Higher education in Leinster · accessibility, cost, emissions and choice
This is individual research and analysis. The views, interpretations and any errors are my own; it is not an official Maynooth University publication and does not represent the University’s position. About this analysis.
How long, how costly and how carbon-heavy is the daily journey from home to college for the young people of Leinster, and does it shape where they choose to study? The study models a weekday commute from every one of Electoral Divisions across Leinster (except Wexford) and Monaghan to 14 higher-education campuses, by public transport and by car, for the people aged 17 to 19 who live there.
Find your area
Type the name of a town or Electoral Division. You get the morning journey to each of the 14 campuses by public transport and by car, what a year of that commute costs, and the carbon it emits.
Morning peak (arrive about 08:30), one representative term Wednesday. Public transport includes the option of driving to a railway station. Costs are for 140 commuting days: student and Young Adult fares (January fare levels) or car fuel plus campus parking and tolls. Carbon is operational CO2 for a single-occupant petrol car or the public-transport mode used; the carbon branch lets you change these assumptions. An Electoral Division is a small statistical area; the figures are for its population-weighted centre.
Methods, data and how to cite
Every figure in this report is regenerated by a numbered pipeline of scripts from dated copies of public data. This section sets out the whole method in the order the work was done, then what the results can and cannot support, where the data came from, and how to reproduce it.
The DOI above identifies this version; 10.5281/zenodo.23108980 always resolves to the latest version. Archived on Zenodo with the code and data; source at github.com/JKeatingMU/commuting-to-college. This is individual research and analysis, not an official Maynooth University publication.
Who. John G. Keating, Professor and Associate Dean for Teaching and Learning, Faculty of Science and Engineering, Maynooth University (ORCID 0000-0002-5063-2773).
Status. This is individual research and analysis. The questions, interpretations and any errors are my own. It is not an official Maynooth University publication, it was not commissioned by the University, and it does not represent the University's position or policy.
Funding. This work received no funding.
Why. The daily journey to college shapes who can study where, what it costs families, and what it adds to transport emissions, yet it is rarely measured campus by campus across a whole region. Every input here is public, so the same questions can be asked of any campus, and the answers checked by anyone.
An interest to declare. I work at Maynooth University, one of the fourteen campuses in the study, and Maynooth is the worked example. Every campus is modelled with the same data and the same methods, every tool works for any of them, and the report states plainly where Maynooth comes out poorly: far fewer young people can reach it within an hour by public transport (42%) than by car (69%), and it recruits fewer students than its commute predicts from some of the counties on its doorstep.
How AI was used. The study was built with Claude Code, Anthropic's AI coding assistant, working under my direction. It wrote most of the data-processing and modelling code and drafted much of the report's text. I set the questions, made the analytical choices and reviewed the results. Every figure in the report is computed by the published scripts from the published data; none is supplied by the AI. Two kinds of check guard against errors: every number in the report was checked against the output tables before release (that check found and corrected a fault in the enrolment model, recorded in the version history), and the travel-time model is compared with the journeys students report in Census 2022. The code, data and checks are all released so that anyone can repeat them.
Corrections. If you find an error, please open an issue on the repository or email john.keating@mu.ie. Corrections are made in a new version, listed in the version history, and each version keeps its own DOI, so earlier citations remain valid.
| Source | Used for | Snapshot | Licence |
|---|
Raw downloads are kept, dated, in data/raw/; the full licence review is in
docs/DATA-SOURCES.md. Attribution. Contains Central Statistics Office data (Census 2022), © CSO,
CC BY 4.0. Contains Tailte Éireann boundary data, CC BY 4.0. Public-transport timetables © National Transport
Authority, CC BY 4.0. Map and routing data © OpenStreetMap
contributors, available under the Open Database Licence. County outlines from geoBoundaries (source: Ordnance Survey
Ireland), CC BY 4.0. Pobal HP Deprivation Index 2022 (Haase and Pratschke), © Pobal, CC BY 4.0. School details
from the Department of Education, Data on Individual Schools, CC BY 4.0. Higher-education entrant counts from the Higher Education Authority, CC BY 4.0.
Each dot is one of 1,197 Electoral Divisions: the cost of a round trip per day, peak schedule, against straight-line distance to the campus. Axes start at 0 to 150 km and €0 to €80 a day so campuses compare directly. Scroll to zoom, drag to pan, ↻ to reset; click a legend entry to hide a series.
