<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Charles Stewart on Medium]]></title>
        <description><![CDATA[Stories by Charles Stewart on Medium]]></description>
        <link>https://medium.com/@cstewartiii?source=rss-b3b359353b26------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*e9RuesuYeCsNWmRk.jpg</url>
            <title>Stories by Charles Stewart on Medium</title>
            <link>https://medium.com/@cstewartiii?source=rss-b3b359353b26------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 07:19:22 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@cstewartiii/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Amid a Pandemic and Social Unrest, Turnout Holds Its Own in the Primaries.]]></title>
            <link>https://medium.com/mit-election-lab/amid-a-pandemic-and-social-unrest-turnout-holds-its-own-in-the-primaries-d718335cd0e7?source=rss-b3b359353b26------2</link>
            <guid isPermaLink="false">https://medium.com/p/d718335cd0e7</guid>
            <category><![CDATA[voting]]></category>
            <category><![CDATA[covid19]]></category>
            <category><![CDATA[elections]]></category>
            <dc:creator><![CDATA[Charles Stewart]]></dc:creator>
            <pubDate>Mon, 15 Jun 2020 18:46:56 GMT</pubDate>
            <atom:updated>2020-06-15T18:46:56.153Z</atom:updated>
            <content:encoded><![CDATA[<h4>What this Means for November</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vIGgByJ66rwE3GIqzKr-KQ.jpeg" /></figure><p>The COVID-19 pandemic has pushed the 2020 presidential primary season far down the nation’s political agenda. Without large public gatherings to focus the public attention’s on the messages of the main candidates, and with the Democratic presidential nomination coming to a surprisingly swift conclusion, the primary season has quite literally gone underground, to Joe Biden’s basement and Donald Trump’s bunker.</p><p>Because Americans’ minds have been on other things — COVID-19 and racial injustice — it would be unsurprising if voters stayed away from the polls during the primaries. And yet, just the opposite has occurred. With a notable dip in Ohio’s primary late April, turnout in Democratic primaries have been at about the same level as 2016. Republican primary turnout has been about half of what it was four years ago, but this is understandable, given the lack of a serious challenge to President Trump’s renomination.</p><p>At the <a href="https://electionlab.mit.edu/">MIT Election Data and Science Lab</a>, we have been studying turnout and how voters cast their ballots, with an eye toward understanding how they are adapting to these challenging times. We have been following turnout levels in the primaries, comparing them to 2016, to see whether the pandemic has dampened voter interest. This comparison makes perfect sense for the Democratic primary, since 2016 was fiercely contested, as was 2020 — at least until Super Tuesday. It’s a little less informative on the Republican side, since 2016 also saw intense Republican competition, and thus competition-driven turnout, unlike this year.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FpRiyab_Qw_w7D3eDAx6gQ.png" /></figure><p>As the accompanying figure shows, turnout in the Democratic primaries mostly exceeded the figures seen in 2016 during the months of February and March, in some cases, dramatically so. (Turnout in Virginia and Washington were over 50% greater in 2020 than in 2016, for instance.)</p><p>Republican turnout was about half what it was in 2016, with a few notable exceptions. Two of those exceptions — Alabama and California — were states that yoked the presidential primary with the state primary that chose nominees for state and local races to face off in the fall. The third, Washington, mails ballots to all registered voters, which could make it subject of another post in the future.</p><p>March 17 marked a break in the primary calendar, as all but one state that had originally scheduled primaries in late March and early April postponed them to later dates. Wisconsin (in)famously failed to postpone its primary, which resulted in ballots being cast on April 7. <a href="https://www.washingtonpost.com/politics/2020/04/15/wisconsin-primary-had-extraordinarily-high-voter-turnout/">As I wrote at the time</a>, despite the swirl of court cases that cast doubt over whether the Wisconsin primary would be postponed, the rapidly accelerating health crisis in the Badger State, and difficulty keeping in-person polling places open in the cities, turnout in the primary was almost exactly what we would expect, in statistical terms. (The statistical model took into effect how competitive the presidential primary and supreme court election were.)</p><p>As primaries have come back into view in May and June, turnout has returned to what we were seeing in March — Democratic turnout has generally exceeded 2016 levels, whereas Republican turnout has been about half of what we saw four years ago.</p><p>The presidential primaries will continue into early July, but increasingly they will give way to state and local primaries that will determine general election match-ups in the fall. These primaries will not only help us gauge overall interest in electoral politics, but help us to see interest in moving away from Election Day to early in-person voting and mail balloting. The Election Lab will be releasing a series of reports in the coming weeks that drill down into the details of voting during the primaries, with an eye to what they can tell us about what to expect in November.</p><p>Until then, it is clear that Americans continue to be engaged in the 2020 election, and that we should continue to see robust turnout in the upcoming primaries, as well as solid — and maybe even record-breaking — turnout in November.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d718335cd0e7" width="1" height="1" alt=""><hr><p><a href="https://medium.com/mit-election-lab/amid-a-pandemic-and-social-unrest-turnout-holds-its-own-in-the-primaries-d718335cd0e7">Amid a Pandemic and Social Unrest, Turnout Holds Its Own in the Primaries.</a> was originally published in <a href="https://medium.com/mit-election-lab">MIT Election Lab</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What “Should” We Expect about Turnout in Wisconsin?]]></title>
            <link>https://medium.com/@cstewartiii/what-should-we-expect-about-turnout-in-wisconsin-a49ac28d818?source=rss-b3b359353b26------2</link>
            <guid isPermaLink="false">https://medium.com/p/a49ac28d818</guid>
            <category><![CDATA[election-2020]]></category>
            <category><![CDATA[wisconsin]]></category>
            <category><![CDATA[wisconsin-primary]]></category>
            <category><![CDATA[voter-turnout]]></category>
            <dc:creator><![CDATA[Charles Stewart]]></dc:creator>
            <pubDate>Mon, 13 Apr 2020 14:57:08 GMT</pubDate>
