Abstract
A Divvy dock is the smallest unit of access in Chicago's bikeshare system. A bike you cannot reach is a bike you cannot ride, so the question of who the system serves starts at the curb, with whether a dock stands within a short walk and whether it holds enough bikes to be worth that walk. This paper joins the published bikeshare-equity literature to our own descriptive analysis of real, downloadable public data, namely the City of Chicago's roster of Divvy stations, a four-month 2024 sample of Divvy trip records, and ACS five-year 2023 demographics summarized by community area [1]. We make no causal claim and no rider-level demographic claim, because Divvy reports only whether a rider holds an annual membership or rides as a casual user, and never the rider's race, income, or age [1]. Read against the literature, the data shows three things. Dock supply per resident falls by more than half across the city, from 111.38 docks per 10,000 residents in the whitest quartile of community areas to 44.00 in the most-residents-of-color quartile, and from 115.85 in the highest-income quartile to 43.66 in the lowest [1]. Trips are concentrated to an extreme that outruns even the dock gap, with 92.9 percent of 1,575,899 station-matched trips beginning in higher-income, whiter areas, and the busiest tenth of stations accounting for 62.7 percent of all matched trips [1]. And the annual-member share is highest, not lowest, in lower-income, more-residents-of-color areas, at 70.2 percent, because casual riding in Chicago turns out to be largely a downtown-tourist behavior that empties the casual side of the ledger out in the neighborhoods [1]. The supply gradient and the trip concentration are the sturdy findings, holding across both demographic axes and reproducing from the public files. The member result is real but noisier, and we report it with its own contradiction visible. We state the limits plainly, hold to a strictly geographic and descriptive reading, and let the two sturdy findings carry the weight.
Start at the dock
Picture the thing itself before any number. A Divvy dock is a low steel rail bolted to a sidewalk or a parkway or the lip of a plaza, a row of locking points, a small payment screen, sometimes a bike or two left in it and sometimes none. For the block around it, that rail is the whole bikeshare system. The promises Divvy makes, the quick hop to the Red Line, the lakefront ride on a Saturday, the grocery run that would otherwise be a bus and a transfer, all of them route through one prior fact, which is whether a dock sits close enough to walk to with a few minutes to spare. Ask any other question first and you have skipped a step the rider cannot skip.
So the plain question comes before the interesting ones. Not who rides, not how often, not on which kind of bike, but where the docks are. That question has a few parts, and the data answers some of them cleanly and refuses others outright. One part is supply, the physical placement of docks across the city's neighborhoods. A second is use, the record of where trips actually begin. A third is the single behavioral split the system itself discloses, whether a given trip belongs to an annual member or a casual rider. Each part is a description of a real pattern in real public files. None is a verdict on why the pattern exists.
A little background helps the question land. Divvy opened in 2013 as the city's public bikeshare program, owned by the Chicago Department of Transportation and operated under contract, most recently by Lyft, which also bought the equity-membership obligation along with the system. From the start the network was densest downtown and along the North Side lakefront, the parts of the city with the most office workers, the most tourists, and the most existing bike infrastructure, and it expanded outward from there over the following decade onto the West and South sides. That history is not the subject of this paper, and the analysis does not try to reconstruct it. But it is the reason a snapshot taken now finds the heaviest dock coverage where the system first grew, and it is why the city eventually attached an income-eligible membership to a network that had been built, in its early years, around the lakefront core [9]. The map this paper reads is the cumulative result of more than a decade of those siting decisions, frozen at one moment.
What this document is matters as much as what it finds, so the boundaries are worth drawing exactly. It joins two kinds of work. The first is a reading of the published research on bikeshare equity, a body of studies built over the past decade in Chicago and in other American cities by transportation scholars who asked, in various ways, whether shared bikes reach everyone. The second is our own count of real, downloadable public data about Divvy and about Chicago neighborhoods. The list of things it is not runs longer. We ran no experiment and changed nothing about the system, so nothing here tells you what a denser network would do. We surveyed no one and interviewed no one, so the residents around the docks never get a voice in their own words on these pages. We did not scrape anything private, and we built no model that forecasts a single trip. Every Divvy figure in these pages is a count, a share, or a plain correlation, taken from files anyone can download and recompute [1].
Three sources carry the analysis, and the methods section introduces each properly, but they deserve a first naming so the reader knows the ground. The first is the City of Chicago's public roster of Divvy stations, which gives each station a location and a dock count. The second is a four-month sample of Divvy's own public trip records from 2024, one month from each season. The third is the American Community Survey five-year estimates for 2023, summarized by Chicago community area, which supplies the neighborhood demographics. All three are cited together to the analysis record [1], which points to the reproducible files.
Two honesty points have to come before anything else, because the entire paper hangs on them.
The first is about demographics. Divvy's trip data says nothing about the people who ride. No race field. No income field. No age field. The only personal fact in a trip record is whether the rider was an annual member or a casual user, alongside the type of bike taken [1]. That caveat governs every sentence that follows. When the paper reports that docks are thinner in more-residents-of-color neighborhoods, it is describing the population that lives around those docks, as the Census counts it, not the people who unlock the bikes. Every demographic word in this document refers to a neighborhood's residents. None refers to a rider. The point is easy to misread, so the paper repeats it on purpose.
The second is about cause, and it is the harder discipline. A pattern is not an explanation. Showing that docks thin out as a neighborhood gets poorer is a statement about the map as it stands in the sampled window. It is not a claim that anyone meant the result, not a forecast of what adding docks would do, and not a theory of why riders in any given neighborhood ride more or less. Bikeshare systems grow through a tangle of sponsorship deals, capital budgets, ridership projections, real-estate constraints, and political pressure, and pulling those threads apart is causal work this analysis did not attempt. The numbers say where things sit. They do not say how they got there.
With those boundaries fixed, the rest of the paper keeps a simple promise. It shows three patterns, in the order the analysis builds them. Supply leads, because it carries the most weight. Counting docks against the residents who live near them, the per-resident supply drops sharply as neighborhoods get poorer and less white, on both the race axis and the income axis, and the gradient is the load-bearing result [1]. Concentration follows, because it is the most vivid. Of the 1,575,899 trips matched to a community area, the overwhelming majority start in a small, central, well-served core, out of all proportion even to where the docks are [1]. The member split comes last, because it breaks the obvious story. One would guess that casual, pay-as-you-go riding would dominate the lower-income neighborhoods while committed annual membership clustered among the affluent. The data shows the reverse, and the reason has more to do with tourists than with equity [1].
The subject baits two bad tones, and the paper tries to dodge both. This is not a takedown. Divvy moves a great many people, it is a genuine public good, and the city has attached an explicit equity program to it, which the paper describes fairly [9]. Neither is it a brochure. The supply gradient is steep and real, and softening it would be its own dishonesty. The register the paper aims for is the one a careful Chicagoan would use at a ward meeting, concrete, fair, willing to name a gap without inflating it, and willing to stop where the data stops.
