Transit Access and Neighborhood Income Across Chicago

Abstract

Chicago's rapid transit map is a fact a person can stand inside. The "L" reaches some neighborhoods with a station every few blocks and skips others for miles. This paper sets the published transit-equity literature beside a descriptive analysis of real, downloadable public data, drawn from the Chicago Transit Authority list of "L" stops, the CTA station-entry totals, and the City of Chicago community-area income file. There is no causal claim here, no demand model, and no door-to-door travel time. The exercise counts stations, measures where they sit, and cross-tabulates that footprint against neighborhood income. The pattern is plain. Chicago has 144 unique rail stations, yet only 42 of its 77 community areas contain one, which leaves 35 with none.^[1] Station density climbs from 0.35 stations per square mile in the lowest-income quartile of neighborhoods to 1.24 in the highest, more than three times as dense, and 2023 entries per station rise across the same gradient from about 328,000 to about 991,000.^[1] Step-free access runs the other way, highest where stations are scarcest. The Red Line touches 14 community areas while the Yellow touches one.^[1] These numbers describe a geography that a Chicago-specific and national literature already predicts, from peer-reviewed measurement of accessibility across the city to recent mapping of its transit deserts.^[2][5][6] The limits of a decade-old income measure and a point-in-polygon geography get stated plainly, and the paper closes on what the map asks of policy.

Where the train goes, and where it stops short

Stand on the platform at Clark/Lake and you are inside the densest tangle of rail in the Midwest. Several CTA lines pass within a few blocks. Trains stack overhead on the Loop elevated, dive into subway tunnels under the river, and fan out toward the two airports and the edges of the city. Ride forty minutes south or west and the map changes character. The stations thin out. The lines straighten into single spines. Then the rail ends, and the neighborhoods keep going for miles without it.

That contrast is the whole subject of this paper, and it is not a figure of speech. It is a count. Chicago has 144 unique CTA rail stations, recorded across 302 platform-and-direction entries in the agency's own list of "L" stops, since each direction of travel at a station is logged as its own row.^[1] Those 144 stations do not spread evenly across the city. Only 42 of Chicago's 77 community areas contain even a single rail station, which means 35 community areas, nearly half the city's official neighborhoods, have none at all.^[1] A resident in one of those 35 places does not live near a slow train, or an infrequent train. There is no train.

Picture the city the way the map draws it. A dense downtown core holds stations within a few blocks of one another, the Loop and the near neighborhoods that ring it. From that core the lines radiate outward like spokes, and as they travel they lose stations and lose company. The far reaches of the North Side keep their rail. Long stretches of the South Side and the West Side do not. The gap is not random noise scattered across the map. It has a shape, and the shape runs along income.

That gap is a set of real places, and abstraction lets the reader off too easy. Riverdale, at the city's far southern edge, has the lowest per-capita income of any community area in our data, $8,201, and it has no rail station.^[1] New City, on the Southwest Side, has none. Humboldt Park, on the West Side, has none. Auburn Gresham, on the South Side, has none. West Pullman, deep south, has none.^[1] These are not marginal slivers of the city. They are large, named, populated neighborhoods, and on the rail map they read as blank. A worker in Auburn Gresham who needs the train starts the trip on a bus, or on foot, to reach a station that sits in someone else's neighborhood.

Precision about what this paper is matters here, because precision is the house rule and because the subject invites overreach. The method pairs published research with a descriptive read of real public data, and that is the whole of it. Nothing was changed and no intervention was measured, so the work is descriptive rather than experimental. No rider was interviewed, and no quotation in these pages comes from a person we contacted. Nothing was modeled, simulated, or predicted, so there is no forecast of demand and no estimate of how many people would board a station that does not yet exist. And no causal conclusion is drawn. The paper does not claim that low income produces thin rail service, or that thin rail service produces low income, or that either one drives the other. The work counts what is on the ground, sorts it by neighborhood income, and reports what the sorting shows. The arrows of cause stay out of it.

What we can do responsibly is map. The data is real and downloadable, and we name every file. The CTA list of "L" stops gives us each station, its coordinates, the lines that serve it, and whether it is wheelchair accessible.^[1] The CTA station-entry file gives us boardings.^[1] The City of Chicago's community-area income file gives a per-capita income figure for almost every neighborhood.^[1] We place each station inside the community area whose boundary contains its coordinates, group those community areas by income, and tabulate. The result is a description of a geography, built from public records, that anyone can reproduce from the files we cite. The reproducible files sit on the project's data archive, named and downloadable, so the count in this paper is not a claim a reader has to take on faith.^[1]

The piece follows the map the way a rider would. The network gets walked color by color, with an eye to how far each line reaches and which parts of the city it touches. The neighborhoods get sorted into income quartiles to see how rail density changes as income changes. Then the question of who can actually board, since a wheelchair user stopped by a flight of stairs gets no use from a platform overhead. Then where the boardings pile up. And finally what a literature built by other researchers makes of patterns like these, set against a plain statement of what our own numbers can and cannot support.

The central question runs underneath all of it. How far does the train reach, and who lives where it does not? The CTA map already holds the answer. This paper reads it out, station by station, and sets it beside the published record so the pattern is documented and not merely glimpsed.

The thirty-five blank neighborhoods

The absences are the cleanest finding here, so they come first. Thirty-five of Chicago's 77 community areas have no rail station of any kind, and they are not scattered evenly around the city.^[1] They lump together on the South and West Sides, the same parts of the map that thin out when you trace the lines by color. To see who lives in the blank spaces, rank the rail-less areas by the City's per-capita income figure and read from the bottom up.

Riverdale sits at the bottom, a far-Southeast-Side community area with a per-capita income of $8,201, the lowest of any neighborhood in the data, and no station.^[1] The City's hardship index, which the income file carries alongside the per-capita figure and on which a higher number marks a more distressed neighborhood, puts Riverdale at 98, among the highest in the city.^[1] Above Riverdale on the income ladder but still rail-less come Burnside at $12,515 with a hardship index of 79, New City at $12,765 with a hardship index of 91, and Humboldt Park at $13,781 with a hardship index of 85.^[1] Keep climbing and the list continues through South Deering at $14,685, Hermosa at $15,089, Belmont Cragin at $15,461, and Auburn Gresham at $15,528, every one of them without a train and every one carrying a high hardship index.^[1] Archer Heights, West Pullman, South Chicago, West Lawn, East Side, Oakland, and South Shore round out the lowest stretch of the rail-less list, none above roughly $19,400 in per-capita income.^[1] These are not edge cases. They are a long roll call of the city's lower-income and higher-hardship neighborhoods, and the common thread among them is that the "L" does not stop there.

