What is the effect of reduced street lighting on crime and road traffic injuries at night? A mixed-methods study

Ethnographic data

Within each area, we interviewed key stakeholders (local authority lighting professionals and councillors); collected documentary evidence (including local authority plans, blogs, e-mails and letters to residents’ associations, local newspapers and local authorities); reviewed local authority consultations (if available); and conducted focused ethnographic visits. The data from these visits included field notes from ‘walk-arounds’ of areas with street lighting reductions, including informal intercept interviews and in-depth individual and group interviews (which were recorded and transcribed) with a mix of residents, visitors and workers. A topic guide for the formal interviews included prompts on whether or not changes had been noticed; what impact they had on mobility, sleep, security and feelings about their neighbourhood; what they did differently after lights had changed; and whether or not they had taken part in any formal consultations. Participants were asked to also complete a brief questionnaire covering demographic information [age range, gender, (last) occupation and ethnicity]. Group interviews were all with natural groups of participants who knew each other already. Intercept interviewees were told we were researching street lighting, and asked if they had noticed changes in their area and what they thought about those changes, if noticed. Fieldwork was conducted between April 2013 and December 2014.

Household survey

The aim of the household survey was to estimate the prevalence of reported negative and positive well-being impacts of reduced street lighting and identify whether or not reduced street lighting did have an impact on well-being in core domains identified in ethnographic work. We identified one area in Shropshire where lighting reductions at night had been introduced in selected streets and were scheduled for other streets. Using data provided by the local authority on implemented and planned lighting changes, the roads were divided into 12 strata based on whether or not lighting reductions had been introduced (yes/no), tertile (low/medium/high) of deprivation of the census lower super output areas (LSOAs) in which the roads were located, and proportion of postcodes within the LSOAs that were pubs, restaurants, shops or other businesses (higher/lower than 2%, the median proportion). We then randomly selected roads within each of the 12 strata and obtained the addresses for every residence on the selected roads to achieve a total sample of 500 houses in streets where lighting had been reduced at night and 500 in streets unaffected to date. Even if only half of households responded to the survey, the study would have 90% power to detect a 10% absolute difference between affected and unaffected streets in prevalence of well-being impacts at the 5% significance level. A self-completed questionnaire was designed to assess feelings about neighbourhood and local authorities, safety, fear of crime, impact of street lighting on sleep and the importance of seeing the night sky (see Appendix 1). After piloting, this was mailed to residents, with a choice of return by enclosed stamped addressed envelope or online, followed up by one reminder. A total 483 responses (476 postal and 7 online) were received, comprising 250 (i.e. 50%) responses from the part-night lighting group and 233 (47%) from the unaffected streets. There were no significant differences in the distributions of respondents from the affected and unaffected streets according to gender, age group or the number of adults in the household.

Analysis

Qualitative data (interviews transcripts, field notes, documentary data and open comments on the questionnaire) were analysed using thematic content analysis,22 with a coding frame drawing on both the literature and inductive coding of early data. The project team met to open code an initial transcript and generate a draft coding frame (see Appendix 2). This was applied to early data, with emerging analysis used to identify further sampling.

Data from the survey questionnaire were entered into a database and each record was then checked for accuracy. We created binary variables for each questionnaire item using the extreme of the response options (e.g. when asking about personal safety we dichotomised using the response ‘very safe’; when asking ‘how often do you walk alone in your neighbourhood at night’ we dichotomised using the response ‘at least once a week’, etc.). The difference in the prevalence of each negative or positive well-being impact between respondents in affected and unaffected streets was estimated using odds ratios with 95% confidence intervals (CIs). The significance of any differences in prevalence was tested using a chi-squared test. Results were compared with national surveys using similar indicators to estimate representativeness. Data were analysed using Stata 13 statistical software (StataCorp, College Station, TX, USA).

