Skip to content

Establishment Versus Change in the St. Louis Mayoral Election of 2017

The recent mayoral election in St. Louis has received a fair amount of attention in local media. To recap, Lyda Krewson, the main white candidate in the race, won with 32% of the vote, beating out three black candidates, Tishaura Jones (30%), Lewis Reed (18%) and Antonio French (16%). Several observers have concluded that a lopsided field with several prominent black candidates split the black vote and allowed the one high-profile white candidate to win. As the reasoning goes, if one or more of the other black candidates had dropped out then Jones would have won the election. Similarly, Steven Hill argues that if St. Louis used different election rules that allow voters to rank their candidates in order of preference, then Jones might have won the election.

However, we might want to resist jumping to conclusions about how the election might have ended differently. First, it is difficult to assess hypothetical campaign scenarios. That is, if the circumstances were different (fewer candidates or different voting rules) then the candidates would have altered their behavior – basic decisions involving campaign messages and voter targeting strategies would have changed – which could affect the choices voters make. Second, even though St. Louis has a long history of racially polarized voting, black voters and white voters in St. Louis are not monolithic blocs that uniformly support candidates of their own race. The city’s two black mayors, Freeman Bosley, Jr. and Clarence Harmon, each received significant support from white voters. Francis Slay, the long-serving white mayor who leaves office today, garnered around 20 to 30 percent of the vote from black residents in his re-election campaigns. Third, we don’t really know who voters would prefer as mayor in the absence of one or more of the major candidates unless we ask them.

Several UMSL students and I did just that – we conducted an exit poll in the March 7 primary election. We got around 900 voters to complete our survey at 20 randomly chosen polling places throughout the city. Naturally, we asked who the voters selected in the mayoral contest. Furthermore, to address the hypothetical questions raised above, we also asked voters to indicate their second choice for mayor. The table below summarizes the second choice preferences for mayor among Krewson voters, Jones voters, and other voters in separate columns.

The first thing that is evident in the numbers is that second choice preferences are distributed widely among the candidates, rather than being concentrated among Krewson and Jones. As the final column indicates, those who voted for a mayoral candidate outside the top two finishers almost evenly split between Tishaura Jones (28%) and Lyda Krewson (25%) in their second choice preferences. Our exit poll also shows a clear division between preferences for continuity versus change in City Hall. Lewis Reed was the top preference for second place among Krewson voters, while Antonio French was the overwhelming second choice among Jones voters. Similarly, among Reed voters Krewson is preferred as a second choice more frequently than Jones. Among French voters it is the other way around. Reed and Krewson have been on the Board of Aldermen since the late 1990s and Reed has been President of the Board for the past ten years. Furthermore, Krewson was endorsed by Mayor Slay. In contrast, French and Jones are much more recent arrivals to elected positions in city government and both offered the clearest campaign messages for change among the candidates running for mayor. It appears that Krewson and Reed appealed to voters who valued experience and continuity, while Jones and French attracted voters who wanted a new direction in city government. It is not clear that Jones would have prevailed if other candidates had dropped out or if ranked choice voting rules were used in the mayoral election. If French had dropped out, then maybe Jones would have won the election. But if Reed had dropped out it is possible that Krewson’s margin of victory would have been larger. Race is very important in St. Louis politics, but by focusing solely on race we may overlook other important sources of division in the city.

2nd choice 2017


St. Louis: Vote Yes on Prop B

In next week’s election the St. Louis ballot will include Proposition B, a measure to change the schedule of municipal elections in the city. Under the proposition, municipal elections would no longer be held in odd-numbered years. The March primary would shift to coincide with the August primary election, and the April general municipal election would instead be held in November of even-numbered years. Thus, the municipal elections would move away from “off-cycle” dates to coincide with “on-cycle” state and federal elections.

One of the main impacts of Prop B will be to substantially increase voter turnout for municipal elections in the city of St. Louis. The March and April elections in St. Louis are notorious for low turnout and an unrepresentative electorate. My students and I conducted an exit poll for the March 7 mayoral primary a few weeks ago. Approximately 900 voters at randomly selected polling places completed our survey (a response rate over 60%). Thus, our exit poll is a good representation of people who voted in the March election. The table below compares the demographic profile of the March 7 primary voters to the adult population of the city. As the table makes clear, voters in the March election are substantially older, wealthier, whiter, and more educated than adults living in the city. This is a common feature of low-turnout off-cycle municipal elections.

