IN THE UNITED STATES DISTRICT COURT FOR THE EASTERN DISTRICT OF WISCONSIN BETTYE JONES, et al., Plaintiffs, v. Case No. 2:12-cv-00185-LA JUDGE DAVID G. DEININGER, in his official capacity, Defendants. 26(a)(2); EXPERT DISCLOSURE FOR LELAND BEATTY Leland Beatty submits this expert disclosure pursuant to Federal Rule of Civil Procedure 1. The Declaration of Leland Beatty (Exhibit A to this Disclosure) provides a complete statement of all opinions I plan to express, the basis and reasons for them, and the facts and data I considered in forming those opinion. 2. I do not at present intend to use an exhibit to summarize my opinions. Should I form such an intention, I will promptly make that exhibit available. 3. My Curriculum Vitae is attached to my Declaration, and provides a statement of my qualifications. My only publication in the last ten years is attached as Exhibit B. 4. There are no other cases in which, during the previous four years, I have testified as an expert at trial or by deposition. 5. I am being compensated for my time in connection with my Declaration and testimony at the rate of $125 per hour. I will be compensated for my reasonable out-of-pocket expenses.
Dated: Apnl22,20l2 n $il#
Exhibit A
IN THE UNITED STATES DISTRICT COURT FOR THE EASTERN DISTRICT OF WISCONSIN BETTYE JONES, et al., Plaintiffs, v. Case No. 2:12-cv-00185-LA JUDGE DAVID G. DEININGER, et al., (all sued in their official capacity), Defendants. DECLARATION OF LELAND BEATTY Leland Beatty hereby declares as follows: 1. My name is Leland Beatty. I am a resident of Austin, TX. I am currently employed by commercial and political clients as a statistical marketing consultant. Previously, I served as general manager of Texas Rural Communities, Inc., a non-profit economic development corporation, and as director of research at the Texas Department of Agriculture. My resume is attached. 2. I have done extensive research on voting behaviors including minority participation in Texas, Minnesota, New York and Wisconsin, including study of the impact of photo identification requirements on the voting age population and on registered voters in Texas. 3. I have been asked to test the effect of Wisconsin voter identification requirements by race, using individual-level data from voter registration, driver's license and state-issued identification card records. This data was augmented with Census voting age population counts
at the county and zip code levels, and with commercially obtained individual-level ethnicity identification. 4. As explained below, my work primarily relies on data provided by the State of Wisconsin, specifically by its Department of Motor Vehicles and its Government Accountability Board. The State s data show a disparate impact that is stark and clear, bearing out what earlier analyses showed: minority voters are at a substantial disadvantage under Wisconsin s voter ID law, and the effect of that law imprints an unavoidable disparate impact on minority election participation. 5. My opinions are based on technical and specialized knowledge that I have gained from my education, training and experience and are developed based on widely accepted and reliable quantitative social science methods. I. Conclusions 6. Non-white Wisconsin residents will be significantly and adversely impacted by the Wisconsin voter identification law. Non-White Wisconsin registered voters are significantly less likely to possess a driver's license or state identification that matches their voter registration data. 7. 9.5% of registered White voters do not have a matching driver's license or state identification, compared to 15.8% of registered Asian voters, 16.2% of registered African American voters, and 24.8% of registered Hispanic voters. 8. Overall, some 11.1% of Wisconsin registered voters do not have a matching driver's license or state identification. African American registered voters are 1.7 times as likely as White registered voters to be without a matching driver's license or state identification. - 2 -
