Richman Report Page 1 STEVEN WAYNE FISH, et al., V. KRIS KOBACH, US. District Court for the District of Kansas, Case No. 16-2105 PARKER BEDNASEK V. KRIS KOBACH, US. District Court for the District of Kansas, Case No.2: Expert Report Dr. Jesse T. Richman Old Dominion University Political Science and Geography Department BAL 7000 Noifolk VA 23529 Jesse I: Richman Date Richman Report Page 2 Introduction The focus of this report is on an analysis of data related to the question of whether a substantial number of non-citizens in Kansas registered to vote prior to the implementation of the State?s proof-of-citizenship requirement or are attempting to register to vote after implementation of the requirement. This report draws upon several lines of evidence: several survey datasets collected by a national polling ?rm at the request of the state of Kansas, matches of datasets, and data collected by election workers in the state of Kansas. The overall conclusion of this report is that the proof-of-citizenship requirement being litigated in this case which was put in place by the State of Kansas is functioning as intended to prevent non-citizens from registering to vote. In addition, a substantial number of non-citizens registered to vote prior to its implementation, and a substantial number are still attempting to register to vote. Biography I am an associate professor in the Department of Political Science and Geography at Old Dominion University in Norfolk VA. 1 received my B-Phil in History and Political Science from the University of Pittsburgh, and a Masters and in Political Science ?01n Carnegie Mellon University. I have published papers in a wide range of political science journals including the American Political Science Review, the Journal of Politics, Legislative Studies Quarterly, State Politics and Policy Quarterly, and Electoral Studies. I served as an American Political Science Association Congressional Fellow from 2011 to 2012. In this report I write as an individual and in no way should my assessment be interpreted as speaking for or expressing the opinions of Old Dominion University. I have extensive experience in the ?eld of public opinion. As director of Old Dominion University?s Social Science Research Center I supervised numerous public opinion and evaluation studies. In the ?eld of electoral participation, one of my published papers is particularly relevant: my 2014 peer-reviewed study of non?citizen voting: Richman et a1. (2014).1 My CV is attached. I have served as an expert witness on one case in the past four years. In 2016 I provided an expert report and gave a deposition in the case of Barbara H. Lee et al. v. Virginia State Board of Elections et al., Case No. 3:15cv357 National Evidence of Illegal Involvement in US Elections bv Non-Citizens In this section I brie?y summarize the results of several analyses and studies concerning the question of Whether and to what extent non-citizens register to vote and vote in elections across the United States. The question of non-citizen involvement in elections across the United States is relevant because in the absence of distinctive rules, requirements, or enforcement efforts in Kansas one might well expect that the rate of non-citizen registration and voting would be broadly similar in Kansas to that in other states. 1The full citation for the article is: Richman, Jesse, Gulshan Chattha, and David Earnest. 2014. ?Do Non-Citizens Vote in US. Elections?? Electoral Studies. Volume 36, December 2014, Pages 149?157. Richman Report Page 3 Since the 19203 voting by non-citizens has been prohibited in most US elections.2 However, recent studies have found evidence of a low, but non-zero and potentially substantially consequential, rate of non-citizen involvement in US elections. I focus here on summarizing the broad results of that literature. My 2014 co-authored study in Electoral Studies Quarterly and a more recent working paper are attached to this report and provide more detailed analysis. The Richman et. a1. (2014) article published in the peer reviewed journal Electoral Studies analysed the Cooperative Congressional Election Study (CCES) dataset for the years 2008 and 2010.3 Non-citizens were identi?ed on the basis of self-reported citizenship status on the survey. Non-citizens were weighted to better match nationwide non-citizen demographics. The article found that a substantial number of individuals who reported that they were non- citizens indicated that they were registered to vote or had a Catalyst voter ?le match. In 2008, 19.8 percent had either a ?le match, a self-reported registration, or both. 3.3 percent of non- citizens had both a voter ?le match and self-reported registration status. At either the high end (some indication of registration) or the low end (both indications) to the extent that the CCES sample can be extrapolated to the national non-citizen population, these are potentially substantial numbers of non-citizens. Richman et. a1. (2014) also found that a substantial number of non-citizens reported voting. For example, 1.5 percent both said they had voted in the 2008 presidential election and had a voter ?le match, while 11.3 percent had at least one indicator of voting (a ?le match or a self-report of voting). Ansolabehere, Luks, and Shaffner raised a methodological concern about the analysis in the Richman et. a1. 