Population-weighted, peak schedule. "Car fuel" is fuel only; "car all-in" adds campus parking and any motorway toll. Public transport (PT) is the cheapest of the student rail or bus fare or the Maynooth commuter coach.
| County | km each way | Car fuel /day | Car all-in /day | PT /day | PT /year |
|---|
Choose Maynooth and the public-transport dots split into three: two flat horizontal bands and one rising line. That is not a plotting artefact; it is three different fare structures in one series.
| Fare type | Applies to | Shape on chart | Cohort share |
|---|
Caveat on the coach band. It credits the season-ticket price to any area within of a weekday Maynooth coach route's starting town, or of any stop along it, and assumes a seat is available. These services run one to three vehicles per route, so the coach band is a lower bound on cost for those areas, not a statement about capacity.
Every figure in this branch is recomputed from routed distance and the emission factors below. The chart sits next to the sliders so you can watch it move; its scale is fixed, so a change in position is a real change, not a rescaled axis.
Annual commute CO2 against one-way road distance. Each dot is one of 1,197 Electoral Divisions; the line is the car. Click the chart to enlarge it.
Cohort-weighted for the selected campus, at the current settings. The last column is the whole-cohort scale: if every 17 to 19 year old in that county commuted to this campus, the annual difference between all of them driving and all of them taking public transport.
| County | km each way | Car kg/yr | PT kg/yr | Saved kg/yr | Cohort | Cohort saving t/yr |
|---|
The flows are Higher Education Authority new-entrant counts by county of domicile and institution, undergraduate, 2018/19 to 2024/25, pooled, for each field of study in turn. The accessibility figures are the study's commute model aggregated to county, cohort-weighted, and are the same for every field (the drive to Maynooth is the drive to Maynooth whatever the subject). The region is the twelve counties of the study area (the four Dublin council areas count as one, as they do in the HEA data), with eight HEIs: Maynooth, DCU, UCD, Trinity, TU Dublin, Dundalk IT, TU Shannon and SETU.
Read this as a first look, not a finding. County is a coarse unit: Dublin is a single cell hiding everything inside the M50, and a large county straddles several commute bands. Field-level flows from small counties are thin and the HEA rounds every figure to the nearest 5. The model is cross-sectional, so it cannot separate the commute from everything correlated with it (course reputation, points, the pull of a home-county campus). What it can do is show whether the commute carries signal, whether that signal is the same size in every field, and where an institution's recruitment sits above or below what its geography would predict.
Domicile is not term-time address. The flows and the commute are both keyed to the student's home county. But a large share of students relocate: the Eurostudent VIII survey (Ireland, 2022) found about 45% living with parents, 19% in student accommodation and 36% in the private rented sector, so roughly half live away from home during term. Relocation rises with distance and is more common at the nationally recruiting universities. For those students the daily commute is far shorter than their county of domicile suggests, which makes commute friction look like a weaker deterrent here than it is for the students who genuinely commute from home. The true commuter-only decay is likely steeper than the figures below. Separating the two needs a student-level record with both addresses (see How to sharpen the answer).
The test is a gravity model of the flows, run separately for each field. It asks: within a single county, once you account for how big that county's cohort is and how large and well-known each HEI is, do more of its students in that field end up at the HEIs that are easier to reach? Each row of the table below is one run of that model: for a given field, using a given measure of the commute (car time, public-transport time, or straight-line distance), with a fixed effect for the county. What you are looking for is whether the commute column comes out negative, whether it is large enough to matter, and whether it is beyond the reach of chance.
Per 10: the change in a county's flow to an HEI for ten more minutes of commute (or ten more km), holding the county and the HEI fixed. A negative number means a longer commute goes with fewer students. z: how many standard errors the estimate sits from zero; above about 2 is "unlikely to be chance". Deviance drop: how much better the model fits once the commute term is added, on top of the county and HEI controls; above about 11 is significant at p < 0.001, and it is the fair way to compare the commute measures. AIC: overall model quality with a penalty for complexity; lower is better. The table shows the county-fixed-effects specification; a second specification that swaps the fixed effect for two plain county variables gives the same pattern with slightly smaller effects in most fields (see the methods).
