Noise does not stop at the window. What research says about road and air traffic around schools

The lesson runs at its own pace until a lorry pulls away from the lights outside the window or an aircraft passes overhead on approach. The teacher breaks off mid-word, waits two seconds, repeats the last phrase more loudly. A moment later everything is back to normal and nobody records it. But if such events recur dozens of times a day for a whole school year, it is worth asking whether their significance really ends the moment the sound fades. What do we actually know from research about children taught in schools exposed to long-term transport noise?

External noise does not vanish when the door closes

Sound from the street or the sky reaches a classroom by several paths at once. The weakest link is usually the windows: most of the acoustic energy comes through them, and the difference between a closed and a tilted window can be greater than the difference between two schools. The façade itself matters too - its construction and airtightness, the presence of trickle vents, and which side of the building the room is on. A classroom facing a busy road and one facing the courtyard can have entirely different conditions, though only a corridor separates them. Distance from the source, the storey and whether anything between road and building breaks the direct path all play a part.

The distinction we return to throughout this series bears repeating, because here it is crucial. Sound insulation describes how much sound passes through the building fabric - it governs external noise in the room. Acoustic treatment shapes conditions inside the room, above all reverberation. These are two independent parameters and two different sets of solutions. A room can have an exemplary reverberation time and still let in every passing bus - absorbing material on the ceiling does not stop sound coming through a window. It works the other way too: a superbly insulated room with hard surfaces will still be a place where words run together.

There is no single insulation value right for every school - requirements depend on the external noise level, the room’s function and the regulations applying to that building. This is a job for a design, not for a catalogue.

The average does not tell you everything

Exposure to transport noise is usually described by a long-term averaged level - a single number summarising the day, evening or night. It is a useful indicator for planning and for noise maps, but a great deal is lost in it: the number of individual events, their maximum levels, the variability of the signal over time and the intervals between events.

Picture two schools with similar averaged exposure. The first stands on an artery with heavy but fairly even traffic - its environment is more of a steady hum, easier to get used to, but one that constantly raises the background level and erodes the teacher’s advantage over it. The second lies under an approach path: most of the time it is quiet, and every few minutes an aircraft passes that makes conversation impossible for a dozen seconds or so. The same number in the report, two entirely different working days.

It would be an overstatement, though, to say either profile is always worse. How they act depends on the task - a steady hum disturbs silent reading differently from a sudden overflight during dictation - and on the character of the sound and its predictability. That is why, further on, we will see that studies of road and air traffic do not produce identical results.

What the research shows

The first large international study

The RANCH project (Stansfeld et al., 2005). This study still sets the benchmark for the whole field [1]. It covered 2,844 children aged 9-10 from 89 schools near major airports in the Netherlands, Spain and the United Kingdom. Exposure to both aircraft and road noise was assessed, outcomes were measured with standardised cognitive tests and questionnaires, and the analyses adjusted for some of the socio-economic factors that might explain differences between schools.

The findings were not uniform - and that is the most interesting thing about them. For reading comprehension and recognition memory an association with aircraft noise exposure was found: children at more exposed schools scored lower on these tests, and the association remained visible after adjustment for some confounders. The association between exposure and children’s reported annoyance was clear too. But an association was not observed for every function analysed - not every dimension of attention studied, nor every mental-health indicator, was related to noise level. Popular accounts often omit this, and it matters: if noise affected everything equally, the result would be more suspect, not less.

What the study does not allow us to say: RANCH did not “prove” that aircraft cause poorer school results. It was an observational project - children were not randomly assigned to schools with different noise levels, so a school’s location could be linked to other environmental features the model did not capture. The results indicated an association, not a sole cause.

Exposure-effect relationship (Clark et al., 2006). This analysis within the same project asked a more precise question: does higher exposure correspond systematically to poorer reading [2]. It covered 2,010 children aged 9-10 from 89 schools around Schiphol, Barajas and Heathrow, using multilevel models that accounted for pupils being nested within schools. For aircraft noise the association held after adjustment for some variables.

The second finding matters most, though: in this particular analysis no analogous association was found for road traffic noise. This is one reason why it is wrong to write that “transport noise impairs reading” - you have to say which source. The limitations are those of the whole project: an observational design, modelled exposure estimates, and results concerning the areas around three specific airports.

