DRAFT REPORT 

Voting Irregularities in Florida During the 2000 Presidential Election

(Approved by the Commissioners on June 8, 2001)


Appendix IX


Statement of Dr. Allan J. Lichtman Regarding The Racial Impact of The Rejection of Ballots Cast in the 2000 Presidential Election In The State of Florida Presented Before The United States Senate Committee on Rules and Administration

July 16, 2001

This statement responds to the presentation before this Committee by Abigail Thernstrom, a Commissioner for the United States Commission on Civil Rights, and by her statistical consultant John R. Lott, who also submitted a statistical report. The Thernstrom-Lott statement attempts to challenge findings of the report “Voting Irregularities in Florida During the 2000 Presidential Election,” adopted by a majority of the United States Commission on Civil Rights. I am responding to their statement as it addresses a study I prepared for the Commission, documenting major racial disparities in ballot rejection rates in Florida’s 2000 presidential election. My statement demonstrates the following:

 

  1. The initial finding of major racial disparities in ballot rejection rates in Florida’s 2000 presidential election set forth in my original report is confirmed by additional evidence of what actually happened in voter precincts and is not challenged by any alternative findings in the dissenting statement.

 

2.      The dissenters’ statistical consultant admitted before the Senate Committee on Rules that “a greater percentage of black and Hispanic people are turned away than, or don’t get to vote, than white people.” The dissenters concede in their report that African-Americans in Florida probably had their ballots rejected at a rate at least triple that of non-African Americans.

 

3.      Racial disparities in ballot rejection rates cannot be explained by differences between blacks and non-blacks in education, income, or any other factor pointed to by the dissenters.

 

  1. The relationship between race and ballot rejection remains substantial and statistically significant even within a comprehensive model with much greater explanatory power than any of the models presented by dissenters.

 

  1. Although the dissenting statement offer education and literacy as explanations for ballot rejection, their proposed models exclude these key variables, include redundant variables, explain relatively little of the variation in ballot rejection among counties, and are contradicted by precinct-level results.

 

The Finding Of Major Racial Disparities In Florida’s 2000 Presidential Election Is Confirmed By Additional Evidence And Is Not Contradicted By Any Alternative Findings

 

My initial report analyzed county and precinct-level data in Florida that revealed major differences in the rate of ballot rejection for blacks and non-blacks in the 2000 presidential election. There is not a single alternative finding in Dr. Thernstrom’s statement or Dr. Lott’s statistical report that even purports to show the lack of racial disparities in ballot rejection. Indeed, Dr. Lott. admitted the existence of such disparities under questioning by Senator Charles Schumer during the hearings before this Committee. The following is the account of that examination in the New York Times: “Mr. Schumer then wrung out of Mr. Lott a defeated “yeah” to the question of whether ‘a greater percentage of black and Hispanic people are turned away than, or don’t get to vote, than white people?’”[1]Thus the dissenters’ expert confirmed under examination the key finding of my initial report.

Likewise Dr. Thernstrom conceded in her statement that there was at least a 300 percent discrepancy in rejected ballots for black as compared to non-black voters. She says that racial disparities in ballot rejection “was likely on the order of three to one” in Florida, not on the basis of any independent study, but rather by choosing the lowest within-county ratio of the counties I studied with precinct data.

Since completing my original report I have since gathered additional precinct-level data for Broward and Escambia Counties. Broward is the second most populous county in Florida. Escambia is a county with optical scanning technology recorded by precinct. Escambia County had the largest number of rejected ballots (4372) for such counties, accounting for about a quarter of all rejected ballots for counties with optical scanning technology recorded by precinct. The analysis of Escambia County offers the first detailed glimpse using precinct-level data of the relationship between race and ballot rejection for counties with optical scanning technology recorded by precinct. It confirms the finding suggested in my first report of racial disparities in ballot rejection rates even among counties with the best available technology.

Broward County had an overall rejection rate of 2.5 percent. However, 6.5 percent of ballots were rejected in precincts that were 90% or more black in their voters, compared to 2 percent in precincts that were 90 percent or more non-black. Escambia County had an overall rejection rate of 3.6 percent. However, 13.7 percent of ballots were rejected in precincts that were 90% or more black in their voters, compared to 2.2 percent in precincts that were 90 percent or more non-black.