The hardship index earns a place in the description because it tracks something a bare income figure can miss. It is the City's own summary of neighborhood distress, and a high value flags the kind of acute need that the transportation-equity literature treats as the case where transit access matters most.^[7][8] Riverdale at 98, New City at 91, Humboldt Park at 85, Burnside at 79, Oakland at 78, and South Chicago at 75 sit at the high end of the index, and all six neighborhoods are rail-less.^[1] When the people with the sharpest need for a cheap, reliable way across the city are also the people without a station, the access question stops being abstract. It is a specific set of households, in a specific set of named places, starting every rail trip with a transfer.

Two cautions keep this from being overstated, and both belong right here rather than buried later. First, the rail-less neighborhoods are not uniformly the poorest in the city by the median. The median per-capita income of the unserved areas is $21,301, and the median of the served areas is $22,008, a gap of only about seven hundred dollars.^[1] On that one summary number, the rail-less neighborhoods are barely poorer than the rest. The pattern lives in the bottom of the distribution, where the very poorest neighborhoods cluster among the rail-less, and not in the middle, where served and unserved areas look much alike. Second, a neighborhood without a station still has transit. Every one of these places is served by CTA buses, and a bus can carry a rider to a rail station in a neighboring area. The finding is about fixed rail specifically, the high-capacity, high-frequency, grade-separated spine of the system, and a rail-less neighborhood is one whose residents reach that spine only after a bus ride or a walk.^[1] We hold to the narrower, defensible claim. On the rail map, thirty-five neighborhoods are blank, and the blank ones skew toward the bottom of the income and hardship distributions.

This map was not drawn at once, and a little framing makes the present shape legible without adding a single new number to the analysis. The "L" was not planned as one system. It was assembled, piece by piece, from separate elevated lines built by competing private companies, each laying track toward the downtown it wanted to capture and out into the neighborhoods it expected to pay. The loop of elevated track that gave the system its name came later, tying the private routes together at the center. The City did not run the trains at first. Private operators did, until a public agency was created to buy out and consolidate the failing private lines into the single system the CTA runs today. By the time the public agency took over, the basic geometry was already set, a hub of overlapping routes downtown and on the North Side, with thinner reach into the South and West Sides that the private companies had found less profitable to serve. The later additions leaned on the expressways, with branches laid down the medians of highways that were themselves cut through particular neighborhoods and not others. The rail map a Chicago rider stands inside today is the residue of those choices, private and public, profit-driven and highway-driven, accumulated over more than a century. We do not measure that history here, we attach no number to it, and we draw no causal weight from it. The point of naming it is narrower. The thirty-five blank neighborhoods are the outcome of a long sequence of building decisions, not a fact of nature, and a description of where the rail sits now is also a description of where all that building did and did not reach.

What other people found before we counted

This is not the first look at the question, and the descriptive numbers below only mean something against the work that came before. A real research literature, some of it written about Chicago specifically, has spent years asking how transit access is distributed and who ends up on the wrong side of the distribution. That literature supplies the framing and the stakes. Our own contribution is narrower, a plain map. Setting the two side by side is the point of this section.

The first question is what to even measure. It is tempting to count service mileage, to add up how many bus and rail miles a city operates, and to call the place with more miles better served. The distributive-justice scholarship in transportation argues that this is the wrong yardstick. What matters for equity is accessibility, meaning the ease with which a resident can actually reach the places they need to go, not the raw quantity of service an agency provides.^[7] Two neighborhoods can sit beside the same number of route-miles and still have wildly different access to jobs, clinics, and groceries, depending on where those routes go and how often they run. The move from counting service to measuring access is the conceptual shift that organizes the modern field, and it is why our station counts are a starting point rather than a conclusion. A station is service. Whether it gets you somewhere is accessibility, and accessibility is the harder, truer measure. Pereira, Schwanen, and Banister make the normative version of this argument, drawing on theories of justice to contend that the fair distribution of accessibility, not vehicles or track, is what a transport system owes its residents.^[7] This paper borrows that yardstick and stops short of their measurement, counting stations rather than reachable destinations, but the yardstick is theirs.

Once accessibility is the metric, a pattern recurs across cities. Accessibility gaps tend to track income. Bocarejo and Oviedo, studying transport accessibility and social inequity, find that the residents with the least access to opportunity by transit are disproportionately the lower-income ones, and that the relationship is systematic rather than a local accident, visible when you line accessibility up against income across a metropolitan area.^[8] The finding travels. It shows up in different cities, with different transit systems, measured by different teams. That recurrence is exactly why a single-city description like ours is worth doing carefully. If Chicago's rail footprint sorts by income, it is not an anomaly. It is the local instance of something the literature has documented elsewhere, and the comparison to the published cases is what tells us whether our blunt count is picking up signal or noise.

Chicago has been studied directly, and the Chicago-specific work gives our map its closest comparison. Ermagun and Tilahun measure the equity of transit accessibility across Chicago and find it unevenly distributed across the city's neighborhoods and across demographic groups, with the gaps falling along lines of both geography and population.^[2] That is an established result about this city, not a claim this paper introduces, produced with accessibility methods more careful than anything attempted here, and our descriptive station counts sit underneath it as a coarser view of the same terrain. The 35 community areas with no rail station, and the tilt of the rail-less ones toward the lower-income end, describe with a blunter instrument the same distribution that careful Chicago measurement has already characterized. Their work computes what a resident can reach. Ours counts what stands nearby. The second is a shadow of the first, and the shadow falls the same way.

The Chicago evidence extends past rail counts into what rail counts are supposed to buy, namely the ability to reach the destinations that hold a life together. Liu, Kwan, and Kan, analyzing transit-based access to healthcare in the Chicago metropolitan area, find that a higher share of minority population in an area is associated with lower transit-based accessibility to key destinations, healthcare among them.^[5] Read that against our map and the stakes sharpen. A neighborhood without a nearby station is not merely missing a convenience. The same places that come up short on rail tend to come up short on transit access to the clinic, the hospital, the specialist. The deficits compound on the same ground, so the blank space on our map is also, in their measurement, a longer trip to a doctor.

More recent Chicago work makes the geography explicit and gives it a name. Rozhkov, Zandiatashbar, and Seyrfar map the city's transit deserts, the areas where transit supply falls short of transit need, and locate those deserts on the South and West Sides, concentrated in low-income, predominantly minority community areas.^[6] That is the same corner of the map where our station count goes blank. When we list Riverdale, New City, Humboldt Park, Auburn Gresham, and West Pullman as rail-less, lower-income places, we are pointing at neighborhoods that this transit-desert work has already flagged through a different method built on different data. Two roads arrive at one destination. Our descriptive blanks and their measured deserts overlap because they are describing a single city, and the overlap is the kind of corroboration a coarse method earns when it lands where a finer one already stood.

