes along the way. As a result, Table 1 shows an EG (total
wasted votes as a percentage of statewide votes) of 11.690% favoring
Republicans.
An EG above +7% or below -7% is considered suspect,4 so the gap of 11.690%
in Table 1 shows that the state of Wisconsin under Act 43 was very likely
gerrymandered according to the theory. To show how Wisconsin can be
redistricted fairly under federal and state requirements, Mayer introduces a
demonstration plan. Table 2 shows this demonstration plan. Democrats still
have 51.145% of the vote, very near what they did in Table 1, but they now
have 51 Assembly seats, or 51.515% of the total seats, a much fairer result.
This fairness is supported by far fewer total wasted votes and an EG of 2.195%,
still favoring Republicans, but well within the +/- 7% margin.
3
https://docs.legis.wisconsin.gov/2011/related/acts/43
4
Gill vs. Whitford Complaint and Exhibits: US District Court for the Western District
of Wisconsin, Case: 3:15-cv-00421-bbc Document #: 166 Filed: 11/21/16
5 Equal Districts Efficiency Gap
Partial Table 1 . Mayer's Efficiency Gap Calculation for Wisconsin's Act 43 (see Appendix 2 for full table)
Total Total
Democratic Democratic Efficiency
Vote: Seats: Gap:
51.138% 42 11.690%
Total Total
Democratic Democratic Efficiency
Vote: Seats Gap:
51.145% 51 2.195%
6 Equal Districts Efficiency Gap
This paper will use Democratic numbers to compute the EG though either party
can be used. Following Professor Mayers lead, this paper uses a positive EG to
represent a Republican advantage and a negative EG to represent a Democratic
advantage. For brevitys sake, the tables presented within the body of this paper
show representative districts 19-24 instead of all 99 districts in the tables. The
complete tables are included in Appendix 2.
There may be some confusion with Table 1, as it calculates a 42-57 seat
makeup while in the actual 2012 Assembly election there was a 39-60 seat
makeup. Mayer factored in incumbency into the Democratic and Republican
district votes, and since we are only looking at the differences between Tables 1
and 2, as he states, the substantive inferences are identical.5
The low EG calculated in Table 2 seems to indicate a fair election, but the EG
measures wasted votes not the fairness of how district lines are drawn. In fact,
the makeup of the individual districts barely comes into play in such
calculations and this makeup can be manipulated if simple rules are met.
MANIPULATING DISTRICTS6
In Table 2, the slightly over 50% Democratic seat share aligns with the slightly
over 50% Democratic vote share. This seems fair and is confirmed by the low
EG. Table 3 manipulates the vote makeup of the districts to get lopsided
districts that are either nearly 100% Democratic or 100% Republican. Table 4
manipulates the districts to where each race is close to a 50-50 margin. The EG
remains at 2.195% even though both scenarios are unnatural.
Say a redistricter wants a plan that produces lopsided, packed, districts. As long
as the seats-votes criteria is met, that is, the percentage of statewide seats won
by each party and the percentage of statewide votes won by each party remains
constant, the EG will remain constant. This is demonstrated in Table 3 where
all districts are packed but the EG does not change.
Between Tables 2 and 3, there is no difference between the total number of
votes per district, the total wasted votes, the total lost Republican votes, the
total lost Democratic votes, the seat count, or the EG. Each district has retained
its representative, so no seat-flipping has occurred, and the total number of
wasted votes is the same. But each district in Table 3 has been reconfigured to
be so lopsided that each candidate either wins or loses by nearly a 100%
margin. Under this redistricting plan, seats will flip only under the most
5
Rebuttal Report: Response to Expert Reports of Sean Trende and Nicholas Goedert.
Gill vs. Whitford Complaint and Exhibits: US District Court for the Western District of
Wisconsin, Case: 3:15-cv-00421-bbc Document #: 64 Filed: 01/25/16
6
Wendy K. Tam Cho recently published a similar analysis: Tam Cho, Wendy K.
(2017) "Measuring Partisan Fairness: How Well Does the Efficiency Gap Guard
Against Sophisticated as well as Simple-Minded Modes of Partisan Discrimination?,"
University of Pennsylvania Law Review Online: Vol. 166 : Iss. 1 , Article 2.
7 Equal Districts Efficiency Gap
extreme circumstances. Surely packing is involved. Surely this state has been
gerrymandered. But according to the EG of 2.195% the state has been
redistricted in a very fair manner.