2014 article.4 They noted the possibility that the level of non-citizen voting observed might potentially be accounted for by response errors made by citizens who erroneously claimed they were non-citizens my mistake. Richman et. a1. (2017) comprehensively responded to this critique.5 They began by noting that the focus of the Ansolabehere, et. a1. analysis on 85 individuals who twice reported that they were non-citizens in the 2010 and 2012 waves of the CCES panel study lacks power. Indeed, a substantial portion of the time one would expect to get precisely the results obtained by Ansolabehere et. a1. despite a true state of the world in which the Richman et. al. estimates were precisely correct. The analysis goes on to show that the error rate assumptions made by Ansolabehere et. al. are likely incorrect, and that even in the face of their assumptions, there is solid evidence of non-citizen involvement in US elections. For example, in 2012 the CCES panel study examined by Ansolabehere et. a1. (2015) contained 5 individuals who had twice con?rmed that they were non-citizens and had a validated voter registration (5 out of 85 is 5.9 percent). 2 Aylsworth, LE, 1931. The passing of alien suffrage. Am. Polit. Sci. Rev. 114-116. 3 Jesse Richman, Gulshan Chattha, and David Earnest. 2014. ?Do Non-Citizens Vote in US. Elections?? Electoral Studies. Volume 36, December 2014, Pages 149?157. 4 Stephen Ansolabehere, Samantha Luks, and Brian F. Schaffner. 2015. ?The perils of cherry picking low frequency events in large sample surveys.? Electoral Studies. 40, 409-410. 5 Jesse T. Richman. Gulshan A. Chattha, and David C. Earnest. 2017. Valid Analysis of a Small Subsample: The Case of Non-Citizen Registration and Voting? Working Paper dated 1/29/2017. Available at 1 5 1 -AnsolabehererResponse1-28- 7.pd'f Richman Report Page 4 Analyses of other types of data from various states also provide some evidence of non? citizen involvement in US elections. For instance, Richman (2016a) discusses the report of a matching analysis done by the state of North Carolina which indicated that nearly one percent of individuals who signed up for the DACA program (popularly known as Dreamers) appeared to be illegally registered to vote in the state.6 Another report found that substantial numbers of individuals had been terminated for non?citizen status in a selection of Virginia Counties.7 Defining a ?Substantial Number? The rates of non-citizen involvement with the US electoral system are potentially large enough to have consequence. The plurality rule used in the US. electoral system and in Kansas ensures that in a close election even a single fraudulent ballot can have enormous consequence. On this issue, the Carter Baker Commission observed ?The problem, however, is not the magnitude of the fraud. In close or disputed elections, and there are many, a small amount of fraud could make the margin of difference.?8 Hence, a number need not be large to be substantial, as the electoral system can magnify the in?uence of even a small amount of fraud. The bipartisan Carter Baker Commission recognized disputes concerning the frequency of fraud, but forcefully articulated the view that the specific volume of fraud is not relevant because any level of fraud can change election outcomes. Indeed, it is not hard to ?nd instances in which fraud has been con?rmed as the cause of an election outcome and election outcomes have consequently been overturned by courts.9 Richman et. a1. 2014 show that non-citizen votes have plausibly in?uenced a number of close election outcomes including the 2008 Minnesota Senate election in which Democrat A1 Franken?s victory gave Democrats a critical 60th vote that helped make possible passage of major legislative achievements in 2009 and early 2010. Non-citizens? preference for parties and candidates do not tend to divide evenly on party lines, creating additional reason for concern about the potential impact of non-citizen votes in elections. Richman et. a1. (2017 attached) reported that 80 percent of non-citizen respondents to the 2012 CCES cross-sectional study indicated that they supported Barak Obama for president. And 92 percent of validated voting non-citizens supported Obama in 2012. Evidence of substantial non-citizen involvement in Kansas elections This section details several different measurement strategies and lines of evidence that were used to evaluate the degree to which non-citizens have registered to vote or attempted to register to vote in Kansas. These include the Cooperative Congressional Election Study, pre- 6 Jesse. T. Richman. 