The 1.5-mile section of Memphis’s Union Avenue on which 10 pedestrians have been killed since 2020 is the empirical case the Wall Street Journal built the article around. Bennett and Checkel’s process-tracing protocol structures the analysis in five moves: construct competing causal hypotheses before evidence is read in; classify each evidence piece by what its presence or absence would do to each hypothesis; apply the tests; update hypothesis status; and reconstruct the causal chain in temporal sequence. The substrate supports five candidates.
H1 — Stroad design (road geometry as proximal cause). Roads that try to function as both high-speed arterials and local-access streets concentrate pedestrian fatality risk. H2 — Sprawl as upstream cause. Postwar land-use patterns place housing far from jobs, forcing the construction of high-speed arterials (i.e., stroads); the geometry is downstream of land use. H3 — Pedestrian behavior in its strong form. Midblock crossings, inattention, and nighttime visibility are the proximate causes. H4 — Vehicle fleet composition. Larger, heavier vehicles strike pedestrians with greater force than the cars of prior decades. H5 — Crossing-infrastructure scarcity. Half-mile crosswalk spacing, narrow sidewalks, and inadequate lighting channel walkers into the most dangerous configuration.
Per-evidence tests.
In Bennett and Checkel’s terminology, a “smoking gun” is an evidence piece whose presence would discriminate decisively between hypotheses; a “hoop test” is an evidence piece a hypothesis must pass to remain viable; a “straw-in-the-wind” is an evidence piece consistent with a hypothesis but not decisive.
E1 — “Stroads … account for only a small share of the city’s roads, but 78% of pedestrian fatalities occur on roads of such widths,” per a WSJ analysis. The disproportionality is what Bennett and Checkel term a smoking gun for H1. H3 and H5 in their pure forms do not predict it; it falls out cleanly from geometry.
E2 — AAA data: a pedestrian struck head-on at 32 mph faces about a 25% chance of death; at 42 mph, twice the risk. A hoop test for H1; it passes.
E3 — Tennessee DOT’s 2022 report on pedestrian fatalities in the state finds urban roads with speed limits of 35 mph or higher are roughly three times as deadly as urban roads with lower speed limits. A hoop test for H1; the speed-mortality mechanism is confirmed at the relevant jurisdiction.
E4 — After a fatality on a state-owned road, the Tennessee Department of Transportation plans lighting and crosswalk repainting only. According to Steve Allen, TDOT’s director of local programs and community investments, “The state cannot add more crosswalks, reduce the number of lanes, or lower the speed limit on Union Avenue … because those changes would take time and money.” Allen placed the responsibility for pedestrian safety on walkers themselves: “Pedestrians think you have the right of way all the time. But that doesn’t matter if you’re not paying attention. You assume that responsibility when you’re walking out on that kind of roadway.” Allen attributed the no-redesign constraint to cost and time. According to Heidi Simon of Smart Growth America, the lighting-and-repainting plan probably will not make Union Avenue safer. The TDOT response is a smoking gun against H3 in its strong form on documented-conduct grounds: the corrective action the agency chose is confined to the right-of-way and does not include geometry changes.
E5 — In Milwaukee, Kevin Muhs, the city engineer, reports that dangerous roads were narrowed, given more crossing points, and had raised sidewalks installed. According to Jessica Wineberg, coordinator of Milwaukee’s Vision Zero program, pedestrian deaths fell 19% between 2024 and 2025, and “Wherever we see a redesign, we see reduction or total elimination of fatalities.” The substrate pairs the Milwaukee redesign data with a measured fatality change. The Tennessee lighting plan is paired only with Simon’s forecast, not a measured outcome. The comparison is asymmetric on what the substrate establishes.
E6 — Robert Schneider, co-chair of urban planning at the University of Wisconsin–Milwaukee, attributes part of the rise to fleet composition: “Americans have also gravitated toward larger vehicles in recent years, which strike pedestrians with greater force.” A hoop test for H4; the substrate confirms it. H4 sits on the causal map but is not weighed against H1 in the substrate, because the WSJ analysis establishes concentration on stroad-class roads without controlling for vehicle mix.