The data stops in specific places, and the paper marks each as it arrives. For now the shape is enough. Seventy-seven community areas cover the city. A current roster of stations sits across them. A four-month slice of 2024 trips lays on top, each trip joined to a neighborhood through its start station. From that join the paper reads supply, use, and the member split. It starts where the riders do, at the dock.
What other cities already learned about bikeshare and access
The literature comes first, before our own numbers, because Chicago is not the first city where researchers asked whether bikeshare reaches everyone, and the published answers are consistent enough that the Chicago snapshot is best read as a local test of an established pattern rather than a fresh discovery. The aim here is to lay that pattern out fairly, crediting each piece of evidence to the study that produced it, so the reader can judge for themselves how well our snapshot fits. Bikeshare equity has become its own small field over the past decade, and the studies in it use different cities, different data, and different methods. The agreement across them, despite that variety, is what makes the pattern worth trusting. No single paper proves it. The repetition does.
The broadest siting evidence comes from a four-city study by Ursaki and Aultman-Hall, who mapped the service areas of bikeshare systems in Boston, Chicago, New York, and Seattle and compared the people living inside those service areas with the people living outside them [8]. The share of Black residents inside the service areas was significantly lower than the share outside [8]. The strength of that design is its directness. It does not infer siting bias from ridership or from survey self-reports. It draws the actual footprint of where the stations are, counts who lives within it and who lives just beyond it, and finds Black residents disproportionately on the outside of the line in all four cities. That is a finding about geography, about where the stations go before anyone rides, which is the same axis our supply gradient measures. It is the natural backdrop for everything reported later, and it is worth holding onto that the multi-city result is not about a single city's politics. The same tilt showed up in four systems run by four different operators under four different local governments.
Chen and colleagues, working with disaggregated data from southern Tampa, supply the conceptual vocabulary [6]. They show bikeshare accessibility distributed unevenly across both physical geography and sociodemographic groups, and they separate two ideas worth keeping. Horizontal equity asks whether similar people get similar access. Vertical equity asks whether the people with the greatest need get adequate access [6]. The distinction earns its keep later in this paper, because a neighborhood can sit technically inside a service area, with a station somewhere within it, and still be poorly served when the docks are too few and too far apart. A city can pass a horizontal-equity test, with stations scattered into most neighborhoods, and fail a vertical-equity test at the same time, if the neighborhoods that most depend on cheap mobility are the ones left with the sparsest coverage. Tampa names the gap between a station merely existing and a station being usable. That gap, between presence and density, turns out to be exactly where the Chicago story lives, and the analysis returns to it by name once the dock counts are on the table.
Seattle adds the rider's side through survey work by Hirsch and colleagues, who studied a free-floating bikeshare system, a model with no fixed docks at all [5]. That detail sharpens the result. Even when the bikes can be left anywhere, removing station distance as a barrier, awareness of the system ran high across groups while actual use did not follow. Use skewed toward riders who were younger, male, White, living centrally, and already cycling before bikeshare arrived [5]. Non-White respondents reported more barriers tied to cost and to spatial access [5]. The lesson is that knowing the system exists is not the binding constraint. People know. The friction shows up in what a ride costs and in where the bikes actually are, and it lands unevenly even in a dockless system where, in principle, the bikes float free. Divvy gives us no rider demographics, so a finding about who rides is one the Chicago data cannot reproduce, but Seattle tells us what kind of mechanism might sit behind a usage skew, and it warns against reading low use as low interest. The point about awareness matters for Chicago specifically. It means a thin map cannot be excused as a marketing problem, as a neighborhood that just needs to hear about Divvy. People hear. What they encounter is the dock that is too far and the fare that is too high.
McNeil and colleagues address lived barriers most directly, through a survey across Philadelphia, Chicago, and Brooklyn [4]. Among residents of traditionally underserved neighborhoods, concerns about traffic safety, the cost of a membership, and the distance to the nearest station fell hardest on lower-income residents of color [4]. The traffic-safety finding is easy to skip past and should not be. It is about more than whether a dock is near. It is about whether the streets between home and the dock, and between the dock and the destination, feel survivable on a bike, and lower-income neighborhoods of color in many American cities have less protected bike infrastructure and more dangerous arterials to begin with. The same respondents supported discounted-membership remedies, which cuts against any lazy reading of low participation as disinterest [4]. People named concrete, addressable obstacles, cost and distance and safety, and endorsed concrete fixes. That study is part of why this paper declines, later, to read thin ridership in lower-income Chicago neighborhoods as a sign that residents do not want bikeshare. They were asked, in three cities including this one, and they said they wanted in.
Two Chicago-specific studies carry the most weight here, because they examine the very system this paper analyzes. Qian and Jaller studied Divvy and found low-income households, households of color, and transit-dependent households underrepresented among users, and they found station use more unequal in disadvantaged community areas than in advantaged ones, with a higher concentration of activity on fewer stations [2]. Their unevenness result is worth pausing on, because it is more specific than a simple count of who rides. They measure how concentrated the riding is within a neighborhood's own stations, and they find that in disadvantaged community areas the activity bunches onto a smaller share of the docks. A neighborhood can have several stations and still see almost all its trips run through one of them, which is a different and harder problem than a low total. It means the few docks that do get used are working, and the rest are close to idle. That is a Chicago result about both who uses the system and how thinly the use spreads across the stations a struggling neighborhood does have, and it lines up closely with the trip-concentration pattern reported below from our own sample [2].
Bergman, Allenspach, and Bergman take a qualitative approach in a study titled, pointedly, a tale of two Divvys, arguing that the system's expansion model can systematically under-serve poorer neighborhoods and less-privileged residents even while the network as a whole grows and succeeds [3]. The value of that second study is its focus on expansion as a mechanism, the way a system that is genuinely popular and genuinely growing can still hold a durable gap in place. A network expanded by demand will add docks fastest where ridership is already high, which is the affluent core, and that logic compounds. The places with the most bikes generate the most trips, the trip data then justifies the most new docks, and the gap widens through ordinary operational reasoning rather than any single decision to neglect a neighborhood. That is precisely the situation a descriptive snapshot like ours can document but cannot, on its own, explain, because the snapshot sees the result of the expansion process and not the process [3]. The two Chicago studies together frame what our data can and cannot add. Qian and Jaller already established the underrepresentation with rider-aware methods we cannot match, and Bergman and colleagues already named the expansion mechanism. What a fresh public-data snapshot contributes is a current, reproducible measurement of how steep the geographic gap is right now, on the two axes the public files can actually support.