Ethics

In quoted extracts, we have removed identifying place and other information, with tags indicating local authority area (except for stakeholder interviews, where this might breach confidentiality) and source of data: in-depth individual or pair interview (I), group interview (G), stakeholder interview (S), document (D) or field notes, including informal intercept interviews (F). Written consent was obtained for transcribed in-depth, group and stakeholder interviews; intercept interviewees were told that we were ‘conducting research on street lighting’. Approval for the study was provided by the London School of Hygiene & Tropical Medicine Ethics Committee (number 6341).

Public views: strong and polarised

Unsurprisingly, views expressed about switch-off or part-night lighting in public domains such as local authority consultations and newspaper letters pages tended to be strong and polarised, with more people expressing strong views on the negative impacts on well-being than giving positive views of street light reductions. Key domains of concern on the potential or experienced negative impacts were impact on personal security when walking at night, fear of crime and potential for increased road traffic injuries. Letters to one local newspaper in Hertfordshire, a county that had introduced part-night lighting, are typical of both the polarisation and the strength of the individual views expressed in public domains:

. . . this is great for the criminal fraternity that prowl our streets . . .

Letter to The Comet, 16 August 2012, p. 19, D923

I am a serving police office and work long shifts . . . I am not impressed why, after I pay my taxes, that I now have to walk home in pitch black.

Letter to The Comet, 24 May 2012, p. 18, D1424

The great switch-off has begun [. . .] Never have I felt so frightened to walk back to my house from the road after a night out.

Letter to The Comet, 17 May 2012, p. 19, D2325

Am I alone in thinking it’s a fantastic idea? . . . the money saved . . . can be ploughed back into schools and lollipop ladies . . . Sleeping in the pitch black will have amazing health benefits for the people of this town. Not to mention the impact on the environment.

Letter to The Comet, 10 May 2012, p. 18, D2726

Local authorities that had engaged in consultations over planned reductions solicited comments from both representative groups (parish councils, residents associations), and individual members of the public. One phase of Buckingham County Council’s consultation over ‘switching off’ lighting, for instance, attracted 34 comments from individual residents, of whom 24 protested the decision, and eight supported it, with two making other or mixed comments. In addition to noting details of particular roads or junctions of concern, negative comments also questioned likely cost or carbon emission savings and focused on safety and the competence of the authority to make decisions:

This is a regressive plan that will make roads more dangerous. Endangering lives to save money is gross mismanagement.

Anonymous comment, public consultation, Buckinghamshire D104

Positive impacts reported or expected in public comments were benefits in sleep with less light pollution, impact on environment through reductions in carbon emissions and improved ability to see the night sky. Both the polarisation of views and the preponderance of negative views were reflected in open comments on the household survey. Around half of all respondents added comments: a higher number of these expressed negative views of street light reductions than gave positive views (see Appendix 3).

Negotiating views: deliberative settings

If views expressed in public domains suggest strongly held public opinion, a somewhat different tenor was evident in more deliberative settings. When asked about street lighting reductions in contexts such as group interviews, participants were typically more equivocal or hesitant in offering their assessments and more willing to consider potential benefits as well as risks. For instance, despite some in this group of neighbours being concerned about the dark streets at night, they all went on to agree on the benefits of being able to see the night sky:

I really notice being able to see the stars.

Yes, yes, I have, too.

And one night I even . . .

It’s lovely.

I woke up because we had our curtains open, and I woke up and it was really bright outside . . . It was just so beautiful.

G3 Hertfordshire

In groups, participants were also more likely to explicitly note that there might be trade-offs between different outcomes likely to accrue from reductions in street lighting and that these would impact differently across their communities. Here, for instance, after rehearsing the negative impacts, particularly (they felt) for older residents, these members of a residents’ association then went on to discuss potential benefits of changes:

Uh, it’s, there’s a balance to strike, isn’t there? . . .

[indicates agreement]

I’m a reasonable man but I know if there was no cost to it I, I’d probably rather have more lights than less. And I find it quite hard. I probably haven’t got very good night vision . . .

I do feel slightly better for thinking, well, we’re doing our bit towards cutting down carbon.

