the margin of victories were often larger

METHOD & ENERGIES
Christy Vaughn
Sachet Bangia
Bridget Dou
Sophie Guo
The way we redraw districts (beginning with actual districting);
we are using Markov Chain Monte Carlo Simulations for our distribution:
1. Propose a precinct on a district boundary to move into the neighboring district.
2. Accept or reject the change based on new vs. current district energies (see below) and the
Metropolis Hastings Algorithm.
3. After many changes, re-tabulate election results using actual vote data
The factors we consider for each district when redrawing plans:
Compactness Energy
ensures geographically
compact districts.
so this is better
The populations of all
districts are about equal and
minority populations are
in accordance with the
Votings Rights Act of 1965.
than this
Minimize the number
of split counties
RESULTS
Projected Number of Democrats
Elected to US House in NC, 2012
Distribution of Average Margin of Victories vs.
Actual Average Margin of Victories,
Based on 2012 House Data
Our results on the left also so far seem
to prefer bipartisan committees
(IA) plans over those drawn solely by
the state legislature (NC, MD).
There is some evidence of bias in the redistricting
process. Even in cases where the seats-split matched
that of our randomly drawn samples, the margin of
victories were often larger, as seen above.
In addition, it is interesting to note the
effect of the Majority-Minority
Energy on the distribution of Democrats elected to office, as seen above.