Diagnosis is not where the field ends. It also asks what happens when a city tries to close the gap, and the most useful evaluation comes from Greater Boston. Soto and colleagues measured access and use before and after Boston paired an income-eligible membership with a physical expansion of the system, the same two-lever combination most cities reach for, a cheaper way in plus more stations [7]. Access and use rose across the city, the encouraging half of the result. The sobering half is that the gains were smallest in the communities with the greatest need [7]. A program that helped everywhere helped least where help was needed most, so the gap narrowed without closing [7]. The result is not a verdict that the intervention failed. Access did rise. It is a calibration of how much a discount-plus-expansion approach can be expected to do, which is to compress the gap rather than eliminate it, at least over the window Boston studied. For any policy conversation about Chicago this is the single most important piece of context, because it sets a realistic expectation against which Chicago's own program can be read. Targeted programs are worth doing and they move the numbers. The best before-and-after evidence available says they narrow the gap rather than erase it, which means a city that wants the gap gone has to do more than offer a cheaper membership and add a few stations.
Chicago has its own version of that targeted program, and it sets up a thread this paper returns to at the end. Divvy for Everyone, run by Divvy and Lyft with the Chicago Department of Transportation, offers an annual membership for five dollars to residents who qualify through programs like SNAP, WIC, LIHEAP, or public housing [9]. The program is an official acknowledgment that price and access are not evenly distributed across the city, and that some residents need a different door into the system [9]. Reporting by Greenfield in Streetsblog Chicago puts numbers on the contrast the program means to address. About 84 percent of regular Divvy members are white, while 72 percent of Divvy for Everyone members are people of color, and program participation runs highest on the South and West sides [10]. Those figures belong to that reporting and to the program, not to our analysis, and the paper keeps them separate throughout [10]. Taken with the program's design, they sketch a system that knows it has an access problem and has built at least one tool to address it, a standard membership base that is overwhelmingly white set beside an equity membership whose riders are mostly people of color and concentrated in the very parts of the city where, as the supply data will show, the docks are thinnest.
Stand those studies together and a single shape emerges across methods that share almost nothing. Fixed-station bikeshare tends to be placed where fewer Black residents live [8]. Access is distributed unevenly across geography and across sociodemographic groups [6]. Users skew younger, whiter, more male, and more central, even where awareness is broad [5]. The hardest barriers, cost and distance and traffic safety, fall on lower-income residents of color, who want in regardless [4]. In Chicago specifically, disadvantaged communities are underrepresented and their stations ridden more unequally [2], and the expansion model itself can entrench that result [3]. The best-studied remedy narrows the gap without closing it [7]. Against that backdrop the task here is narrow and clear. Bring real Chicago data to the two questions the data can actually answer, where the docks sit and where the trips begin, and read the answers as a local test of a pattern the field has already documented from several directions.
How we read the map, and what we refused to pretend
This is the methods section, though it is better understood as the set of rules we held ourselves to, because the hardest part of an honest analysis is naming what it will not pretend to know. The mechanics are simple. The discipline behind them is the work.
The join runs in one arc. Take the City of Chicago's public roster of Divvy stations, each with a location and a dock count. Place each station inside one of Chicago's 77 community areas using its coordinates. Attach to each community area its demographics from the ACS five-year estimates for 2023, summarized at the community-area level. Then lay the Divvy trip sample on top, assigning each trip to a community area by its start station [1]. From that one stack, station to neighborhood to demographics to trips, every number in this paper is computed.
Each neighborhood gets described along two axes, the same two throughout. The first is the percent of residents who are people of color. The second is the share of households earning under 50,000 dollars a year, which stands in for income. That share is used rather than a median household income for a specific and honest reason. The City's ACS extract reports income in brackets, not as a single median figure, so a clean per-area median was not available in the table at hand, and none was invented [1]. The share of households below 50,000 dollars is a real number in the source, and it orders the neighborhoods sensibly from higher income to lower. Where the paper says lowest-income quartile, it means the quartile of community areas with the largest share of households under 50,000 dollars [1]. The median split that separates the area types falls at 85.4 percent residents of color and 30.7 percent of households under 50,000 dollars, the citywide midpoints across the 77 areas [1].
The scale of the data is worth stating precisely, because precision here is part of the honesty. The station roster holds 1,153 stations and 19,318 total docks [1]. The spatial assignment that places each station in a community area is a point-in-polygon test, asking which of the 77 community-area boundaries contains a station's latitude and longitude. Of the 1,153 stations, 1,139 fall inside one of the 77 community areas and feed the neighborhood analysis, while 14 sit in suburban locations beyond the city's boundaries and are left unmatched, since Divvy's footprint now spills into a few neighboring municipalities that have no Chicago community area to join to [1]. All 1,153 stations carry valid coordinates, so nothing drops out for missing location data, which is a real and useful property of this dataset because a station with a bad coordinate would be silently misplaced rather than flagged [1]. Of the roster, 1,147 stations show a status of in service and six show another status, a small enough share that the distinction does not move any figure here [1]. On the demographic side, all 77 community areas have ACS 2023 data, and the citywide population summed across them is 2,647,621 [1]. Those are the denominators behind every per-resident figure that follows. The community area is the unit of geography throughout, which is the right grain for this question. Chicago's 77 community areas are stable, well-known boundaries that residents and planners actually use, large enough to carry reliable ACS estimates and small enough to separate, say, the Near North Side from Englewood, and they are the geography the City's own published demographic table is keyed to.
The trip layer is a sample, and the kind of sample matters. Rather than a full year, it is four real monthly files from 2024, one drawn from each season, January, April, July, and October [1]. One month per season was a deliberate choice, made to avoid the distortion of leaning on summer alone, when casual and tourist riding peaks and would overstate the casual share across the whole analysis. Bikeshare ridership in a four-season city is sharply seasonal, with far more trips in July than in January, so a sample built only from warm months would not just be larger. It would be biased toward exactly the recreational and tourist riding that this paper has to keep separate from everyday neighborhood use. Spreading the sample across a winter month, two shoulder-season months, and one peak summer month holds that bias down, at the cost of not capturing any single month in full. Those four months together hold 1,925,141 raw trips [1]. That raw count is the starting point, not the number analyzed, because two exclusions come first.