G5 Buckinghamshire

Similarly, this group of teachers in Wakefield debated the likely impact of reductions in street lighting in their neighbourhood, drawing on news coverage, their own experience and anecdotes from others to offer and test out rather more tentative views than those expressed in public consultations:

And they’re, they are actually energy saving . . . there were a big piece on [local TV news programme] or something about them . . .

You wouldn’t want them to just totally switch things off. I guess if they did it at a certain time, you know, when the majority of people aren’t on the roads?

I don’t know, because it’s, on estates, because I know what [colleague] was saying, since she got burgled, because it’s so well lit where she is. She says, the, the policeman said if, if it hadn’t been able as well lit and they hadn’t been able to see into your garden, your house, your garage.

I suppose you’re never going to please everybody are you?

G9 Wakefield

Despite one group interview participant suggesting ‘I think people seem to be either very for lighting or very against lighting’ (G4 Buckinghamshire), in deliberative settings, the tenor of most discussion was that there were likely to be both positive and negative health and well-being outcomes and that the balance of these would depend on individual circumstances.

Accounting for well-being consequences: social attributes

In accounting for who would be most affected by the potential consequences of street lighting reductions on safety and mobility, participants drew on normative assumptions about how social attributes such as age and gender affected mobility, safety and risk in public spaces after dark. Older adults, for instance, often cited their age as a reason for not noticing or being affected by street lighting, as they reported rarely going out after dark. Younger adults, even when negative about reduced street lighting, couched their concerns in terms not of impacts on their own mobility, but the potential for harm to older neighbours. One couple in their thirties, for instance, recalled moving from an urban area to a commuter town which had part-night lighting and their initial surprise and discomfort on realising that their local streets were dark by the time they returned from a night out:

. . . we might be coming home at gone midnight . . . once you’ve turned off [main road] you would yeah, literally around 12 o’clock, you would watch lights going out in front of you as you walked down the road . . . it would be pretty pitch black.

. . . no street lights at all!

I7 Hertfordshire

However, despite reportedly having to take care not to trip over unseen hazards and having to walk in the road to avoid bumping into cars parked on the pavement in the narrow streets late at night, they deflected any suggestion that the 12 a.m. switch-off had constrained their activities or particularly affected their well-being. Indeed, despite a long list of inconveniences and, at times, dangerous situations (including nearly being run over on an early morning walk), they cited their (young) age to deflect any question of significant impact on their routines:

Does the lack of street lighting affect what you choose to do?

I wouldn’t say it bothers me . . . if I want to go out, I’ll go out . . . but I’m not a 70-year-old person . . .

. . . it doesn’t impact on anything as in we’d not do something because of the lighting.

I7 Hertfordshire

Symmetrically, older participants disavowed any impact on their own behaviour, in their case because they were at less risk than younger people, being less likely to be out and about later, but also because they described themselves as more likely to be adequately prepared to cope with navigating dark streets:

I think it needs to be said that, um, someone spoke to me about their teenage children coming back from London, nights out – And they were concerned [about] picking up the, the children coming home from the station and it was dark. Now this has been said to me.

Well, I’ve walked up from the bottom in the dark when it’s been the blackout and I felt perfectly safe . . . I just walk in the middle of the road . . . I found it quite surprising the first time I was coming back in the dark, but actually I’ve, oh, because I’m, I’m fairly sober and not wearing silly shoes.

G3 Hertfordshire

Thus, in public, participants were often careful to claim that, even where lighting reductions were unwelcome, that they themselves could ‘cope’ and that their own competence would offset any impacts on well-being. In a group interview with a group of neighbours who were all deeply opposed to the local switch-off at 12 a.m., one woman, for instance, describes how she has ‘taught’ younger workers how to cope with darkness:

I gave two of my apprentices torches . . . Because they were complaining about coming home drunk from [town] and couldn’t walk and find their way. So I gave them torches [brief laugh] . . . and I gave them a torch and was like put that in your handbag because it’s, they’re really light . . . I mean they thought it was quite funny because they’re about 18 and 19, but at the same time they were like we really can’t see.

