This Week In Public Transit

Transcript 43 lines

Cold Open Stats Overview Paper Walkthrough free_promo

Cold Open

Jenny When you think about getting people out of cars, what actually changes minds: better service, safer streets, or just making it easier?

Davis I want it to be better service, but I think the switch flips when the trip gets simple and reliable, like you don’t have to plan your whole day around it.

Jenny Okay, but when people say they “support transit,” do they mean they’ll pay more, give up parking, or just say yes to the idea while keeping the lane space for cars?

Davis That’s the funny part, because one new Montréal survey basically says transit wins across the board on support, and car-first spending loses pretty broadly, which is a real clue about what’s politically moveable.

Jenny If the public’s actually that consistent, the question becomes what agencies do with that permission slip, and what they keep blaming on “public opinion”…welcome to This Week In Public Transit on paperboy.fm.

Stats Overview

Davis Quick stats check for the week: we’ve got 4 papers in the feed, 16 unique authors, and work coming out of 2 countries. That’s a tight little bundle, but it still lines up with our through-line—what actually moves people toward transit, from public preferences to job access to keeping rail running.

Jenny And it’s fewer papers than last episode: 4 this week versus 5 before, a 20% drop. I don’t see a smoking-gun reason in the metadata, so I’m wondering—did we just have a quieter week in the usual venues, or did our query net pull in fewer transit-adjacent pieces?

Davis The authorship shrank too: 23 down to 16, so about a 30% dip. When I look at the methods—one survey, one comparative analysis, one bi-objective optimization, and one case study—it feels like four very different toolkits, which can mean fewer big multi-author consortium papers and more small teams shipping focused studies.

Jenny Country coverage narrowed from 3 to 2, down a third, and that matters because transit findings travel badly when governance and land-use don’t match. We’ve got Canada showing up twice and Iran once, and I’d want to know what dropped out—was it a missing region, or just a week where the indexing didn’t surface it?

Davis One more texture point: the author mix is very new this week—9 of the 16 authors, so 56%, are first-time, meaning it’s their first-ever paper we can see, not just new to our feed. Then 2 emerging authors, about 13%, and 5 experienced, about 31%, which is a nice reminder that a lot of transit evidence is still getting built by people early in their publishing arc.

Jenny Theme-wise it’s a clean sweep: urban transportation and public preferences are in there, plus accessibility and land-use—basically, who supports transit, what choices they face, and whether the system gets them to jobs. And with only 4 papers, every one of those themes is basically a quarter of the week, so we should treat any “trend” as a hint, not a verdict.

Paper Walkthrough

Paper 1 Understanding Public Preferences for Urban Transportation: Evidence from Montréal

Jenny Alright, let’s get into the papers, starting with “Understanding Public Preferences for Urban Transportation: Evidence from Montréal.”

Jenny It’s a Montréal-region survey where they ask people what kinds of transportation investments they actually support, but they don’t treat “support” like one blob.

Jenny The simple headline is: Montréal residents are consistently pro-transit, and they’re pretty cold on car-oriented investment, and that pattern changes depending on whether you ask about safety, space, money, equity, or infrastructure.

Jenny They compare five modes—car, bus, rail, walking, and cycling—across six dimensions: general support, safety, space, financial investment, equity by income, and infrastructure.

Jenny And when they run a repeated-measures ANOVA—which is basically “the same people rating multiple options, so you test the differences within-person”—public transit comes out high across every dimension, while car-focused spending is broadly unsupported.

Davis When you say “support,” what did people actually have to choose between across those six dimensions—like were they rating each mode separately for “safety” versus “financial investment,” or making tradeoffs?

Jenny They had respondents evaluate each of the five modes across each of those six dimensions, so you can see, for example, how someone feels about cycling on safety versus cycling on infrastructure, without forcing a single either-or choice.

Jenny Then the repeated-measures ANOVA tests whether those average ratings shift meaningfully by mode and by dimension, and they find big, significant variation on both axes, with transit staying strong and cars staying weak.

Jenny The clean limitation is that this is strong evidence about Montréal opinions, but it doesn’t automatically tell you what people in, say, Calgary or Houston would back when the politics and street design are different.

Davis The practical takeaway is kind of liberating if you’re a transit agency: you don’t have to sell transit as only “more service,” you can justify it on safety, space, equity, and infrastructure and still be aligned with what people say they want.

