is a nonpartisan redistricting web gateway for citizens, professionals, academics,
and anyone interested in the congressional, legislative, and local process of redistricting




Laws Affecting the
 Redistricting Process

14th Amendment

15th Amendment

• Voting Rights Act of 1965

• State Constitutions

Additional Resources

Redistricting Law 2010

U. S. Department of Justice Civil Rights Division, Voting Section

Census Data Links

Current State P.L. 94-171 Data Releases

2010 Block Assignment and
Block Relationship Files
2010 TIGER/LINE Shape Files 

Census 2010 Redistricting Maps

Census Tech Tips

NCSL National Redistricting Seminar Highlights: Measuring Minority Vote Dilution



   January 25, 2011                                                                Share

The preliminary factors any plaintiff must show when suing a governmental entity for minority vote dilution include a large and compact enough minority population to form a district, cohesive voting among the minority population, and consistent racially polarized voting by white voters. Accordingly, it is has become good practice for states to perform this analysis prior to redrawing district lines to avoid claims of minority vote dilution.


Dr. Lisa Handley, Director of Frontier International Electoral Consulting demonstrated to seminar participants how to measure the extent of racially polarized voting when performing a minority vote dilution analysis. There are 3 different statistical techniques that professionals use:


1.      Homogeneous Precinct Analysis

The simplest and least rigorous method requires identifying voting precincts comprised of at least 90% of one particular race. For example, if one were to compare the election results of a group of homogenous African-American precincts with the election results of homogenous white precincts and found that both racial groups would have elected two different candidates, racially polarized voting does exist. Handley warns however, that the existence of racially polarized voting may or may not be legally significant.


This method is not useful for areas that have no or too few homogeneous precincts to compare. Also, the voting precincts involved in homogeneous precinct analysis are often not a representative sample of voters.


 Data Needed for Homogeneous Precinct Analysis


•Election results

•Precinct level total voting age population

•Precinct level minority voting age population

•Precinct level white voting age population

•Precinct level election turnout

•Precinct level vote counts for each candidate



2.      Bivariate Ecological Regression Analysis

Handley assures us that the title of this method is not as complex as it sounds; “bivariate” refers to the fact that it measures two variables, “ecological” refers to the use of aggregate level data as opposed to precinct level data (as in homogeneous precinct analysis), and “regression” refers to the statistical method which assumes a linear relationship between the two variables.


The method plots percentage minority voting age population and the minority’s percentage vote for a candidate in each precinct in the region under analysis. The result is an estimate of how many whites and how many minorities voted for a particular candidate.  


One weakness of this method is that it can produce voting percentage estimates that are negative or over 100%. The data needed for this analysis is the same as for homogeneous precinct analysis.


3.      Ecological Inference

This method is a modification of the bivariate analysis discussed above. It was designed in part to address the problem of out of bounds results in bivariate analysis. This method uses more information than the previous two methods, and involves identifying the range of possible values for the percentage of minority and white voters that voted for a given candidate on a tomographic map. The most likely estimate of the proportion of white and minority voting for a candidate is then selected using standard statistical algorithms.


Drawbacks to this method include the inability to replicate the exact results for the same data set since the method itself is a computer simulation. The method is also difficult to explain in the courtroom although it has been used and accepted by courts. Handley explained that the Supreme Court however, is more familiar and accepting of the previous two methods.


 Data Needed for Ecological Inference Analysis 

•Precinct level turnout data

•No. of minority voters (estimate)

•No. of white voters (estimate)

•Total votes for candidates

•Minimum maximum no. of minority voters (calculated)

•Minimum & maximum no. of white voters (calculated)



Important Considerations


Handley reminds us of the following caveats when using the results of these analyses:


All three of these techniques should be performed as part of a vote dilution analysis. The results should all trend in the same direction, if they do not; the value of the data diminishes accordingly. Confidence in the data output increases when polarization is found across different methods and different elections.


These methods must be performed across “several” elections to confirm any polarization trend. Also, some elections may not yield reliable results, such as judicial elections; which tend to have minimal voter engagement, and school board elections; which often do not get much voter interest.


A finding of polarization must be accompanied by a losing minority preferred candidate to legally significant.









Related Articles

»Federal Laws Governing the Redistricting Process

» 2011 NCSL Redistricting Seminar: Voting Rights Law Changes