Transcript Gerrymandering and Legislative Efficiency John Mackenzie FREC/CANR University of Delaware Gerrymandering involves the redrawing of election district boundaries to maximize the number of legislative seats that.

Gerrymandering and
Legislative Efficiency
John Mackenzie
FREC/CANR
University of Delaware
Gerrymandering involves the redrawing of
election district boundaries to maximize
the number of legislative seats that can be
won by the political party in charge of
redrawing the district boundaries.
Gerrymandering creates “safe” seats for the
party’s incumbent legislators.
Opposition legislators can be “kidnapped”
into hostile districts and defeated.
Do Congressmen in gerrymandered “safe”
districts accumulate more seniority,
committee chairmanships, clout, etc. that
benefit their constituents…
…or do they get lazy?
What is the net benefit (or cost) of
gerrymandering?
Satire of the original 1812 “Gerrymander”
The Constitution only guarantees that your
actual vote is secret.
Your age, sex, household composition,
address, phone number, party registration
and voting history are all public record, and
made available to political parties and
candidates.
Voter databases are now routinely
geo-coded and can support high-precision
gerrymandering via GIS.
Basics of Gerrymandering:
packing & cracking
Data Sources
GIS shapefile of district boundaries for the
109th Congress obtained from the US
Census Bureau website
www.census.gov/geo/www/cob/bdy_files.html
Consolidated Federal Funds Report
Using geometric compactness as a
gerrymandering index:
G = area / perimeter squared
G ranges from zero (for a true fractal) to
1/(4π), or ~0.079577 (for a circle). This can
be rescaled as desired.
Some state polygons are more compact than
others, and a better gerrymandering index
would exclude the complexities of state
boundaries.
A simple refinement of the index would be to
normalize each district’s compactness Gd by
the state’s compactness Gs
A better approach is to distinguish natural
(black) from artificial (red) boundary complexity
for each individual district polygon:
Let P = g + n denote total district perimeter
as the sum of artificially-drawn interior
political boundary length g plus natural
exterior boundary length n.
The artificially-drawn proportion is g/P, so
the polygon complexity attributable to
political manipulation is
G* =
2
(g/P)(P /A)
= gP/A.
GIS Procedures:
•Dissolve Congressional districts by state ID to
obtain a shapefile of exactly congruent state
boundaries.
•use Polygon-to-Line utility to crack the 435
district polygons into 2,232 component
boundary segments. These line features
include attribute table fields LEFT_FID and
RIGHT_FID that identify the polygon feature
ID’s on the “left” and “right” sides of each line.
•use Select-by-Location “are within
(Clementini)” criterion to select the interior
district boundaries, identifying 1,318 district
boundary line features as not coincident with
any state boundary.
•Create a LENGTH field in this line shapefile’s
attribute table
•use Calculate Geometry determine the
lengths of these interior boundary segments.
The unselected segments on state boundaries
were assigned LENGTH values of zero.
•Create Summarize tables to sum the LENGTH
values for each LEFT_FID and RIGHT_FID.
•Join these two Summarize tables to the
original district polygon shapefile
•Create a new INTLENGTH field in the polygon
attribute table
•Sum LENGTHs from the two Summarize
tables into that field. The revised index:
G* = INTLENGTH*PERIMETER/AREA.
Caveats:
This index still omits other natural boundary
elements (rivers, etc.)
Measurements of natural boundary lengths
are scale-dependent (Mandelbrot)
Thus any geometric index of gerrymandering
must have some degree of arbitrariness.
Maryland 2nd
G* = 410 (red shoreline)
G* = 318 (black shoreline)
G* = 210 (blue convex quasi-hull)
State
States with Highest
Average
Gerrymandering Scores
Seats
Average G*
MD
8
150.3
NC
13
115.5
FL
25
90.4
PA
19
89.1
CA
53
80.6
NJ
13
77.6
IL
19
76.6
TX
32
68.6
AL
7
64.8
TN
9
62.9
MA
10
62.0
VA
11
55.7
NY
29
54.9
OH
18
51.0
LA
7
47.8
CO
7
47.2
WV
3
45.5
GA
13
44.5
AZ
8
44.4
SC
6
43.0
ln(FEDEXP/TAX)
$10.00
$1.00
0.01
0.10
1.00
$0.10
ln(ADJCOMPACT)
10.00
100.00
Conclusions:
This analysis does not indicate any
correlation between the geometric complexity
of US Representative district boundaries and
the efficiency of Representatives in
recapturing federal dollars for their districts.
The results do not support the “seniority”
hypothesis or the contrary “laziness”
hypothesis. Gerrymandering is not shown to
cause net harm to district constituents.