Davis And it tees up our through-line for the episode—what drives transit use—because this paper is about stated public support, not ridership behavior, but it tells you where the public permission structure is, at least in Montréal.

Paper 2 Which access matters? A comparative analysis of accessibility metrics and their impacts on commuting

Davis So Montréal told us what people say they’ll back, but that doesn’t tell you what actually moves bodies onto buses and trains.

Davis This next one, "Which access matters? A comparative analysis of accessibility metrics and their impacts on commuting," asks a nerdy planning question with a real payoff: which accessibility number best lines up with transit mode share across Census Tracts in Toronto, Montréal, and Vancouver.

Davis Plain version: the simple “how many jobs can you reach by transit in a reasonable time” measure predicts transit use as well as, or better than, fancier models that try to account for everyone competing for those same jobs.

Davis They compare non-competitive opportunity measures to competition-based ones, and the winners are cumulative opportunities calculated at the mean regional transit travel time and a gravity-based measure, meaning a measure that down-weights farther jobs instead of counting them equally.

Jenny When they say “higher explanatory power,” what is that in practice—are we talking a noticeably better fit, like a real jump in variance explained, or is it basically a rounding error that sounds impressive in a table?

Davis They run statistical comparisons at the Census Tract level, lining up each tract’s transit mode share with multiple job-accessibility metrics by public transit, then checking which metric has the stronger association and better model fit across all three metros.

Davis And the punchline is kind of deflationary in a good way: the simple cumulative-at-mean-travel-time measure and the gravity-based measure are basically neck-and-neck, and both beat the more complex competition-based measures, at least in these Toronto–Montréal–Vancouver contexts.

Davis The big limitation is baked in too: it’s three Canadian cities, so if you drop this into a region with very different job distribution, service patterns, or fare barriers, the ranking of metrics could shift even if “access to jobs by transit” still matters.

Jenny I love this because it’s a permission slip for planners who don’t have a modeling shop the size of an airline ops center.

Jenny If the simple access number tracks transit mode share better than the competition one in three big metros, then for the “what drives transit use” thread, it says the basic story might be enough: make more jobs reachable by transit in a typical commute time, and you’re more likely to see mode share move.

Paper 3 Security Actions in Urban Rail Transport Networks: A Multi‐Objective Optimization Approach

Jenny That “permission slip” point about not needing an airline-sized modeling shop is still ringing in my ears, and it tees up a different kind of model-heavy paper: Security Actions in Urban Rail Transport Networks: A Multi‐Objective Optimization Approach.

Jenny It’s Kouroshniya and Hasany in Risk Analysis, twenty twenty-six, and they’re asking a very specific question: if someone tries to disrupt a rail network, which links matter most, and how should a defender spend a limited security budget?

Jenny Plain language first: they show that putting more resources into defenses you can move around fast—like shifting guards—can sharply cut how much an attacker can slow the network down.

Jenny And the way they define “damage” isn’t just longer trips; it’s also less predictable trips, because the attacker in their model tries to maximize average travel time and a weighted variance, meaning the spread in travel times so riders can’t rely on the schedule.

Davis So in their world, what counts as “success” for the attacker—are they basically just trying to make trains slow, or are they explicitly trying to make the system feel unreliable, like you can’t plan a commute anymore?

Jenny It’s both, by construction: they set it up as a bi-objective optimization, which just means two goals at once, where the attacker chooses which arcs to hit to push up mean travel time and that weighted variance, while the defender chooses where to place security to pull both down.

Jenny They focus on defenses that are rapidly reallocated and partly unobservable—guards you can move and not fully telegraph—so they model attacker and defender simultaneously, then solve the minimax problem with Lagrange relaxation and Frank–Wolfe to turn it into something convex they can actually compute on a big network.

Jenny In the Iran railway case study, they find you can get these threshold effects—linear, nonlinear, even peak-like behavior—where once the defender budget crosses certain points, attacks start to become ineffective in terms of those travel-time outcomes, but it’s still a model-based result tied to their assumptions and that specific network.

Davis This lands for the “keeping rail resilient” thread because it’s basically a prioritization tool: protect the few arcs that blow up average time and reliability first, especially if your defenses are mobile.

Davis I also hear the caveat loud and clear: it’s strong as decision support inside its assumptions, but until you see it stress-tested on other networks and real incident patterns, it’s more like a smart map for where to look than a promise that spending X dollars buys Y minutes of reliability back.