Why does any of this matter beyond the inconvenience of a long walk to the train? Because access to transit is tied to access to work, and the link is among the most durable findings in the field. Blumenberg and Ong, comparing transit and automobile access to jobs among welfare participants in Los Angeles, find that residents who depend on transit reach far fewer jobs within a reasonable commute than residents who can drive, and that the mobility gap falls hardest on the low-income households least able to absorb it.^[11] Grengs deepens the point with his study of the modal mismatch in Detroit, where the spatial distribution of jobs and the reach of transit do not line up, so car-less residents reliant on the bus face large job-accessibility penalties and the people caught in the mismatch are those who can least afford a car to bridge it.^[4] A neighborhood without rail is, through this lens, a neighborhood whose car-less residents face a steeper climb to employment than their cross-town neighbors, and the climb is measured in jobs they cannot reach in the time they have.

The employment connection is not theoretical hand-waving. Sanchez measured it at the level of individual residents, studying the connection between public transit and employment among TANF recipients in Portland and finding that proximity to transit was linked to better employment outcomes for the people who depended on it, with a person's ability to reach a job hinging on whether transit could carry them there.^[10] The chain runs from a station on a map to a household's odds of holding work. We do not measure that chain in this paper, and we say so carefully below. But it is the reason the map is worth reading. The blank spaces are not blank in their consequences.

So what exactly are we adding? Not a new causal finding, and not a new accessibility model. We are adding a clean descriptive map built entirely from public data, and the choice to build it that way has its own precedent. Karner shows that credible transit-equity analysis can be carried out with publicly available data and route-level accessibility measures, rather than requiring proprietary feeds or bespoke surveys.^[3] That is our approach, real downloadable files and simple transparent measures, with the reproducibility that public data allows, though Karner computes accessibility along routes while we stop at counting stations. The distinction matters for honesty about what we deliver. Karner's route-level method captures how often service runs and where it can take a rider, which is a step closer to accessibility than a station count, and our decision to stop at counts is a decision to trade resolution for reproducibility and transparency. Anyone can verify a count against a published list of stops. Fewer readers can audit an accessibility surface. We chose the measure that a skeptical reader can check line by line, and we accept that it sees less than Karner's does.

There is also a precedent for relating transit access to social need across neighborhoods and watching the relationship over time. Foth, Manaugh, and El-Geneidy do exactly that for Toronto across a decade, pairing accessibility with the socioeconomic profile of the places it serves and tracking whether the relationship between transit access and social need grew more or less equitable between 1996 and 2006.^[9] We run a one-period, single-city version of that pairing, with station counts where they use accessibility surfaces and a single income snapshot where they use change over time. The time dimension they capture is one we cannot, since our income file is a single decade-old window, and that is a real difference between a study built to detect change and a snapshot built to describe a present state. We describe a state. They measured a trend. Both are legitimate, and ours is the more modest.

One more strand of the literature concerns what a station count can stand in for. The conceptual core of the field is the gap between mobility, meaning the ability to move, and accessibility, meaning the ability to reach. A neighborhood can have high mobility, plenty of buses and trains passing through, and still have low accessibility if those vehicles do not go where the jobs and clinics and schools are, or do not run often enough to be useful. Pereira, Schwanen, and Banister make accessibility the object of justice precisely because it, rather than raw mobility, is what determines whether a resident can participate in the life of the city.^[7] A station count, the measure we use, is closer to mobility than to accessibility, since it records whether a high-capacity vehicle stops nearby without recording where that vehicle can take a rider or how often. We are honest that this places our measure a rung below the field's preferred one. What rescues the exercise is that the Chicago-specific accessibility studies, which do measure reach, find the same geographic tilt our cruder mobility proxy finds, so the proxy is tracking the thing it is a proxy for.^[2][5][6] When a coarse measure and a fine one agree about the shape of a city, the coarse measure has earned a limited trust.

The literature does two jobs for us at once. It tells us what to expect, that rail access in a city like Chicago will be uneven and will sort by income and race, and it tells us that a public-data description is a legitimate way to look. What it does not do is our looking for us. The published Chicago findings are several years old, use their own geographies, and answer their own questions about reachable destinations rather than nearby stations. The contribution here is to take the current station roster and the City's own income file, lay one over the other, and report the cross-tabulation plainly. We add the map. The literature tells us, in advance, roughly what the map will say, and our job is to check whether it says it.

Reading the network one color at a time

The quickest responsible way to see the shape of Chicago rail is to take the lines one color at a time and ask how much of the city each one touches. Stand at Howard, the northern terminal, where three lines share one canopy. Ride south and the Red Line keeps going, under the river, through the Loop, down the median of the Dan Ryan Expressway to 95th Street. No other line touches as much of the city. The CTA runs eight rail lines, and they are not equals. Some thread through a dozen neighborhoods. One barely leaves a single one. Counting the community areas each line reaches turns the familiar transit map into a ranking, and the ranking is where the geography starts to talk.

The Red Line reaches the most. It touches 14 of Chicago's community areas, more than any other line, running the length of the city on a north-south spine from the far North Side down through the Loop and on toward the South Side.^[1] That spine is the backbone of the whole system, the one line that genuinely crosses the city end to end, and its reach into 14 neighborhoods reflects that. If a single line had to stand for Chicago rail, it would be this one, because it comes closest to spanning the entire map.

Behind the Red Line sit two lines tied for second. The Blue Line touches 12 community areas, and the Green Line touches 12 as well.^[1] The Blue runs the long northwest-to-southwest diagonal, O'Hare to Forest Park, cutting across the street grid the way an expressway does because for much of its length it follows one, an expressway median on the way out of downtown and a subway and a second highway median on the way to the airport. The Green is among the oldest alignments in the system, an elevated east-west and south route descended from the earliest South Side and Lake Street elevated lines, and it reaches into parts of the West and South Sides that other colors miss, running out to Oak Park on one end and splitting toward Ashland/63rd and Cottage Grove on the other. Twelve neighborhoods each is real coverage, and between them the Red, Blue, and Green carry the bulk of the system's geographic spread. They are also the three lines that most clearly cross between the wealthier center and north and the lower-income south and west, which is why they do the heavy lifting in any account of how the network reaches across income.

The middle of the ranking thins quickly. The Orange Line touches 9 community areas, a single arc southwest toward Midway Airport that threads a corridor without branching, and it is among the newest lines in the system, which is part of why it serves a stretch of the Southwest Side that older alignments skipped.^[1] The Brown Line touches 7, looping up through Lincoln Square and Ravenswood on the North Side and back, dense where it runs but narrow in the count of neighborhoods it enters.^[1] The Purple Line touches 6, most of them on its Evanston end outside the city, the Pink Line touches 5, and then the drop becomes a cliff.^[1] The Yellow Line touches exactly 1 community area.^[1] It is a short shuttle running north of the city, and its reach in the count matches its reach on the ground. One line crosses fourteen neighborhoods. Another crosses one. Both are the "L," and a rider in the wrong place gets the second kind, or gets nothing.