In an earlier post I explain how low-turnout elections tend to produce municipal governments that are less responsive to the needs of their residents. There is additional research to add to this list. Sarah Anzia shows that low-turnout local elections tend to be dominated by special interest groups. Zoltan Hajnal finds that minority residents, particularly African Americans, tend to be underrepresented in city offices when local elections are scheduled in off-cycle months. In St. Louis the schedule of municipal elections also determines who votes in the partisan municipal primary election. Since St. Louis has a weak Republican Party, city GOP voters tend to vote in the Democratic primary when the municipal primary is held in March. However, in the August primary election city Republicans vote in the Republican primary because there are usually competitive races for state and federal nominations on the Republican ballot in that election.

It is more difficult for a black candidate to win a citywide Democratic primary election when Republicans vote in the Democratic primary because Republican voters tend to heavily support white candidates in city elections. Recent elections bear this out. In the March 2017 primary, for example, Tishaura Jones (a black candidate) narrowly lost the mayoral race to Lyda Krewson (a white candidate) in a multi-candidate contest. Our exit poll indicates that Krewson got a majority of the vote from Republicans and voters with a favorable opinion of President Trump. However, in August of 2012 Tishaura Jones easily won the Democratic primary race for Treasurer in a multi-candidate contest. In the August 2016 Democratic primary for Circuit Attorney Kim Gardner (a black candidate) easily won in a multi-candidate contest as well. If St. Louis voters want a more representative city government then they should support Prop B.

unrepresentative electorate

Where Trump Performed Poorly in St. Louis

Missouri has now certified the results of the 2016 election, in which Donald Trump won a historic statewide victory. Trump improved on Mitt Romney’s 2012 raw vote total in all but 5 of Missouri’s 116 local jurisdictions. Trump improved on Romney’s 2012 share of the major-party vote in all but 4 local jurisdictions. Basically, the election was a wipeout for Democrats everywhere outside of Boone County (Columbia), as well as St. Louis and Kansas City and their near suburbs.

There is one slice of the St. Louis region where Trump fared poorly. The map below shows the change in the major-party presidential vote in St. Louis City and St. Louis County from 2012 to 2016. The city portion of the map is broken down by ward and the county portion of the map shows townships. There are 28 wards in the city and 28 townships in the county. On the map, the areas in red indicate where Trump improved on Romney’s 2012 performance by at least 2 percentage points. Trump outperformed Romney in three south city wards (11, 20, and 25) and two south county townships (Lemay and Oakville). The map also shows some GOP growth in northwest St. Louis County (Airport, Midland, and Northwest townships). The areas in blue show where Trump lagged behind Romney by at least 2 points. The blue areas underscore the fact that Clinton’s winning margin in St. Louis County (roughly 84,000 votes) exceeds Obama’s margin of 72,000 votes in 2012. When people in St. Louis talk about the “central corridor” they are basically referring to the dark blue portions of the map, which extends west from downtown St. Louis through the central west end, the inner suburbs of Clayton and Maplewood, out to the western suburbs of Chesterfield and Wildwood. Interstate 64 runs through the middle of the blue portion of the map. The central corridor contains the largest concentrations of education and wealth in the state. This is consistent with other data showing a strong correlation between education and presidential voting in the rest of the country.

The blue portion of the map is also similar to the areas where John Kasich did well in Missouri’s GOP presidential primary. Some moderate Republicans in the St. Louis area may have voted for Clinton in 2016. Of course, Kasich did very poorly in the rest of the state, as did Hillary Clinton. This election highlights the increasing isolation of St. Louis’ central corridor in Missouri politics. Among the local political scientists and pundits who failed to foresee a Trump presidency, how many live outside the central corridor?


Vote No on Amendment 6 (with hyperlinks)

Yesterday the Columbia Daily Tribune published my op-ed opposing Missouri Amendment 6, along with some ideas for more effective voting reforms. Below is the same piece with hyperlinks to evidence and studies supporting my claims.