Hispanic registered voters are 2.6 times as likely as White registered voters to be without a matching driver's license or state identification. 9. Although African-American voters represent 5.3% of the registered voters in Wisconsin, they constituted 7.8% of the registered voters who lack a driver s license or a State identification card. Although Hispanic voters represent 1.6% of the registered voters in Wisconsin, they constituted 3.6% of the registered voters who lack a driver s license or a State identification card. 10. The number of registered voters without matching driver's license or state id is large enough to change the outcome of many Wisconsin elections. For example, the number of registered voters without matching driver's license or state id is larger than the vote margin that decided the elections of the U.S. Senator, Governor, Attorney General, Secretary of State, Treasurer, 3 of 8 Congressional districts, 9 of 17 State Senate districts, and 34 of 99 State Assembly Districts. The deciding margin in some districts was less than 5% of the number of registered voters without matching driver's license or state id. (See Tables 1 thru 4, attached at end of document). II. Data Sources 11. I worked with the following data sources in forming my opinions. 12. I obtained voting age population (VAP) data from the 2010 Census Redistricting Data Summary File. The U.S. Census reports total Wisconsin voting age population to be 4,347,494. 13. I obtained driver's license (DL) and State issued identification card (State ID) data from the Wisconsin Department of Motor Vehicles. - 3 -
14. Wisconsin Department of Motor Vehicles initially provided a county-level report of the number of persons with either a driver s license or a State ID. This report states 4,356,681 voting age Wisconsin residents possessed either driver s licenses or State IDs. 15. Subsequently, WI DMV produced a list of all Wisconsin driver's license holders, and a list of State ID holders. These files included, for each person with driver s license or State ID, thirteen variables: 1. Run Time Stamp (of administrative value only) 2. First name 3. Middle initial 4. Last Name 5. Gender 6. Date of Birth 7. Race 8. Zip Code 9. County name. 10. Current DL or State ID number 11. Date of issuance 12. Date of expiration 13. Current DL/State ID status III. Methodology and Results 16. To determine which registered voters possessed a matching driver's license or state identification, I created match strings from basic voter and driver's license/state identification records, then counted the matches between the two files. Three different strategies were used to qualify matches between registered voters and driver's license/state identification holders: 1) If the voter's first name, last name, date of birth, residence county and zip code uniquely matched the same information in a driver's license record; 2) if the voter's first name, last and date of birth uniquely match the same information in a driver's license record; or 3) if the - 4 -
voter's last name, date of birth and zip code uniquely matched the same information in a driver's license/state identification record, that voter was considered to have an identification match. 1. Creating Records That Allowed Me to Match Persons With Wisconsin Driver s Licenses or State Identification Cards With Registered Voters 17. I wanted to create data files that would allow me to match Wisconsin residents who had either a Wisconsin driver s license or a State identification card with registered Wisconsin voters. To do that, I created the following data files. a. Persons With Driver s Licenses or State Identification Cards. 18. The Wisconsin Driver s License list included 4,394,270 people, all with unique identification numbers. I created a unique identity string (IDstring0) for each record, composed of First Name, Last Name, Date of Birth, Gender, Ethnicity, County and Zip Code. In doing so, I discovered that 13 DL holders had an identical Idstring0 with another person. I eliminated the duplicate records. 19. The Wisconsin State ID list included 420,416 people, all with unique ID numbers. A unique identity string (IDstring0) was created for each record, composed of First Name, Last Name, Date of Birth, Gender, Ethnicity, County and Zip Code. Three State-ID holders had identical IDstring0 with one other ID holder. I eliminated the duplicate records. 20. 112,397 State ID holders had matching identification numbers with persons in the driver s license file, and well as identical IDstring0. I considered these records to be duplicates and eliminated them, leaving 308,016 unique State ID holders. 21. I then combined the driver s license and State ID files into one list with 4,702,273 unique driver s license or State ID holders (DL_ID). - 5 -