The exclusions get stated before anything rests on them. Of the 1,925,141 raw trips, 338,070 are dockless trips with no start-station identifier, which is 17.6 percent of the raw rows [1]. Divvy's e-bikes can be locked to a public bike rack rather than returned to a dock, and a trip that begins from such a free-floating start carries no station to anchor it. A dockless trip therefore has no fixed origin dock, so it cannot be placed at a station or, through that station, in a neighborhood, and every figure in this paper is built on station origins. All 338,070 dockless trips are dropped from the station-based analysis [1]. This is the largest single exclusion and the one with the most uncertain consequences, because nearly one in five raw trips leaves through it, and there is no way from these files to know whether those rides share the geography of the docked ones or reach somewhere different. A further 11,172 trips begin at stations outside the 77 community areas, the suburban fringe again, and those are dropped as well [1]. What remains is 1,575,899 station-matched trips, which is 99.3 percent of the trips that carried a usable start station, and that is the figure behind every trip number reported here [1]. The 99.3 percent is reassuring about one thing only, that almost no trip with a real station origin gets lost in the join. It says nothing about the 338,070 dockless rides set aside before that step, and the limitations section returns to them, because a block of riding that size sitting outside our view could carry a geography we simply cannot see.
Two choices trimmed the analysis, and they are named rather than buried. The first concerns the commute layer. The original plan brought in the Census journey-to-work data, the table that records how people get to work, including by bicycle, so that latent cycling demand could be lined up against dock supply. That layer was dropped, because the Census API now requires a key that was unavailable at the time of the analysis, and rather than improvise around it, the analysis used the City's published ACS-five-year-by-community-area table instead [1]. That table draws on the same underlying ACS source and is cleaner for community-area geography, so the demographic backbone is sound, but the commute-demand comparison first intended is not in this paper [1]. The second choice concerns distance. Where trip length is reported, it is the straight-line haversine distance between the start dock and the end dock, not the distance along streets that a rider actually travels [1]. Straight-line distance understates real travel, since no one rides through buildings, so every distance figure is treated as a floor and not over-read.
One discipline line closes the section. Every count here describes docked, in-city activity only, and the station figures come from the current active roster, not the network exactly as it stood during the 2024 months sampled [1]. Stations get added and retired over time, so a handful present in the 2024 trip file no longer appear in the current roster, and a handful in the roster did not yet exist during parts of the sample [1]. The mismatch is small but real, so every station figure is dated to a snapshot rather than presented as the permanent shape of the system. That is the rule behind all of it. The paper reports what the current roster and a four-month slice of trips can honestly support, joined to real Census demographics, and stops there.
The supply gradient, dock by dock
The first finding holds the most weight. Count docks against the residents who live near them, sort Chicago's community areas from whitest to most residents of color, and the supply falls off a shelf. In the whitest quartile there are 111.38 docks for every 10,000 residents. In the quartile with the most residents of color there are 44.00 [1]. More than half the per-resident supply is gone by the time you cross the city along that axis.
The quartile-by-quartile path makes the slope legible. Reading from the whitest quartile down to the most-residents-of-color quartile, docks per 10,000 residents run 111.38, then 63.99, then 44.47, then 44.00 [1]. The steepest drop is the very first step, from the whitest quartile to the second, where per-resident supply nearly halves. The gap then narrows and flattens across the bottom three quartiles, which cluster from the mid-forties into the low sixties [1]. The top quartile is the outlier, sitting far above the rest. That one fact, that the whitest quartile holds more than two and a half times the per-resident docks of the most-residents-of-color quartile, is the spine of the paper.
The raw counts behind those rates make the gap concrete rather than abstract. The whitest quartile is 20 community areas holding 905,046 residents, and it carries 465 stations and 10,080 docks [1]. The most-residents-of-color quartile is 19 community areas holding 470,274 residents, and it carries 166 stations and 2,069 docks [1]. So a part of the city with nearly twice as many people holds nearly five times the docks. The two middle quartiles fall in between in the expected order, 284 stations and 4,264 docks in the second, 224 stations and 2,695 docks in the third [1]. Put the endpoints side by side and the asymmetry is stark on its own terms. Roughly 905,000 residents in the whitest areas can reach 10,080 docks. Roughly 470,000 residents in the most-residents-of-color areas can reach 2,069. The per-resident rate compresses that into a single ratio, but the underlying counts are where the gap is easiest to feel.
Divvy Dock Supply Thins as Neighborhoods Get More Residents of Color
The stations-per-resident line in the same chart tells a more precise story. Stations per 10,000 residents slide much more gently than docks, from 5.14 in the whitest quartile to 3.53 in the most-residents-of-color quartile [1]. That is a real decline, but a modest one beside the dock collapse from 111.38 to 44.00. The two lines moving at such different rates point to something specific about the shape of the gap. The number of stations a neighborhood has does not fall nearly as fast as the number of docks those stations hold. In plain terms, a lower-income, more-residents-of-color neighborhood is not far less likely to have a Divvy station somewhere in it. What it is far less likely to have is a station, or a cluster of them, holding many bikes.
That distinction, between a station existing and a station being dense enough to matter, is the analytical heart of the section, and it is the gap the Tampa framework named [6]. The dock is the unit of access, not the station. A six-dock station that empties out by 8 a.m. offers far less real access than a twenty-five-dock station a few blocks away, even though a map of stations counts them the same. Presence and density are different things, and the Chicago map separates them cleanly. Counted by stations, the city looks close to evenly covered. Counted by docks against people, it does not. The supply gradient lives in density, in how many bikes sit at the stations that exist, more than in whether a station exists at all.
This is the central-versus-periphery tilt the multi-city siting work describes, arriving on a Chicago street. The Tampa study documented uneven accessibility across both geography and sociodemographic groups [6], and the four-city study by Ursaki and Aultman-Hall found the share of Black residents inside service areas significantly lower than outside [8]. The Chicago numbers are a local instance of that same tilt, with a wrinkle the density-versus-presence split adds. In Chicago the gap is less about service-area boundaries shutting neighborhoods out entirely, and more about how thinly the docks are spread once you leave the affluent, whiter core. The Chicago-specific finding from Qian and Jaller belongs here too, since they reported station use concentrated more unequally in disadvantaged community areas [2]. Thin dock supply and concentrated, unequal use are two faces of one condition, a system whose real capacity bunches where residents are whiter and richer.
The bare-presence picture deserves its own number, because it is the part that looks reassuring and the part that is most often quoted. Of the 77 community areas, 74 have at least one Divvy station [1]. Only three have none, and the three do not line up neatly along the equity axis the way the dock counts do. They are Oakland, a small South Side lakefront area that is 96 percent residents of color, with 57 percent of households under 50,000 dollars; Clearing, a working-class Southwest Side area near Midway that is 64 percent residents of color; and Edison Park, the far Northwest Side area on the city edge that is 20 percent residents of color and among the highest-income in the city, with only 10 percent of households under 50,000 dollars [1]. One unserved area is poor and heavily of color, one is middle-income, and one is affluent and white. That spread is the clearest sign that bare coverage is not the right lens. If having no station at all tracked race or income tightly, the three blank spots would cluster at one end. They do not. The inequality lives less in which neighborhoods get a station than in how many docks the stations hold once they are there, which is why the dock-per-resident rate, not the station count, is the measure this paper leans on.