            <atom:updated>2020-04-13T14:57:08.219Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/369/1*hRSSp3DSEo4l1q3j-yk3iQ.png" /><figcaption>Source: Wisconsin “Blue Book,” 1973.</figcaption></figure><p>While we wait for the election results to be released later tonight, it would be useful to predict <em>a priori</em> what we would expect turnout to end up being. With an expectation established, based on past voting patterns in Wisconsin, it will be easier to assess how the unusual circumstances surrounding the primary may have affected turnout and the vote shares for candidates.</p><p>Over in <a href="https://electionupdates.caltech.edu/?p=6827&amp;preview=true">Election Updates</a>, I have posted an extensive analysis of what statistical analysis reveals we would normally expect about turnout in the Wisconsin primary that was conducted last Tuesday. This is an interesting question, because the unusual circumstances surrounding the primary — held amidst the escalating coronavirus crisis, confusion over conflicting legal proceedings, and apparent collapse of the availability of in-person voting in some jurisdictions — raises questions about whether the election can be considered legitimate.</p><p>My piece in Election Updates doesn’t address the legitimacy question, but rather, provides an insight into what we would expect turnout to be in the primary, based on prior experience with the Wisconsin primary since 1948. That analysis suggested that if this were a “typical” primary held during this period, we would expect that between 1.1 and 1.3 million people voted last Tuesday. There is a caveat to this prediction — the recent history of heightened partisanship in Wisconsin may have ratcheted up turnout in Wisconsin elections, as was evidenced in the 2016 presidential primary. In that event, that is, Wisconsin is in a new hyper-partisan era, we might expect turnout to be in the range of 1.8–1.9 million voters.</p><p>Already, Wisconsin voters have returned 1.1 million absentee ballots, so we know that the lower level of the estimate has been reached. The results, which will begin to be released Monday evening, will determine where in the “expected” range of voter turnout Wisconsin ended up landing.</p><p>If you’re interested in the details of the analysis, I invite you go to over to <a href="https://electionupdates.caltech.edu/?p=6827&amp;preview=true">Election Updates </a>and check it out.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a49ac28d818" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Believe It or Not, Americans Are Confident about the Security of the Upcoming Election]]></title>
            <link>https://medium.com/@cstewartiii/believe-it-or-not-americans-are-confident-about-the-security-of-the-upcoming-election-8be4cf789419?source=rss-b3b359353b26------2</link>
            <guid isPermaLink="false">https://medium.com/p/8be4cf789419</guid>
            <category><![CDATA[cybersecurity]]></category>
            <category><![CDATA[elections]]></category>
            <dc:creator><![CDATA[Charles Stewart]]></dc:creator>
            <pubDate>Tue, 30 Oct 2018 14:41:02 GMT</pubDate>
            <atom:updated>2018-10-30T14:41:02.264Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*o_uBRQZg8HMQXeH4iBpJZw.jpeg" /></figure><p>You can’t swing a dead cat without reading a story or an op-ed, often from reputable source, bemoaning the level of cyber preparation to defend the upcoming election. Given the level of alarm often expressed in print, it’s natural to ask whether the public is similarly alarmed that the upcoming election is about to be stolen by hackers and the inattention of election officials.</p><p>The answer appears to be “no,” at least according to some public opinion work I’ve done over the past few months into this question. I recently wrote an extended post over at Election Updates on this question, and I invite you to read the whole story there by clicking on <a href="https://electionupdates.caltech.edu/2018/10/10/americans-are-slightly-more-confident-about-fending-off-computer-hacking-in-the-upcoming-election/">this link.</a></p><p>The Election Updates article was based on a recent study about attitudes related to election cyber security and election hacking. I had done a series of parallel studies in the spring, and one thing I was interested in was how things had changed over the precedent half year or so. Here are the main take-aways:</p><ul><li>Over 3/5 of respondents (63%) were confident that local election officials were prepared to guard against election interference due to computer hacking in the upcoming election. This was up about 9 points since last June. (The great majority of respondents gave a “don’t know” response, rather than expressing a lack of confidence.)</li><li>Although both Democrats and Republicans had grown more confidence since June, Republicans had grown more confident by a greater amount.</li><li>Respondents who pay closer attention to the news are more confident than those who don’t.</li></ul><p>Americans aren’t responding to a steady diet of alarming stories by concluding that all is lost. At the very least, attitudes are holding firm, and there is some evidence that people are gaining some assurance about the resilience of the system.</p><p>I plan on re-asking these questions, this time retrospectively, next week after Election Day.</p><p>I suspect that the public’s answers to security-related questions will follow what happens Tuesday night and Wednesday morning. Assuming no major confirmed election cyber attacks occur between now and then, I suspect that “confidence” numbers will shift according to which party is seen as having done better in the election. The wildcard is the fact that control of the House and the Senate may revolve around the resolution of a few close elections, and what happens there may color attitudes more broadly.</p><p>In any event, check out Election Updates in the days following the election to discover what I find out in detail.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8be4cf789419" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Auditing: The Little Indicator that Could]]></title>
            <link>https://medium.com/mit-election-lab/auditing-the-little-indicator-that-could-ba7f42aa234c?source=rss-b3b359353b26------2</link>
            <guid isPermaLink="false">https://medium.com/p/ba7f42aa234c</guid>
            <category><![CDATA[politics]]></category>
            <category><![CDATA[elections]]></category>
            <category><![CDATA[auditing]]></category>
            <dc:creator><![CDATA[Charles Stewart]]></dc:creator>
            <pubDate>Thu, 11 Oct 2018 19:48:24 GMT</pubDate>
            <atom:updated>2018-10-11T19:57:56.285Z</atom:updated>