2016a. DACA Non-Citizen Registration Rate Estimate for North Carolina. Downloaded on 12/8/2016 from 7 Jesse T. Richman. 2016b. Non-Citizen Terminated Registration Rates in Virginia Counties. Downloaded on 12/ 8/2016 from counties/ 8This quote is from the Report of the Commission of Federal Election Reform. 2005. Building Con?dence in US. Elections. Page 18. 9The Heritage Foundation outlines several such instances in ?Does your vote count? Ensuring Election Integrity and Making Sure Every Vote Counts.? Downloaded January 5, 2016 from: http://thf Richman Report Page 5 existing registration by newly naturalized citizens, matching of temporary driver license and voter registration lists, surveys of non-citizens on the suspense list, surveys of non-citizens with a temporary driver license, and an analysis of incidentally contacted non-citizens. Cooperative Congressional Election Study The 2006 through 2012 pooled Cooperative Congressional Election Study contacted a small number of self-reported non?citizens who resided in Kansas (14). According to my analysis of this data four of the 14 (roughly 28 percent) indicated that they were registered to vote in the state of Kansas. These numbers suggest that a substantial portion of Kansas non? citizens may have been registered to vote during the period from 2006 through 2012 prior to the implementation of the rules being contested in this case. Indeed, if extrapolated to the current state non-citizen adult population of roughly 114,000 this suggests registration by roughly 32,000 non-citizens (0.285 114,000). Pre?existing Registration by Newly Naturalized Citizens A message and spreadsheet made available to me by the Kansas Secretary of State?s of?ce summarized speci?c non?citizens discovered on the voter rolls in Sedgwick County Kansas. That spreadsheet lists cases for 21 non-citizens who were registered to vote in Sedgwick County (the sheet is attached). All were discovered at naturalization ceremonies attended by Sedgwick County election of?ce staff. The newly naturalized citizens ?lled out applications to register to vote, but the election of?ce determined that those individuals had already registered to vote or attempted to register to vote prior to becoming citizens. In order to put these numbers in context one must have a sense of the total number of new citizens registering to vote. According to information forwarded to me by the Kansas Secretary of State of?ce from Tabitha Lehman, Sedgwick County Election Of?cer, ?Since January 1, 2016, staff from the Sedgwick County election of?ce have attended 62 naturalization ceremonies. We do not know how many individuals applied to become US Citizens at these ceremonies but we do know that 791 new citizens completed voter registration applications and submitted their new naturalization documents. When the registrations and POC documents were brought back to our of?ce, 8 had already applied for voter registration in Sedgwick County and were already in According to the Sedgwick County data roughly 1 percent of newly naturalized citizens since January 1 2016 (8/791) turned out to have previously registered to vote while non-citizens. From 2008 through 2015 roughly 23 thousand individuals naturalized in Kansas according to DHS statistics. If a similar percentage applies across the state, this suggests that prior to naturalization one thousand four hundred non-citizens (0.010114 23,619 239) may have been registered to vote prior to naturalization since 2008. If all current Kansas non-citizen adults registered at the rate found in Sedgwick County, this would imply roughly 1153 (0.010114 114,000) non- citizens registered to vote across the state. Non-citizen registration on this scale has the potential to alter suf?ciently close election outcomes. It is worth noting that the Sedgwick County measure potentially undercounts the portion of newly naturalized citizens in the county who had previously been registered to vote in Richman Report Page 6 the county while non-citizens: one might expect some who had previously been registered to avoid re-registering at the naturalization ceremony. Matching of Temporary Driver License and Voter Registration Lists Staff with the state of Kansas Division of Vehicle Of?ces provided me