There is a tempting misreading of these counts that the data itself corrects. A line that touches few community areas is not necessarily a failure of coverage, since the Brown Line, with only 7, is one of the busiest and most useful lines in the system for the dense North Side neighborhoods it threads. Reach across many neighborhoods and intensity of service within a few are different virtues, and the count rewards the first, not the second. The right way to read the ranking is geographic, namely as a measure of how widely each color spreads across the city's neighborhoods, not as a verdict on how good any one line is for the people already beside it. What the spread shows is that the lines built to traverse the city, the Red above all, do most of the work of connecting its far parts, and the lines built to serve a single dense corridor, however well they serve it, do little of that connecting work.

Counting the stations on each line sharpens the same picture. The Red Line carries 33 stations, the Blue 28, the Green 28, the Brown 27, the Pink 20, the Purple 18, the Orange 16, and the Yellow just 1.^[1] Reach and station count do not move in perfect lockstep, because some lines pack many stops into a few dense areas while others spread fewer stops across more ground. The Brown Line is the clearest case, 27 stations but only 7 community areas, a line that threads a lot of platforms through a compact slice of the North Side and the Loop.^[1] The Orange Line runs the other way, 16 stations across 9 areas, fewer stops covering more map.^[1] The lesson is that station count and geographic reach are two different measures of a line, and a line can be generous on one while stingy on the other.

A fair read of the network has to set the suburbs aside, and the data does. Of the 144 unique stations, 20 sit outside Chicago entirely, in the suburbs the "L" reaches.^[1] They include the Evanston stops along the Purple Line, namely Davis, Dempster, Main, Noyes, Foster, Central, South Boulevard, and Linden; the Skokie ends of the Yellow at Dempster-Skokie and Oakton-Skokie; the Oak Park and Forest Park stops on the Green and Blue, namely Harlem/Lake, Ridgeland, Austin, Harlem, Forest Park, and two distinct stations both named Oak Park; the Cicero-area Pink Line stops at 54th/Cermak and Cicero; and Rosemont on the Blue.^[1] Those 20 stations are real and carry real riders, but they fall outside the 77 Chicago community areas, so we set them aside for every neighborhood-level table here. That count includes the two separate suburban stations that both carry the name Oak Park, which is why a reader tallying the names should land on twenty rather than nineteen. Holding the suburbs out leaves 124 stations inside the city.^[1] When the later sections sort stations into income quartiles and compute density, they work with those 124 city stations, not the full 144. We flag the distinction here so the figures line up later and so no one wonders where 20 stations went. They went to the suburbs, and the community-area analysis is a Chicago analysis.

Translated back into geography a reader can picture, the colors themselves are not the point. The point is direction. The lines that reach farthest, the Red above all, run the north-south length of the city and converge downtown. The lines that reach least cover short or single corridors. Stack the colors on the map and they pile toward the center and lean toward the North Side, where multiple lines overlap and a rider can choose among them. Move toward the South and West Sides and the colors do not overlap so much as fan out, single lines reaching thinly into large territories, with long gaps between them where no color runs at all. The Green Line's southern and western branches and the Red Line's south leg do reach into those parts of the city, but they reach as lone lines, not as the thick overlapping mesh that covers downtown and the North Side. Coverage is not just present or absent. It is layered in some places and bare in others, and the layering favors the center and the north.

The line-by-line read sets up everything that follows. More colors, longer reach, and thicker overlap cluster toward the center and the North Side. Fewer colors, shorter reach, and long bare gaps mark the South and West Sides. That is a statement about line geography, made entirely from counting which neighborhoods each color enters. The next section takes the same 124 city stations and sorts the neighborhoods not by compass direction but by income. The striking thing is how closely the income sort reproduces the geographic one. The lines lean toward the center and the north. Income leans the same way. The two maps, color and income, turn out to be nearly the same map.

How density tilts uphill with income

Counting which neighborhoods a line touches tells you where rail goes. It does not tell you how thick the rail is once it gets there. For that, sort the city by income and measure density, and the gradient comes into focus. This is the analytical heart of the paper, and the method is deliberately plain so that anyone can check it against the files we name.

Here is the grouping, stated without ornament. We take the 77 community areas, attach the City's per-capita income figure to each one, and sort them into four income quartiles, roughly 19 neighborhoods per group.^[1] Within each quartile we sum the city rail stations that fall inside those neighborhoods and divide by the group's total land area, which gives stations per square mile.^[1] We use density rather than a raw count because a quartile with more land would hold more stations even if rail were spread perfectly evenly, and dividing by area removes that distortion. The question density answers is simple. For the same square mile of ground, how much rail does a neighborhood at this income level have?

The answer climbs, and it climbs unevenly. The lowest-income quartile holds 20 stations across 57.9 square miles, which works out to 0.35 stations per square mile.^[1] The second quartile is a touch denser at 0.41, with 22 stations over 53.1 square miles.^[1] The third quartile dips to 0.27, 17 stations across 64.1 square miles, the sparsest of the four.^[1] Then the top quartile jumps to 1.24 stations per square mile, 65 stations packed into 52.3 square miles, more than three times the density of the lowest group and more than four times the third.^[1] Read the ends against each other and the gap is stark. For every square mile of the poorest quartile there is roughly a third of a station. For every square mile of the richest there is more than one. A rider in the top quartile is, on the ground, surrounded by more than triple the rail of a rider in the bottom one.

The dip in the middle should not be smoothed away. The progression is not a clean staircase from low to high. It rises from 0.35 to 0.41, falls to 0.27 in the third quartile, then jumps to 1.24 in the fourth.^[1] The third quartile is the low point of the four, below even the poorest group, because several of the city's middle-income community areas are large in land and thin in rail, and the number reflects that rather than hiding it. We report it because it is what the data shows and because tidying the non-monotonic step would be the kind of smoothing this project refuses. A truthful description is not a smooth uphill line. It is a jagged climb that ends far higher than it starts, with a real trough in the upper-middle before the top quartile pulls sharply away from everyone else. The headline is the spread between bottom and top, and the spread is genuine, but the path between them wobbles, and we say so plainly.

A single number summarizes the tendency, with its guardrail attached. Across all 77 community areas, the Pearson correlation between per-capita income and station density is +0.398.^[1] The correlation between the City's hardship index and station density is -0.183.^[1] A positive sign on income and a negative sign on hardship describe the same pattern from two directions, that density tends to run higher in higher-income, lower-hardship neighborhoods.^[1] These are descriptive Pearson coefficients across the community areas, computed with no controls and carrying no causal claim.^[1] A correlation of +0.398 is a moderate, real association, not a deterministic law. Plenty of individual neighborhoods sit off the trend, and a few of them surface in a moment. The coefficient says that, on the whole, richer neighborhoods tend to be denser in rail. It does not say income put the rail there, and neither do we.

