a. Dr. Kenneth Mayer

Plaintiffs tendered Dr. Kenneth Mayer as an expert witness in political science, quantitative analysis, election administration, and voter behavior. Tr. 327:24–328:6. Dr. Mayer is a full professor at the University of Wisconsin-Madison, having completed graduate studies in econometrics and statistics. Tr. 325:6–23. He has testified as an expert witness in numerous cases where courts credited his opinions and relied on his analysis. PX 15, 11; Tr. 327:16–21. He has never been excluded as an expert witness under Daubert or other expert standards. PX 15, 11. Defendants did not object to Mayer being tendered as an expert in this case. Tr. 328:4–6. The Court has reviewed Mayer’s qualifications and observed him during his testimony. He testified clearly, reliably, and objectively. The Court has no reason to doubt that he used sound methodology in reaching his conclusions, consistent with the standards in his professional field. The Court credits his expert report and testimony, and affords it great weight.

In this case, Plaintiffs asked Mayer to assess True the Vote’s challenge file of potentially ineligible voters and to assess the reliability of this data in relation to the Georgia voter file. Tr. 330:2–5. Mayer only assessed TTV’s file (not the Davis and Somerville list (Tr. 372:21–23)) and only looked at the counties that had a volunteer to make challenges. Tr. 330:2–3 (“I was asked to analyze the challenge files that True the Vote offered in 65 Georgia counties.”). He concluded that TTV’s file was filled with errors, many of which were “obvious” to even an untrained eye upon “immediate inspection.” Tr. 331:1–332:8, 390:6–25.

Mayer first criticizes TTV’s premise of making the challenges, expressly disclaiming any empirical evidence of voter fraud or large-scale instances of ineligible individuals casting ballots. PX 15, 13; Tr. 332:13–334:2. Specifically, in Georgia, Mayer points out that the Secretary of State’s office audited the absentee ballots cast in Cobb County and concluded that not one invalid absentee ballot had been cast in 2020. Tr. 332:22–25. Hence the underlying rationale for TTV’s voter challenges is, in Mayer’s view, empirically unsupportable.

For TTV’s actual data set, Mayer testified that TTV’s “record linkage” process was inadequate under professional standards. PX 15, 1; Tr. 334:4–20. Namely, TTV attempted to take two large datasets and determine if a voter on one list (i.e., a potentially ineligible voter based on NCOA data) is the same as a voter on the other list (i.e., the Georgia voter file), which is a difficult analysis ripe with potential problems. Tr. 334:9–20. In academic study, when a scholar undertakes this analysis, he or she clearly defines the exact procedures used for assurance of reliability and replication. Tr. 334:12–20. Such description was missing from TTV’s process. Tr. 417:8–11 (“[I]f I had a complete and accurate and reliable explanation of what True the Vote actually did and what that process looked like, I would have used that information in my analysis. But that didn’t exist.” (emphasis added)).

What Mayer did have access to, however, showed that TTV used the NCOA file and the Georgia voter file in their record linkage process—which was insufficient for proper analysis. Not only has the NCOA file been criticized as carrying a high risk of false positives39 (PX 15, 33; Tr. 344:20–25), it also lacks any unique identifier—“some data field or combination of data fields that uniquely identifies an individual, such that when we see those values in those fields, that we can be certain or very confident that any other time you see those fields we’re talking about the same person.” Tr. 339:14–18; PX 15, 1 & 16–17. Of the two datasets, only the voter file contains a single unique identifier: the voter ID number. Tr. 340:4–25. TTV, however, did not use the voter ID number in making their list. Tr. 341:4–6. Mayer testified that the risk of this omission is that the process could “be matching or linking to the wrong person” and “by definition [it] increases the probability that you’re linking to a different individual in the [other] file.” Tr. 341:12–13, 341:25–342:1.

And indeed, Mayer found evidence of such mistakes in TTV’s list. He found numerous duplicated names and addresses, meaning more than one person had the same first and last name and address. PX 15, 4; Tr. 342:2–11. Mayer found over ten thousand instances of challenged voters where the data lacked any information about the address where the voter moved. PX 15, 5; Tr. 349:13–15. On the list of potentially ineligible voters in Henry County specifically (which included over 9,000 voters), each record had the municipality of the voter’s registration where a voter’s zip code was supposed to be. PX 15, 5 & 27; Tr. 350:17–25. The TTV file generally contained misspellings and variations in recounting city names, indicating “a problem or a lack of quality control.” Tr. 351:4–9; PX 15, 6 & 28. He found 240 instances where the voter in the voter file had a different name than the person in the NCOA file. PX 15, 6; Tr. 35113–19. In 5 cases, the registration address of the voter and the “moved-to” address were the same. PX 15, 6 & 28; Tr. 352:10–15, 363:13–16. A number of voters in the challenge file had “moved-to” addresses within the same county as their prior registration address, which under Georgia law allows the voter to “retain[ ] their eligibility to vote in a presidential or Senate election.” Tr. 353:13–15; PX 15, 6 & 28. In over 6,000 cases, the challenged voter re-registered at the address to which they moved, and in several hundred instances TTV challenged a voter who was not registered in Georgia. PX 15, 6 & 29; Tr. 353:23–25, 354:4–7. Mayer’s overall conclusion based on these errors was that “the challenge file is just rife with errors” and “it just took [his] breath away how sloppy it was.” Tr. 355:11–14; see also PX 15, 42–43 (summarizing the errors found).