Replace the Photo ID Requirement with More Effective Election Reforms

Next month Missourians will vote on a constitutional amendment that would require voters to show photo identification when they cast their ballots. The proposed photo ID requirement will cost the state $17 million to implement, will not prevent fraud, and may prevent eligible voters from casting a ballot. There are more effective and less expensive ways to prevent fraud and improve election procedures in Missouri.

Several nationwide investigations have determined that voter fraud is extremely rare, with roughly 31 cases for every billion votes cast. For some perspective, consider that thousands of Missourians did not vote in the last presidential election because of registration problems. Voter impersonation, the type of fraud a photo ID requirement addresses, has not occurred in Missouri in more than 15 years. Voter impersonation is extremely rare because Missouri voters are already required to show identification, such as a voter identification card, utility bill, or driver’s license, and sign their name in the poll book.

Furthermore, a photo ID requirement may impose a barrier to voting by eligible voters who lack a valid Missouri driver’s license. The Secretary of State estimates that 200,000 eligible Missouri voters don’t have a valid driver’s license. Several studies indicate a photo ID requirement would disproportionately burden college students, racial and ethnic minorities, the elderly, and the poor. Thus, courts have struck down photo ID requirements in several other states.

Here are four more effective election reforms that would prevent voter fraud, reduce costs, and improve Missouri election procedures.

  • Enact full online voter registration. Missouri still relies on an antiquated registration process where voters fill out paper forms and election officials then type the information into a computer. The Secretary of State created a web site where people can register to vote online, but the registration information is then transmitted on a paper form to county election officials, where it still must be manually entered into another computer. State law should change to allow for a completely paperless online voter registration system, where voters’ addresses and other information can be matched against the state’s motor vehicle records and any errors or inconsistencies can be corrected in real time. The new registration information then should be sent electronically to the relevant county election office to update the official voter list. Most other states already have this type of online voter registration and studies show that it reduces administrative costs, improves the accuracy of voter lists, and increases voter participation.


  • Missouri should join the Electronic Registration Information Center (ERIC), a partnership of many states that compare their registered voter lists and match them against other state and federal data to find and correct errors. ERIC also notifies eligible unregistered adults with instructions on how to register to vote in their state. Missouri is part of another data sharing program with Kansas, but ERIC is a more extensive system and includes data not used by the Kansas program, such as Department of Motor Vehicle records, the Social Security Deceased Index, and the Postal Service National Change of Address information. Participation in the ERIC program reduces errors in voter lists and increases voter registration and turnout rates.


  • Revive “Motor Voter.” The “Motor Voter” law of 1993 requires that states let citizens register to vote when they conduct transactions at motor vehicle agencies or state benefit offices. But as government services have moved online, fewer voters are coming to state offices and the efficacy of Motor Voter is fading. Registration options need to be incorporated into state websites. When citizens go online to pay state taxes, update licenses or motor vehicle records, or apply for social benefits, their addresses and basic voter information should be automatically updated. If they are not registered, they should be given an immediate opportunity to do so. Improving the voter registration process will make the voter list more accurate, which cuts down on fraud and other irregularities on Election Day.


  • Take a closer look at absentee voting. A recent investigation by the St. Louis Post-Dispatch revealed absentee voting irregularities in the Democratic primary election for the 78th district seat in the state house. The problems were serious enough that a judge ordered a new election for that contest. Absentee voting tends to lose more votes and produce other irregularities because it is less secure than in-person voting methods. For what it’s worth, the proposed photo ID requirement in Missouri does not apply to absentee voting. The legislature should review absentee voting procedures and consider changes, such as routine audits of absentee ballots, to make absentee voting more secure.

These four reforms would better serve Missouri voters and taxpayers at a lower cost than the $17 million price tag for giving voters a free photo ID.

RINOs Endangered in St. Louis County

For decades St. Louis County was the base of the Missouri Republican Party. From at least the 1960s to the 1990s Republican candidates won statewide offices by running up big margins in St. Louis County. During that same period the St. Louis County GOP was led by moderates, politicians like U.S. Senator John Danforth and County Executive Gene McNary. These leaders typically spoke in measured tones, had an instinct for compromise, and held centrist positions on a variety of issues from immigration to consumer issues and environmental protection. Overall, moderate Republicans were a major force within the state party (although McNary lost a close primary election to John Ashcroft for the party’s gubernatorial nomination in 1984, foreshadowing the future direction of the party).