Named extremes anchor the gradient so it stops being arithmetic. At the top of the density ranking sits the Loop, with 9.63 stations per square mile across sixteen stations, a figure that reflects downtown's tangle of overlapping lines packed into a small, intensely built area, on a per-capita income of $65,526.^[1] The Near North Side follows among the leaders at 2.55 stations per square mile, carrying the highest per-capita income in the city at $88,669.^[1] Lake View holds seven stations at 2.24 per square mile on $60,058.^[1] Against those numbers, set the rail-less lower-income neighborhoods named at the start, Riverdale and New City and Humboldt Park and Auburn Gresham and West Pullman, each carrying a density of zero because each holds no station at all.^[1] The distance from 9.63 to zero is the full range of the city, and the two ends are not scattered at random across the income map. The dense end clusters where incomes are high and the ground is expensive. The empty end sits in lower-income neighborhoods on the South and West Sides.

The full top of the density ranking fills the picture in. Below the Loop, the Near North Side, and Lake View come Edgewater and Grand Boulevard, tied at 2.30 stations per square mile, on per-capita incomes of $33,385 and $23,472 respectively, a pairing that already shows the trend is not airtight, since Edgewater is a middle-income North Side neighborhood and Grand Boulevard a lower-middle-income South Side one and they land at the same density.^[1] Rogers Park follows at 2.18 on $23,939, the Near West Side at 1.76 across a striking ten stations on $44,689, and Armour Square at 2.01 on $16,148.^[1] The list mixes incomes more than the quartile averages do, which is the honest texture of a +0.398 correlation. The averages tilt clearly uphill; the individual leaderboard does not march in income order.

The density ranking also carries its exceptions, and they matter for keeping the story straight. Fuller Park, a low-income South Side community area with a per-capita income of $10,432, ranks second in the entire city for station density at 2.80 stations per square mile, because it is a very small area that happens to contain rail.^[1] East Garfield Park, with a per-capita income of $12,961, ranks high as well at 2.07.^[1] These are lower-income neighborhoods that are dense in rail, and they are part of why the income-density correlation is +0.398 rather than something far stronger. The relationship is a tendency with real outliers, not a clean rule. We name the outliers on purpose, because a +0.398 correlation includes them by definition, and a paper that buried Fuller Park to make the gradient look cleaner would be doing the thing we promised not to do. The arithmetic of density also rewards small neighborhoods, which is worth saying plainly. Fuller Park ranks high not because it has many stations, it has two, but because its land area is tiny, so two stations divide into a small number of square miles and the ratio shoots up.^[1] Density is a fair measure of how much rail sits underfoot per unit of ground, but it says nothing about how many people that rail can serve, and a small dense neighborhood and a large dense one are different facts wearing the same number.

The income figures themselves deserve to be in view, since the quartiles rest on them. Per-capita income across the community areas ranges from a low of $8,201 to a high of $88,669, with a median near $21,668.^[1] That is more than a tenfold spread from the poorest neighborhood to the richest, and the quartile boundaries fall at roughly $15,906 at the 25th percentile and $29,642 at the 75th.^[1] The income data is real, it is the City's own, and it carries an important limit we treat at length later, namely that it comes from the 2008 to 2012 American Community Survey and is a community-area average rather than a household median.^[1] We hold that caveat for its own section rather than letting it interrupt the gradient here, but we flag it now so the quartiles are read for what they are, a decade-old per-capita measure, blunt but real.

One nuance in the data complicates the easy version of the story, and it belongs here rather than in a footnote. If you compare the median per-capita income of the served community areas against the unserved ones, the gap is small. Served areas run a median of $22,008, unserved areas $21,301.^[1] On that single summary statistic, the rail-less neighborhoods are barely poorer than the ones with stations. The sharper pattern lives in the tails and in the density, not in the medians. The very poorest neighborhoods are heavily represented among the rail-less, the very richest among the dense, and the gradient that the density numbers capture is far steeper than the modest median gap suggests. Both belong on the page, because the modest median gap is true and the steep density gradient is also true, and a description that showed only one of them would mislead. The honest summary is that having any station at all does not sort cleanly by income, but how much rail a neighborhood has, once it has any, sorts steeply.

What the density read establishes is the second of the two maps. The line-by-line section drew a map by compass direction, with thick overlapping coverage toward the center and the north and thin single lines toward the South and West Sides. The density section draws a map by income, with 1.24 stations per square mile in the top quartile and 0.35 in the bottom. Lay the two over each other and they nearly coincide. The neighborhoods that are dense in rail are the higher-income ones, and the higher-income ones cluster where the lines already overlapped. The +0.398 correlation is the numeric statement of an overlap a rider can see by eye, that the train is thickest where the money is and thinnest where it is not, with real exceptions at both ends and not one word of cause attached.

Who can actually get on the platform

A station on the map is not the same as a station you can board. Stairs sit between many Chicago platforms and the street, and whether a rider can reach the train without them is its own map, drawn in a direction that complicates any simple story about wealth buying better transit.

Sort the city stations by income quartile again and look at the share that are ADA-accessible. The lowest-income quartile is the most step-free, 19 of its 20 stations, 95.0 percent.^[1] The second quartile drops to 86.4 percent, 19 of 22.^[1] The third falls to 76.5 percent, 13 of 17.^[1] The highest-income quartile is the least accessible of the four, 46 of 65 stations, 70.8 percent.^[1] Citywide, 97 of the 124 city stations are ADA-accessible, 78.2 percent.^[1] The gradient runs backward from density. The places with the fewest stations have the most step-free ones; the place with the most stations has the largest share a wheelchair user cannot board unaided.

The reason sits downtown. The wealthiest quartile contains the Loop and the old North Side elevated, which means it contains the oldest steel in the system, platforms built generations before the Americans with Disabilities Act and not yet rebuilt to it. Density and age travel together there, a lot of stations and a lot of stairs. The lower-income quartiles, by contrast, hold a higher share of stations that were built or reconstructed more recently and came out step-free.^[1] So the same neighborhoods that have few stations tend to have accessible ones, and the same downtown core that has the most stations drags its quartile's accessibility share down.^[1] The equity picture here is layered, not flat. It runs scarce-but-accessible in some low-income areas and dense-but-partly-inaccessible at the center, and a rider who uses a wheelchair experiences the two halves of the city very differently from a rider who does not.