22. Finally, I deleted 133,836 persons who were under 18 years of age from the combined driver s license/state ID (DL_ID) list, leaving 4,568,437 voting age persons with DL or State ID. b. Persons Registered to Vote in Wisconsin 23. The Wisconsin Government Accountability Board, which oversees voter registration, produced a complete list of 3,255,377 Wisconsin registered voters with 74 variables for each voter. No race or gender identification variables are included in the voter registration file. Each voter has a unique voter identification number. 24. Among the variables in the file were: a. An administrative id number b. First Name c. Middle Name d. Last Name e. Name Suffix f. Date of Birth g. State ID h. Voter Registration Number i. 13 Address variables, including variables for each address part, as well as the address parts combined into Address1 and Address2 variables. j. Zip Code k. Five voter status variables, including current status, status reason, application source, application date and effective date l. 23 jurisdictional and district variables, including county, ward, assembly districts, court districts and municipalities. m. Phone n. Permanent Absentee o. 22 election participation history variables 1 1 The State ID field appears to be identical to the State ID field in the DL_ID file, and 66.1% of registered voters have a State ID number. These voters uniquely match to specific DL_ID holders. - 6 -
2. Using These Data Files to Calculate the Percent of Wisconsin Registered Voters, By Race, With a Driver s License or a State Identification Card 25. I next wanted to determine how many of the individuals in the Wisconsin registered voter files I could match to those in the file of individuals who had a Wisconsin driver s license or State ID. By eliminating the individuals for whom I could make this match, I would identify the Wisconsin registered voters who did not have a driver s license or a State ID. a. Creating Sub-Files to Compare 26. I started by sorting the data in both the voter registration file and the combined driver s license/state ID file (DL_ID) into three sub-files of data. The first sub-file (IDstring1) identified individual in both sets of files (voter registration and combined driver s license/state ID (DL_ID) by their first name, last name, date of birth, county and zip code data. This data sort uniquely identified 99.9% of registered voters, and 99.9% of DL_ID holders. 27. The second sub-file of data (IDstring2) identified individuals in both the voter registration data and the combined driver s license/state ID file (DL_ID) by their first name, last name and date of birth. This string uniquely identified 99.7% of registered voters, and 99.9% of DL_ID holders. 28. The third sub-file of data (IDstring3) identified individuals in both the voter registration data and the combined driver s license/state ID file (DL_ID) by their last name, date of birth and zip code. This string uniquely identified 99.4% of registered voters, and 99.5% of DL_ID holders. - 7 -
b. Running the Comparison 29. I then did three rounds of comparisons. Individuals who appeared in the voter registration data but could not be matched to the driver s license/state ID data after all three rounds of comparison represent registered voters who have neither a Wisconsin driver s license nor a Wisconsin state identification card. 30. First Round Match. The first round of matching I conducted was based on the sub-file that identified individuals in both the voter registration and combined driver s license/state ID (DL_ID) by their first name, last name, date of birth, county and zip code data. (IDstring1). 31. 1,209 registered voters (0.04%) had an identical IDstring1 with at least one other voter. I eliminated these duplicate entries. I eliminated these duplicate entries. Only 2 persons in the DL_ID file had identical IDstring1 with another DL_ID holder. I also eliminated these duplicate entries. 32. 71.8% of registered voters matched at least 1 DL_ID record on IDstring1, or 2,338,535 voters, leaving 916,505 voters unmatched. 33. Second Round Match. I next attempted to match the 916,505 registered voters that I did not match in the first round using the second data sub-file that was based on individuals in both the voter registration data and the combined driver s license/state ID file (DL_ID) by their first name, last name and date of birth (IDstring2). 34. Of the 916,505 registered voters unmatched after the IDstring1 comparison, some 99.9% (915,362) had a unique IDstring2. I eliminate the duplicate records. - 8 -