with the data table displayed as Exhibit 1 below via the of?ce of the Kansas Secretary of State. It appears to indicate that a substantial number of individuals on the Suspense List they had registered to vote without providing proof of citizenship had provided proof of permanent residency status to the Kansas DOV at some time. In a number of instances (29) the proof of permanent residency was provided within 45 days of the voter registration application date. This table implies that a substantial number of permanent resident aliens in the State have also attempted to register to vote in the State. Exhibit 1 - Breakdown of KS SOS Voter Registration Suspension List for NVRA applications forwarded via Kansas Division of Vehicle Of?ces Alien Lawful Permanent Residency Serving as Basis for KDOV lawful presence determination. Percentage of Voter Registrants Number of Voter that Had Registration Applicants Before/ ame wherein Permanent Permanent Time/After Residency Documents Residency Provided KS were provided to KS Birth Total Numb er Status DOV with DOV near or around Year of on Suspense Provided to Documentation of Voter Registration Voter List KDOV at Permanent Application Date 45 Registrant (1 1 .3 .20 1 6) Some Time Residency days) 1998 34 13 3 8% 2 1997 1995 168 34 20% 4 1994 74 36 49% 9 1993 108 21 19% 1 1992 1990 77 12 16% 1 1989 140 17 12% 1 1988 1 71 23 1 3% 4 1987 21 8 24 1 1% 2 Totals 1353 269 20%* 29 *If the applicant subsequently naturalizes/becomes citizen, the applicant will not normally supplement such information with KS DOV. The production of the permanent residency status documentation may have occurred at a prior KS DOV encounter and prior to a subsequent visit to KS DOV that resulted in applicant making a voter registration election. Richman Report Page 7 As I had access to a version of the suspense list as well as the TDL list, I also completed some of my own matching efforts between the lists. While I did not have the detailed information used by the creator of Exhibit 1, I was able to identify 16 individuals I was con?dent were on both lists. Surveys to Assess Non-Citizen Involvement Methodology In January 2017 the State of Kansas hired Issues and Answers, a national polling ?rm to Survey three groups of individuals in order to develop a more speci?c sense of non-citizen involvement in Kansas elections. The number of completes for each group is as outlined below in Exhibit 2. All available numbers from each list were made available and calls continued until the budget had been exhausted. The ?nal Dialing Report from Issues and Answers is attached as document 19. Each sample will be analysed and discussed in turn. Exhibit 2 Number of Completes by Survey Group 1. Suspense Voter List 1621 2. Temporary Driver?s License Holders 104 3. Registered voters in Ford, Seward, 576 Finney, and Grant counties In order to streamline the survey process, the same survey was used for all three groups. Copies of the survey in both English and Spanish are in the appendix. Weights were computed using a statistical procedure known as raking. Identi?cation of non-citizens was primarily on the basis of a pair of survey questions. The ?rst question was ?Are you a citizen of the United States?? Individuals who responded no were then asked the follow-up question ?Of what country or countries are you a citizen?? Individuals were coded as non US citizens if they responded ?no? to the ?rst question and gave an answer of a country other than ?United States? to the second question. For group 2, a secondary strategy for identi?cation of citizens was presence on the Temporary Driver License (TDL) list individuals are on this list because they have applied for a temporary driver license as a non-citizen. As an additional check, respondents from the Temporary Driver List were checked by the Department of Homeland Security (DHS) against their list of naturalized citizens to ensure that individuals on this list had not become US citizens -- naturalized without yet acquiring a new driver license or removing their name from the TDL list. The ?les resulting from this and other DHS matches are attached (Documents 14 through 16). These were forwarded to me by the Kansas Secretary of State Of?ce a?er they received them from DHS. Suspense Voter List The suspense voter list consists of individuals who have applied to register to vote in Kansas but do not yet have veri?ed citizenship status. Analysis of this list serves two key functions. (1) it can identify if and to what extent the requirement to verify citizenship has Richman Report Page 8 prevented non-citizens from registering to vote and (2) it can provide a basis for estimating the degree to which citizens on the list have ready access to the means