G2 Hertfordshire

Similarly, this young man describes in detail the importance of street lighting for security in his neighbourhood (‘if it’s more, it’s more dim, it’s more dark, it’s easier for crime to take place’) but again claims competence in the form of both maturity (‘I’m getting to the stage now where that sort of thing shouldn’t bother me’) and local familiarity; others, in contrast, he suggests, might be more vulnerable:

Personally I don’t feel unsafe around here because this is my neighbourhood to be honest, I’m quite familiar around everyone and the surroundings, however that doesn’t mean everyone who lives here or lives around here or comes here feels safe . . . street lighting . . . it’s a mental thing but I just think it gives people a kind of a reassurance that . . . nobody wouldn’t try anything because they’ve got light.

I64 Hackney

Concerns were, then, typically rationalised as concerns for others: of different age groups or those lacking local familiarity. The exception was female gender, which was on occasion used in discussion to justify concern participants’ own concerns. Many women of all ages reported that personal security was an issue for them, or at least that they could flag up concerns ‘as women’ about security. Some did report avoiding travelling after dark on their own. However, the role of street lighting per se, rather than street lighting as an indicator of time of day, was ambiguous. Two railway workers, for instance, first suggested lack of street lighting early in the morning when they travel to work as an issue, particularly for women, but then went on to suggest that the problem is the time of night, rather than the darkness itself, being inherently dangerous:

I take a taxi. As a young woman I wouldn’t want to be walking out at that time in the morning.

Does it make a difference, no street lights?

No, I’d get a taxi anyway. It does put you off, not knowing who’s around, especially as it’s dark. But I’d come by taxi anyway – 10 years ago, would have been different – you don’t think about it when you’re young do you? But you know, as a woman, you don’t want to be out on your own walking around.

No, not as a woman. Mind you, same for a lot of men – wouldn’t want to be walking, you never know who’s out and about.

F3 Shropshire

Direct reports of reduced lighting affecting mobility were rare and reported impacts on well-being were largely marginal. They included reports from a few people who had changed their behaviour to avoid walking, running or cycling in the dark:

What it has changed for me is, because I’m a runner, I run, I can’t run early in the mornings like I used to. Or not in the winter anyway. It’s too dangerous, with all the narrow roads and that. I have to drive somewhere now to go for a run.

F33 Swansea

However, as for this woman, the impacts of lighting reductions had, in general, entailed manageable changes to routines rather than the abandonment of activities that would have positive impacts on well-being. Strategies included taking a torch when walking the dog at night, or using main roads for as long as possible when walking home rather than taking short cuts through unlit side roads.

Lack of opinion

Expressed public views were, then, strong and polarised, and those in more deliberative settings were more considered. However, this suggests a level of engagement with the topic of street lighting that was not evident in the walk-arounds of affected areas. Most intercept interviews during fieldwork suggested that, for the majority in affected areas, changes in street lighting had had little impact or had not been noticed. Changes such as white light and dimming had gone largely unnoticed, and even residents in streets with lights switched off typically said they did not know if there had been any changes, or had not noticed, or that it would not affect them as they were rarely out after 12 a.m. A group of customers and barman in a local pub in a town in Shropshire, for instance, were typical in that although they reported going home after 12 a.m. (the time of local switch-off), they were unsure if their streets were affected, or, if aware, were unconcerned about the implications for their own mobility or safety:

I don’t notice to be honest.

That’s because you’re pissed! [All laugh]

They’re all off in my area – I think they go off at 2 or something.

People are leaving about 2, 2:30 – we kick them out then and, to be honest, most of them are getting taxis home anyways. It’s just one of them things, street lights, doesn’t really make much difference . . . for most people, no one’s going to notice – in my street, it’s only really me coming back past midnight, no one else.