35. 2,230,887 DL_ID holders were unmatched after IDstring1. 99.99% (2,230,584) had a unique IDstring2. I eliminated the duplicate records. 36. In the second round match based on IDstring2, some 483,709 voters matched IDstring2 with at least one DL_ID holder. 37. After this second round match process, some 87% of registered voters had matched at least 1 DL_ID holder, leaving 432,796 registered voters unmatched with an entry in the driver s license/state ID data. 38. Third Round Match. My final effort to match registered voters with an individual identified in the combined driver s license/state ID data was based on the sub-file of data including individuals in both the voter registration data and the combined driver s license/state ID file (DL_ID) by their last name, date of birth and zip code (IDstring3). 39. 99.6% of registered voters unmatched after IDString2 had a unique IDstring3. I eliminated the duplicates. 40. 1,747,520 DL_ID holders remained unmatched after IDstring2, and 99.6% of DL_ID holders unmatched after IDstring2 had a unique IDstring3. 41. 76,283 voters matched on IDstring3, leaving 356,512 voters (10.95%) unmatched to a DL_ID holder. 3. Identifying the Race of the Un-matched Individuals in the Voter Registration Data. 42. The individuals in the voter registration data that I was unable to match to a driver s license or a State ID after three rounds of trying represent registered Wisconsin voters who have neither a driver s license or a State identification card. - 9 -
43. I submitted the 356,512 unmatched voters registered voters that I could not uniquely match to a driver s license or a voter identification card to Ethnic Technologies, a company that maintains proprietary data on race and ethnicity. One of the services Ethnic Technologies offers is the ability to identify the race of individuals based on identifying data such as that contained in the Wisconsin voter registration data. 2 44. Ethnic Technologies was able to identify the race of 91.6 percent of the unmatched voters those who are registered to vote in Wisconsin but who lack a driver s license as demonstrated by my inability to match them in the driver s license/state ID file (DL_ID). 45. The end result of my work is summarized in the following table: Registered Voters by Race, With Count Not Matching Driver's License or State ID, and Race Share Without Matching Driver's License or State ID Share of Race Race ASIAN Registered Voters 37,597 Share of All Voters 1.2% Voters Without Matching Driver's License or State ID 5,929 Share of All Unmatched Voters 1.6% Without Matching Driver's License or State ID 15.8% BLACK 172,251 5.3% 27,938 7.8% 16.2% HISPANIC 51,974 1.6% 12,879 3.6% 24.8% INDIAN 15,519 0.5% 872 0.2% 5.6% OTHER 32,373 1.0% 32,373 9.0% 100.0% WHITE 2,945,663 90.5% 280,396 77.8% 9.5% Total 3,255,377 100.0% 360,387 100.0% 11.1% 2 I also submitted to Ethnic Technologies some 3,650 entries from the voter registration data file were not unique records because they had multiple matches in the driver s license/state ID file (DL_ID). My hope was that by identifying these entries based on race, I could isolate unique entries. This was successfully in only a small number of entries. Where it was successful, I added the unmatched entries to my calculation of registered voters who lack either driver s license or a State identification card. - 10 -
IV. The Results of My Work are Consistent With Prior Research 46. Disparate effects by ethnicity, age and education status from requirements to produce driver's licenses and state-issued identification were documented in a 2005 study by John Pawasarat, Director of the Employment of Training Institute at the University of Wisconsin- Milwaukee. (Pawasarat, Employment and Training Institute, Univeristy of Wisconsin- Milwaukee, June 2005) 47. Pawasarat's study was based on complete data on all driver's license holders in the State of Wisconsin, provided by the Wisconsin Department of Transportation, as well as complete statewide counts of persons with state id cards. This data allowed Pawasarat to compare counts of persons with driver's licenses by county and zip code, broken out by age, gender and ethnicity, with counts of voting age Wisconsin residents from the United States Census. 48. Pawasarat concluded, Minorities and poor are the most likely to have driver s license problems, with African Americans and Latinos almost four times as likely as whites to be without a valid driver s license. Statewide, 55% of African-American males and 46% of Latino males lack a driver s license, compared with 16% of White males. 49. Among females, 49% of African-Americans, 59% of Latinas, and 17% of Whites lacked a driver s license. 50. Minority young adults ages 18-24 are impacted even more severely. Pawasarat documented that 78% of young African American males and 66% of young African American females did not have driver s licenses at the time of his study, compared to 43% of young White males and 31% of young White females. - 11 -