            <content:encoded><![CDATA[<p><em>The </em><a href="https://electionlab.mit.edu/"><em>MIT Election Data and Science Lab</em></a><em> manages the </em><a href="https://elections.mit.edu/"><em>Elections Performance Index</em></a><em> (EPI), an objective, nonpartisan assessment tool evaluating U.S. election administration. Following the recent </em><a href="https://medium.com/mit-election-lab/u-s-election-management-is-improving-5a7cc18256c8"><em>update</em></a><em> to the index with data from 2016, we are dedicating a series of posts to exploring the EPI’s underpinnings.</em></p><p><em>Our post today was penned by Charles Stewart III, who is the founding director of the MIT Election Data and Science Lab.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Yo0pfuV_o5Oah8_sYC8IEA.jpeg" /><figcaption>Source:<a href="https://www.flickr.com/photos/gringer/5096129532"> David Eccles</a></figcaption></figure><p>Of the seventeen indicators that make up the <a href="https://elections.mit.edu/">Elections Performance Index</a>, post-election auditing is the one that’s tailor-made for elections geeks. The term “auditing” is drawn straight out of the most buttoned-down profession, and it suggests a no-nonsense approach to making sure that everything was done according to Hoyle. Election audits aren’t identical to financial audits. However, the idea behind election audits is similar — to give assurance to voters and candidates that the election returns accurately depict the outcome of an election.</p><p>The post-election auditing indicator in the EPI is pretty simple and basic. It records whether a state requires a post-election audit of the vote total, either through statute or administrative directive. It is not based on the type of audit performed — risk-limiting vs. fixed-percentage, for instance — nor is it based on the outcome of the audits. All it asks is: “audits, yes or no?”</p><p>But, that might change. It is now widely recognized that risk-limiting audits are the preferred way to test the accuracy of the outcome. Efforts are also under way to extend auditing to other parts of the election system. As a consequence, one shouldn’t be surprised to see the details of this indicator change for the 2020, or even 2018, EPI.</p><h3><em>Why audit elections?</em></h3><p>The idea behind auditing election results is an old one. As early as 1965, the California election code required that the accuracy of an automated vote count be verified using a public process of manually tallying the votes in 1% of precincts.</p><p>Greater attention began to be paid to post-election audit requirements in the immediate aftermath of the 2000 presidential election. Two events spurred this interest. The first was the poor performance of the punch-card voting machines in Florida during the 2000 election. The second was growing distrust of electronic voting machines that had no paper record of the individual ballots cast. In the former case, audits were seen as a way to identify machines that might be failing. In the latter case, advocacy in favor of paper “back-ups” of electronic voting systems, called “voter-verifiable paper audit trails,” would be effective only if the paper records were actually compared in a systematic way to the electronic tallies.</p><p>By the time the EPI was developed around 2010, a robust field of post-election auditing was beginning to take shape. More and more states were writing post-election audit requirements into their codes. Scientists and engineers were beginning to develop practical procedures that would allow election officials to apply increasingly sophisticated sampling techniques to post-election audits. <a href="http://electionaudits.org/files/Brennan%20Center%20Report%20on%20Post-Election%20Audits.pdf">A very helpful summary</a> of the state of thinking about post-election audits in the late 2000s was published by the Brennan Center in 2007.</p><h3><em>Auditing has become more common</em></h3><p>The <a href="https://elections-dev.mit.edu/#indicatorProfile-PEAR">EPI</a> can be used to show the growth in the utilization of post-election audits over the past decade. As the accompanying figure shows, the number of states with an auditing requirement has grown from 23 in 2008 to 34, plus the District of Columbia, in 2016. As I have written in a <a href="https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3249830">recent retrospective essay</a> on the EPI, the rise in the adoption of post-election auditing over the past decade is one of the most important reasons why the average overall EPI score has increased during this period.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/461/1*VWP6IBYmCeD6_fEgHlaa5A.png" /><figcaption>Increase in the number of states requiring post-election audits. Source: <a href="https://elections.mit.edu/#indicatorProfile-PEAR">Elections Performance Index</a></figcaption></figure><p>Focusing only on the rise in post-election auditing over the past decade misses the variability of the auditing programs that the states have adopted. <a href="https://www.verifiedvoting.org/state-audit-laws/california/">California,</a> for instance, still maintains its 1% precinct audit, although <a href="https://leginfo.legislature.ca.gov/faces/billTextClient.xhtml?bill_id=201720180AB2125">recent legislation</a> opens the door for other methods. <a href="https://www.scvotes.org/data/AuditDesc.html">South Carolina</a>, which uses electronic voting machines without a voter-verifiable paper audit trail (VVPAT), checks to see if the various electronic reports generated by the voting machines are consistent with each other. <a href="http://bcn.boulder.co.us/~neal/elections/corla/">Colorado</a> has begun implementing risk-limiting audits, and other states have passed legislation to mandate them in the future. A challenge for the future of the EPI is to reflect the variety and sophistication of auditing requirements as they evolve in the states.</p><h3><em>The new kid in town: risk-limiting audits</em></h3><p>The most typical form of post-election auditing is called the “fixed-percentage audit.” Like the name implies, it involves selecting a fixed percentage of the ballots cast (or precincts), hand-counting the ballots, and then comparing the hand-count to the automated count.</p><p>With “risk-limiting audits,” officials select a randomly chosen set of ballots and hand-count them carefully. Then, the officials check to see whether the winner reflected in the sample of ballots is consistent with the winner from the unofficial tally of all the ballots.</p><p>An important difference between risk-limiting audits (RLAs) and fixed-percentage audits is that the number of ballots chosen to audit varies under RLAs. If the unofficial tally shows the winner prevailing by a large margin, fewer ballots are sampled, compared to a situation where the unofficial tally shows a narrow margin.</p><p>The term <em>risk-limiting</em> refers to one important feature of RLAs. With an RLA, the election jurisdiction has to decide the confidence they wish to have that the automated vote-count is the correct one. Said another way, the jurisdiction has to decide how much risk they are willing to take that the audit will incorrectly conclude that the automated vote-count is correct. If a state sets a risk limit at 10%, for instance, then if the automated count declares the wrong winner — either because of fraud or error — the sample of ballots will fail to catch the error 10% of the time, but will catch it 90% of the time. If the risk limit is set at 1%, there is only a 1% chance that the audit will fail to catch an erroneous result.