Every demographic word in this section refers to the people who live in these community areas, as the Census counts them, not to anyone who rode a Divvy bike. The data offers no way to know who unlocked the 2,069 docks in the most-residents-of-color quartile or the 10,080 in the whitest [1]. What it can say is where the docks are and who lives around them, and on that geographic measure the gradient is steep and consistent.
The race axis is one way to cut the city, and it is bound up with another. Neighborhoods with more residents of color in Chicago also tend to have lower incomes, which is not an accident of the present but the residue of a century of housing policy, of restrictive covenants, redlining, contract selling, and the routing of expressways and public housing that fixed the color line into the map. Chicago is one of the most segregated large cities in the country, and that segregation is the reason the race cut and the income cut sort the 77 community areas into nearly the same order. The same South and West Side neighborhoods that the federal mortgage maps of the 1930s shaded red, and that decades of disinvestment followed, are the ones that now show up at the bottom of the dock-supply ranking. None of that history is measured in this paper, and the supply gradient is not evidence about its causes. But it is the backdrop against which a present-day map of thin docks in those same neighborhoods has to be read, because the geography Divvy inherited was not a blank one. The natural next question is whether the supply gradient looks the same sorted by money instead of race. It does, almost exactly, and that is the next turn.
Money tracks the same line
Sort the same 77 community areas by income rather than race and the supply gradient barely moves. Docks per 10,000 residents fall from 115.85 in the highest-income quartile to 43.66 in the lowest [1]. That is the same slope just traced on race, reaching very nearly the same floor, by a different road.
The quartile values walk down in the same shape. From the highest-income quartile to the lowest, docks per 10,000 residents run 115.85, then 63.19, then 46.14, then 43.66 [1]. As with race, the first step is the steepest, the top quartile stands well above the rest, and the bottom three settle in the forties and low sixties. Stations per 10,000 residents again move gently, from 5.33 in the highest-income quartile to 3.75 in the lowest [1]. The presence-versus-density split from the previous section returns unchanged. Stations are spread fairly evenly. Docks are not.
The counts mirror the race cut almost line for line. The highest-income quartile is 20 community areas with 875,089 residents, carrying 466 stations and 10,138 docks, while the lowest-income quartile is 19 areas with 554,768 residents, carrying 208 stations and 2,422 docks [1]. The two top quartiles, sorted one way by race and the other way by income, are nearly the same 20 areas and nearly the same 10,000-odd docks, which is the first hint, before any correlation, that the two rankings are tracking one underlying geography rather than two independent ones. The middle quartiles hold 222 stations and 3,445 docks, then 243 stations and 3,103 docks [1]. The slight non-monotonic wobble there, where the third quartile has more stations but fewer docks than the second, is exactly the kind of small irregularity a real distribution shows and a smoothed one hides, and it does nothing to disturb the overall slope from top to bottom.
Divvy Dock Supply Falls With Neighborhood Income
The interpretive point is short and it is the whole reason this section is short. Because race and income are correlated across Chicago neighborhoods, these two gradients are not two independent pieces of evidence. They are two views of one tilt. The same affluent, whiter community areas sit atop both rankings, and the same lower-income, more-residents-of-color areas sit at the bottom of both. The analysis does not separate the effect of race from the effect of income, and it should be clear about declining to, because separating them would be a causal exercise, an attempt to say which trait drives dock placement, and a descriptive snapshot cannot answer that [1]. What the snapshot can say is that whether you slice the city by race or by income, per-resident dock supply collapses by more than half from top to bottom, and it is very likely the same underlying geography showing up twice.
The simple correlations across all 77 community areas add supporting texture, reported plainly as descriptive and carrying no causal weight. The Pearson correlation between percent residents of color and docks per 10,000 residents is -0.254, and between the share of households under 50,000 dollars and docks per 10,000 residents it is -0.232 [1]. Both are negative, meaning per-resident dock supply tends to run lower in higher-residents-of-color and lower-income areas, and both are modest, computed across served and unserved areas alike [1]. They point the same direction as the quartile tables, which is the useful confirmation. A correlation of that size is a gentle tendency across a noisy set of 77 areas, not a tight line, and the paper presents it as exactly that.
One more correlation earns its place, because it nails the density-versus-presence distinction with a single number. The correlation between percent residents of color and stations per 10,000 residents is -0.002, essentially flat [1]. Whether a neighborhood has stations per resident is, across the city, almost unrelated to its racial composition. Whether those stations hold many docks is a different matter. The supply gap is a density gap. That flat station correlation, sitting next to a clearly negative dock correlation, is the cleanest evidence in the analysis that the two are not the same thing.
The thin-dock map has a counterpart in the survey work, on the ground, in what residents report. McNeil and colleagues found station distance and cost among the barriers falling hardest on lower-income residents of color across Philadelphia, Chicago, and Brooklyn [4], and Hirsch and colleagues found non-White respondents in Seattle reporting more cost and spatial-access barriers even where awareness ran high [5]. A map with few docks per resident is, from the sidewalk, precisely the experience those surveys describe, a longer walk to a station that may hold fewer bikes, in a neighborhood where the cost of a casual ride bites harder. The analysis cannot link its map to those riders directly, having no rider data, but the supply gradient measured here and the access barriers surveyed there are plausibly two descriptions of one condition, one taken from the dock and one from the resident.
That covers supply, on both axes. The map of where bikes can be reached tilts hard toward the affluent, whiter core, mostly through density rather than the bare presence of stations. The next question is what people do on that map, where the trips begin, and there the tilt sharpens.
Where the rides actually begin
Move from the map of where docks sit to the record of where wheels turned, and the tilt traced in supply does not merely persist. It collapses toward a single corner of the city. Of the 1,575,899 station-matched trips in the four-month sample, 92.9 percent start in higher-income, whiter community areas. Lower-income, more-residents-of-color areas account for 3.7 percent. Mixed areas account for 3.4 percent [1].
Nearly All Divvy Trips Start in Higher-Income, Whiter Areas
The area types behind those shares are defined by splitting the city at its own midpoints on both axes at once, an area counting as higher-income and whiter only if it sits above the median on income and below the median share of residents of color, and as lower-income and more of color only if it sits the other way on both [1]. The two-axis split is deliberate, and the mixed category that falls out of it carries real weight. It holds the community areas that are high on one axis and low on the other, an affluent area that is also heavily of color, or a lower-income area that is also majority white, and keeping it separate is what lets the higher-income, whiter and lower-income, more-of-color groups stay clean. Lumping the mixed areas into one side or the other would blur exactly the contrast the paper is trying to measure. By that double split, 32 community areas with 642 stations land in the higher-income, whiter group and produce 1,464,559 of the matched trips; 31 areas with 344 stations land in the lower-income, more-of-color group and produce 57,918; and 14 mixed areas with 153 stations produce 53,422 [1]. The lower-income, more-of-color group has more than half as many stations as the affluent group, 344 against 642, and yet produces about one twenty-fifth of the trips. That juxtaposition, a comparable number of stations and a wildly smaller number of trips, is the trip-concentration finding in its rawest form, and it is worth sitting with for a moment. The two groups hold the same order of magnitude of stations. They do not hold the same order of magnitude of riding. Whatever turns a station into a trip is happening at the affluent docks and barely happening at the others.