51. Among young Latinos 57% of males and 63% of females did not possess valid driver s licenses. 52. Statewide, 167,687 African Americans and Latinos were found to be without a driver s license in 2002, or 4% of the voting age population. Many Wisconsin elections are decided by margins no larger than this, including the 2010 Secretary of State contest. 53. Pawasarat concluded that the number of African Americans without driver's licenses would continue to grow over time. 54. A number of studies conducted in other states also found that photographic voter identification laws created disparate impact by ethnicity. 55. The Disproportionate Impact of Indiana Voter ID Requirements on the Electorate by Barreto, et al determined that in Indiana 17.5% of registered voters with annual incomes under $40,000 did not possess qualifying identification, compared to 11.2% of registered voters with annual incomes over $40,000. 2 In Wisconsin, only 38% of white persons have incomes under $40,000, while 68% of African Americans and 55% of Latinos have incomes under $40,000. Minority voters in Wisconsin are significantly impacted by income disparity, according to the U.S. Census, which means they are significantly more likely to be without acceptable identification for voting. Wisconsin Household Income Group Share by Ethnicity African Income Group White American Latino Under $40,000 38.4% 68.0% 55.4% $40,000 and Over 61.6% 32.0% 44.6% - 12 -
56. Barreto's estimates, if applied to Wisconsin, would mean that African Americans are 14% more likely than Whites to be without acceptable voting identification, and Latinos are 8% more likely. 57. Most research shows additional disproportionate impacts on minority voters. Barreto, et al, discovered that 18.1% of minority voters in Indiana did not possess acceptable photo identification, compared to only 11.5% of white voters. 58. From a study of Georgia registered voters, M.V. Hood III and Charles S. Bullock III of the University of Georgia concluded that minority voters are almost twice as likely to be disenfranchised by photo identification requirements as were white voters. 3 59. In Voter IDs Are Not the Problem: A Survey of Three States, Pastor, et al discovered that African American voters in three states were almost 2.5 times as likely to be without acceptable photo identification as white voters. 4 60. The table below shows estimates from all three studies of the share of the voting age population without acceptable id for voting, by race. Share Without Acceptable Voter ID: Three Studies Ethnicity Hood Study (Georgia) Pastor Study (Indiana, Maryland and Mississippi) Barreto Study (Indiana) African American 6.8% 2.2% 18.1% Latino 7.3% Not Calculated Not Calculated White 3.7% 0.9% 11.5% Other/Missing 4.9% Not Calculated Not Calculated - 13 -
61. I declare under penalty of perjury that the foregoing is fiue and correct. Executed on April 22,2012 in Austin, Texas. -t4-
Tables 1 thru 4: 2010 General Election Statewide, Congressional and Legislative Contests Decided by Margin Smaller than Number of Registered Voters Without Matching ID Table 1: Statewide Offices Congressional Districts, Registered Voters Without Matching State ID Compared to 2010 Election Margin Ratio of Registered Voters Without Contest Congressional District 3 Registered Voters Without Matching Driver's License/State ID 43,064 2010 General Election Vote Margin 9,542 Matching Driver's License or State ID to 2010 General Election Margin 4.5 Congressional District 7 43,648 19,533 2.2 Congressional District 8 41,604 25,352 1.6 Table 2: Congressional Districts Statewide Offices, Registered Voters Without Matching State ID Compared to 2010 Election Margin Ratio of Registered Voters Without Contest Governor Registered Voters Without Matching Driver's License/State ID 360,155 2010 General Election Vote Margin 124,638 Matching Driver's License or State ID to 2010 General Election Margin 2.9 AG 360,155 330,711 1.1 SecState 360,155 21,948 16.4 Treasurer 360,155 142,852 2.5 US Senate 360,155 105,041 3.4-15 -
Table 3: State Senate Districts State Senate Districts, Registered Voters Without Matching State ID Compared to 2010 Election Margin Ratio of Registered Voters Without Contest State Senate 3 Registered Voters Without Matching Driver's License/State ID 10,059 2010 General Election Vote Margin 8,605 Matching Driver's License or State ID to 2010 General Election Margin 1.2 State Senate 5 11,791 3,150 3.7 State Senate 7 13,164 607 21.7 State Senate 15 9,893 9,737 1.0 State Senate 21 10,650 3,106 3.4 State Senate 23 9,881 5,073 1.9 State Senate 25 11,725 1,583 7.4 State Senate 29 8,672 2,898 3.0 State Senate 31 10,266 403 25.5-16 -