</p><p>There is a trade-off in setting the risk limit and the number of ballots that must be randomly chosen and hand-counted. The smaller the risk accepted, the more ballots that need to be hand-counted. At the extreme, if a state accepts 0% risk, then it would hand-count all ballots, which is what a recount is.</p><p>The mechanics of an RLA are such that if enough discrepancies are revealed in the counting of the sampled ballots, one of two things can happen. First, a larger sample can be drawn, to increase the confidence in the audited results. Second, all ballots can be recounted.</p><p>The idea of RLAs has been developed by statisticians working for many years. A well-titled introduction to RLAs, “<a href="https://www.stat.berkeley.edu/~stark/Preprints/gentle12.pdf">A Gentle Introduction to Risk-limiting Audits</a>,” is a good place to start in learning more about them.</p><p>Colorado passed a law in 2009 requiring risk-limiting audits, implementing it in 2017. New Mexico has had a type of adjustable-percentage audit that functions like a hybrid between fixed-percentage and risk-limiting audits. At least five other states — California, Colorado, Ohio, Rhode Island, and Virginia — have passed laws moving those states in a direction that favors RLAs.</p><p>An ad hoc committee of the National Academies of Science, Engineering, and Medicine recently released a report, entitled <a href="https://www.nap.edu/catalog/25120/securing-the-vote-protecting-american-democracy"><em>Securing the Vote</em></a><em>,</em> which called for all states to implement RLAs within the next ten years.</p><p>Under most circumstances, RLAs have the advantage of requiring the inspection of a smaller number of ballots in an audit than fixed-percentage audits. RLAs don’t solve all auditing problems, however. For instance, the small number of ballots reviewed make it hard to uncover certain types of fraud, unless the discrepancies are very large. For that reason, I would argue that RLAs are a powerful tool, but should not be the only post-election auditing tool in the election administrator’s belt.</p><h3><em>Wait: There’s more!</em></h3><p>Vote-counting is a critical task in election administration, but it is not the only critical task. While post-election audits have gotten the most attention within the topic of assuring the quality of election administration, other problems can arise in election administration that undermine the integrity of outcomes. Voters can be given wrong ballots. Candidates can be missing from ballots. Voter registration rolls can be incorrect. The wrong addresses can be assigned to election districts. The list could go on.</p><p>Elections are a detail-rich activity, and efforts have been made to develop methods of auditing all those details. For instance, Michigan runs “procedural audits,” to ensure that local jurisdictions are following good procedural practices before, during, and after Election Day. Other states run logic and accuracy tests on ballot counters after the election, to check whether a “test deck” of ballots is properly tallied. New methods are being developed by academics to help assess whether election districts were drawn correctly, while academics and private organizations have worked on ways to ensure that voter registration records are correct.</p><p>The National Conference of State Legislatures has a very useful <a href="http://www.ncsl.org/research/elections-and-campaigns/post-election-audits635926066.aspx">Web page</a> that summarizes what states are up to on the broader election auditing front.</p><h3><em>Keeping up with the world of auditing through the EPI</em></h3><p>As I noted at the outset, the post-election auditing indicator in the EPI is very simple, reflecting the world of election auditing around 2010. Over the past decade, this world has become much more complex. On just the topic of post-election ballot audits, the consensus is emerging that RLAs are the gold-standard for providing assurance that an election’s outcomes were correct. Shouldn’t the EPI reflect which states are using RLAs and which aren’t? Should the EPI reflect more thoroughly the full range of election auditing procedures that states mandate?</p><p>The EPI has served a useful purpose in identifying the states that are committed to post-election ballot audits through their laws and regulations. To stay relevant, the EPI now needs to dig deeper into what states are actually doing in those audits, and identify which have the most thorough programs of checking that the details were right.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ba7f42aa234c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/mit-election-lab/auditing-the-little-indicator-that-could-ba7f42aa234c">Auditing: The Little Indicator that Could</a> was originally published in <a href="https://medium.com/mit-election-lab">MIT Election Lab</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What’s a Residual Vote, and What’s It Doing in the Elections Performance Index?]]></title>
            <link>https://medium.com/mit-election-lab/whats-a-residual-vote-and-what-s-it-doing-in-the-elections-performance-index-e455289acb30?source=rss-b3b359353b26------2</link>
            <guid isPermaLink="false">https://medium.com/p/e455289acb30</guid>
            <category><![CDATA[voting]]></category>
            <category><![CDATA[politics]]></category>
            <category><![CDATA[medsl-guest]]></category>
            <dc:creator><![CDATA[Charles Stewart]]></dc:creator>
            <pubDate>Wed, 29 Aug 2018 14:59:21 GMT</pubDate>
            <atom:updated>2018-08-29T14:59:21.903Z</atom:updated>
            <content:encoded><![CDATA[<p>What does it mean when a voter’s ballot doesn’t contain a vote for president? The answer to this question reveals what we know, or think we know, when we look at the <em>residual vote rate,</em> one of the seventeen indicators in the Elections Performance Index.</p><p><strong>What <em>Is</em> the Residual Vote Rate?</strong></p><p>But first, let’s review what a residual vote is and why it might be important. The measure was invented by the Caltech/MIT Voting Technology Project in the weeks following the 2000 presidential election, as a way to assess how many “lost votes” were caused by poorly performing voting machines.</p><p>A residual vote is either an over- or under-vote for a particular race, usually for president. The residual vote <em>rate</em> in a state is simply the total number of over- and under-votes divided by the number of people who turned out to vote. (Election-geek detail: Because most states do not separately account for over- and under-votes, the number of residual votes is usually calculated by taking the number of people who turned out and subtracting the number of votes cast for candidates.)