Hold the pie beside the supply charts, because the two are not measuring the same thing and the difference is the point. Dock supply per resident ran about two and a half to one across the racial quartiles, 111.38 down to 44.00 [1]. Trip origins run better than twenty-five to one between the two area types. The places that hold most of the docks do not merely host most of the riding. They host nearly all of it. The use gap is far steeper than the supply gap that sits underneath it, and that distance between the two is itself worth marking. If access were the whole story, trips would track docks. They do not. Trips fall away much faster than docks do, which means something beyond bare dock supply, latent demand, trip purpose, the presence or absence of tourists, is doing additional work that the supply numbers alone cannot see.
Pull the camera in from area types to single docks and the concentration sharpens once more. The top decile of start stations, 159 of them, accounts for 62.7 percent of every matched trip in the sample [1]. All ten of the busiest origin stations sit in just four community areas, Near North Side, Loop, Lincoln Park, and Near West Side [1]. The single busiest is Streeter Dr and Grand Ave, a lakefront dock near Navy Pier and the river mouth, at 22,401 trips. That one dock draws about 39 percent as many trips as the entire lower-income, more-residents-of-color category collects across its hundreds of stations, since that whole category logs 57,918 matched trips in the sample [1]. Behind Streeter come DuSable Lake Shore Dr and Monroe St at 13,894, Michigan Ave and Oak St at 13,223, Kingsbury St and Kinzie St at 13,004, DuSable Lake Shore Dr and North Blvd at 12,883, Clinton St and Washington Blvd at 11,609, Clark St and Elm St at 11,577, Clinton St and Madison St at 11,197, Millennium Park at 11,178, and Wells St and Concord Ln at 10,529 [1].
The list reads like a walking tour of the lakefront and the central business district, because that is what it is. Streeter Dr and Grand Ave sits where the tourist city is densest, a few minutes from Navy Pier, the Ohio Street Beach, and the hotels of Streeterville. Michigan Ave and Oak St is the top of the Magnificent Mile. Millennium Park and the two DuSable Lake Shore Drive stations, at Monroe and at North Boulevard, sit on the lakefront path itself, which is the single most ridden recreational corridor in the city and a magnet for visitors on day passes. Then come the commuter docks. Kingsbury and Kinzie sits in River North, and the pair on Clinton Street, at Washington and at Madison, sit at the western edge of the Loop beside Union Station and Ogilvie, where Metra trains and the expressway pour suburban commuters into downtown each morning, many of them finishing the last half mile by bike. The top of the list is two economies stacked on the same few blocks, tourism along the lake and white-collar commuting through the West Loop gateways, and both of them live in the affluent, central core. None of the ten busiest origin docks sits south of Roosevelt Road or west of the river's industrial belt. The geography of heavy riding is, quite literally, the geography of the postcard.
Usage concentration outruns the supply gradient, and the temptation is to read cause into the gap between them, to say the thin docks on the South and West sides are choking off rides that would otherwise happen. The data cannot carry that claim. A trip count records what happened, not what would have happened under a denser network. The 3.7 percent could reflect docks too sparse to ride from, or lower latent demand for bikeshare in those neighborhoods, or commute patterns that do not run where the docks run, or some braid of all three that this slice cannot pull apart. The concentration gets reported because it is real and large. Its cause goes unnamed because naming it would be a different study, one with rider-level data and a comparison the trip files do not contain.
The top-decile number deserves a moment on its own, because it is the cleanest single measure of concentration the analysis produces. One station in ten, ranked by volume, accounts for nearly two-thirds of all matched trips [1]. A perfectly even system would put about a tenth of the trips in the top tenth of stations. Chicago puts 62.7 percent there. The riding lives in a thin band of very busy docks while the long tail of stations, the hundreds that serve the rest of the city including nearly all of the South and West sides, splits the remaining third among them. Concentration on that scale is consistent with what Qian and Jaller measured inside disadvantaged community areas, where activity bunched onto a few stations, but here it is the citywide shape, the whole system tilted toward its busiest corner [2].
What the trip geography does confirm is that the published pattern holds in Chicago on the measure that matters most for access. Qian and Jaller's Chicago analysis found station use more unequal in disadvantaged community areas, with activity concentrated in already-advantaged places [2]. Seattle's survey work found actual users clustered in the central, already-cycling core even where awareness ran broad [5]. Our snapshot is a fresh draw from that same well. Whatever else bikeshare is in Chicago, in the record of where rides begin it is overwhelmingly a downtown and North Side lakefront system.
That raises the obvious next question. If almost every trip starts downtown, and downtown is where the tourists are, then who is riding from the relatively few docks out in the lower-income, more-residents-of-color neighborhoods? The answer is where the paper's method earns its keep, because it inverts the easy guess.
The member surprise
Run the member-versus-casual split by area type and the result lands the wrong way around from intuition. The lower-income, more-residents-of-color neighborhoods, where one would expect annual members to be scarcest, are where they cluster most thickly. The member share reaches 70.2 percent in those areas, against 65.1 percent in mixed areas and 63.9 percent in higher-income, whiter areas [1]. Citywide across all matched trips the figure is 64.1 percent [1]. The neighborhoods with the fewest docks and the fewest trips have the most committed riders per trip taken.
Annual-Member Share Is Highest in Lower-Income, More-POC Areas
The mechanism is not mysterious once you remember where the trips start. Casual riding in Chicago is largely a downtown-tourist behavior, not a neighborhood one. The casual category, in Divvy's own terms, is anyone riding on a single trip or a day pass rather than an annual membership, which is the natural choice for a visitor who will use a bike a handful of times and never again. A tourist in for a weekend buys a day pass, rides Streeter Dr and Grand Ave along the lake to Navy Pier and back, and never becomes a member. That single station drew 22,401 trips, and a heavy share of casual riding piles up at exactly those lakefront and central docks, the ones nearest the hotels and the attractions [1]. Out in the lower-income, more-residents-of-color neighborhoods the casual tourist is simply not there, because the tourists do not go there, so the relatively small number of trips that do begin in those areas come disproportionately from people who bought an annual membership and ride from where they live, the way a regular commuter or errand-runner would. The higher member share in those neighborhoods is the residue of who is absent, the tourists, not a sign of how well the neighborhood is being served. A place can post the city's highest member share precisely because it has been left off the tourist map, and that is close to what is happening here.