Citizens may complain that the lopsided districts of Table 3 are not practical,
that they want to see close, competitive districts. By keeping the seat and vote
shares constant, the redistricter can again move votes around to get evened out
districts, where each race is competitive. The result is shown in Table 4.
Partial Table 3. Efficiency Gap Calculation for Demonstration Plan with Votes Shifted to Create Packed Districts
Total Total
Democratic Democratic Efficiency
Vote: Seats Gap:
51.145% 51 2.195%
8 Equal Districts Efficiency Gap
Partial Table 4. Efficiency Gap Calculation for Demonstration Plan with Votes Shifted to Create 50-50 Districts
Total Total
Democratic Democratic Efficiency
Vote: Seats Gap:
51.145% 51 2.195%
Table 4 shows a state that has been redistricted so each race is very close to 50-
50. Compared to Table 2, the demonstration plan, we have the same number of
wasted votes, votes per district, lost Republican votes, lost Democratic votes,
the same election results, and the same EG. No districts have been flipped and
the total number of wasted votes remains the same. Since this plan produces a
tight race in each district, its makeup is surely fair. The EG of 2.195% agrees.
We have produced a very fair state.
But what happens next cycle? What if there is a slight shift and Democrats pick
up votes across the state? According to Table 4, if Democrats pick up 1
percentage point in each district, they will win every race in the state, a gain of
51 seats and the new EG will calculate close to -50%, its maximum, far beyond
the -7% threshold. How can a measurement tool jump from near zero to its
maximum with such a small change in input?
Of course, all districts will never be packed with either 100% Democratic or
Republican votes. But some could, and when they are, the EG will not measure
it. All districts will never be evened out to produce 50-50 races, but if they are,
the EG will not measure it. Between these two extremes, packed districts and
evened districts, the redistricter can manipulate the votes as long as the seats-
votes criteria is met. One must ask what exactly the EG is measuring. It is not
measuring gerrymandering.
9 Equal Districts Efficiency Gap
8
(Positive favors Republicans)
Efficiency Gap Percentage
0
48
49
50
51
52
53
-2
Percentage of Statewide Democratic Vote
10 Equal Districts Efficiency Gap
In Figure 1, one by one, votes are flipped until that critical vote is cast and a
seat flips, causing the EG to jump. Each jump is from one vote, in one district,
in a sea of tens of thousands. No theory should be so sensitive as to produce a
one percent jump from one vote.
When several seats flip in a narrow vote share range, the EG can jump by even
larger margins. The EG in Figure 1 is 9.7% at a 49.0% share and 2.3% at a
49.7% share, a jump of 7.4 percentage points from a change in input of only 0.7
percentage points. If a charismatic governor or president caused the
Democratic vote share to shift those 0.7 percentage points from one election to
the next, it could produce that 7.4 point jump in the EG, larger than the 7%
threshold offered as a cutoff for gerrymandering. A state can be deemed well
under the 7% threshold one election cycle and well above the next. The
question becomes, where do we presume the EG is valid? Is it at 2.3%, or is it
at 9.7%? Again, what is the EG actually measuring? It cannot be measuring
gerrymandering with such swings in output.
A smooth method of measurement is possible. Figure 2 uses the same input
data as Figure 1 but shows a result of this authors forthcoming paper that
addresses a fundamental flaw in the EG theory (not included in and not
necessary for this paper) by using Gaussian distributions as the basis of
analysis. The results are similar but there are no jumps or discontinuities in the
Figure 2 results.
8
Gap Percentage
0
48
49
50
51
52
53
When all districts have an equal number of votes, the EG table calculations
reduce to a simple equation given by the authors of the EG7 and presented by
Mayer to the District Court for the Western District of Wisconsin.8
Considering the negative sign at the beginning of Equation 1, some authors use
Democratic wasted votes minus Republican wasted votes to reach total wasted
votes. Others use Republican wasted votes minus Democratic wasted votes.
Only the sign of Equation 1 changes between the two methods. The sign of
Equation 1 matches results from the tables.
With 99 seats in the Wisconsin Assembly, each seat contributes 1/99,
approximately 1 percentage point to the percentage of seats in Equation 1. This
corresponds to the one percentage point jump in Figure 1 each time a seat flips.
Between jumps in Figure 1, the slope of the line is 2, from the 2x relationship
between seats and votes in Equation 1.
In the Wisconsin Assembly, a flipped seat results in a one percentage point
jump but the EG has been used to calculate bias in congressional redistricting
plans where a state may have only a half-dozen to a dozen districts. Wisconsin,
for example, has 8 congressional districts and each seat flip will cause the EG
to jump by 1/8, or 12.5 percentage points, far more than the 7% threshold.
