Background

In-Person Absentee Voting (IPAV) is available to City of Madison voters two Tuesdays before any election through the Sunday before the election. It offers voters a secure and convenient way to vote “early” at sites around the community. Between absentee voting methods, mail-in and in-person, Clerk’s Office data shows that the popularity of in-person absentee voting is steadily increasing in Madison.

The rate at which voters frequent IPAV sites may fluctuate drastically throughout the 13-day period. The Clerk’s Office records day-end totals at the sites, but currently there is no documentation of hourly volume or temporal trends. Aside from anecdotal data and assumptions surrounding voters’ schedules (e.g. greater availability at the lunch hour and in the evening hours), the Office does not have comprehensive data to show:

Project Goal

The Clerk’s Office would like to better understand temporal IPAV volume trends. Visually mapping out the time stamp data is an approach to understanding the IPAV schedule, staffing, and community impact through a data-driven framework. Rather than producing a recommendations report, this project is intended to guide further exploration surrounding the design and execution of IPAV.

It is possible that findings may be useful later on for:

Data

This report exclusively concerns IPAV data from the November 5, 2024 election (the presidential election). The templating, data manipulation, and analysis set forth in this report can and should be applied to historical IPAV data and future IPAV data to provide a fuller picture of temporal trends and variances.

Temporal site-specific data can be derived from joining datasets accessible to and used by the Clerk’s Office. The datasets used in this report are:

A total of 33 sites of varying location, size, and intended function/purpose for the community were used for IPAV. In this report, sites are categorized by type:

Four additional sites were designated as official IPAV locations on an appointment-only basis. Since staffing for these sites was not scheduled, this report does not consider these sites.

IPAV Schedule

The heat map below provides an at-a-glance view of the relative number of hours each site operated for every day it was open. Note, some sites operate on weekdays only, weekends only, or both weekdays and weekends.

Nov 2024 hours heatmap

Apr 2025 hours heatmap

Apr 2026 hours heatmap

Turnout

The plot below illustrates two main patterns known to occur during any IPAV period:

  • Substantially increased traffic during the second week of IPAV (beginning Monday, October 28)
  • Less traffic during the weekends, as only a limited number of sites are open, each for reduced hours

Nov 2024 daily totals

Apr 2025 daily totals

April 2026 daily totals

Site-specific Turnout

The table below shows the most trafficked sites in descending order by Rate, the average number of IPAV voters served every hour. An interactive visual exploration of this data is mapped out in the two plots that follow.

Nov 2024

Site Totals
IPAV Site Voters Hours Rate
Memorial Union 5350 58.5 91
Alicia Ashman Library 5395 66.0 82
Pinney Branch Library 6598 84.0 79
UW Madison Union South 3897 54.0 72
Sequoya Library 4888 84.0 58
Meadowridge Library 4593 84.0 55
Central Library 4132 83.5 49
Lakeview Library 648 16.0 40
Hawthorne Library 3210 84.0 38
Badger Rock Center 127 4.0 32
Monroe Street Library 1594 54.0 30
Warner Park Community and Recreation Center 3316 116.0 29
South Madison Goodman Library 2562 88.0 29
Olbrich Gardens 1286 45.0 29
Madison Municipal Building 1685 63.0 27
Edgewood University - The Commons 496 20.0 25
UW Health Sciences Learning Center 1383 58.5 24
Fountain of Life Church 159 8.0 20
Global Market & Food Hall 1651 90.0 18
Elver Park Neighborhood Center/WI Youth Company 407 22.0 18
SS Morris AME Church 143 8.0 18
Boys and Girls Club 749 45.0 17
Urban League Hub 351 20.5 17
Lussier Center 322 21.0 15
Eagle Heights Community Center 161 14.0 12
Madison College Truax Campus 254 22.5 11
Bridge-Lakepoint Waunona Community Center 184 22.0 8
East Madison Community Center 249 45.0 6
Madison College Goodman South Campus 140 22.5 6
Catholic Multicultural Center 92 16.0 6
Hmong Institute 145 29.0 5
Freedom, Inc 53 10.0 5
Centro Hispano 35 10.0 4