Two accessibility counts run through the analysis, and they differ on purpose, so each has to be named clearly. At the platform and stop-record level, the level at which each direction of each station is its own row, there are 224 accessible records and 78 that are not, 74.2 percent, which is the figure that matches the source file directly.^[1] Deduplicate those records down to unique stations and 106 of the 144 are ADA-accessible.^[1] Restrict that to the 124 stations inside city limits and 97 are accessible, the 78.2 percent cited above.^[1] The stop-record share and the unique-station share answer slightly different questions, one weighted by platforms and directions, the other by stations on the ground, and both belong here rather than collapsed into a single tidy number. They differ by a few points, and the difference is real arithmetic, not a rounding artifact, so flattening it would hide a genuine feature of how the source counts its platforms.

The backward gradient is a useful corrective against a too-simple reading of the density numbers. It would be easy to take the density finding and conclude that wealthy neighborhoods get the good transit and poor ones get the bad, full stop. The accessibility numbers break that symmetry. On the single dimension of step-free boarding, the lowest-income quartile is the best served in the city, and the highest-income quartile is the worst, because step-free access is a function of when a station was built and last rebuilt, and the wealthiest quartile is saddled with the oldest stations.^[1] Wealth bought density downtown a century ago, and that same age now means stairs. The two facts are not in tension once you see that they are measuring different things, one the quantity of rail and the other the vintage of it, but a reader who saw only the density chart would guess the accessibility gradient exactly backward.

None of this softens the density gradient from the previous section. It sits beside it. A rider in a wealthy North Side neighborhood is surrounded by stations, several of which still require stairs. A rider in a low-income South Side area that has a station at all is likely to find it step-free, but is far less likely to have a station nearby in the first place.^[1] Two different barriers, pulling in two different directions, both real, and a full account of who the system serves has to hold both at once. For a rider who does not use a wheelchair, the density gradient is the one that governs the daily trip, and that gradient tilts toward the wealthy. For a rider who does, the calculus is harder still in the lower-income areas, where the nearest accessible station may be both rare and far, even though the few stations that exist are likely to be step-free. The system does not hand any one group an unambiguous advantage on every measure, and saying so is part of describing it honestly rather than fitting it to a slogan.

Where the boardings pile up

A station can stand on the map and sit nearly empty, so the turnstiles are the next thing to read. An entry count says where the system actually gets used, and in 2023 the boardings piled up where the stations were densest and the incomes highest. Total CTA rail entries for the full year came to 99,842,880, very nearly a hundred million trips through the turnstiles.^[1]

Spread those entries across the income quartiles and the climb is steep. Stations in the lowest-income quartile averaged 328,297 entries each over the year.^[1] The second quartile averaged 506,980, the third 714,613, and the highest-income quartile 990,599 entries per station.^[1] A station in the wealthiest quartile took in roughly three times the boardings of a station in the poorest, on average, across the year. The ridership gradient is even smoother than the density one, rising cleanly from quartile to quartile with none of the middle dip that marked the density numbers.^[1]

The concentration at the top is stark. The highest-income quartile alone drew 64.4 million of the roughly 99.8 million full-year 2023 entries.^[1] Close to two of every three rail boardings in Chicago happened at a station inside the wealthiest quarter of community areas. That quartile holds 65 of the 124 city stations, so it is carrying a heavy share of stations and an even heavier share of trips.^[1] The boardings concentrate harder than the stations do, which is its own small finding, since it means the busy stations are more numerous in high-income areas and busier per station besides. The bottom of the ladder shows the same fact from the other end. The lowest-income quartile's stations together logged about 6.6 million entries across the whole year, and the second quartile about 11.2 million, so the two poorest quartiles combined drew well under a third of what the richest quartile pulled in alone.^[1] The turnstiles tell the density story over again, in trips rather than track.

What this is, and what it is not, matters. These are entries, boardings only, not alightings, so they count the trips that begin at a station rather than the full traffic through it.^[1] Each station's entries are attributed to whichever community area the station physically sits inside, the same point-in-polygon placement used everywhere else here.^[1] This is a descriptive ridership-by-income profile, a picture of where boardings happen, not a demand model and not a statement about who rides.^[1] A station in a wealthy downtown-adjacent area is also a station that workers from across the whole city pour into on their way to jobs, so heavy boardings in a high-income quartile do not mean only high-income people are boarding. The number describes the location of the turnstile, not the income of the rider, and the distinction is the difference between an honest reading and an overreach.

The shape of the ridership gradient differs from the density gradient in a way that carries meaning. Density wobbled, with the third quartile dipping below the first before the fourth pulled away. Ridership does not wobble. It rises monotonically, 328,297 to 506,980 to 714,613 to 990,599 entries per station, each quartile busier than the last.^[1] So even where the middle-income quartiles are not especially dense in stations, the stations they do have are busier than the poorest quartile's, and the richest quartile's stations are busier still. Two things move together up the income ladder, the number of stations per square mile and the boardings per station, and they compound. The wealthiest quartile has both the most stations and the busiest ones, which is why its share of total entries, close to two in three, runs even further ahead of its share of stations, just over half.^[1]

Read against the published work, the pattern fits without being stretched. The heaviest boardings cluster where high-frequency rail concentrates downtown, which is also where the jobs concentrate, and the jobs-access literature has long held that transit's value turns on reaching employment, with low-income car-less workers facing real penalties when the network does not carry them to it.^[4][11] Blumenberg and Ong's Los Angeles finding and Grengs's Detroit finding both point the same way, that the spatial mismatch between where workers live and where jobs sit falls hardest on transit-dependent low-income households.^[4][11] A ridership pattern in which boardings concentrate downtown, where the office jobs are, is exactly what one would expect of a network whose busiest nodes are its job-rich center, and it is consistent with a city where the residents best positioned to reach those jobs by rail are the ones living along the dense central and northern lines. Our entries describe a usage pattern that sits comfortably alongside that literature. They do not measure opportunity, and they do not prove who reached a job. They show where the system fills up, and where it fills up is downtown, on the lines that the higher-income quartile happens to own the densest share of.

A further caution applies to the ridership figures specifically, and it is the kind of thing a careful reader should want stated. Attributing every entry to the community area the station sits inside is a clean rule, but it makes downtown stations look like they belong to the downtown neighborhoods, when in truth they belong to the whole city. A worker who boards at a South or West Side station in the morning and boards again at a Loop station in the evening is counted once in a lower-income area and once in the Loop, and the Loop tally swells with riders who live nowhere near it. So the concentration of boardings in the high-income quartile partly reflects the simple fact that downtown is where everyone's trips converge, not that downtown residents do most of the riding. The honest reading of the ridership gradient is geographic, namely that the turnstiles spin hardest in the dense, job-rich, higher-income center, and it is not a claim about the household incomes of the people pushing through them.

How the numbers were built

A descriptive paper earns its trust by showing its work, so this section walks through how each figure above was produced, in enough detail that a reader with the same files could rebuild the tables. Nothing here is sophisticated. That is the point. The whole analysis is a set of joins, counts, and ratios over three public files, and its credibility rests on being checkable rather than clever.