Equation 1 also explains the majority of the 8.5 percentage point drop in the EG
from Table 1 to Table 2. Democrats won 8 more seats in Table 2 than in Table
1. This 8.1 percentage point seat difference in Equation 1 moves the efficiency
gap by 8.1 percentage points, nearly all of the 8.5 percentage point drop
between the two plans. Any effort put into re-drawing district lines to
compensate for cracking, packing, or fairness had little effect on the EG in
Table 2 as the EG calculations saw only the seats flipping.
7
Eric McGhee and Nicholas Stephanopoulos. Partisan gerrymandering and the
efficiency gap. 82 University of Chicago Law Review, 831, 2015. 70 pages. U of
Chicago, Public Law working Paper No. 493. Available at SSRN:
https://ssrn.com/abstract=2457468.
8
Analysis of the Efficiency Gaps of Wisconsin's Current Legislative District Plan and
Plaintiffs' Demonstration Plan. Gill vs. Whitford Complaint and Exhibits: US District
Court for the Western District of Wisconsin, Case: 3:15-cv-00421-bbc Document #: 1-2
Filed: 07/08/15
12 Equal Districts Efficiency Gap
In a gubernatorial election, it does not matter if one district has more votes than
another, what part of the state votes come from, or if a Democrat or Republican
wins an Assembly seat in this district or that. All that matters is the final
statewide vote percentages. In an Assembly election, it does not matter how
many voters show up in a district or who wins the governorship. All that
matters is which candidate has more votes in that district. These may be simple
statements but they contain subtleties overlooked in many analyses.
The fundamental question when forming a basis for analyzing gerrymandering
is, are seats or votes being measured? Seats are measured by district and votes
are independent of district, which means both seats and votes cannot be used in
the same equation as they have different bases. Seats and votes are used in
Equation 1, so it is valid only when all districts have the same number of votes,
that is, when all districts are equivalent, just as all seats are equivalent.
Table calculations use total wasted votes which are calculated differently
depending on which party wins a district. When these calculations are
combined and added statewide, the result keeps both seats and votes
information. Appendix 1 begins with the general formulas for wasted votes
used in the tables, combines them into total wasted votes, and hones them down
to a simple equation, our second EG equation.
Equation 2 is equivalent to the wasted votes calculations in the tables, and the
only difference from Equation 1 is that seats has been replaced with weighted
seats. In Table 2, Democrats won 51 districts and each seat has a value of 1, so
the Democrats won 51 seats. But, in Equation 2, if one district has 50% more
votes than the statewide district average, its seat has a value of 1.5 as it
represents more votes. A seat no longer represents a district, it represents the
total votes in that district, so both terms in Equation 2 come from the same
basis. In Table 2, Democratic-won districts have 2.756% fewer votes than
average, diluting the power of their seats, making them less efficient, and when
their seats are weighted, we find there are 2.756% fewer seats than the 51 actual
seats, which is 49.594 seats, or 50.095% of the 99 total seats. Equation 2
becomes,
2.195% = ( 50.095% 50% ) 2 ( 51.145% 50% ) ,
weighted seats also show why the jumps in Figure 1 are non-uniform as they
mimic their corresponding district size.
One can now understand the importance of finding representative numbers for
uncontested districts, both as a percentage and to the extent that all votes
contribute to the district total which is the weight of its seat in Equation 2.
In the same way that seats are weighted in Equation 2, wasted votes are
weighted by the district vote total. Consider that a district can contribute a
maximum of half its votes to the states total wasted votes. If one district has
50% more votes than average, its number of possible wasted votes is 1.5 times
the average and that district has more influence over the statewide wasted votes.
Both the weighted seats of Equation 2 and the wasted votes of the tables
unfairly bias the EG toward larger districts, skewing the results. And in the
demonstration plan, where more voters turn out in Republican districts, the
results are skewed unfairly against Republicans.
To exploit this disparity, in Table 5 we decrease both the Republican and
Democratic votes by 20% in Democratic-won districts and increase both the
Republican and Democratic votes by 20% in Republican-won districts. The
Democratic vote share has dropped below 50% but Democrats still hold more
than 50% of the seats. This is not fair and the EG should reflect this with a
negative value, showing the plan favors Democrats. But the EG in Table 5 has
actually jumped in the positive direction to 7.73%, above the 7% threshold,
indicating the state surely has been gerrymandered to favor Republicans. This
result is terribly wrong, showing again how wasted votes can be skewed.