The precinct-level data I examined (for Broward, Duval, Escambia, Miami-Dade and Palm Beach Counties) now includes about 58 percent of the statewide total of rejected ballots. Within each county, there were major disparities in ballot rejection rates for African-Americans and non-African-Americans, ranging as high as 18 percentage points, with an average of about 10 percentage points. If black ballot rejection rates had equaled white ballot rejection rates in these counties alone, more than 30,000 additional black votes would have been counted in Florida’s presidential election.           

 

Racial Disparities In Ballot Rejection Rates Cannot Be Explained By Differences Between Blacks And Non-Blacks In Education, Literacy, Income, Poverty Or Any Other Factor Cited By Dr. Thernstrom.

 

The statistical report presented by John R. Lott allegedly tests the hypothesis presented in Dr. Thernstrom’s statement that differences in education and literacy account for racial disparities in ballot rejection. Yet Dr. Lott’s analyses fail to include literacy or any other educational measure in any of the eight models purported to explain differences in the ballot rejection rates. Instead, the models include only measures of income and poverty, even though Dr. Lott admits that these variables are only “roughly associated with education” or “the ability to read and follow voting instructions.”

            Table 1 below reports the results of a county-level analysis that, unlike the statistical report presented by dissenters, controls for literacy, education, poverty, and income for all counties. In addition, these equations also control for differences in technology: whether a county uses punch cards, optical scanning centrally recorded, or paper ballots or machines. The county-level findings reported in Table 1 show that racial differences in ballot rejection rates in the 2000 presidential election are not reducible to differences between blacks and non-blacks in income, poverty, education, or literacy. The coefficient measuring the influence of the percentage of black voters on ballot rejection

rates, controlling for all variables in Table 1, is .140, which corresponds to a difference in ballot rejection rates of 14.0 percentage points between blacks and non-blacks.[2] This means that independent of income, poverty rates, education, literacy, and the technology for voting, there is a double-digit difference in ballot rejection rates between African Americans and non-African Americans.      

 

TABLE 1: THE INFLUENCE OF RACE ON BALLOT REJECTION RATES 2000 PRESIDENTIAL ELECTION IN FLORIDA, CONTROLLING FOR LITERACY, EDUCATION, INCOME, AND POVERTY, COUNTY DATA

 

 

 

 

VARIABLE

REG.

COEFF.

T

VALUE

SIGNIF.

BLACK

VOTERS

.140

4.2

.000

POVERTY

PERSONS 18+

-.047

-.644

.522

MEDIAN INCOME

-.00009

-.1.24

.220

LITERACY

-.077

-1.14

.261

% UNDER

9TH GRADE

.192

2.65

.010

PUNCH

CARD

3.20

6.81

.000

OPTICAL

CENTRAL

4.42

7.70

.000

PAPER/LEVER

3.02

2.65

.011

R2  = .805

 

 

 

 

 

 

 


The Relationship Between Race And Ballot Rejection Remains Substantial And Statistically Significant Even Within A Comprehensive Model With Far Greater Explanatory Power Than Any Models Presented By The Dissenters.

 

The results of estimating a more comprehensive model of ballot rejection in Florida counties are presented in Table 2. This model does a far better job than any of the models in Dr. Lott’s statistical supplement in accounting for changes from county to county in the percentage of rejected ballots in Florida’s 2000 presidential election. With an R2 value of .866, this model accounts for 86.6 percent of the variation from county to county in ballot rejection rates, compared to from 73.1 percent to 78.6 percent for the models presented in Dr. Lott’s report. The coefficient measuring the influence of the percentage of black voters on the percentage of rejected ballots -- statistically significant at a level beyond the stringent .01 standard used in social science -- is .143, which corresponds to a difference in ballot rejection of 14.3 percentage points between blacks and non-blacks, controlling for the variables in the equation.