The station roster begins with the CTA list of "L" stops, which records one row per platform and direction.^[1] Each row carries a stop identifier, a station identifier, the station name, a wheelchair-accessibility flag, a set of per-line boolean columns marking which colors serve the platform, and a latitude and longitude. Counting rows gives 302 platform-and-direction records.^[1] Collapsing those rows to unique stations by their station identifier gives 144 stations, the figure used throughout for the system as a whole.^[1] The accessibility flag is read at both levels, which is why two ADA shares appear, the 74.2 percent computed over all 302 records and the unique-station shares computed after the collapse.^[1] Keeping both is a deliberate choice to match the source at the row level while also reporting the station-level figure a rider would recognize.

Assigning stations to neighborhoods is done by point-in-polygon. Each station's single latitude and longitude is tested against the City's community-area boundary polygons using ray casting, a standard geometric test for whether a point lies inside a closed shape, implemented in plain Python without a geographic-information-system library.^[1] A station falls inside exactly one community area or, if its coordinates lie outside all 77 city polygons, inside none, in which case it is flagged as suburban. Twenty of the 144 stations fall outside the city this way, and they match the known suburban stops in Evanston, Skokie, Oak Park, Forest Park, Cicero, and Rosemont, which is the check that the geometry is working.^[1] The remaining 124 stations sit inside the city and carry the neighborhood analysis. This method counts a station as belonging to the single area its dot lands in, and it cannot split a boundary-straddling station between two areas, a limitation the next section returns to.

The income and hardship figures come from the City of Chicago's "Selected socioeconomic indicators" file, joined to community areas by area number.^[1] Each area receives its per-capita income and its hardship index from that file. Seventy-six of the 77 areas carry income, with Chicago Lawn the lone gap, and since Chicago Lawn has no station the gap does not disturb any served-area count.^[1] To build the income quartiles, the areas are sorted by per-capita income and cut into four groups of roughly nineteen, then the city stations and the land area falling in each group are summed and divided to give stations per square mile.^[1] The same groups carry the ADA shares and the ridership totals, so every quartile table rests on one income sort applied consistently.

Ridership is the CTA station-entry file, aggregated to full-year 2023 totals per station and then summed within each income quartile by station location.^[1] These are entries, boardings recorded at the turnstile, not alightings, so they count trips that begin at a station. The full-system 2023 total comes to 99,842,880 entries, and 143 of the 144 stations carry a 2023 total in the file.^[1] The two Pearson correlations, income against density and hardship against density, are computed across all 77 community areas, treating each area as one observation, with no weighting and no controls.^[1] A correlation here is a single summary of how two columns move together across the neighborhoods, nothing more. The files, the join keys, and these steps are the entire method, and they are posted in reproducible form so the path from raw data to every number above can be retraced.^[1]

What this map can and cannot tell you

Every number above rests on two real public files and a deliberately plain method, and the responsible thing is to say exactly where each one bends. The point of this section is not ritual modesty. It is to draw the line between what the data supports and what a reader might be tempted to read into it.

Income carries the whole gradient, so start there. The income figure here is per-capita income from the City of Chicago "Selected socioeconomic indicators" file, a community-area dataset built from the 2008 to 2012 American Community Survey.^[1] That is a per-person average across a whole community area, not a household median, so it answers a coarser question than a tract-level household figure would. It is also roughly a decade older than the 2023 ridership and the current station roster it is paired with.^[1] A neighborhood's income has moved since 2012 in places, sometimes sharply, and this measure cannot see that movement. The gradient it draws is real in its own terms, but it is an older and broader income picture than the ideal, and a reader should hold the 2023 ridership and the 2008 to 2012 income in mind as two clocks that do not quite agree.

A finer measure exists, and naming it plainly is part of being straight about the method. A tract-level pairing of the ACS transit-commute share against median household income is the finer-grained join the literature usually favors, and it is the join this analysis would have preferred. We did not run it. The Census Bureau's data interface now requires a free access key for the programmatic pull this would need, and an unkeyed request redirects to a missing-key page rather than returning data, so the tract-level join could not be executed on real numbers within this environment.^[1] The community-area income file is what we used instead, because it was downloadable here without a key, and the project's own archive already ships the identical ACS income structure for another county, so the Cook County version is obtainable through that same documented path by anyone who supplies the key.^[1] The honest position is that the finer measure exists and was not used here, not that the question is closed. A reader who wants the tract-level version can build it; we simply did not, and we will not pretend the coarser file is the better one.

Geography is approximate, and in a specific way. Each station is placed by a single latitude and longitude point, and it is assigned to whatever community-area polygon that point falls inside, tested by plain ray casting.^[1] That counts the stations physically standing inside an area.^[1] It does not count the population that lives within a half-mile walk of a station, which is the measure an accessibility study would want, because a station near a boundary serves the people on both sides of the line regardless of which polygon its dot lands in. No walk-shed was drawn and no door-to-door travel time was computed, which is consistent with a descriptive thesis and is also a real ceiling on it.^[1] A station counted in one community area does real work for the neighborhood next door, and our method cannot credit that spillover. The half-mile-walk population share that a fuller analysis would compute was left undone here because the tract or block population layer it requires could not be pulled without the same Census key the income join needed.^[1]

The choice of the community area as the unit of analysis shapes the numbers in ways worth surfacing. Chicago's 77 community areas vary widely in land area and population, so treating each as one observation in a correlation gives a small dense neighborhood the same weight as a large populous one, and computing density across a quartile lumps together areas of very different size. A measure built on census tracts, which are smaller and more uniform, would draw a finer-grained gradient and might shift the exact coefficient, since the way a city is carved into units can change the apparent strength of a relationship measured across those units. We use community areas because the City's income and hardship figures are published at that level and because they are the unit a Chicago resident actually thinks in, but the reader should know that the +0.398 is a community-area correlation, not a tract one, and that a different geography could move it.^[1] The boundary effect compounds this. A neighborhood scored as rail-less may sit directly across a street from a station in the next area, and our method, which credits a station only to the single polygon its coordinate lands in, cannot see that a resident on the near edge of a blank neighborhood may live a five-minute walk from a train.^[1] The blank on the map is a blank in the count, not always a blank in a resident's real reach.