Expressing this in terms of Equation 2, the total weighted seats in this example
comes in at 39.8 seats, far less than the 51 seats actually won.
14 Equal Districts Efficiency Gap
Partial Table 5. Efficiency Gap Calculation for Demonstration Plan with Votes Shifted to Give Lower Turnout in
Democratic-Won Districts
Total Total
Democratic Democratic Efficiency
Vote: Seats Gap:
48.904% 51 7.730%
Equal Districts Efficiency Gap = - [ ( Percentage of Seats - 50% ) + 2 ( Average Percentage of Vote - 50% ) ]
Equation 3: The Equal Districts Efficiency Gap with Average Percentage of Vote
The average percentage of vote in Equation 3 is the average of the final column
in the tables, the district percentage of votes, which is not influenced by district
size. From Table 2, the demonstration plan yields a 51.781% average district
vote, and Equation 3 gives us,
2.046% = ( 51.515% 50% ) 2 ( 51.781% 50% ) .
The equal districts efficiency gap accurately captures the gap between election
results and what is considered fair. A key to understanding the fundamentals of
this or any gerrymandering analysis is to understand why the Democratic share
of votes is 51.781% and not the 51.145% used in the tables. Granted, in the
Wisconsin Assembly case, it is a subtle change, but the reasoning behind it has
deep implications.
As shown in Appendix 1, Equation 2 is defined classically as the total wasted
votes over the total statewide votes,
99
=1
= = ,
99
=1
where the district wasted votes, Wk, and total district votes, Tk, are summed
separately. Equation 3 is defined as the average number of wasted votes per
district over the total number of districts,
99
=1
= ,
where STOT in our case is 99, the total number of seats in the state. Notice that
the original efficiency gap is defined by state vote totals divided by state vote
totals and the equal districts efficiency gap is defined by district averages
divided by districts. Are we focusing on votes or districts? The equal districts
efficiency gap (EDEG) normalizes each districts wasted votes and each district
has equal influence on the statewide EG. The EG becomes equal to the equal
districts efficiency gap when all Tks are equal.
16 Equal Districts Efficiency Gap
Again, when looking at election results, then, yes, votes take center stage. But
when looking at gerrymandering, we want to understand how individual
districts influence a state. Theories that use a partys statewide votes are biasing
their results. Theories that use district percentages often seem unaware that
these percentages properly weight the votes by district and cannot be used with
statewide vote totals.
This mini-state has two districts. District X elected a Republican and district Y
elected a Democrat, so the states seats are split evenly between the parties.
From the vote totals on the bottom of Table 6, we see that the state leans
slightly Democratic, 14 to 13. But, from the percentages in the right columns,
we see that the state is split evenly between the parties. Which is correct? Is the
state more Democratic or is it a wash? Should the EG favor Republicans,
Democrats, or come out even?
And that is the crux of the problem. Is one vote, one vote? Or is one district one
district? If a voter moves from district Y to district X, suddenly his vote carries
more weight. This is accounted for in Equation 3 but not in Equation 2. And, if
the district Y seat flips, it will have more impact on Equation 2 than if district X
flips. Why should one seat flipping have more sway on a redistricting analysis
than another seat flipping? Equation 3 treats all flipping seats equally, as should
be the case.
We can now understand why the three equations give such different results.
[ ( 50% 50%) 2 (51.852% 50%) ] = 3.704 (. 1)
come full circle all the way back to Tuftes seminal work of 19739 where he
presented the equation,
Percentage Seats for a Political party = b (Percentage Votes for that Party) + c
Equation 4. Tuftes Seats-Votes Equation
CONCLUSION
The efficiency gap has helped raise public awareness of gerrymandering issues,
and the story behind it, of wasted votes verses no wasted votes, is compelling.
The popularity of the efficiency gap underscores the publics and the courts
need for a way to understand gerrymandering from a mathematical point of
view. The efficiency gap is a positive addition to the literature but the structure
behind the method and its results are not nearly dependable enough to be
presented as evidence in a trial. When it is required, the equal districts
efficiency gap provides more accurate results, but even these results can be
skewed if not watched closely.
We have established a foundation to build on, but even if the concerns
presented in this paper were rectified, tables and equations will not solve the
problem. More comprehensive methods are needed, perhaps several methods
that work together. It is amazing that we can be shown pictures of obscenely
carved out political districts but somehow we desire a number to point to and
say hah, there it is. When we vote for an official, especially at the state house
level, our decision affects those around us most. Each of us is part of a local
community. It would be difficult not to support a geographically based measure
that would stop redistricters from cutting up neighborhoods to look like swiss
cheese, visual proof that our communities are being torn apart for political gain.