</p><p>Generally speaking, the residual vote rises and falls over time for one of three reasons:</p><ul><li>voters might abstain from casting a vote for that office at all;</li><li>voters might be confused and either fail to mark their ballot or mark too many choices; or</li><li>a problem with the voting technology might cause the voter’s choice to go unrecorded.</li></ul><p>When the EPI was first developed, research suggested that the abstention rate for president was roughly constant over time and across states. Therefore, changes in the residual vote rate were most likely caused by fluctuations in voter confusion and technology failure. Because these both lead to lost votes, the inclusion of the residual vote rate in the EPI is intended to quantify the degree to which the machinery of ballot-marking and vote-counting serves voters, for better or worse.</p><p><strong>How Does the Residual Vote Rate Vary?</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EZKTY1QIejh_Zr0SKzuasg.png" /><figcaption>Variability in Residual Vote Rates in 2016</figcaption></figure><p>In 2016, the residual vote rate ranged from 0.0039% in Nevada to 3.1% in Maine. (Nevada’s rate was the lowest observed since the EPI project began; more on that in a moment.) The accompanying map shows the states with the lowest rates (colored darkest green) and the highest rates (darkest orange), with shades in between. (The grey states on the map do not reliably and uniformly report turnout, and so are excluded from the calculations.)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*wXZpVX1oExYqMTFo_-kqdQ.png" /><figcaption>Nationwide Residual Vote Rates from 1980 to 2016</figcaption></figure><p>The residual vote rate not only varies across states; it has varied across time, as well. Take a look at the accompanying graph, where I have plotted the nationwide residual vote rate since 1980. Two things bear noticing. First, the biggest drop occurred between 2000 and 2004, when it fell by nearly a full point. Second, there was a noticeable up-tick in 2016.</p><p>The stories behind these two big changes help illustrate what drives the residual vote rate, and what it can tell us about elections.</p><p><strong>Why Does the Residual Vote Rate Change?</strong></p><p>The residual vote rate is driven by the number of over- and under-votes in a presidential election. What would affect these numbers? Let’s start with under-votes.</p><p>There are many reasons why a vote doesn’t register on a ballot. Generally speaking, the most common reason is that voters abstain. They don’t like the candidates running for president, or they are indifferent between them. The less excitement there is about presidential candidates, the more under-votes there will be, causing the residual vote rate to rise.</p><p>The 2016 uptick in the residual vote rate is probably due to an increase in abstentions in the last presidential election. In a conference paper that I co-wrote with collaborators last January, we showed that more moderate voters — from both parties, but especially Republicans — were more likely to leave their choice for president blank. (The paper is being revised, but there is a link to it <a href="https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3225197">here</a>. Note that an update will be posted shortly.) The result was a small increase in the residual vote rate of around half a percentage point compared to the recent past. This might seem trivial on its face, but when the residual vote rate has recently averaged around 1%, it’s a big deal.</p><p>Voter confusion and malfunctioning machines can also lead to under-votes. The punch card problems that were the centerpiece of the 2000 Florida recount are the best-known examples, but they aren’t the only ones. In old lever machines, for instance, the gears of the “odometer” counting wheels used to get stripped, causing some votes not to be registered, unbeknownst to the voter..</p><p>Over-votes can also be caused by poor ballot design or subtler machine problems. Since HAVA became law, most optical scanners have been required to return any ballot that contains an over-vote to the voter, but not all scanners are properly programed. Smudges can be inadvertently made on ballots, then inadvertently interpreted as over-votes when scanned.</p><blockquote><strong>A reduction in machine problems probably explains the one-percentage-point decline in the residual vote rate between 2000 and 2004.</strong></blockquote><p>In a <a href="http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.522.6316&amp;rep=rep1&amp;type=pdf">paper</a> Steve Ansolabehere and I published in the Journal of Politics in 2005, we showed, for instance, that just shifting from punch card to scanned paper ballots would reduce the residual vote rate in a county by 1.3 percentage points.</p><p>The years between 2000 and 2004 saw a huge shift in voting machine types, and that benefitted voters by capturing more of their votes. However, not all jurisdictions in America shifted from punch cards to paper ballots in these years. Thus, some of the reduction during that time, which persisted into the future, probably occurred because election officials just became more careful in counting ballots.</p><p><strong>A Word about Nevada</strong></p><p>One of the anomalies in the 2016 EPI is that although the residual vote rate went up significantly overall, it declined to a record-low level in Nevada. What is that about?</p><p>The short answer is: “none of these candidates.” In the 1970s, in the wake of Watergate, Nevada added this choice to its ballot. Unlike the rest of the nation, if a Nevadan doesn’t like the options on the ballot, she can still express that dislike in a positive way by voting “none,” rather than abstaining. The use of the “none” option skyrocketed in 2016, to its highest level since its first use on the presidential ballot in 1976.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XudVpsq7j_UhhDxxx2I2rA.png" /><figcaption>Nevada Residual and “None” Vote Rates from 1976 to 2016</figcaption></figure><p>The accompanying figure shows the full story. In 1976, 2.47% of Nevada voters chose the “none” option. This is on top of the 2.20% of voters who chose no option at all, including “none.” If we add these two together, that sums to 4.68%, which is what the residual vote rate would have been in 1976, assuming the “nones” would have just abstained instead.</p><p>Over the next three decades, both the “none” choice and the residual vote rate in Nevada declined in parallel, until 2008, when a slight wedge was opened between the two. This gap opened up even further in 2016, when 2.56% of voters chose the “none option,” while the residual vote rate declined to a record-low 0.0039%.