The commute bites in all eight fields, but not by exactly the same amount. The bars below compare its size field by field; the text after them asks whether the differences mean anything.
Each coloured line is an exponential fit for one field: how a county's share of new entrants going to an HEI falls as the car commute to it grows. The grey cloud behind is every individual county-HEI point. A steeper line is a field where the commute matters more.
Pick a field, an HEI and what to shade. Click a county to open its panel; for Computing you can then click an HEI row to see its course list and 2025 points. What "computing" means differs sharply by HEI: the universities offer it through a handful of broad codes (Maynooth 4, DCU 3, Trinity 2, UCD 1), the technological sector through many named ones (TU Dublin 18, TU Shannon 20, SETU 18, Dundalk 6).
Click a county to open its panel.
This analysis gives a strong, consistent association and shows where it matters. It is a first look, held back by four specific limits. Each can be lifted with a bounded, well-defined data request, and each lift has a concrete payoff. This section doubles as the rationale for making those requests. The analysis pipeline is already built, tested and documented; the data is the only missing piece.
| Data addition | Removes | Enables |
|---|---|---|
| Eircode routing key in place of county | Dublin as one cell; large counties spanning several commute bands | A recruitment target list of named districts, not a county; far tighter estimates |
| Home and term-time address per student | Commuters and relocators mixed together | The true deterrent for daily commuters; the size of the addressable "study locally" market |
| Individual records plus a socio-economic marker | County averages; time and money inseparable | A proper choice model; a direct test of whether the commute is a bigger barrier for lower-income students |
| Year-by-year counts, not pooled | Any trend hidden | Whether commute sensitivity is rising with the cost of living and study-from-home norms |
The gap. County is coarse. Dublin is a single cell that hides everything inside the M50; Kildare spans a twenty-minute Maynooth commute in the north and an hour in the south. The friction coefficient is estimated from twelve county averages.
The fix, and what it buys. Resolving flows to Eircode routing key (the first three characters, about 140 non-identifying districts) turns Dublin into roughly thirty cells and splits every large county into commute bands. The commute term would then be identified from real within-city and within-county variation, and the standard errors would fall sharply. "Maynooth under-recruits Wicklow" becomes "Maynooth under-recruits these named districts, where its commute advantage is largest and its enrolment share is smallest": a targeting list, not a county.
The ask. The HEA's student data already holds domicile at Eircode level. A research extract of new-entrant counts by routing key by institution by field by year, with small cells suppressed at source, is aggregate statistics and never leaves as identifiable data. Request through statistics@hea.ie.
The gap. The flows and the commute are both keyed to the student's home address. About half of Irish higher-education students live away from home during term. For a relocating student the daily commute is nothing like the distance from home, so the model understates the deterrent for the students who genuinely commute, and cannot tell the two groups apart.
The fix, and what it buys. A student record carrying both a home address and a term-time address splits the population into daily commuters and relocators. The commute model then applies to the right people, and the true commuter-only decay can be measured directly. It answers the question a recruitment strategy actually needs: of the Kildare and Wicklow students Maynooth is missing, how many would commute daily if they enrolled, and are therefore reachable with a "study locally" message, and how many were always going to move away? The first group is the addressable market.
The ask. Maynooth's own student records hold term-time address. An internal research extract of home routing key plus term-time routing key plus course plus year, aggregated with small-cell suppression, handled as a research request through the Data Protection Office. This is not a Freedom of Information request against your own institution.
The gap. County aggregates cannot support a proper choice model or separate the effects that travel with distance: points, school type, family background, the pull of a home-county campus. And because cost and distance are almost perfectly correlated at county scale, the analysis cannot say whether it is time or money that deters, which is the equity question. This matters more now: from January 2027 the Young Adult and Student public-transport fare rises about 15%, the first broad increase since 2018, so the money side of the commute is growing against a cost-of-living backdrop.
The fix, and what it buys. Individual new-entrant records allow a destination-choice model with student-level controls, much tighter estimates, and an interaction between commute cost and a socio-economic indicator: HEAR or DARE status, school DEIS band, or the Pobal deprivation index for the student's area. That gives a direct test of whether the commute is a bigger barrier for lower-income students, the finding that would justify a commuter bursary, targeted transport support or a "study locally" access initiative with evidence rather than assumption.