Apr 2025

Site Totals
IPAV Site Voters Hours Rate
Sequoya Library 4197 84.0 50
Alicia Ashman Library 3295 66.0 50
Pinney Branch Library 3973 84.0 47
Lakeview Library 483 14.0 34
Memorial Union 1699 58.5 29
Meadowridge Library 2348 86.0 27
Central Library 2056 83.5 25
UW Madison Union South 1223 54.0 23
Hawthorne Library 1699 84.0 20
Monroe Street Library 450 30.0 15
Warner Park Community and Recreation Center 1421 116.0 12
South Madison Goodman Library 1098 88.0 12
UW Health Sciences Learning Center 562 58.5 10
Madison Municipal Building 572 63.0 9
Urban League Hub 171 20.0 9
Edgewood University - The Commons 166 20.0 8
Christ the Solid Rock Baptist Church 60 8.0 8
Global Market & Food Hall 585 90.0 6
Fountain of Life Church 51 8.0 6
Mt Zion Church 42 7.0 6
Madison College Truax Campus 112 22.5 5
SS Morris AME Church 21 4.0 5
Lussier Center 79 21.0 4
Sherman Avenue United Methodist Church 30 8.0 4
Bridge-Lakepoint Waunona Community Center 59 22.0 3
Hmong Institute 63 29.0 2
East Madison Community Center 59 25.0 2
Elver Park Neighborhood Center/WI Youth Company 46 22.0 2
Catholic Multicultural Center 29 16.0 2
Eagle Heights Community Center 27 14.0 2
Badger Rock Center 18 8.0 2
Boys and Girls Club 33 25.0 1
Madison College Goodman South Campus 30 22.5 1
Centro Hispano 10 10.0 1
Freedom, Inc 11 10.0 1
Olbrich Gardens 308 -35.0 -9

Apr 2026

Site Totals
IPAV Site Voters Hours Rate
Sequoya Library 2579 84.0 31
Alicia Ashman Library 1772 66.0 27
Pinney Branch Library 1964 84.0 23
Monroe Street Library 175 12.0 15
Lakeview Library 184 14.0 13
Meadowridge Library 1017 84.0 12
Central Library 877 84.0 10
Hawthorne Library 791 84.0 9
Memorial Union 540 58.5 9
UW Madison Union South 386 54.0 7
Warner Park Community and Recreation Center 710 116.0 6
South Madison Goodman Library 414 86.0 5
Madison Municipal Building 326 63.0 5
UW Madison School of Human Ecology 104 26.0 4
Edgewood University - The Commons 53 16.0 3
Christ the Solid Rock Baptist Church 22 8.0 3
Fountain of Life Church 24 8.0 3
Global Market & Food Hall 215 90.0 2
UW Madison School of Nursing 68 30.0 2
Urban League Hub 47 20.0 2
Lussier Center 31 15.0 2
Sherman Avenue United Methodist Church 19 8.0 2
SS Morris AME Church 8 4.0 2
Madison College Goodman South Campus 22 40.5 1
Madison College Truax Campus 59 40.5 1
Hmong Institute 31 29.0 1
Freedom, Inc 12 10.0 1
Elver Park Neighborhood Center/WI Youth Company 6 8.0 1
Mt Zion Church 8 7.0 1
Boys and Girls Club 9 25.0 0

Absolute Turnout

The plot below shows the volume of voters attributed to each site across the 13-day period, where the radius is a measure of absolute turnout.

Nov 2024

April 2025

April 2026

Relative Turnout

The plot below shows the normalized totals per site, where the radius is a measure of the turnout divided by the total hours the site was open. This plot may more equitably display the “impact” that each site has. Note the reduced size differentials in this plot between some of the larger sites, like libraries and campus sites, compared to affinity sites and churches.

Nov 2024

April 2025

April 2026

Findings & Data-maintenance Recommendations

Largely due to data quality concerns, there are no obvious takeaways from looking at the temporal trends plots. Peaks and valleys seem to fluctuate at random. It is hard to extrapolate too much from the plots because many data points are not represented and there is no inclusion of historical data on which to base trends.