The exclusions and edge cases deserve restating so none of them hides. Twenty suburban stations are set aside from every community-area table, leaving 124 city stations.^[1] One community area, Chicago Lawn, is missing from the income file, so 76 of 77 areas carry income, and since Chicago Lawn has no station the gap does not move the served-area counts.^[1] The two ADA counts, 224 accessible against 78 not at the stop-record level and 106 of 144 unique stations once deduplicated, are kept separate on purpose because they answer different questions.^[1] The 2023 ridership is entries only, boardings without alightings, attributed by station location.^[1] And every correlation reported here, the +0.398 on income and the -0.183 on hardship, is a descriptive Pearson coefficient across the community areas with no controls and no causal claim.^[1]

What survives all of that is a description, not a verdict. Richer accessibility methods exist, and the literature has built them, route-level measures on public data and full neighborhood accessibility surfaces that this analysis does not attempt.^[2][3] Run on this city, those methods would sharpen the picture and might complicate it in places, since a station-poor neighborhood with a nearby high-frequency line could be better off than its station count suggests, and a station-rich one served only by an infrequent line could be worse. The plain map drawn here still lands where that work predicts it would, which is the most this method can fairly claim and enough to be worth claiming.^[2][3]

Reading the city by its rail lines

The threads make one shape. The train reaches farthest and packs in densest where incomes are highest, more than three times the station density in the top income quartile as in the bottom.^[1] The boardings pile up there too, close to two-thirds of a year's rail entries at stations in the wealthiest quarter of community areas.^[1] Step-free access runs the other way, highest where stations are scarce and lowest downtown where the old steel still has stairs.^[1] And the places with no station at all are heavily lower-income South and West Side community areas, Riverdale and New City and Humboldt Park and Auburn Gresham and West Pullman among them.^[1] Different layers, one tilt.

Set that next to the published record one more time. This analysis did not prove a cause, did not model demand, did not compute a travel time. What it did was document, with real and downloadable public files, the geographic pattern that a Chicago-specific and national equity literature already describes.^[2][5][6] Ermagun and Tilahun found transit accessibility unevenly distributed across Chicago's neighborhoods and groups.^[2] Liu, Kwan, and Kan tied higher minority population share to lower transit-based access to destinations like healthcare.^[5] Rozhkov and colleagues located the city's transit deserts on the South and West Sides, in low-income, minority community areas.^[6] The plain count of stations against neighborhood income lands in the same place that body of work does, which is the modest thing a coarse method can offer, a second witness in agreement with the careful ones.

What the map asks of policy, it asks without needing a single program named. Decisions about where to extend a line, which platforms to make step-free, and where to add frequency all intersect with neighborhood income in ways a resident can already read from a platform. The 35 community areas with no station are a record of choices, made and unmade across more than a century of building track, about where the city put its rail.^[1] The downtown stations that still require stairs are another such record. A rider does not need a model to see the gradient. They feel it in how far they walk to the train and whether they can climb to it once they arrive.

A description does not write a policy, and we will not pretend it does. What it can do is set the table that any honest policy conversation has to start from. A debate about where to extend service ought to begin from a clear-eyed account of where service is missing and who lives there, and the map of absence supplies it, the thirty-five neighborhoods with no rail and the lower-income, higher-hardship places that crowd the bottom of that list.^[1] But absence is only half the table. The density gradient adds that the question is about how thickly rail reaches, not just whether it reaches at all, since a single line touching a neighborhood is a far cry from the overlapping mesh that covers the center.^[1] And the accessibility map, running backward from density, adds a third fact that cuts against the easy reading of the other two, that closing a gap can mean rebuilding old stations as much as laying new track, since the wealthiest, densest quartile is also the one with the most stairs.^[1] None of that is a recommendation. Each is a fact a recommendation would have to reckon with, and the value of a plain description is that it puts the facts where a reader, or a planner, or a resident at a hearing, can pick them up and argue from them. The equity literature supplies the principle, that the fair distribution of access is something a transit system owes its residents.^[7] Our map supplies the local measurements that principle would be applied to.

There is also a discipline in refusing the causal leap, and it is worth naming as a finding in its own right. It would be easy, and wrong, to read the +0.398 correlation as proof that the city steers rail toward wealth, or that rail draws wealth toward itself. The correlation is consistent with both stories and with several others, including the plain historical one, that a private system built toward a profitable downtown a century ago left a hub-and-thin-spoke geometry the public agency inherited and never fully rebalanced. We cannot adjudicate among those accounts with a cross-tabulation, and we do not try. What the description establishes is the pattern that any of those explanations would have to fit, namely that today, on the ground, rail is thickest where incomes are highest and absent across a band of lower-income South and West Side neighborhoods. The why is a separate study. The what is the thing a person can stand on a platform and verify.

The paper has been one thing throughout, a synthesis of published research set beside a descriptive read of real public data, reproducible from the CTA stop file, the ridership totals, and the community-area income file named here.^[1] A useful next step is the work the caveats flag as not yet run, the tract-level income join, the half-mile walk-sheds, and the door-to-door travel times that would turn a count of stations into a measure of reach.^[1] Until then, the network itself is the clearest statement available. Trace it color by color, station by station, and it draws a line across the city between where rail got built and where it did not, a line a rider can stand inside.

^[1]: Author analysis of CTA System Information - List of 'L' Stops. Chicago Transit Authority via City of Chicago Data Portal (Socrata). Reproducible files at rooted-forward.org/research/data/chicago-transit-access-income.

^[2]: Ermagun, A., & Tilahun, N. (2020). Equity of transit accessibility across Chicago. Transportation Research Part D: Transport and Environment, 86, 102461.

^[3]: Karner, A. (2018). Assessing public transit service equity using route-level accessibility measures and public data. Journal of Transport Geography, 67, 24-32.

^[4]: Grengs, J. (2010). Job accessibility and the modal mismatch in Detroit. Journal of Transport Geography, 18(1), 42-54.

^[5]: Liu, D., Kwan, M.-P., & Kan, Z. (2021). Analyzing disparities in transit-based healthcare accessibility in the Chicago Metropolitan Area. The Canadian Geographer / Le Géographe canadien, 65(3), 343-361.

^[6]: Rozhkov, A., Zandiatashbar, A., & Seyrfar, A. (2025). Tackling Transportation Inequality through Unveiling Chicago's Transit Deserts and Vulnerable Communities. Journal of Urban Planning and Development, 151(3).

^[7]: Pereira, R. H. M., Schwanen, T., & Banister, D. (2017). Distributive justice and equity in transportation. Transport Reviews, 37(2), 170-191.

^[8]: Bocarejo S., J. P., & Oviedo H., D. R. (2012). Transport accessibility and social inequities, a tool for identification of mobility needs and evaluation of transport investments. Journal of Transport Geography, 24, 142-154.

^[9]: Foth, N., Manaugh, K., & El-Geneidy, A. M. (2013). Towards equitable transit, examining transit accessibility and social need in Toronto, Canada, 1996-2006. Journal of Transport Geography, 29, 1-10.

^[10]: Sanchez, T. W. (1999). The connection between public transit and employment, a transit access analysis of TANF recipients in Portland, Oregon. Journal of Public Transportation, 2(4).

^[11]: Blumenberg, E., & Ong, P. (2001). Cars, Buses, and Jobs, Welfare Participants and Employment Access in Los Angeles. Transportation Research Record, 1756(1), 22-31.