by which to verify their status. In addition to phone numbers attached to the list, information was matched by national survey research ?rm MSG to identify additional phone numbers. In order to ensure that contacted individuals were in fact on the list, responses were matched to the ?le by ?rst name, age, and gender. A total of 1380 such reSponses were received from individuals on the suspense list. Suspense list respondents were weighted to match the overall list across the following variables: gender, age group, party af?liation, geographic location (?rst three digits of zip code), year of registration, and foreign name. Foreign name was coded by two coders (Dr. Richman and a graduate student hired for the purpose) based upon ?rst, middle, and last names. Inter- rater agreement was 95.9 percent across 5015 cases coded by both. Non-citizens on the suspense list Seven individuals listed on the suspense list who were contacted indicated that they were non-citizens. Each was then asked a follow-up question conceming country of citizenship to con?rm citizenship status. We can be con?dent that the individuals who self-identi?ed as non- citizens were in fact non-citizens as these individuals con?rmed their citizenship in a follow-up question which asked them what country or countries they were citizens of. Responses were as follows, with the number of responses in each category in parentheses: Congo (1), Cuba (1), and Mexico (5). Once respondents to the suspense list survey were weighted to match overall list characteristics in terms of age, gender, party identi?cation, geographic region, year of registration, and name code, the weighted total number of non-citizens increased to 9. These results imply that 0.7 percent of the overall suspense list consists of non-citizens. With the entire suspense list to which the data was weighted containing 17905 individuals, the implication is that roughly 0.009*17905 161 suspense list individuals would have self-identi?ed as non-citizens had the survey questions been responded to by the entire list. Although given the small sample size any inference about the precise magnitude of the non-citizen presence on the suspense list is fraught with substantial uncertainty, this evidence corroborates other evidence g. the Temporary Driver?s License to Suspense List matching discussed above) that non-citizens are present on the suspense list, and in numbers substantial enough to potentially tip a close election. In addition, as the Sedgwick County data makes clear, some non-citizens have removed their names from the suspense list by ?ling a form to cancel their application. The survey of suspense list individuals also points to the limitations of relying exclusively on matches of other lists available to the state in order to prevent non-citizens from registering to vote. Of the seven non-citizens identi?ed through the survey of the suspense list, none were identi?ed through a match of the Temporary Driver?s License list and the Suspense list. Thus use of the Temporary Driver?s Licence list to periodically identify non-citizens on the voter rolls would be only partially ineffective as an alternative to Kansas?s proof-of-citizenship requirement. A large percentage of non-citizens residing in Kansas do not obtain Temporary Richman Report Page 9 Driver?s Licenses (roughly four out of ?ve non-citizen adults have no TDL) and therefore cannot be detected on the voter rolls through this method. Self-reported citizens on the suspense list The second purpose of the suspense list survey was to assess the degree to which self- reported citizens would have dif?culty completing the identity veri?cation requirements. Weighted numbers are reported in the analysis below with unweighted numbers in parentheses. Among self?reported citizens on the suspense voter list, 88.5 percent (89.7 percent unweighted) reported possessing at their home, of?ce, or other location at least one of the three most common identifying documents: a US passport, a US birth certi?cate, or a US naturalization document. This number rises to 91.3 percent (93 percent unweighted) if those who indicated that ?someone else keeps this for you? are included. Another 5.8 percent (4.2 percent unweighted) indicated that they were born in Kansas and thus eligible to receive a free birth certi?cate in conjunction with registering to vote. An additional 0.7 percent indicated that they had at least one of the items from a long list of other types of citizenship documentation. Thus, only 2.2 percent (2.3 percent unweighted) of self?reported citizens on the suspense list indicated that they did not have immediate access to the documentation required to prove their citizenship status to the state of Kansas. A