The ask. The same HEA or records-office extract as items 1 and 2, at individual rather than aggregate level, with the socio-economic marker attached and direct identifiers removed before release.
The gap. Seven years are pooled to make the county-by-HEI counts large enough to model, which hides any trend.
The fix, and what it buys. The finer geography in item 1 makes single-year cells large enough to keep separate. That shows whether commute sensitivity is rising. With cost-of-living pressure and the post-pandemic normalisation of studying from home, this is the difference between a stable background factor and a growing one that policy needs to act on now.
For the sector, the same data quantifies how much geography shapes who reaches higher education, and where new provision or better regional transport would most widen participation. None of this requires new fieldwork or new methods. The pipeline that produced this report runs on the finer data unchanged; only the extract is outstanding.
Share of the study area's 17 to 19 cohort within N minutes (morning, one way) of the selected target, a group of institutions or one named HEI, for each mode and each timetable. "A university" = MU, DCU, UCD, Trinity, TU Dublin, TUS, SETU; "a regional HEI" = TUS Athlone, Dundalk IT, SETU Carlow. For a named HEI the figure is the best trip to any of its campuses.
Public transport PT + drive-to-rail Car
| Timetable | Mode | 30 min | 45 | 60 | 75 | 90 |
|---|
The students genuinely weighing two institutions on commute grounds are those who can reach both within a reasonable trip: the contested cohort. For them, how do the two compare, in journey time and in daily cost, by public transport and by car? Home defaults to Maynooth; pick any comparator. Times are one-way morning minutes; cost is the round-trip daily figure from the cost model (student rail season or coach, or car fuel, parking and toll).
An example. Set this to Maynooth against DCU at 60 minutes: many students can reach both within the hour. The county breakdown shows where each institution wins the commute. The figures below update as you change the selectors.
Cohort-weighted medians and shares over the contested cohort. "Better" means faster or cheaper by more than a rounding margin. Potential accessibility only: real choice also turns on course availability, points and reputation (see the enrolment model).
| County | Contested cohort | Home faster: PT | Home faster: car | Home cheaper: PT | Home cheaper: car |
|---|
Minutes a full private-car commute saves over the best public-transport option (PT + drive-to-rail), to the nearest HEI, at the morning peak. This is the reachability gap that a car closes, largest in exactly the rural areas where car ownership is highest and the alternatives thinnest. It is a measure of dependence on the car, not of good provision.
| County | Median car saving (peak) | Within 60 min by car | By public transport |
|---|
1,197 Electoral Divisions, one dot each. Scroll to zoom, drag to pan; hover a dot for its campuses; click a campus marker to isolate the areas it wins (its modelled catchment). The "most accessible campus" colour is the shortest modelled trip, not where a student would choose to go.
| Campus | AM | Round trip |
|---|
Daily commuter coach (Monday to Friday, one timed morning run) Weekend residential coach (Friday out, Sunday back) Hover a line for the timetable. These services are already in the public-transport times.
Best morning one-way trip to any HEI, a university or Maynooth, for the mode and schedule chosen on the map. Type to filter; click a column to sort. With a campus catchment selected on the map, the table narrows to it.
| Electoral Division | County | 17-19 | Any HEI | University | Maynooth | Most accessible |
|---|
The tools above support asking not just which campus wins an area but why. Here is one pattern worth knowing about.
| From | NCAD | Next best |
|---|
Two lessons. First, for the intercity-rail catchment a campus's position relative to the terminus matters more than anything about the institution. Second, "most accessible" is the shortest trip, not the chosen one: a real applicant trades ten minutes against the course and the campus they actually want, so this map describes reach, not demand.
Every figure uses the morning peak and public transport with the option of driving to a station, unless a column says otherwise. "Young people" is the college-entry cohort, 17 to 19 year olds, placed where they live at home.