A natural experiment

Munich (Hygge, Evans and Bullinger, 2002). This study has a design nobody could plan [3]. The relocation of Munich’s airport - closing the old one and opening a new one elsewhere - meant that one group of children stopped being exposed to aircraft noise and another started. The researchers studied 326 children with a mean age of about 10.4 in three waves: before the change and twice afterwards, with matched control groups at both airports.

Changes in reading, long- and short-term memory and speech perception were observed. The direction of change at the old airport is especially interesting: once exposure ceased, some scores improved. Reversibility is a stronger argument than co-occurrence alone, because it is harder to explain by fixed features of a district or the social composition of its schools.

This does not, however, make it a randomised experiment. Nobody assigned children to conditions at random; the surroundings changed, not the allocation of participants. Other things may have changed along with the airport - local road traffic, the social composition of districts, the attention paid to schools in the region. A natural experiment strengthens inference but does not remove every possible confounder.

More recent European studies

NORAH (Klatte et al., 2017). This German study covered 1,243 second-grade pupils from 29 schools in the Frankfurt am Main area and matters for one particular reason: the exposure levels in it were lower than in many earlier works [4]. Multilevel modelling was used, and the analysis covered reading, noise annoyance, quality of life and lesson disruption as reported by teachers.

The publication converts the observed effect into months of difference in reading development. That figure needs very careful reading: it is a statistical conversion of an effect size observed on a particular test, referenced to a specific difference in exposure level given in the publication. It does not mean that any particular child “loses two months of learning”, and it describes no biological developmental delay. It is a way of expressing a small statistical difference in units a reader can grasp - and nothing more. We give the exact value of this conversion and the corresponding difference in exposure only after verification in the full text, which is why we do not quote them here.

As in the earlier studies, not all the outcomes analysed were associated with exposure - the picture varied by the dimension measured. A systematic limitation: the children’s exposure at home and at school may have been related, which we discuss in a separate section below.

Barcelona (Foraster et al., 2022). This cohort study covered 2,680 children aged 7-10 from 38 schools and followed them for about twelve months, with repeated measurements of cognitive function [5]. Road traffic noise was assessed both outside the school and inside classrooms, and the analysis covered not only the mean exposure level but also its variability - an important methodological contribution, because it distinguishes steady hum from an environment with pronounced peaks. The outcomes studied were working memory, complex working memory and inattentiveness.

Three caveats are needed. It was an observational design, so a statistical association does not identify a sole cause. The “inattentiveness” measured is not a diagnosis of ADHD - it is the score on a behavioural scale, not a clinical diagnosis. And finally: the results concern the specific urban conditions of Barcelona, with its particular building density, traffic structure and type of school buildings.

When the single event is what counts

London (Shield and Dockrell, 2008). This study covered children aged about 7 and 11 in London primary schools and set standardised test results against measurements of external noise and of noise inside classrooms, adjusting for socio-economic factors [6]. It matters in our set because it separates two things other studies often combine: the average acoustic background and individual external events. The findings suggested that not only averaged levels but also individual loud external events may matter - and that is a different problem, with a different solution, from lowering the background.

Limitations: we do not quote the exact numbers of schools and pupils or the specific noise levels from this publication. The levels it derives are also not universal safety thresholds or legal limits - they are the results of statistical analysis in a particular sample, not normative values.

Noise or air pollution?

Secondary analysis of the British sample (Clark et al., 2012). This work touches on a problem that either undermines or reinforces all the earlier results [7]. Schools on busy roads experience noise, air pollution and a whole set of other urban features at once - dense built-up surroundings, less greenery, a particular social structure. Since these factors occur together, an observed association cannot automatically be attributed to any of them alone.

The authors re-analysed the British RANCH sample: 719 children aged 9-10 from 22 schools around Heathrow, including nitrogen dioxide in the models as an indicator of traffic-related air pollution. The result should be stated carefully: in this sample, at the moderate pollution levels present in it, adjusting for that factor did not explain away the previously observed associations between aircraft noise and some cognitive outcomes. The authors also stressed the need for studies in places with higher pollution levels, where the picture may look different.

So one study does not show that air pollution does not matter - it shows that in this particular sample it did not explain the effect attributed to noise. That is an important difference.