None of the socio-economic variables in this model, however, have a statistically significant influence on ballot rejection rates. Neither does the variable measuring changes in voter turnout between 1996 and 2000, indicating no support for the hypothesis that differences in first-time voting among black and non-blacks account for difference in

TABLE 2: THE INFLUENCE OF RACE ON BALLOT REJECTION RATES 2000 PRESIDENTIAL ELECTION IN FLORIDA,

COUNTY DATA

VARIABLE

REG.

COEFF.

T

VALUE

SIGNIF.

BLACK

VOTERS

.143

4.77

.000

POVERTY

PERSONS 18+

-.014

-.218

.828

MEDIAN INCOME

-.00002

-.229

.820

LITERACY

-.0003

-.002

.998

% UNDER

9TH GRADE

.012

.158

.875

INCREASE

VOTE 96-2000

-.014

-.638

.526

TURNOUT 2000

-.075

-2.42

.019

PERCENT

DEM.

-.049

-2.01

.050

VOTERS PER

PRECINCT

-.002

-2.78

.008

DEM

SUPERVISOR

-.345

-.466

.686

REP

SUPERVISOR

-.317

-.406

.643

PUNCH

CARD

3.46

8.09

.000

OPTICAL

CENTRAL

4.31

8.18

.000

PAPER/LEVER

2.35

2.30

.026

R2  = .866

 

 

 

ballot rejection rates.[3] Similarly, the political identity of election supervisors has no discernible influence on ballot rejection rates in the comprehensive model. Variables measuring whether the supervisor is Republican or Democrat both have negative signs (relative to non-partisan supervisors), and fail to approach conventional levels of statistical significance.

                Based on analysis of Florida individual voters files for 2000, Professor Philip Klinkner of Hamilton College has also provided county-by-county percentages of first-time voters. Substituting this direct measure of first-time into the model for change in voting between 1996 and 2000 produces virtually no change in the relationship between race and ballot rejection.[4] The values reported in Table 3 show that the coefficient for the percentage of black voters is statistically significant beyond conventional levels and has a value of .137, which corresponds to a difference in ballot rejection of 13.7 percentage points between blacks and non-blacks, controlling for the variables in the equation. The coefficient measuring the relationship between first-time voting and ballot rejection is negative and falls far short of statistical significance.     


 

TABLE 3: THE INFLUENCE OF RACE ON BALLOT REJECTION RATES 2000 PRESIDENTIAL ELECTION IN FLORIDA,

COUNTY DATA

VARIABLE

REG.

COEFF.

T

VALUE

SIGNIF.

BLACK

VOTERS

.137

4.56

.000

POVERTY

PERSONS 18+

.011

.167

.868

MEDIAN INCOME

.00001

.149

.882

LITERACY

-.009

-.142

.887

% UNDER

9TH GRADE

.024

.326

.746

% FIRST-TIME

VOTERS

-.015

-.319

.751

TURNOUT 2000

-.071

-2.50

.016

PERCENT

DEM.

-.047

-1.98

.054

VOTERS PER

PRECINCT

-.002

-2.20

.033

DEM

SUPERVISOR

-.452

-.637

.527

REP

SUPERVISOR

-.475

-.640

.525

PUNCH

CARD

3.32

8.27

.000

OPTICAL

CENTRAL

4.47

8.94

.000

PAPER/LEVER

2.32

2.24

.030

R2  = .866

 

 

 

The Statistical Report Presented By Dissenters Provides No Credible Models Of Ballot Rejection In Florida’s 2000 Presidential Election.

 

                The statistical report commissioned by dissenters includes eight models that purport to explain the relationship between race and ballot rejection. Not only do these models omit key variables that are essential to hypotheses advanced by Dr. Thernstrom, but they also include duplicative measures of the racial composition of counties, destroying the integrity of the effort to gauge the independent influence of race on ballot rejection rates. Dr. Lott’s models examine a far less inclusive set of variables than the model developed in Table 2 above, excluding among other factors, literacy, education, voter turnout, and changes in voter turnout. His models also produce results that are internally contradictory and conflict with what we actually know happened at the precinct level in Florida’s 2000 presidential election. Why would the dissenters’ statistical models exclude the variables that dissenters affirm to be most relevant while including redundant variables that provide no new information, but only cancel each other’s effects? One can only speculate that the dissenters might have been less interested in accurately modeling the ballot rejection process in Florida and more interested in coming up with models -- however invalid -- that at least appeared to challenge the relationship between race and ballot rejection rates.