Each area is placed in one of four bands on the Pobal HP Deprivation Index 2022, the same bands the Higher Education Authority uses for its socio-economic profiles of students. Does college sit further away, and cost more to reach, for young people from disadvantaged areas?
| Deprivation band | Young people | Within 60 min: public transport | Within 60 min: car | Typical public-transport time (median) | Public transport € a year | Households with no car |
|---|
The rest of this report shows the car commute "if a car is available". Here each area's share of households with no car (Census 2022) is taken into account: for those households, public transport is the commute.
| Area | County | Households with no car | Young people without a car (est.) | Public transport | Car |
|---|
SUSI pays the higher "non-adjacent" maintenance grant only when a student lives 30 km or more from college, measured as the shortest road route. Which areas sit under the line, so qualify only for the lower rate, yet face a long public-transport commute?
| Area | County | Road km | Public transport | Car | Young people |
|---|
Road distance here is the routed distance from each area's population-weighted centre; SUSI measures each student's own Eircode with Google Maps' shortest route avoiding tolls. Areas within about 3 km of the line could fall either side for an individual student, and are shown separately on the map.
Residual from the county-fixed-effects model with car commute minutes as the friction term, one column per field. Positive (teal) means Maynooth recruits more students in that field from that county than its cohort size, the county's other options and the commute predict; negative (coral) means fewer. Counties are ordered by the Computing residual; ⌂ marks the home county. Treat the exact numbers as indicative, given the pooled counts and the rounding to the nearest five.
Maynooth has something the other campuses do not: a fan of dedicated commuter coaches run by private operators (JJ Kavanagh, Kearns, Streamline, Slevins, Wexford Bus and others) that collect students each morning, run to the North Campus for first lectures, and return in the late afternoon. They reach and account for much of Maynooth's public-transport presence in the midlands and south-east.
These services are already in the model: they sit in the Transport for Ireland national feed and feed the public-transport times throughout this report. Some are near-direct (Athboy 45 min, Dundrum 90); others are long milk-runs that collect across a whole county before turning for Maynooth (Portlaoise and Birr are both over two hours). Either way each route is a single timed departure, so a student catches the morning coach or falls back on a multi-leg rail-and-bus trip; and because the study reports the median over a 60-minute departure window, the modelled time blends the coach with those slower fallbacks. Read the schedule below as the best case a coach town actually has. The cost model credits the coach fare to every area within reach of a stop.
| Route | Corridor | Operator | Runs | Boards | On campus | Run time |
|---|
UM04 (Wexford) and UM15 (Carrick-on-Suir) run only Friday out and Sunday back. They serve students who have relocated to Maynooth for the week, not daily commuters, and are shown dashed on the map. The line between "commutes" and "moves and travels home at weekends" is exactly what the enrolment model cannot yet see (see How to sharpen the answer).
1,197 Electoral Divisions, one dot each. Scroll to zoom, drag to pan; hover a dot for its campuses; click a campus marker to isolate the areas it wins (its modelled catchment). The "most accessible campus" colour is the shortest modelled trip, not where a student would choose to go.
| Campus | AM | Round trip |
|---|
Daily commuter coach (Monday to Friday, one timed morning run) Weekend residential coach (Friday out, Sunday back) Hover a line for the timetable. These services are already in the public-transport times.
Post-primary schools whose Electoral Division puts them within a chosen commute of a chosen HEI. A school inherits its area's population-weighted travel time, so the figure is the commute its typical intake faces, not a door-to-door route for the school building. This is reach, not enrolment: a school being commutable does not mean its students choose that HEI. Every school attribute is from the Department of Education's "Data on Individual Schools" for 2025/26 (roll number, type, gender, ethos, Irish-medium classification, fee-paying status, DEIS status, enrolment by year group), CC BY 4.0. Pick any HEI: the tool works the same way for all of them.
| School | Town / area | Type | Gender | DEIS | Irish | Fee | Coach | Pupils | Min |
|---|
| select a school |
Catchment schools ranked by a simple opportunity score: the school's Leaving Certificate pupils (the two senior years, the students who will apply to college next) weighted by how easy the commute is to the selected HEI, so large senior cycles with a short commute rise to the top. It shows the kind of prompt an institution's recruitment team could draw from the data; it is not a prediction of where students go, and not a plan.
| # | School | Town / area | Leaving Cert | Min | Score |
|---|
Tick "Post-primary schools" on the map above to plot the filtered catchment as diamonds, sized by enrolment and coloured navy, coral for DEIS or sand for fee-paying. With "Maynooth commuter coaches" also on, schools within about 2 km of a daily coach corridor get a violet ring. Clicking a diamond opens the detail panel.