That reframes what the member share even measures in this system. It is tempting to read a high annual-member rate as a marker of good access, a sign that a neighborhood has folded bikeshare into daily life. Here the reading runs closer to the opposite. In Chicago a neighborhood's member share tracks its distance from the tourist core more than its depth of access. The whiter, higher-income center looks less member-heavy precisely because it is awash in casual day-pass riders the periphery never sees. A measure that seems to describe rider commitment is, in this geography, mostly describing the absence or presence of tourists.
The reframing carries a small but real warning for anyone tracking equity through this number. Member share is exactly the kind of metric that could be misread in a dashboard, where a high annual-member percentage in a South or West Side community area might look like a sign the system is working there, that residents have committed to bikeshare and built it into their routines. The geography says to be careful. The committed riders are real, and the next section returns to them, but the high percentage is driven as much by the denominator emptying out, by the casual trips that never happen in those neighborhoods, as by the members filling it. The honest version of the metric is not member share alone but member share read alongside trip volume, and the trip volume out there is very low. A neighborhood with a 70 percent member share and a few thousand trips is not better served than the lakefront with a 64 percent member share and well over a million.
This is the spot where the honesty line has to hold hardest, because the headline number is genuinely counterintuitive and a counterintuitive number is the easiest kind to oversell. So the limit goes down beside the finding. The trip-weighted area-type table that produces 70.2 percent is dominated by a few very high-volume downtown community areas. Set those aside and look at member share across the 77 community areas one at a time, giving each area an equal vote, and the relationship is weak, noisy, and actually pointed the other way. The community-area-level correlation is -0.205 for percent residents of color against member share and -0.076 for the income measure [1]. A negative correlation there means higher-residents-of-color areas lean, faintly, toward fewer members per trip, the reverse of what the area-type table shows. Both numbers are real. They disagree because they weight the city differently, one by trip volume and one by giving each area equal say. The paper does not launder that tension into a clean claim. The findings that survive every reasonable way of cutting the data, and that reproduce from the public files, remain the dock-supply gradient and the trip concentration. The member surprise is a real and interesting pattern in the trip-weighted data, reported with its own contradiction left in plain view.
Set the neighborhood pattern beside the one program figure that bears on the same question from another direction. Streetsblog reports that about 72 percent of Divvy for Everyone members are people of color and that D4E participation runs highest on the South and West sides [10]. That fits the texture of what we see, the committed riders out in lower-income, more-residents-of-color neighborhoods being real and signing up. The figure has to be handled for exactly what it is. It is a program-level statistic about who enrolls in a five-dollar income-eligible membership, reported by a third party [10][9]. The analysis did not compute it, and it cannot be stacked onto the 70.2 percent as though the two were one measurement. They point the same way. They are different instruments, read off different data, and the paper keeps them apart.
The signals that stayed quiet
A fair analysis has to be willing to come up empty, and on two of the questions put to the data, it did.
Compare e-bike share across area types and the range is narrow. Higher-income, whiter areas run 42.0 percent electric, mixed areas 48.0 percent, and lower-income, more-residents-of-color areas 42.8 percent [1]. Compare mean trip distance and it is just as flat, 2.14 km in the whiter, higher-income areas, 2.30 km in mixed areas, and 1.84 km in the lower-income, more-residents-of-color areas [1]. Neither shows a clean gradient. Whatever separates Chicago's neighborhoods in bikeshare, it does not surface in which kind of bike people grab once they reach a dock, or in how far they ride.
These get reported as a minor descriptive note and nothing more, because two cautions apply and both push toward claiming less. The differences are small to begin with, well inside the range a different sample of months could reshuffle. And the distance is haversine straight-line between the start and end docks, not routed distance over real streets, so every figure understates how far anyone actually traveled and none should be read as a precise trip length [1]. A six-block dogleg around a rail yard and a six-block straight shot land in this measure as the same number. They are not the same ride.
The absence is still informative, which is why the section stays rather than getting buried. If the equity story lived in how people ride once they reach a dock, in a tilt toward classic bikes here or longer trips there, this is where it would surface, and it does not. The story sits upstream of the ride itself. It is in how many docks stand within walking distance and how many trips ever begin, not in the rideable type or the kilometers logged after someone has already reached a bike. Saying that quiet result out loud is part of keeping the loud ones honest.
There is a methodological reason to expect this kind of flatness, and naming it guards against over-reading even the small differences that do appear. The riders captured in the trip data are, by definition, people who reached a dock and took a bike. They have already cleared the access barrier that the supply gradient measures. Conditioning on that, on the act of riding, washes out a great deal of between-neighborhood difference, because the people riding from a thinly served neighborhood are a self-selected group who found a way to the bike. So the place to look for inequality is not in the behavior of people who rode but in the count of how many got to ride at all, and that count is the trip-concentration finding, not the e-bike mix. The quiet of this section is consistent with the loudness of the earlier one. The gap is in access, measured before the ride, and once the ride begins the neighborhoods look much more alike.
What this can and cannot tell you
Everything above describes a snapshot, and the snapshot has edges worth drawing plainly, because a reader deserves to know where the map runs out.
The boundary that governs all the rest comes first. This is dated, descriptive, and geographic. Every demographic word in the paper describes a neighborhood's resident population from the ACS five-year 2023 tables, never a rider [1]. Divvy reports whether a trip was taken by an annual member or a casual rider and nothing else about the person on the bike, so member versus casual is the only rider category anywhere in this analysis, and percent residents of color is a community-area population share laid underneath the docks and trips, not a description of who mounted them [1]. No number here proves a cause. The figures describe where access and activity sit. They do not explain why.
The concessions stack up from there, and they are real ones. The trip layer is a four-month slice across the seasons, not a full year, chosen to dodge the summer casual-rider bulge but a slice all the same [1]. The 338,070 dockless trips, 17.6 percent of the raw rows, are absent from every station-based figure because they carry no fixed origin to place, and there is no ruling out that those rides hold a different geography than the docked ones, perhaps reaching into neighborhoods the dock network thins out of [1]. The station roster is the current active network of 1,153 stations and 19,318 docks, so it does not line up perfectly with the docks that stood in 2024, and a handful of stations fall on one side of that seam and not the other [1]. Income is proxied by the share of households earning under 50,000 dollars because the City extract reports brackets rather than a single median, and no median was invented to paper over the difference [1]. Distances are straight lines [1]. The two supply gradients, race and income, are not independent evidence, since the two neighborhood traits move together across Chicago and no attempt was made to separate their effects, which would be causal work this analysis did not do [1]. And the member-share correlations across community areas are weak and point against the trip-weighted table, which is why the paper rests its weight on supply and trip concentration instead [1]. One ambition set down along the way deserves its own line. The plan had been to lay the Census journey-to-work bike-commute layer beside the dock map, to ask where latent cycling demand and infrastructure agree and disagree, and the Census API key that pull required was unavailable, so the analysis used the City's cleaner community-area ACS table and left the commute comparison unbuilt [1]. It is a real gap, and naming it beats implying a comparison that was never made.