Table 4: State Assembly Districts State Assembly Districts, Registered Voters Without Matching State ID Compared to 2010 Election Margin Ratio of Registered Voters Registered Voters Without Matching 2010 General Without Matching Driver's License or State ID to Driver's Election Vote 2010 General Election Contest License/State ID Margin Margin Assembly 1 4,277 4,060 1.1 Assembly 5 4,358 3,451 1.3 Assembly 7 3,788 3,126 1.2 Assembly 15 3,593 387 9.3 Assembly 20 4,427 1,341 3.3 Assembly 26 3,237 151 21.4 Assembly 28 3,577 3,136 1.1 Assembly 36 3,618 3,435 1.1 Assembly 37 2,856 1,013 2.8 Assembly 42 2,309 287 8.0 Assembly 43 3,659 1,001 3.7 Assembly 44 2,746 515 5.3 Assembly 45 3,488 1,519 2.3 Assembly 47 4,313 3,670 1.2 Assembly 49 3,310 2,540 1.3 Assembly 51 2,494 891 2.8 Assembly 54 3,345 2,797 1.2 Assembly 57 3,636 1,006 3.6 Assembly 62 3,708 1,883 2.0 Assembly 68 2,966 92 32.2 Assembly 70 3,288 1,717 1.9 Assembly 71 3,739 3,035 1.2 Assembly 72 2,877 1,069 2.7 Assembly 73 4,066 2,503 1.6 Assembly 74 3,998 1,301 3.1 Assembly 75 3,661 415 8.8 Assembly 77 6,978 4,380 1.6 Assembly 80 2,989 1,302 2.3 Assembly 85 3,167 1,838 1.7 Assembly 88 4,057 267 15.2 Assembly 90 3,585 3,137 1.1 Assembly 91 4,423 3,375 1.3 Assembly 92 2,854 312 9.1 Assembly 94 4,412 4,211 1.0-17 -
Leland Beatty Predictive Analytics 1103 Upland Drive. Austin, TX 78741 (512) 619-8732 leland@austin.rr.com Employment 1/00-present Consultant, Predictive Analytics and Business Process. Austin, TX Voter Participation and Preference Modeling Highly accurate predictive modeling. In the past four election cycles, actual results varied less than 2% from predicted results. This information advantage has consistently helped candidates win against much better funded opponents. Services include innovative low-cost polling, detailed market segmentation, and integration with campaign resources, including budgeting, message, canvas, mail and phone programs, and media buying. Marketing Planning and Management Create targeted marketing plans based on data mining analysis of complex consumer data. Product includes detailed market segments with strategies and metrics for each segment. Clients include national big-box retailers, insurance marketers and direct mail vendors. 6/89-12/99 Texas Rural Communities, Inc. Austin, TX General Manager Highlights: Created adult continuing education programs, including a farm financial management training program which became a requirement for all Texas farmers seeking a federal farm loan. Delivered program via community colleges and the Internet. Customer satisfaction levels topped 90% in a course expected to be unpopular Operated a grant program targeting rural, minority children which funded innovative social programs, such as heritage-based mentor programs for at-risk rural minority children. Investment fiduciary for Heartland Lloyds, a start-up insurance company targeted to the insurance needs of underserved rural counties; served on Heartland s Board of Directors for five years Operated an active small business loan and technical assistance program Managed operations of five endangered Texas state parks. Tripled income and reduced deficits by 50% in two years Created a set of private-industry tourism promotions that increased revenues of participants by more than 33% in first season. 5/83-5/89 Texas Department of Agriculture Austin, TX Director, Farmer Assistance Programs Created and delivered programs to increase income and provide legal and mental health assistance to farmers and other rural businesses during the worst agricultural economic downturn since the Great Depression Led staff of 6 to 12 researchers and program specialists Producer for Willie Nelson s FarmAid II Primary speechwriter for Agriculture Commissioner Jim Hightower 3/80-5/83 Sweetwater Reporter Sweetwater, TX Managing Editor of award-winning daily newspaper serving 17,000 subscribers 9/77-5/79 Weslaco Independent School District Weslaco, TX Bilingual Teacher for migrant high schoolers with English language skills at primary levels Education The University of Texas at Austin, McCombs School of Business Master of Business Administration Columbia University, New York, New York Bachelor of Arts in Literature, History and Art History
Exhibit B
Print http://beaumontenterprise.tx.newsmemory.com/eebrowser/frame/check.6246/php-scrip... Beaumont Enterprise Beaumont Enterprise 07/31/2011 B: BE (2-ben_b6_0731_015_ba_bemn.pdf.0) Page A15 August 2, 2011 Powered by TECNAVIA Copyright 2011 Beaumont Enterprise 07/31/2011 8:15 pm 1 of 1 8/2/11 3:20 PM