Today moderate Republicans are an endangered species in St. Louis County, as in the rest of the country. Consider the contest for the Republican presidential nomination. When Missouri held its presidential primary on March 15, 2016, John Kasich was the only viable candidate remaining who made a pitch for moderate GOP votes. The map below shows Kasich’s vote share in St. Louis County precincts. As the map shows, Kasich lost badly to Donald Trump and Ted Cruz in St. Louis County, as in the rest of the state. [In case you are wondering, the lone north county precinct that Kasich carried produced only one GOP primary ballot – a vote for Kasich.] In the ideological fight for survival in the current Republican Party, moderates are losing out to populists (Trump) and hard-right conservatives (Cruz). The RINO habitat in St. Louis County has shrunk so much that it is now limited to central corridor communities like Clayton and Ladue and sections extending south and west into Webster Groves and Kirkwood.

Kasich County Vote 2016

Response to Ferguson Commision – Voter Turnout and Municipal Governance

Several scholars in the St. Louis region who are affiliated with Scholars Strategy Network have written short pieces responding to the Ferguson Commission report. You can find a summary of the pieces from Todd Swanstrom and me, as well as links to each them, on the St. Louis Post-Dispatch web site. My piece, on turnout and municipal governance, is posted here. Unfortunately, the PD ran my piece before including some corrections that I made. The updated version appears below.

Breaking the low-turnout/inattentive government cycle

Two of the action steps in the newly-released Ferguson Commission report are striking. The first calls for limits on municipal traffic fine revenues. The second demands alternatives to prison for people who miss a municipal court hearing.
In recent years, some municipalities in St. Louis County have come to rely heavily on traffic fines and court fees to fund their budgets. Those governments have an incentive to use highly punitive practices when enforcing local traffic and small-time municipal ordinances. And incentives matter. For example, in 2013 the city of Ferguson, Missouri (with a population just over 21,000) issued over 9,000 arrest warrants for incidents that were primarily traffic, parking and housing code violations.
Why do local residents put up with such heavy-handed practices? Because their municipalities are essentially trading revenues for democracy.
Low rates of citizen engagement tend to go hand in hand with substandard and abusive municipal government practices. Voter participation in local municipal elections tends to be low in the United States, especially in St. Louis County. My research on municipal elections in St. Louis County with Terry Jones and Cynthia Palazzolo finds that turnout has averaged 15 percent over the past five years. That’s very low.
When people have to pay fines or go to jail, their experiences with police, courts, or prisons inform their views about how government works, teaching unfortunate lessons about local democracy. Research by Amy Lerman and Vesla Weaver finds that contact with the criminal justice system significantly reduces Americans’ faith in government and makes them less likely to contact public officials about their concerns. And these negative impacts aren’t just limited to the people who are fined, arrested, or jailed. Research by Traci Burch indicates that, no surprise, communities with higher rates of imprisonment tend to have lower rates of voter turnout.
A vicious cycle can ensue. In places with low voter participation, municipal officials may feel they can act with impunity. As an extreme example, in the city of Bell, California, less than 400 voters participated in a special election that changed the city’s governing structure, and later investigations revealed that certain Bell city officials awarded themselves annual compensation packages worth more than half a million dollars.
The cycle can be turned around. Evidence suggests that reforming municipal court and governance practices can modestly improve voter turnout in municipal elections in St. Louis County.
In recent elections, less than one voter in six comes out to vote in St. Louis County municipalities that generate more than 15 percent of their revenues from court fees and municipal fines. By comparison, turnout is slightly better than one in five in municipalities where fees and fines comprise less than 5 percent of general revenues. In a positive first step, the Missouri General Assembly recently passed a law that will cap municipal revenues from court fees and traffic fines at 12.5 percent for municipalities in St. Louis County (the cap is 20 percent for the rest of the state). As municipalities impose fewer fees and fines, a smaller number of people will have negative experiences, and perhaps more people are likely to vote. The impact of this recent law may be small, however, since traffic tickets and fines may be targeted primarily at non-residents who drive through a municipality.
But there is a simple idea that could have an even bigger impact on voter participation in municipal elections. By Missouri law, those elections take place in April, when few people vote. To increase participation in a simple step, lawmakers could move municipal elections to November of even-numbered years, so voters, who come out in large numbers to choose their president, U.S. Senators, and governor, can elect their local officials on the same ballot.