signi?cant concern about these 32 individuals is that they might not in fact be citizens. In an effort to evaluate the validity of these individuals? claims to be citizens without any documentation the Kansas Secretary of State of?ce requested that the Department of Homeland Security bureau of Immigration and Customs Enforcement (ICE) match these individuals against their ?les. '0 According to the ?les I was forwarded by the Kansas Secretary of State, none could be veri?ed as a US citizen, but none could be veri?ed as a non-citizen either. All received the code ?Unknown? which is de?ned as ?Individuals that have never been encountered by DHS, or are possible USCs but databases lack information to verify.? Therefore, these individuals could either be non-citizens who are unlawfully present in the United States, or they could be United States citizens. The demographic characteristics of suspense list individuals without ready capacity to document their citizenship might also be of interest. Unweighted data is reported here. Just over half (53.4 percent) were male. Three (9.4 percent) identi?ed as Hispanic or Latino. Of the 30 willing to answer the question about their race, 21 selected white (65.6 percent), 6 A?'ican- American (18.8 percent), 2 Biracial, and 1 Latino. They divided evenly across parties with 7 Democrats, 7 Republicans, and 18 Unaf?liated according to voter ?le records. They spanned a large age range: 31.3 percent were under 30, and 18.8 percent were over 64. '0 According to the Kansas Secretary of State?s Of?ce, ICE has never before been willing to perform such checks to assist in determining citizenship status of voter registration applicants. ICE agreed to do so in this instance because of the inquiry demanded by the Tenth Circuit in this litigation and because of the relatively limited number of individuals being checked. It is worth noting, however, that where ICE has no record of the individual, the individual could either be an unlawfully present non-citizen or a United States citizen. This ?nding of is limited use to a state attempting to determine whether or not the person is a United States citizen. Richman Report Page 10 Temporary Driver License In order to draw a sample of Kansas non-citizens, Temporary Driver License (TDL) information was matched by national survey research ?rm MSG to identify phone numbers. Ultimately a sample of 104 individuals was reached using those matched phone numbers. Of those individuals, 38 were the named person on the TDL list and gave an age that matched the age given in the TDL ?le while the remainder were another individual who answered that phone number. All respondents were asked about voter registration, and about citizenship status along with demographic and identi?cation questions. Names of matched TDL individuals were matched by the Department of Homeland Security ICE at the request of the Kansas Secretary of State?s of?ce. One individual was identi?ed as a possible United States citizen, and another did not have a match (Unknown) but indicated in the survey that he was not a United States citizen. The remainder were veri?ed by ICE as non-citizens. The analysis below excludes the one individual who said he was a citizen and was identi?ed by ICE as a possible United States citizen. The analysis below reports estimates weighted by age as this is the only demographic variable available from the TDL list. Since the age distribution of the sample and the age distribution of the TDL list were very similar, the estimates are extremely similar to the unweighted estimates, so I have not reported those here. Among the 37 respondents named on the TDL list with veri?ed ICE status as non- citizens, 6 (16.5 percent) indicated that they had either registered to vote or attempted to register to vote in the State of Kansas. This is a substantial rate of registration, and if the small sample analysed can be generalized to the broader TDL list of approximately 21,000 names, then it suggests that more than three thousand (21,090 0.165 3,480) individuals on that list have registered to vote or attempted to register to vote in Kansas. This is a substantial number of registered noncitizens; this large pool of voters could easily determine the outcome of an election in Kansas. The TDL non-citizen list does not include all non-citizens of Kansas. My impression is that it is composed of legal permanent resident aliens and aliens possessing temporary visas (non-immigrant aliens) who applied for a temporary driver license. It does not include unlawfully present non-citizens and it does not include non-citizens who chose not to obtain a driver?s license. Census estimates suggest there are roughly 114,000 Thus, generalization beyond the TDL to the