The scale of the phenomenon in Europe

European Environment Agency estimates. The report Environmental noise in Europe and the accompanying EEA materials make it possible to gauge the scale of the problem across the continent [8]. The agency estimates the population burden associated with environmental transport noise - including reading difficulties in children, behavioural problems and the number of children living in areas of elevated exposure.

How these numbers are produced needs explaining, because it is often misunderstood. They are modelled population burdens, calculated from exposure data (noise maps and population statistics) and adopted risk relationships taken from epidemiological studies. This is not a count of individually diagnosed children or the result of any screening - nobody examined these children individually. The estimate says how many cases would be expected in a population given a certain distribution of exposure, not in whom they occurred.

Why findings for roads and aircraft are not always identical

The divergence visible in the studies discussed is neither an accident nor a measurement error. The two sources have different characteristics.

Road noise tends to be more continuous and depends strongly on the character of the street: an artery with an even flow of vehicles sounds different from a road with traffic lights where vehicles pull away and brake. It can have pronounced peaks - a lorry, a motorcycle, a siren - and often co-occurs with air pollution, which complicates interpretation.

Aircraft noise is usually event-based: it comes in series, with a large difference between background and overflight level, and can make communication impossible for a dozen seconds or so. Its distribution over time depends on the approach direction and the time of day, so it can be predictable in a way road traffic is not.

These descriptions are not rules - there are roads with a highly event-based profile and airports with almost continuous traffic. The point is different: since the sources differ in how they unfold over time, there is no reason to expect every study to produce identical results for them. The analysis by Clark and colleagues, in which the association held for aircraft noise and was absent for road noise [2], is a good example.

When the teacher has to break off mid-sentence

How would street noise translate into a reading test score? The most plausible path is surprisingly prosaic and runs through the lesson itself. External sound masks speech - it overlaps the teacher’s voice and takes away parts of the utterance, especially the quiet ones: endings, short words, numbers. The teacher repeats a phrase, raises their voice or waits for the noise to pass. The lesson stops for a few seconds, the thread is broken, and the class’s attention has to be brought back to the task. A single event costs little. Dozens of such events a day, over a year, add up to a total invisible at any single moment.

This is, however, a hypothetical mechanism, not an established fact. Not all its links have been measured directly: studies more often record exposure and the final outcome than what happens in between. Where studies do describe lesson disruption - as in NORAH [4] - the information usually comes from teachers’ reports rather than an objective count of every interruption. Valuable data, but of a different kind from measurement. We write more fully about how much attention working in noise costs in the article on attention and working memory.

Noise at home and at school can overlap

Children taught at a school beside a busy road often live in the same area. Transport exposure therefore does not end with the bell - it covers home as well, and there it also affects rest and sleep. This creates a methodological problem: in some studies exposure at home and at school were strongly correlated, which makes it hard to establish how much of the observed effect really arose during school hours and how much outside them. NORAH [4] is a good example - an analysis covering both places shows that separating these influences is not simple.

We do not pursue the question of sleep here, as it deserves a separate study. The practical implication is worth remembering, though: when a study speaks of “noise at school”, it is worth checking whether and how it accounted for exposure at home.

What a school can do, since it cannot move the road

The answer depends on the diagnosis, not on a catalogue of solutions - and that is the most important sentence here. Areas worth examining include the distribution of rooms in the building (are the lessons that demand most concentration held on the most exposed side?), how windows are used during the day, ventilation, the insulation of the façade and windows, the airtightness of doors, the presence of buffer spaces between classrooms and the most exposed façade, the siting of playing fields and recreation areas, the treatment of the surroundings, and the overall arrangement of functions in the building.

One trap needs a clear warning. Closing the windows is not a solution in itself. Windows in schools are opened for ventilation, temperature and air quality, and deciding to keep them shut without providing an alternative means of air exchange solves one environmental problem while creating another - sometimes a more serious one. Acoustic and ventilation solutions have to be planned together, not one against the other.

What the research shows and what we still do not know

The overwhelming majority of the works discussed are observational, for an obvious reason: children are not and cannot be randomly assigned to schools with different noise levels. A school’s location comes with a whole set of other features - the social structure of the district, the age and type of the building, access to greenery, air quality. Socio-economic status is the most discussed confounder here and most studies try to adjust for it, but none does so completely. Air pollution co-occurs with road traffic, and the analysis by Clark and colleagues [7] settles that question only for one sample and moderate levels. Exposure at home and at school is often correlated.