After all, what is the purpose of politics if it is not to enhance our communities?
The efficiency gap offers a step in the right direction in understanding
gerrymandering but broader methods must continue the work.
9
E. R. Tufte. The relationship between seats and votes in two-party systems. The
American Political Science Review, 67(2):540554, 1973.
19 Equal Districts Efficiency Gap
APPENDIX 1
The three EG equations used in the paper are derived here. The integer
subscripts i, j, and k will correspond to Democratic districts, Republican
districts, and all districts.
Dividing the state of Wisconsin into two columns, column i containing all
Democratic-won districts, and column j containing all Republican-won
districts, let S be the total number of seats won by Democratic candidates, and
in Wisconsin, (99 - S) is the number of seats won by Republican candidates.
The total number of wasted votes in the state is the total from all Democratic-
won districts plus the total from all Republican-won districts,
99
= + ,
=1 =+1
where the first summation is of column i, and the second, column j. WTOT is the
total number of wasted votes in the state, and Wi is the total number of wasted
votes in district i.
In a Democratic-won district, the number of wasted votes is,
= ,
2
where Di is the number of Democratic votes in district i, and Ri is the number of
Republican votes. In a Republican-won District, the number of wasted votes is,
=
2
Combining the above three equations, the total number of wasted votes in the
state becomes
99
= [ ] + [ ] ,
2 2
=1 =+1
which simplifies to
99 99
= + 2 ,
2
=1 =1 =1
1
= +2 ,
2
=1
where the summation term sums votes in Democratic-won districts and DTOT is
the total Democratic votes in the state. We can further simplify this to,
where a weighted seat no longer has a value of one but is weighted by the
number of votes in the district. This is Equation 2 in the paper.
To isolate the bias term, we use Ti = TTOT/99 i, where TTOT/99 is the average
district vote total, and i is the deviation of a districts vote total from that
average. Our equation becomes,
1 1
= +2 + .
99 2
=1
ESTABLISHING A BASIS
The above analysis was done in relation to the efficiency gap as it is presented
in the literature. Here, we tweak the initial assumptions about where the EG
should come from to help solidify a fundamentally sound basis for analyzing a
gerrymander.
The efficiency gap is defined as,
= = ,
but, as stated in the paper, this method unknowingly weights wasted votes by
that districts vote total. To even the field and treat each district equally, the
efficiency gap needs to be written,
21 Equal Districts Efficiency Gap
1
= = ,
99
Where STOT is the total number of seats, 99 in our case, and Wk/Tk is the percent
of wasted votes in district k. Notice how the efficiency gap is defined as
statewide wasted votes divided by the total number of votes in the state while
the equal districts efficiency gap is defined by a sum of district averages
divided by the total number of districts in the state. One focuses on votes, the
other on districts. Our equal districts equation reduces to,
99
2 1
= + ,
99 99 2
=1
where the summation term sums the percent of Democratic votes, district by
district. This reduces to Equation 3 in the paper,
Equal Districts Efficiency Gap = - ( Percent of Seats - 50% ) + 2 ( Average Percent of Vote - 50% )
where the average percent of vote is simply the average of the percent of the
Democratic vote in each district.
APPENDIX 2
TABLES ONE THROUGH FIVE
Total Total
Democratic Democratic Efficiency
Vote: Seats: Gap:
51.138% 42 11.690%
25 Equal Districts Efficiency Gap
Total Total
Democratic Democratic Efficiency
Vote: Seats Gap:
51.145% 51 2.195%
28 Equal Districts Efficiency Gap
Table 3. Efficiency Gap Calculation for Demonstration Plan with Votes Shifted to Create Packed Districts
Total Total
Democratic Democratic Efficiency
Vote: Seats Gap:
51.145% 51 2.195%
31 Equal Districts Efficiency Gap
Table 4. Efficiency Gap Calculation for Demonstration Plan with Votes Shifted to Create 50-50 Districts
Total Total
Democratic Democratic Efficiency
Vote: Seats Gap:
51.145% 51 2.195%
34 Equal Districts Efficiency Gap
Table 5. Efficiency Gap Calculation for Demonstration Plan with Votes Shifted to Give Lower Turnout in
Democratic-Won Districts
Total Total
Democratic Democratic Efficiency
Vote: Seats Gap:
48.904% 51 7.730%