</p><p><strong>The Role of the Residual Vote Rate in the EPI</strong></p><p>The residual vote rate is one of the few indicators in the EPI that touches on the issue of the accuracy of the vote count. On the other hand, as 2016 reveals, the residual vote rate can change because of the nature of the presidential campaign in the run-up to the election, and in such a way that most would argue has little to do with election administration or policy.</p><p>The overall scores of a few states in 2016 would change noticeably if the residual vote rate were excluded. Montana would change the most if the residual vote rate were removed from the index, increasing by four points; Colorado, Maine, North Dakota, and Oregon would increase by three. Although these seem like small point changes, those small changes in points can yield big changes in rank because so many states are bunched up so close to each other.</p><p>(Of course, I have to add that we discourage a focus on rankings, but that is like King Canute commanding the tide.)</p><p>While a good argument can be made that the abstention-driven rise in the residual vote rate in 2016 had little to do with election administration and policy, the example of Nevada requires me to respond, “not so fast.” The residual vote rate proved itself as a valuable diagnostic tool in the 2000 presidential election, and the availability of the “none of these candidates” option in Nevada shows that election laws can be changed to allow an alternative to abstention on the ballot. I know that the idea behind the “none” choice was not to make it clear that a missing vote was likely due to some type of mistake, but we now know that this is a role that this choice can play.</p><p>Thus, perhaps the thing to learn here is that Nevada is onto something. Giving voters who would like a chance to express their displeasure with the nominees for president also has the salutary effect of making voting machine problems more obvious. That’s a policy choice that’s at least worth considering.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e455289acb30" width="1" height="1" alt=""><hr><p><a href="https://medium.com/mit-election-lab/whats-a-residual-vote-and-what-s-it-doing-in-the-elections-performance-index-e455289acb30">What’s a Residual Vote, and What’s It Doing in the Elections Performance Index?</a> was originally published in <a href="https://medium.com/mit-election-lab">MIT Election Lab</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Blue Shift Comes to the California Primary]]></title>
            <link>https://medium.com/@cstewartiii/the-blue-shift-comes-to-the-california-primary-c474454d7d26?source=rss-b3b359353b26------2</link>
            <guid isPermaLink="false">https://medium.com/p/c474454d7d26</guid>
            <category><![CDATA[elections]]></category>
            <category><![CDATA[ca-top-two-primaries]]></category>
            <category><![CDATA[politics]]></category>
            <dc:creator><![CDATA[Charles Stewart]]></dc:creator>
            <pubDate>Tue, 24 Jul 2018 19:00:21 GMT</pubDate>
            <atom:updated>2018-07-24T19:01:05.931Z</atom:updated>
            <content:encoded><![CDATA[<p>Vote counting doesn’t stop on election night. Even after the polls close on Election Day and the ballots have been counted, there are other votes to account for, in the form of absentee ballots, provisional ballots, and ballots that are simply overlooked. Increasingly, the vote that is counted after election night — what my friend <a href="https://moritzlaw.osu.edu/faculty/edward-b-foley/">Ned Foley</a> calls the “overtime vote” — trends Democratic. This has been true in recent general elections. Now, with California’s “top two” primary system, we can see it gaining a hold in primaries.</p><p>I have recently written about this over in Election Updates, and I’d invite those who are interested to partake of the longer blog post <a href="https://electionupdates.caltech.edu/2018/07/19/the-blue-shift-comes-to-the-california-primary/">here</a>. Here, I’d like to underscore a few points.</p><p>First, the number of votes that are counted after election night is substantial. In 2016, out of the 128.8 million votes cast for president, 9.2 million were counted after election night. These votes are a combination of late-counted absentee votes and provisional ballots.</p><p>Second, some states have more of these overtime votes than others. California stands out in particular. In 2016, 5.4 million votes were counted in the Golden State after election night, which is over half the overtime vote nationwide.</p><p>Third, the vote counted after election night tends to be more Democratic than the vote counted on Election Day, thus the term “blue shift” to describe this set of votes. In 2016, 50.1% of the two-party vote cast on Election Day went to Clinton; 64.3% of the vote that was counted after Election Day went to her, allowing her to open up the popular vote lead.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*QUPNkeYJo7Wwz0cSwa2-2g.png" /><figcaption>Source: MEDSL primary data Github (<a href="https://github.com/MEDSL/primaries">https://github.com/MEDSL/primaries</a>) and Calif. Secretary of State.</figcaption></figure><p>Fourth, it looks like the same phenomenon beset the June 2018 statewide primary in California. For instance, in the gubernatorial primary, 61.3% of all the votes went to one of the Democratic candidates, compared to 37.4% that went to the Republicans. In the post-election-night count, 64.2% went to the Democratic candidates and 34.4% went to the Republicans. The accompanying figure shows how the two-party vote share changed from election night to the final canvass for all the statewide partisan races.</p><p>The fact that the blue shift has extended to the top-two primary is interesting because there is relatively little purely partisan activity in the contest — it’s all about the candidates, not about party strategies.</p><p>The fifth point is more parochial. I was able to do this analysis so easily because my team of student researchers at the <a href="https://electionlab.mit.edu/">MIT Election Data and Science Lab</a> have been working on ways to effortless download election returns right after an election and then post them — first to our <a href="https://github.com/MEDSL">Github depository</a>, as a first draft, and then later to our <a href="https://dataverse.harvard.edu/dataverse/medsl?q=&amp;types=dataverses%3Adatasets&amp;sort=dateSort&amp;order=desc&amp;page=1">Dataverse site</a>, as everything is cleaned. All sorts of analysis is possible with this data, and I encourage people to use it.</p><p>So, please take a look at the Election Updates blog to hear more about the California blue shift, and take a look at our precinct data, to do your own analysis, whatever primary-related topic you’re interested in.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c474454d7d26" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Voters Think about Voting Machines]]></title>
            <link>https://medium.com/mit-election-lab/voters-think-about-voting-machines-b468a5a31b44?source=rss-b3b359353b26------2</link>
            <guid isPermaLink="false">https://medium.com/p/b468a5a31b44</guid>