Methodological differences come on top. Exposure is not modelled identically across studies, and transport sources have different characteristics, so comparing results calls for caution. Some older works were carried out in different building realities - with different windows and different façade insulation from today’s buildings.

And two things that matter most. First, a population-level association does not predict an individual pupil’s result: the fact that average scores tend to be lower in more exposed schools says nothing about any one child. Second, not all the cognitive functions analysed show an association with noise, and the findings for road traffic noise are markedly less consistent across studies than simple media coverage suggests. An honest account of this literature has to show the null results too - otherwise it describes not the state of knowledge but a slice of it selected to fit a thesis.

Summary

Let us return to the room from the start and the lorry outside the window. A single pass lasts a few seconds and does not matter - research on transport noise is not about single events but about repeated, long-term exposure lasting whole school years. That is a fundamental difference and worth preserving in conversation about your own school: one loud lesson proves nothing, while a daily situation over nine months of the year is a different phenomenon.

It does not follow that a building on a busy street is an unsuitable place for education. It follows that its acoustic environment has to be known rather than assumed - and that solutions inside the room do not replace analysis of what happens outside. Conditions for learning are not created only between four classroom walls. They are also created by the building’s surroundings and by how much of their sound reaches the place where a teacher is trying to convey information. On what can be changed inside and how to check whether the change worked, we write in the article on the effectiveness of acoustic treatment.

References

1. Stansfeld S.A., Berglund B., Clark C., Lopez-Barrio I., Fischer P., Öhrström E., Haines M.M., Head J., Hygge S., van Kamp I., Berry B.F., Aircraft and Road Traffic Noise and Children's Cognition and Health: A Cross-National Study, „The Lancet” 2005; 365(9475): 1942-1949. DOI: 10.1016/S0140-6736(05)66660-3; PMID: 15936421.
2. Clark C., Martin R., van Kempen E., Alfred T., Head J., Davies H.W., Haines M.M., Lopez Barrio I., Matheson M., Stansfeld S.A., Exposure-Effect Relations Between Aircraft and Road Traffic Noise Exposure at School and Reading Comprehension: The RANCH Project, „American Journal of Epidemiology” 2006; 163(1): 27-37. DOI: 10.1093/aje/kwj001; PMID: 16306314.
3. Hygge S., Evans G.W., Bullinger M., A Prospective Study of Some Effects of Aircraft Noise on Cognitive Performance in Schoolchildren, „Psychological Science” 2002; 13(5): 469-474. DOI: 10.1111/1467-9280.00483.
4. Klatte M., Spilski J., Mayerl J., Möhler U., Lachmann T., Bergström K., Effects of Aircraft Noise on Reading and Quality of Life in Primary School Children in Germany: Results From the NORAH Study, „Environment and Behavior” 2017; 49(4): 390-424. DOI: 10.1177/0013916516642580.
5. Foraster M., Esnaola M., López-Vicente M., Rivas I., Álvarez-Pedrerol M., Persavento C., Sebastian-Galles N., Pujol J., Dadvand P., Sunyer J., Exposure to Road Traffic Noise and Cognitive Development in Schoolchildren in Barcelona, Spain: A Population-Based Cohort Study, „PLOS Medicine” 2022; 19(6): e1004001. DOI: 10.1371/journal.pmed.1004001; PMID: 35653430.
6. Shield B.M., Dockrell J.E., The Effects of Environmental and Classroom Noise on the Academic Attainments of Primary School Children, „The Journal of the Acoustical Society of America” 2008; 123(1): 133-144. DOI: 10.1121/1.2812596; PMID: 18177145.
7. Clark C., Crombie R., Head J., van Kamp I., van Kempen E., Stansfeld S.A., Does Traffic-Related Air Pollution Explain Associations of Aircraft and Road Traffic Noise Exposure on Children's Health and Cognition? A Secondary Analysis of the United Kingdom Sample From the RANCH Project, „American Journal of Epidemiology” 2012; 176(4): 327-337. DOI: 10.1093/aje/kws012; PMID: 22842719.
8. European Environment Agency, Environmental Noise in Europe - report and briefing on the effects of environmental noise on children’s reading and behaviour. Oficjalny serwis EEA.

This article is for information only. It does not replace an acoustic measurement, a site assessment or a building refurbishment design.