broader population of non-citizens in Kansas has some limitations. However it does provide a basis for estimating the non-citizens who have registered to vote in Kansas. If individuals on the TDL non-citizen list are assumed to be representative of the broader non-citizen population of Kansas, then the results above would imply that a very substantial number and portion of non-citizens in Kansas have registered to vote or attempted to register to vote? more than 18,000 (114,000 0.165). Obviously even registration by a fraction of this number would be substantial, and it might also be potentially consequential in determining the outcome of an election. This does not necessarily imply that this many non-citizens succeeded in registering to vote, however. The speci?c question asked concerning voter registration was ?Have you Richman Report Page 11 registered to vote or attempted to register to vote in the state of Kansas?? Hence, respondents are not necessarily asserting that they are currently registered. Indeed, being informed about Kansas requirements concerning documentation of citizenship status may have prevented some from completing a registration application. But it is also clear that many have attempted to register to vote; as discussed above, a substantial number of TDL individuals have been matched by the State of Kansas with the Suspense list. The TDL list also provides a useful opportunity to evaluate the extent to which those who are in fact non-citizens lie about citizenship status on surveys.ll Indeed, ?ve of the named individuals contacted from the TDL list who subsequently had their immigration status veri?ed by ICE none-the-less claimed on the survey that they were citizens. These 5 TDL respondents to the survey individuals whom we are fairly con?dent are in fact non-citizens for multiple reasons claimed to be citizens on the survey. Misstating citizen status also appears to be associated with indicating that they had registered to vote. Although three members of both groups indicated that they had registered or attempted to register to vote, the difference in the frequency with which individuals admitted to this behaviour was striking and statistically signi?cant (p 0.022 Fisher?s Exact Test) consistent with the conjecture offered by Richman et. al. 2017 that non-citizens sometimes lie about citizen status when admitting to election involvement. Registered Voter List Individuals were sampled from those who registered to vote between 2007 and 2012 in four Kansas counties with large non-citizen populations Ford, Finney, Grant, and Seward counties. All of the named individuals contacted from the registered voter list indicated that they were citizens of the United States. Hence, this survey provides no evidence of non-citizen registration to vote. That said, the power of the test is not very strong, as a low but non-zero rate of non-citizen registration might well have been missed in a survey with this small sample size, and as noted above some non-citizens may lie about their citizenship status on the survey. The names of 12 individuals from this list who indicated they had no documents with which to prove citizenship were analysed were identi?ed as naturalized citizens or US citizens while the other 7 were ?agged as ?unknown.? Incidentally Contacted Non-Citizens An initial screening question was asked of all individuals reached from each of the three lists above: the caller asked ?For validation purposes, can I please get your ?rst name?? If the named individual from the list was not the respondent, the caller asked for that individual. But if that individual was unavailable, the caller asked ?Are you willing to complete this short survey?? Nineteen of these 165 incidentally contacted individuals subsequently indicated that they were non-citizens. The analysis below is of those 19 individuals -- incidentally contacted non-citizens who were willing to complete the survey but were not on the initial list. These individuals came This problem occurs in a variety of surveys, though it may be more pronounced in a survey which precedes the question about citizenship with a question that asks if one is registered to vote. Richman Report Page 12 from all three sample groups. Of the 74 individuals initially contacted with numbers associated with the Suspense List, two indicated that they were non-citizens. Of the 43 called at numbers associated with the Temporary Driver License list, 14 were non-citizens. Of the 48 contacted from the Registered Voter list, three were non-citizens. One of the 19 incidentally contacted non-citizens indicated that he or she had ?registered to vote or attempted to register to vote in the state of Kansas.? Although the sample size is extremely