Bigger Counties Produced Longer Voting Lines in North Carolina

I am late commenting on this, but in a post from last year I warned that some counties in North Carolina may not have enough early voting locations to serve their electorate for the 2014 general election. After the elections, following reports of long voting lines in some North Carolina counties, the state board of elections produced a report on voter wait times in the 2014 election. The report was based on a survey of county election officials about voting lines and included some analysis of the resulting data. I commend the board for producing and posting the report. Among other things, it indicates that long waits were concentrated on the final two days of early voting, especially on the last day, Saturday, November 1, when polls closed at 1pm. The report includes a summary of the data, listing which counties had long lines and which ones did not. The report finally discusses several factors to explain the long lines, as noted by county election officials. These factors include the length of the ballot, elimination of straight-party voting, voting equipment, early voting schedules (set by each county), other local issues on the ballot, and various polling place problems (such as parking, space, technology, staffing). All of these factors may have mattered in some way, but the report is remarkable for ignoring the most obvious factor: county size. The most heavily populated counties were the ones most likely to produce long voting lines.
A table on page 4 of the report lists early voting wait times by county in three categories: 0-30 minutes, 30-60 minutes, and over 60 minutes. The table is basically a list of North Carolina counties stratified by population. The 24 smallest counties (by population) all generated wait times under 30 minutes. The five largest counties in the state, with more than 200,000 registered voters apiece, each had early voting wait times greater than one hour. It is not hard to see why this would be the case. For example, consider Jones County, one of the smallest counties in the state with one early voting location serving almost 7,500 registered voters. Wake County, one of the largest, had 9 early voting locations to serve about 670,000 registered voters (or one voting location for every 75,000 voters). So Wake County had ten times as many registered voters per early voting location as Jones County. It doesn’t take a rocket scientist to figure out that Wake County would have longer early voting lines than Jones County.
Nevertheless, to apply a little science I estimated a statistical model of early voting wait times in North Carolina counties with two predictors: county size and voting equipment. Most North Carolina counties used optical scan voting equipment, where voters mark their choices on a paper ballot that is scanned to count the votes. However, some North Carolina counties use electronic voting machines. DREs (short for direct recording electric machines) resemble ATM machines in that voters touch a computer screen to indicate their ballot choices. Research by Charles Stewart and others indicates that people take longer to vote on DRE machines than on optical scan ballots. The results of my statistical analysis also indicate that voting lines tend to be longer with DRE machines, as shown in the figure below. The graph plots the probability of early voting wait times greater than one hour as a function of county size. Since county size is heavily skewed, I have logged the number of registered voters in each county. The figure indicates that the probability of long voting lines rises sharply for counties with at least 100,000 registered voters (there are 17 such counties in North Carolina). I found a similar, but weaker, pattern of long voting lines in larger North Carolina counties on Election Day as well. On Election Day counties tend to have many more polling places to serve voters (in 2014 Wake County had 200 polling places on Election Day while Jones County had 7).

EV wait times NC 2014

One of the challenges of election administration is that most problems are disproportionately concentrated in heavily populated metropolitan local jurisdictions. This should have been obvious to the North Carolina State Board of Elections. Ignoring county size also minimizes the extent of the problem with long voting lines. In several charts the state report emphasizes that only 13 out of 100 counties had early voting wait times greater than one hour, implying that the problem was not widespread. However, those 13 counties with long lines contain over 42 percent of the state’s registered voters. A lot of voters may have experienced long lines in North Carolina even if the long waits were confined to a relatively small number of counties. If states are going to adopt early voting, then they need to make early voting effective for all voters. That means allowing larger counties to have more early voting locations, as well as configuring and staffing those locations to best handle a large volume of voters.