F4 Shropshire

Indeed, the most common response to general enquiries such as ‘have you noticed any changes in street lighting locally?’ was ‘To be honest, I haven’t’. For those who had noticed change, this was often a vague ‘the colour may be different’, and few had concerns about the impact on their own well-being:

I don’t notice really, as I don’t go out after dark.

F22 Swansea

[T]hey’ve changed the orange lights – they’re not as bright. To be honest I haven’t got an opinion, you just get used to it.

F22 Swansea

It’s not an issue, most of us drive, or get a taxi to [town], or someone’s always driving that way and you can get a lift.

F34 Hertfordshire

Few intercept interviewees mentioned negative well-being impacts and, where street lighting reductions had been noticed, the most common responses were spontaneous comments related to seeing the night sky. Although this was not often an issue in more suburban areas (where light pollution from nearby towns and cities was too great to see stars anyway), it was more frequently raised in the more rural neighbourhoods. One young man, who had moved to a rural area relatively recently, described walking home at 12 a.m. the night before: ‘there was no sun, no moon – but I could still see the way; no torch – just starlight – you get amazing skies here’ (F31 Swansea). Another in the same area had noticed changes: ‘They’ve changed them to what do you call it, low density lights – good for the stargazing – now you can see the sky, see whole . . . you’ve got the whole world in front of you!’ (F23 Swansea). In urban areas with white light interventions, if people had noticed this or had a view about it, this was more likely to relate to light pollution and the impact on sleep from existing lights:

Have Southwark reduced street lighting?

It’s definitely dimmer down the other end of [X] road.

. . . maybe only the quieter roads. Is it only on minor roads? This one outside [light in front of house] is still really bright – we worried about that when we moved in.

Well yes, but it has been OK, we haven’t really noticed it have we?

F1 Southwark

The lack of concern reported in intercept interviews was echoed in responses to invitations to take part in more in-depth interviews, with many expressing surprise that this was a topic for research. Even those who went on to make thoughtful and considered comments about the possible impacts of street light reductions on well-being were often initially unsure if they would have any views to share. One young woman who went on to provide a detailed account of how lighting in her neighbourhood affected how, when and who walked around it commented: ‘I’m not going to lie, when you first suggested it to me, I was like, eh, street lights? That’s a random topic’ (I60 Hackney). It was also reflected by some local authority consultations; one, over a proposed introduction of a white light scheme, was described by the lighting engineer as having gone ‘down like a lead balloon – we didn’t get any response at all’ (S2). It may, then, be reasonable to assume that, for the majority of those in affected areas, there was little discernible impact on well-being and few concerns about potential effects on well-being from changes made.

We deliberately included those likely to be out after dark or before sunrise, including workers such as taxi drivers, hospitality workers and police officers. In general, even those who did report working in or leaving and returning home in dark streets did not report significant impacts from street lighting reduction: largely, they reported using cars to get to and from work and the only impacts of dark streets were marginal, in that taxi drivers, for instance, reported that it could be difficult to see house numbers. The exception was police officers, some of whom did report very strong negative views about the impact of street lighting switch off on their work. Three police officers included in the fieldwork all agreed that switching off street lights had an impact on public perceptions of fear of crime, even if there was no evidence that it had an impact on crime levels. Two had particularly strong views that lights should not be switched off, particularly on housing estates, as dark streets both fostered crime and made it more difficult for them to respond appropriately, given the added time it could take to find addresses in the dark and also to see the perpetrators of low-level crime: ‘we go out and we literally can’t see them – we can hear them making a noise, but you can’t see who it is’ (S10).