The models in Dr. Lott’s report that purport to show the lack of a statistically significant relationship between race and ballot rejection include redundant racial variables that make it impossible to gauge accurate the direction, magnitude or statistical significance of the independent relationship between race and ballot rejection. Model 1, for example, includes the percentage of blacks among registered voters, the percentage of Hispanics in the population, the percentage of whites in the population, and the percentage of African-Americans in the population. Several of the models also predict that within counties, where the technology is the same, there should be random variation in the relationship between the percentage of blacks among voters and the percentage of rejected ballots. This theorizing by the dissenters is contradicted by what we all know actually happened in Florida’s voting precincts. The precinct-level data ignored by dissenters demonstrates a powerful, positive statistically significant relationship between the percentage of black voters and the percentage of rejected ballots, with differences between black and non-black rejection rates that range as high as 18 percentage points and averaging about 10 percentage points. Rarely in social science is a statistical model so decisively rejected by its own predictive results!

The statistical models in Dr. Lott’s report also contradict their hypothesis that the presence of a Democratic election supervisor is associated with higher rates of ballot rejection.  In all models, the variable signifying a Democratic election supervisor has a negative, not a positive relationship with ballot rejection rates. Moreover, Dr. Lott’s analysis of the race of election supervisors identifies four supervisors as African-American, whereas information developed by the Commission staff indicates that there was only one African-American supervisor at the time of the election.

                 The remaining analyses in the statistical report pertain to comparisons between ballot rejection in 2000 and earlier years and provide no insight into the measurement of racially linked ballot rejection rates in 2000. These analyses are based on unverified data, fail to control for changes in technology, and falsely assume that if ballot rejection rates are related to race an increase in black voter registration must be accompanied by an increase in ballot rejection rates.

Dissenters also argue that estimates of rejection rates for black and non-black voters cannot be obtained from aggregate data. In fact, the methods and reliability checks used in my report were designed for the analysis of aggregate data and are sustained by multiple levels of analysis – at both the county and precinct level. The far-fetched examples and strained analysis in the dissenters’ opinion suggests that African-Americans lived in counties with mechanisms that somehow produce high ballot rejection rates for the non-blacks living in those counties, but not for the blacks living in the counties. Yet we know that this assumption is false because we have data for some two thousand precincts within counties demonstrating that African-Americans within these counties, not the non-African Americans, experience especially high rates of ballot rejection.

 

Conclusion

            In sum, the dissenters present no evidence contradicting the finding of my initial report that there were major racial disparities in ballot rejection rates in Florida’s 2000 presidential election. They fail to test empirically their hypothesis that such disparities can be attributed to education, literacy, or first-time voting. In addition, the dissenters rely on a statistical report that provides no credible models of ballot rejection in the 2000 presidential election in Florida. As both my initial report and this supplement demonstrate, there were major racial disparities in ballot rejection rates in Florida.



[1]  Katherine Q. Seelye, “Senators Hear Bitter Words on Florida Vote,” New York Times, June 28, 2001.

[2] The coefficient is statistically significant at a level beyond the stringent .01 standard used in social science. 

[3] Dr. Thernstrom also repeats in her statement, without conducting any reliability tests, two statistics from the media that are certainly false: that 40 percent of African-American voters were first-time voters and that the African-American percentage of the 2000 Florida electorate exceeded its 14.6 percentage of the state’s population. Analysis of individual voter files for the 2000 election by Professor Philip Klinkner of Hamilton College indicates that less than 25 percent of African-Americans were first-time voters in 2000, that the difference between first-time voting among African-Americans and non-African Americans was only about four percentage points, and that African-Americans were only about 10 percent of the 2000 electorate.  

 

[4] This measure will slightly overestimate first-time voting because voters may have voted in other states or have voted prior to 1994, when the records begin.