small, and any estimates are accordingly very uncertain (the margin of error is 10.1 points), the estimate from this data suggest a registration attempted registration rate of 5.3 percent. If extrapolated (and again the uncertainty here is very high) to the Kansas non-citizen population as a whole, this implies that about 6,000 may have registered to vote or attempted to register to vote in Kansas. This number, similar to other estimates provided above, suggests that a substantial number of non?citizens have registered to vote or attempted to register to vote in the state of Kansas. Conclusion This study has brought together multiple lines of evidence that indicate a substantial number of non-citizens are registered to vote or attempting to register to vote in the State of Kansas. Data matches, the experiences of county election of?cials, and surveys of multiple groups of individuals nearly all provide evidence consistent with the idea that a substantial number of non-citizens have registered to vote or are attempting to register to vote. To restate the numbers explained above, an estimated 3,480 non-citizens possessing TDLs have registered or attempted to register to vote. But this is only a subset of the non-citizen population in Kansas. Looking at the total noncitizen population in the State, the total might likely be between 1153 (Sedgwick County data), 6,000 (the number derived from the 5.8 percent registration rate of incidentally-contacted non-citizens) and over 18,000 (the number derived from the 16.5 percent registration rate of TDL non-citizens). If even the lowest estimate was accurate, this is a potentially politically consequential and thence substantial percentage of non-citizens on the voter rolls in Kansas. It is more than enough to potentially alter election outcomes at the local and state level. I examined election outcomes for the State of Kansas searching primary and general election outcomes for the state legislature, state-wide of?cer elections, and congressional elections from 2000 on through 2016 in search of examples of close elections that potentially could have been in?uenced by non-citizen participation. The attached document (18) summarizes the results of this analysis. I found 33 elections decided by less than 100 votes. Of particular note were two general election races for the Kansas House of Representatives (2000 in district 8 and 2006 in district 16). They were decided by 0.04% and 0.036% of the two party vote respectively, and both were won by Democratic candidates who held a three vote advantage. While there is no way to know at this point whether the outcomes in these and other close races were shifted by non-citizen votes it is obvious that a few non-citizen votes could have changed these outcomes, particularly given the evidence discussed above which suggests non-citizens tend to support Democratic candidates at least at the presidential level. Richman Report Page 13 In conclusion, the estimated number of non-citizens registered to vote or attempting to register to vote in Kansas is arguably substantial. Several different lines of evidence suggest that this number is likely in the thousands. If thousands of non-citizens are on the voter rolls in a state the size of Kansas, this potentially poses a substantial problem. While I lack information speci?c enough to attempt an estimate of the likelihood that any speci?c election outcomes have been in?uenced by non-citizen participation, it is entirely possible that in some instances votes cast by non-citizens have altered the outcomes of close elections, and very competitive elections in the future will also be subject to this risk. Richman Report Page 14 Statement of Compensation My rate is $100 per hour for my work preparing this expert report. I will charge a rate of $250 per hour for time assooiated with deposition and testimony, exclusive of travel and other expenses. List of Attachments Exhibits as Numbered in Separate Files 9. 10. 11. 12. 13. 14. 15. 16. Survey Instrument (English) Survey Instrument (Spanish) Richman Earnest and Chattha Electoral Studies Article Richman Earnest and Chattha Working Paper Sedgwick County Spreadsheet Richman CV (contains list of publications). Survey Dataset Full dataset from Issues and Answers Survey Dataset Merged with Suspense File Data Survey Dataset Merged with TDL Age Data Registered Voter List for Selected Counties 2007 Registered Voters for Selected Counties Kansas Voters Suspense List Temporary Driver?s License List TDL Names with ICE Immigration Status Codes Registered Voters with ICE Immigration Status Codes Suspense List Voters with ICE Immigration Status Codes 17. Name Coding File from Suspense List 18. Analysis of Recent Close Kansas Elections