Private concerns

The typical lack of engagement encountered in intercept interviews might suggest that characterising public concern by the expressed views of those who do write into local authority consultation risks exaggerating the strength and negativity of views. However, there were also some deeply held concerns that tended not to be expressed in public. In public, accounts of the impact of street lighting reductions on personal security or mobility were circumscribed by age and gender appropriate presentations of the self, with participants reluctant to admit that fear of the dark, for instance, would curtail their activities. Well-being impacts of reduced street lighting were, therefore, largely presented as rational concerns about road safety or crime and its impact on more vulnerable others. In more private settings, though, a rather different tone was sometimes apparent, with interviewees expressing more deep-seated concerns about impacts on well-being of both darkness itself and, more specifically, the ‘switching off’ of lights. These more private concerns were elicited in a number of contexts: typically, at the end of group discussions when participants asked to speak privately to the team, and in in-depth individual interviews. Here, participants were more likely to reflect on the meaning of dark nights and express more uncertainty and anxiety about the implications of reduced street lighting, not just for their own individual well-being, but also for the social well-being of their neighbourhoods:

[after the public meeting] a couple of women are keen to talk more privately [. . .] One says when she first got here ‘I thought there had been a power cut! I feel quite vulnerable – I won’t wait for a bus, because I don’t like not being seen. In winter, it does stop me going out – I struggle even to put the bins out’ [. . .] Another woman then tells a story of a lively village visited on holiday and [concludes] [. . .] ‘ Here, it is as dead as a doornail, because no one ventures out’.

Field notes, Hertfordshire, December 2013

Three themes characterised these more private views: fear of the dark, wonder at the lights going out and concern about ‘going backwards’. Fear of the dark is perhaps difficult to admit to in public, given the need to present oneself as a mature, coping adult, but in more private settings participants were more willing to speak about how darkness affected their mobility, often prefaced by an apologetic ‘I don’t know if this is just me. . .’. This young woman, for instance, talked about visiting a family member who lived in the suburbs, where ‘come 8 o’clock all the street lights are knocked off’, which affected her willingness to walk there:

It’s quiet, like there’s no shops around there. Like come I think 10 o’clock it’s just dead, you need a car basically to move around there. So yeah, and it’s really scary, like you have to take, when I’m going there, say if I go there late at night I have to take a cab there, I wouldn’t walk there.

I60 Hackney

Women in particular described why they feel unsafe walking at night in unlit streets:

I don’t like to walk in the dark. I could carry a torch – but that shows you up. If there’s a street light, people can say [if something happened] ‘I saw a lady walk past us’. If I’m carrying a torch, all they can see is the torch.

I17 Hertfordshire

It [street lighting] matters hugely – you obviously don’t feel so safe when its dark, you can’t see who is there, and it is probably imagination, but you imagine someone there in the shadows.

F20 Westminster

I worry as a dog walker – I do go out at night. Your eyes don’t get their night vision, because there are these pools of light, and the pavements are very uneven and dangerous – and that’s all apart from the worry about personal safety. My mother won’t go out after dark, she’s afraid of falling.

I29 Buckinghamshire

One woman, who described a recent emergency at night that necessitated driving, discussed how she now preferred living in a relatively well-lit town:

I used to live out in [small village], it’s very rural there, it’s completely different. When I was there I didn’t go out as much at night, I didn’t feel as secure. Here I go out to the theatre, cinema, I feel safe . . . in [village] the lighting wasn’t very good – it was well spaced out, and they didn’t light the whole lane . . . you feel much safer in the town. You do think about that more as you get older.

I20 Wakefield

If the dark was, for some, inherently disturbing and a factor that might shape preferences for more urban settings, for others it was the switch-off itself which was unsettling in some way. First, the instant change from lit streets to dark ones evoked a range of emotions in those living in areas where local authorities had instigated part-night lighting. This included wonder, again something participants often flagged as perhaps an individual, or idiosyncratic, reaction:

I know this sounds weird, but I got quite obsessed by it when they first started turning them off at midnight, I started to stay up to watch them going out – just to see how dark it was. It was so strange watching the lights going out – an odd thing to happen.

I37 Hertfordshire

For others, the anomaly of lights going out at a certain time was simply strange, and was often described as a largely humorous disruption in terms of expectations one might have of either modernity or what a walk home should entail:

They go off at midnight – it’s hilarious, it’s like going back to the Dark Ages!