Mayer also raised concerns about relying on the NCOA data as a basis for concluding that someone had moved permanently for a residency determination of a voter’s registration address. He specified a number of categories of people who might be on a NCOA list but still eligible to vote at their prior address, such as people in the military and students away at college or university.40 PX 15, 30–32; Tr. 359:1–360:8; see also Tr. 356 (stating that 397 of TTV’s challenged voters “resided literally on a military installation”). He testified that while TTV indicated they removed students and military members from its file, the results show that “it was not done . . . with a lot of accuracy.” Tr. 399:5–10. Mayer admits, however, that removing all students or military members who did not reside on a military base from a challenge file would be difficult given a lack of publicly available information. Tr. 396:19–23, 401:1–2. Mayer accepted some ineligible voters may appear in TTV’s file, but that ultimately “there [were] so many glaring examples of errors in [the] process” the file itself was unreliable on the whole. Tr. 367:1–8.

Mayer further opined in his report extensively on TTV’s “selection” of counties in which to submit challenges. PX 15, 34–37; see also Tr. 347:18–348:10. The Court will not weigh these conclusions and testimony heavily given the great weight of evidence showing that TTV made challenges in counties in which they had volunteer challengers. 41 The Court does not fault Mayer for these conclusions given the information he had available, however. The Court specifically credits Mayer’s trial testimony that even if the counties were not intentionally selected by TTV, this fact does not alter Mayer’s conclusions on the sloppiness or the unreliability of TTV’s dataset. Tr. 380:1–381:19, 385:21–24 (“[M]y conclusions do not depend on these being the counties that were selected. Most of my report is about the observable errors and problems with the files that were produced.”).

Mayer was not asked to and did not assess the actual impact of TTV’s challenges on any actual voter. Tr. 391:20–24. He instead connected the unreliability of the TTV’s challenge file to the well-known and well-regarded cost of voting theory. PX 15, 8. Under this theory, the higher the cost of voting, the less likely that a voter will indeed vote in an election. When a voter’s eligibility is challenged, it is possible that the voter would have to submit additional documentation or attend further proceedings in order to prove their residency, which results in “enormously high” “administrative and time and opportunity costs.” Tr. 368:9–19. Given the size of TTV’s challenge file, Mayer concluded that “the almost certain effect, or certainly the likely effect, as there were, there were voters whose eligibility was challenged and it made it much more difficult for them to vote.” Tr. 371:1–5. Mayer admitted, however, that once a challenge is made, the county Board of Election ultimately determines what additional steps (if any) a voter must go through to prove his or her eligibility to vote. Tr. 369:1–19 (“[T]he costs [discussed] [are created by] the local board”). Mayer also stated that the determination of a voter’s eligibility is up to the county Boards of Election. Tr. 389:24–390:1.

Defendants did not submit any expert rebuttal to Mayer’s report and testimony. They instead attempted to rebut his conclusions with the testimony of Engelbrecht herself.42 Not only does Engelbrecht’s testimony not carry the same weight of expertise as Mayer’s (and the Court weighs it accordingly), but Mayer even addressed some of Engelbrecht’s points about TTV’s process, such as using SmartyStreets or fuzzy logic.43 See, e.g., Tr. 1899:2–1901:23. Mayer testified that these processes were inadequately described in the documents he reviewed and, more importantly, even if TTV did use these processes Mayer “can look at the data itself and conclude that whatever they did, it was not sufficient because there are still errors in the actual files that they created.” Tr. 346:16–18. Similarly, despite Mayer’s submission that middle names were absent from TTV’s file (Tr. 415:11–16) and Engelbrecht’s maintenance that the TTV file included middle names (Tr. 1893:10–12), the Court credits Mayer’s testimony that the inclusion of middle names would not “affect my overall conclusion about . . . the flaws in that process and the accuracy and reliability issues that we can directly see in the data[.]” Tr. 415:16–19.

In sum, the Court finds Mayer’s expert testimony and evidence to be reliable and unrebutted, and thereby affords it due weight in rendering its legal conclusions.