Once we improve our internal documentation, we will have a cleaner baseline for making analyses and visualizations that better lend themselves to interpretation.

The Office should:

  • Establish uniform naming practices across databases for ease of integration and manipulation
  • Create tabular data and templating for the IPAV schedule
  • Improve the documentation and validation of worker assignments, including for office staff
  • Better integrate electronic record-keeping practices since many records exist only on paper
  • Streamline the process and criteria for deploying additional peak-rush staff

Challenges

It was a very slow and tedious process joining the data from across the four data sets.

  • It involved lots of cleaning, converting documents to tabular data, validating data points through iterative reporting, digging through emails, and surveying office staff on their whereabouts 11 months after the fact.
  • Due to the inaccuracies in the data, exacerbated by technological issues and extreme turnout, analyzing IPAV data from almost any other election would have been more interpretable.

Some election officials were assigned to multiple shifts in a day or not explicitly assigned to generate ballots (like a courier or line management staff) but they ended up generating ballots anyway.

  • For multiple-shift workers, in addition to grouping by staff and day, grouping by time was necessary to match the ballot to the correct site. This was difficuly because there were narrow buffers between assignments in some cases, and some folks may have stayed late processing ballots or started early at the next site.
  • Attributing the ballot to the correct site meant looking into the individual time stamps for that staff and checking for an appropriate time gap.

The time stamp data may be inaccurate in representing when the voter was at the site even where there were no discrepancies between SharePoint and WisVote data (or they were minimal).

  • Some staff may have been generating ballot instances in bulk from voters who has visited earlier in the day.
  • These instances are non-detectable because the time stamp data still falls within the hours the site operated, making it impossible to differentiate real-time voters from catch-up runs.

Reflection & Next Steps

The Clerk’s Office may, at any point, contend with budget constraints, personnel constraints, or equipment constraints that could limit the current schedule and setup for IPAV. Our goal in navigating any constraint would be to preserve (and maximize) the reach that current IPAV sites have for the community and the voters served. The Office would like to ensure it is adequately staffing IPAV sites and doing so in response to community needs and anticipated site volume. Though the findings from this report are not sufficient to call for concrete scheduling and staffing recommendations, the report provides a template for illustrating and analyzing measures of “impact” and “efficiency” of IPAV.

Hourly rates can be mapped out to better allow for inter-site comparison.

  • There are so many IPAV sites and many components of the sites (location, type, weekend and/or weekday, etc.). Generating a daily plot for each and every site-and-day combination did not feel practical for the scope of this project.
  • It could be fascinating to design a workflow to map any value for any half-hour interval or whole-day onto that of any other site, though having all of these pieces ready to manipulate is an incredible about of work and coding.
  • With a more defined question (e.g. should we consolidate x,y,z sites going forward? or should we shift the hours of operation of x site two hours later?) plots can be created with lots of care and detail, even with respect to factors like geographic proximity. I think this project gives us a good baseline to start asking some of those questions, or at the very least, access to making comparisons based on actual temporal data.

Not all election officials entering ballot information are equally technologically savvy.

  • To a degree, the rate at which ballots are processed at any given site is a product of who is entering them, their level of comfort navigating WisVote, who is managing lines and preparing voters while they are waiting for their ballot, etc. Each site has a different “tempo”.
  • Although WisVote users have overall competency and data-entry experience, perhaps their skills are better fit for other aspects of the IPAV.
  • Future reporting could include determining the average rate at which any given election official generates ballots. Considering which election official is behind the computer screen and prioritizing those with more accuracy and speed could be a reasonable exploration to handle staffing at extremely busy sites.

There are natural barriers to each IPAV site. Integrating these variables into reporting makes for a more robust analysis.

  • We can only fit so many people, tables, chairs, and ExpressVote machines into some spaces. This report did not account for these constraints, nor is there consistent documentation for them.
  • Some sites may only allow IPAV at designated times due to conflicts with their own schedules.
  • There may be less liberty in adjusting the schedule in spaces like churches and affinity spaces over campus locations and libraries.
  • Holiday breaks and other anticipated events should be accounted for in building out IPAV schedule and staffing models.