            <category><![CDATA[voting]]></category>
            <category><![CDATA[medsl-guest]]></category>
            <category><![CDATA[politics]]></category>
            <category><![CDATA[voting-machines]]></category>
            <dc:creator><![CDATA[Charles Stewart]]></dc:creator>
            <pubDate>Thu, 05 Jul 2018 22:48:12 GMT</pubDate>
            <atom:updated>2018-07-06T16:42:02.251Z</atom:updated>
            <content:encoded><![CDATA[<p>The annual <a href="http://bowencenterforpublicaffairs.org/institutes/policy-research/election-admin/conference">State Certification Testing of Voting Systems National Conference </a>gave me an opportunity to put together a talk based on unpublished public opinion research I have done in recent years about public attitudes toward voting machines and election security. I recently wrote a blog post on <a href="https://electionupdates.caltech.edu/">Election Updates</a> about the first part of this topic, attitudes toward voting machines. (I will soon get to writing about election security.) Here are the high-level take-aways from that post. If you’re interested in more, please read the longer blog post <a href="https://electionupdates.caltech.edu/2018/06/26/voters-think-about-voting-machines/">here</a>.</p><p>The research I presented was based on questions I have been able to place on my own module to the <a href="https://cces.gov.harvard.edu/">Cooperative Congressional Election Study</a> (CCES) and on other surveys over the past few years. Starting in 2012 I began asking respondents what they thought about the three major types of voting technologies — optical scanners, electronic voting machines, and hand-counted paper — and which machines they would prefer to use.</p><h4>Opinions in 2012</h4><p>Let’s begin with the opinions in 2012. In that year, I found that electronic machines (often called direct-recording electronic devices, or DREs) were generally preferred by respondents compared to optical scanners, and that both types of technologies were preferred overwhelmingly to hand-counted paper. DREs were preferred over optical scanners by a 56%-25% margin. DREs were seen as being easier to use by voters and easier to produce an accurate count than were optical scanners. While DREs and optical scanners were seen as being equally safe from vote-tampering by bad guys, hand-counted paper was regarded as being easy for dishonest people to tamper with.</p><p>Not surprisingly, respondents who lived in counties that used the electronic machines favored them 74%-13%. <em>Surprisingly,</em> even respondents who lived in counties that used scanned paper ballots also said they would prefer to use the DREs by a 50%-30% margin. In short electronic machines were the preferred way to go for most Americans, and were seen as the most functional voting technology.</p><h4>Opinions today</h4><p>Things have changed in the ensuing years. Prompting these changes has been a new focus on the security of the voting process, and especially the security of the computers that help to manage the process. In 2016, when I asked the question again, about which machine respondents would prefer, optical scanners inched up a bit, but support for DREs plummeted. When I asked the question again a few weeks ago, on the <a href="https://today.yougov.com/find-solutions/omnibus/">YouGov Omnibus</a>, support for DREs had fallen even further. Not only that, but users of optical scanners no longer preferred DREs, but now gave a plurality nod to those scanners. The following graph shows the details.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*ztMgNfQ81k1TiZ3WKco-5g.png" /><figcaption>Preference for voting technologies. (Source: MIT Module to the Cooperative Congressional Election Study (2012 and 2016) and the YouGov Omnibus (2017).)</figcaption></figure><p>Digging down deeper into the responses, it was clear that the edge in perceived functionality of the DREs had declined. In addition, while both DREs and scanners are still regarded as safer from vote tampering than hand-counted paper, the gap has narrowed.</p><p>Thus, as criticism of DREs has grown in public discourage, and computer security has become a more salient issue in election administration, the bloom has come off the DRE rose. This is good news for those who have long advocated that DREs be abandoned for paper.</p><p>There is a caution here, however. Although support for DREs has declined significantly over the past five years, DRE users still believe it is the superior technology compared to optical scanners. This suggests that as election administrators transition away from DREs over the next several years, they may find themselves needing to deal with local public opinion that may be skeptical of the move, and regard opscans as an inferior technology.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b468a5a31b44" width="1" height="1" alt=""><hr><p><a href="https://medium.com/mit-election-lab/voters-think-about-voting-machines-b468a5a31b44">Voters Think about Voting Machines</a> was originally published in <a href="https://medium.com/mit-election-lab">MIT Election Lab</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Last week the Bipartisan Policy Center (BPC) hosted an event to celebrate the four years since the…]]></title>
            <link>https://medium.com/mit-election-lab/last-week-the-bipartisan-policy-center-bpc-hosted-an-event-to-celebrate-the-four-years-since-the-f9c4411d0950?source=rss-b3b359353b26------2</link>
            <guid isPermaLink="false">https://medium.com/p/f9c4411d0950</guid>
            <category><![CDATA[elections]]></category>
            <dc:creator><![CDATA[Charles Stewart]]></dc:creator>
            <pubDate>Fri, 04 May 2018 00:20:14 GMT</pubDate>
            <atom:updated>2018-05-04T00:20:14.270Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Di3D2AQ-Mu-jTkkD.png" /></figure><p>Last week the <a href="https://bipartisanpolicy.org/library/improving-the-voter-experience-reducing-polling-place-wait-times-by-measuring-lines-and-managing-polling-place-resources/">Bipartisan Policy Center</a> (BPC) hosted an event to celebrate the four years since the <a href="http://web.mit.edu/supportthevoter/www/">Presidential Commission on Election Administration</a> (PCEA) issued its report with recommendations to improve the experience of American voters. (You can view videos of the event’s sessions <a href="https://bipartisanpolicy.org/events/making-elections-work/">here.</a>) The two major issues addressed in panels were long lines at the polls and the modernization of voter registration rolls — two of the primary concerns outlined in the PCEA’s report.