E7 — U.S. pedestrian deaths climbed 72% from 2009 to 2024 while declining in most other developed nations, according to United Nations data cited in the article. A hoop test for H1. A behavior-first explanation (H3) would predict pedestrian fatalities elsewhere too; the international divergence is consistent with a design-class variable.
E8 — Nicholas Ferenchak, an associate professor at the University of New Mexico and director of the Center for Pedestrian and Bicyclist Safety, frames the historical chain: “When you plop 200 houses down 15 miles from the place where people have to work, you have to build big roads where they can drive fast.” Heidi Simon of Smart Growth America observes that the businesses lining stroads often employ shift workers who cannot afford a car. Substrate straw-in-the-wind for H2: the postwar-era framing and the shift-worker employment pattern on stroad commercial frontages are both consistent with land use as upstream cause. The Smart Growth America finding that 57% of all roadway fatalities in 2024 occurred on state-owned roads, paired with the WSJ finding that 78% of Memphis pedestrian fatalities occur on stroad-class roads, adds a third H2-consistent substrate datum on wider-road concentration.
E9 — Rebecca Sanders (Fehr & Peers) and the article’s reporting document that midblock and nighttime crossings are the peak configuration for pedestrian death, and that Bessan-Gligo and other pedestrians cross midblock because the crosswalks on Union Avenue are half a mile apart. The behavior is itself a response to crossing-infrastructure spacing (H5); attention is required at the worst moment because design places the worst moment where it does.
What the substrate establishes across the hypotheses.
- H1 (road geometry). The 78% concentration of Memphis pedestrian fatalities on stroad-class roads is the substrate’s central H1-consistent finding. The AAA speed-fatality data, the Tennessee DOT 2022 report, and the international divergence all supply H1-consistent evidence. E4 (the TDOT response) documents that the agency with control over the geometry declined to change it.
- H2 (sprawl as upstream cause). Ferenchak’s postwar-era framing, Simon’s shift-worker observation, and the Smart Growth America fatality concentration on wider roads all bear on H2’s claim that land use is upstream of geometry. The substrate does not adjudicate present-day design choices on H2’s terms, but it documents the upstream channel.
- H3 (pedestrian behavior, strong form). The TDOT response is documented-conduct evidence against the strong form. Allen’s attribution of responsibility to walkers, paired with the agency’s choice not to change geometry, is the substrate’s central H3-relevant datum. The substrate’s other H3-relevant findings — midblock crossings, nighttime visibility, large vehicles — all sit downstream of geometry or H5.
- H4 (vehicle fleet). Schneider attributes part of the rise to fleet composition; the substrate confirms the trend without controlling for it against H1.
- H5 (crossing infrastructure). Half-mile crosswalk spacing on Union Avenue, narrow sidewalks, and six lanes of roadway that Sanders says would require substantial lighting all support H5 as a reinforcing channel.
Causal chain, in temporal sequence.
- Postwar land use places housing far from jobs (Ferenchak).
- State and local jurisdictions build high-speed, multi-lane arterials to connect them — stroads (Marohn).
- Commercial activity clusters along those arterials (Simon: businesses employ shift workers who cannot afford cars).
- Pedestrian infrastructure fails to keep pace — half-mile crosswalk spacing, narrow sidewalks, inadequate lighting across six lanes (Sanders).
- Walkers cross midblock because the designated crossings are distant (the article documents this for Bessan-Gligo and as a general pattern).
- The fleet has shifted toward larger vehicles (Schneider), and pedestrians are struck at speeds of 32 mph and above, on roads posted at 40 mph (AAA, Tennessee DOT 2022).
- Outcome: the WSJ analysis reports 78% of Memphis pedestrian fatalities on stroad-class roads; the Memphis metro ranks as the deadliest U.S. metro per capita in the Smart Growth America analysis through 2024; 15 people have been killed by cars along Union Avenue since 2020, with more than 50 pedestrians involved in traffic accidents on the 1.5-mile section Bessan-Gligo regularly crossed.
Analytical techniques used in this piece
This analysis applies the methods below. Each links to a short, plain-English explainer you can read and reuse.
- Process Tracing
- Reconstructs the step-by-step causal pathway of a specific historical event.