A word is owed on how much the dockless exclusion could move the headline, since it is the limitation most likely to matter. The 338,070 dropped rides are e-bike trips that did not end or begin at a station, and if they reached systematically into the South and West sides, the trip-concentration finding would soften. There is reason for caution in both directions. A dockless e-bike still has to be found and unlocked where it sits, so it does not obviously serve a neighborhood the docked network skips, and the people most able to chase down a free-floating bike are not obviously the residents of the thinnest-served areas. But the honest position is that the public files do not say, and a block of riding nearly one fifth the size of the raw sample is large enough that it has to be flagged rather than waved off. The supply gradient does not depend on it at all, since that finding rests on the station roster and the Census, not on the trip sample. The trip-concentration finding is the one a very different dockless geography could in principle dent, and it is reported with that caveat attached.
Set those limits down and the synthesis still stands, because it does not lean on any single fragile number. Across methods that share almost nothing, national service-area mapping, a Tampa accessibility framework, a Seattle survey, a three-city barrier survey, and two Chicago studies, the literature keeps landing on the same shape, a central, whiter, higher-income tilt in where fixed-station bikeshare sits and a cluster of cost, distance, and safety barriers that fall hardest on lower-income residents of color [2][3][4][5][6][8]. The Chicago snapshot agrees with that body of work on its two sturdiest measures, dock supply per resident and trip origin. That agreement is the paper's real contribution, a fresh, reproducible, public-data confirmation that the pattern the field keeps finding is visible in Chicago right now.
It is worth being exact about what going further would take, because the honest answer is sobering. To move from this description to a claim about cause, or to any statement about who rides rather than where rides begin, you would need rider-level data Divvy does not publish, the race, income, and age of the person on the bike, joined to the trip. That data does not exist in the public record, so the rider-demographic question stays formally out of reach here, and any paper that answers it from these files is guessing. What the public record does support is the geographic exposure read given above. As for the fix, the evidence is measured rather than hopeful. Boston's income-eligible membership and station expansion lifted access citywide, and the smallest gains landed in the communities that needed them most [7]. Chicago's own answer, the five-dollar Divvy for Everyone membership, reaches a heavily of-color base concentrated on the South and West sides [9][10]. In both cities the pattern is that targeted programs narrow the gap and do not erase it. That is the realistic frame, and it argues for claiming less rather than more.
Back to the dock
Return to where the paper began, at a single dock on a single block, the smallest unit of whether bikeshare is real for the people who live there.
Read across the snapshot without piling the numbers up again, and three plain things are true. Docks thin out as a neighborhood gets poorer and less white, from better than a hundred per ten thousand residents in the whitest, highest-income quartiles down to the low forties at the other end [1]. Nearly all the recorded riding begins in a small, central, well-served core along the lake and through downtown [1]. And the people who do ride from lower-income, more-residents-of-color neighborhoods lean disproportionately toward committed annual members rather than the day-pass tourists who fill the lakefront docks [1].
That last fact reframes the question. It is easy to assume the equity work in bikeshare is about converting casual riders into members, about pricing, sign-ups, and outreach. The snapshot points somewhere earlier in the chain. In Chicago the binding question is closer to whether a dock stands within a reasonable walk of where someone lives, and whether it holds enough bikes to be worth that walk, because the riders out past the tourist core are already members. They have done the converting. What many of them lack is the dock.
The sequencing of those two levers, the cheaper membership and the denser dock network, is where the cited evidence becomes most useful. A five-dollar membership lowers the price of a ride to almost nothing, and the Divvy for Everyone enrollment that Streetsblog reports, mostly residents of color, concentrated on the South and West sides, shows that the lower price reaches the people it was built for [9][10]. But a membership is only worth what there is to unlock. If the dock nearest a member's home is far, or small, or often empty, the cheaper fare buys access to a bike that is not there. Boston's experience is the warning written out. Pairing an income-eligible membership with a station expansion lifted access across the city, and still the smallest gains landed in the communities that needed them most [7]. That pattern is consistent with a system where the membership lever moves faster than the dock lever, where the price comes down quickly but the network fills in slowly and unevenly, so the people with the cheapest fares are also the people with the fewest docks to spend them at. The Chicago snapshot cannot prove that mechanism, but it is exactly the shape the snapshot would take if the mechanism were real, a heavily of-color equity membership sitting on top of the thinnest dock supply in the city.
It is worth being plain about why the dock matters more than the bike here, given how the broader literature ties bikeshare to the rest of a city's transit. Qian and Jaller found transit-dependent households among the groups underrepresented in Divvy ridership [2], and that is the population for whom a dense bikeshare network would do the most work, by closing the first and last mile between home and a rail station or a bus line. A dock is more than a place to start a recreational ride. For someone without a car, in a neighborhood where the nearest train is a fifteen-minute walk, a bike at the corner is the difference between a transfer that works and one that does not. The thinness of the dock map in lower-income, more-residents-of-color neighborhoods is therefore a transit problem more than a recreational one, a gap in the part of the mobility system that the people with the fewest alternatives would lean on hardest. That is the vertical-equity question the Tampa framework named, asked of Chicago, and the supply numbers answer it the wrong way [6].
None of this is a verdict, and the system's equity intent is real. Divvy for Everyone exists, it reaches the South and West sides, and the broader literature is clear that programs like it help [7][9][10]. The same literature is just as clear that they help partway, that a five-dollar membership cannot stand in for a dock that is not there. The map drawn here is of where access sits in Chicago today, dated and descriptive, an honest starting point for a policy conversation rather than the last word in one. What it shows is consistent enough with a decade of bikeshare-equity research, and steep enough on its two sturdiest measures, to make one claim safely. The first equity question in Chicago bikeshare is not who rides. It is where the docks are. The dock is still the unit. The question is still whether it is there, and whether it holds enough bikes to matter.
Citations
10 sources cited.
Primary Sources
- 1.Author analysis of Divvy Bicycle Stations (City of Chicago Data Portal) joined to ACS 5-year community-area / tract demographics, with Divvy public trip records as a usage layer. Chicago Department of Transportation / Lyft (Divvy) via City of Chicago Data Portal; U.S. Census Bureau (ACS 5-year); Divvy System Data (S3). Reproducible files at rooted-forward.org/research/data/chicago-divvy-bikeshare-equity.
- 2.
Secondary Sources
- 3.
- 4.
- 5.
- 6.
- 7.
- 8.
- 9.
- 10.