</p><p>On the issue of long lines, the event provided the opportunity to release a report on <a href="https://bipartisanpolicy.org/library/improving-the-voter-experience-reducing-polling-place-wait-times-by-measuring-lines-and-managing-polling-place-resources/">Improving the Voter Experience,</a> which provided results about a major project the BPC and the Caltech/MIT Voting Technology Project (VTP) collaborated on to monitor polling place wait times in 2016. That report is based largely on data provided by 88 counties from 11 states that participated in the line-length program. These counties represented 15.6 million registered voters, 11.1 votes cast (8% of nationwide turnout), and 4,006 precincts. This is the largest study ever conducted of how long voters wait to cast their ballots.</p><p>The report is full of facts that came from an analysis of all the data that was gathered by the 88 jurisdictions, and I encourage you to read the full document. Here are some of the most important findings in brief:</p><ol><li><strong>Most voters wait very little, if at all, to vote.</strong> Consistent with past survey research that’s been done on the subject, the modal (most-common) line length recorded in the project was zero. Just over 2/3 of precincts had average wait times of less than 10 minutes.</li><li><strong>The longest lines, and wait times, are first thing in the morning.</strong> Almost every precinct in the study had people lined up waiting to vote when the polls opened on Election Day. (As an aside, the fact that virtually every polling place has a line when the polls open suggests how easy it is for a news photographer to get a picture of a long line early on Election Day, and how meaningless these pictures are as evidence of problems.) The average precinct had between 30 and 45 minutes’ worth of voters at the door when polls opened. This is the most significant source of wait times, both when things go well and when things go poorly. While we saw evidence of surges in turnout at other times of the day, such as around noon and after work, those surges were minor ripples compared to the tsunami of voters at the start of the day.</li><li><strong>If long lines are resolved after two hours, a precinct is highly unlikely to experience long waits the rest of the day. If the morning rush isn’t cleared up in three hours, count on lines throughout the day.</strong> The findings of the study revealed the critical nature of managing wait times at the opening of the polls. The line at the start of the day isn’t an issue so much as the line at the end of the first hour. If a polling station’s line hasn’t been cut (at least) in half after the first hour, it will be difficult to make progress on wait times for the rest of the day. I have talked to election officials who, after seeing the data from the project, have said they will use new resources they get for staffing to increase the number of staff who work in the morning. (These are officials who live in states where poll workers are allowed to work in shifts.)</li></ol><p>The report also brings to mind several points about process.</p><ol><li><strong>We can improve election administration performance if we put our minds to it.</strong>President Obama was inspired to create the PCEA because of press reports of long waits to vote — some up to six hours long — in 2012. In the PCEA’s report, the commission set a benchmark that no voter should have to wait more than 30 minutes to cast a ballot. The survey research reveals that great progress was made toward hitting that benchmark in 2016. According to responses to the Survey of the Performance of American Elections, in 2012, 13% of in-person voters waited more than 30 minutes to vote. In 2016, that was reduced to 9%. The most dramatic improvements occurred in the states that had the longest wait times in 2012. The percentage of voters who waited more than 30 minutes fell from 39% to 4% in Florida, from 39% to 17% in D.C., and from 28% to 10% in Virginia. We still have more work to do to make the commission’s benchmark a reality: more than 5% of voters waited over 30 minutes in 25 states in 2016. Still, the improvement in 2016 in the most troubling states reveals that election officials can make great strides in improving polling place management if they put their minds to it. (As an aside, I think the wait-time success story bodes well for handling the current cybersecurity concerns, but only if officials put the same effort into addressing the issue.)</li><li><strong>If you don’t measure it, you can’t manage it.</strong> A barrier to managing polling place wait times before 2012 was the lack of detailed knowledge about how long voters waited to cast a ballot. Survey research was valuable to help give the public and policymakers an idea about where the longest wait times were happening, but it didn’t provide “news you can use” to improve wait times. After all, if we learn that wait times are much shorter in Vermont than in Florida, it doesn’t help if I tell you to move to Vermont if you want to vote more quickly in the future. To help election officials pinpoint precisely where and when long wait times emerge, they need to measure wait times directly. This means counting the number of voters waiting in line on a regular basis, gathering data about how many people arrived to vote during a given time, and then using Little’s Law to calculate what the wait times were.</li><li><strong>You can measure it, and you can manage it.</strong> The wait time project highlighted in the Improving Voter Experience report offers a simple, high-impact way for election officials to gather the data they need to find out when and where they long wait times are happening. Election officials can go to <a href="https://bipartisanpolicy.org/library/improving-the-voter-experience-reducing-polling-place-wait-times-by-measuring-lines-and-managing-polling-place-resources/">this webpage</a> and answer the call to participate in the program for 2018.</li></ol><p>Finally, this report, which highlights data-gathering, reminds us that there are tools available to help local officials plan ahead of time to make sure they have enough staff, poll books, voting booths, etc. to handle the number of voters who walk through the doors. The VTP continues to host, at the PCEA’s invitation, <a href="http://web.mit.edu/vtp/">a set of online tools</a> to aid in that management. The Center for Technology and Civic Life’s <a href="http://electiontools.org/">Election Toolkit</a> also contains helpful tools. For those interested in a deeper dive into the science behind line management as applied to elections, I wrote the report <a href="http://web.mit.edu/vtp/Managing%20Polling%20Place%20Resources.pdf">Managing Polling Place Resources</a> a couple of years ago to help translate queuing theory to the polling place.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f9c4411d0950" width="1" height="1" alt=""><hr><p><a href="https://medium.com/mit-election-lab/last-week-the-bipartisan-policy-center-bpc-hosted-an-event-to-celebrate-the-four-years-since-the-f9c4411d0950">Last week the Bipartisan Policy Center (BPC) hosted an event to celebrate the four years since the…</a> was originally published in <a href="https://medium.com/mit-election-lab">MIT Election Lab</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>