F34 Hertfordshire

And literally it was quite funny, because there were some people walking ahead of us, and the light was literally turning off as they got to it . . . and we thought, ha, ha, that’s quite funny!

I7 Hertfordshire

The concern, it is implied, is not the dark per se, but the surprise of the dark, something which, in urban areas, was out of time and out of place. If some described this in humorous terms, others were more disturbed. Private accounts included some reflection on the meaning of this switch-off and the symbolic breach in expectations of modernity that it heralded:

The streetlight thing seemed to me a big step forward in quality of life . . . the thought of actually going backwards seems to be quite appalling. I’ve never lived in a non-suburban or non-town area my entire life. Um, and I think, and I would choose not, I, you know, I would actively choose not to live in the country . . . I would prefer to be somewhere where there are lights.

I2 Buckinghamshire

As this response suggests, location shaped the meaning of light at night. If those who were out and about at night in built-up residential neighbourhoods were made anxious by the switch-off, those in more rural areas presented themselves as ‘hardier’ to the removal of artificial light; for many, living in a rural area meant taking a pride in coping with the absence of such indicators of modernity as street lighting:

It wouldn’t bother me to go out at night. Since I was young I’ve done that, I’ve grown up in the country.

G5 Hertfordshire

Indeed, lighting was an important marker of the division between ‘town’ and ‘country’. First, ‘bright lights’ signify ‘the city’:

I’m always relieved when I’ve been driving along country roads to get back to the bright lights of the city.

F19 Buckinghamshire

Second, lighting operates as a marker of normative expectations attributed to residents of town and country and, thus, as the second quote suggests, operates as a potential political faultline in localities which bridge the urban and rural in commuter villages:

If you’re used to being in [city] with loads of streetlights, it’s kind of weird getting here and finding it in darkness.

I7 Hertfordshire

The city people want street lights – there is a new-build estate, and its second generation people coming from London, with different expectations – they want them. It is a divide – people do talk about it as something people come to the country and then expect it to be like the city.

F38 Buckinghamshire

Light demarcates time and place

Key stakeholder interviews with lighting professionals also focused on place and time. Lighting was explicitly used to signal appropriate use of public space in time, and reductions in hours of street lighting were often justified in terms of normative accounts of the ‘proper’ hours for doing things, as suggested by this parish councillor’s discussion of a plan for 12 a.m. switch-off, which he described as having been pushed through despite some residents’ disapproval:

Between the hours of midnight and 5.30 what is anyone doing anyway? Because people are asleep at that time. The only time I’ve been out at that time is Christmas Eve, coming back from church. And that was about 12.30 and I did think, oh, it’s dark!

S9

Outside urban centres, being out and about after dark still risks raising questions about propriety. Deliberately leaving places dark could, therefore, demarcate places where people were not ‘meant’ to be: a reading understood by both lighting professionals and residents:

We do switch down park lights . . . because we don’t want to encourage people into the parks . . .

S2

We’re going into a park or something, a park after dark like to just go and mess about and chill and whatever. That’s the only time we use our phones as torches because there’s no lighting in parks, people assume that nobody’s in the park at that time, so there’s obviously going to be no lighting . . .

I59 Hackney

Lighting engineers discussed in detail how ‘the right light, in the right place, at the right time’ (S1) – a mantra of their professional aims, repeated during the fieldwork many times – was essential not only to the security of the communities they served, but also to creating appropriate lighting, with cues to the proper use of space as well as time. In the city centres, this often implied a welcoming light, which encouraged residents and visitors to use services and facilities. As the introduction to Westminster’s guidance on lighting puts it: ‘Light changes everything: it illuminates our way and makes spaces inviting to enter’ (City of Westminster, undated, D45). For many, lighting was not just about making places more visible, but marking them as identifiable locales which enhanced community well-being, through selecting the right quality, intensity, direction and colour of public lighting. Commercial interests play a large role in this, with a rising number of companies now vying for local authority contracts to provide lighting. In Wakefield, for instance, in an advertising brochure Philips Lighting describe their brief as not only to help the council to save money but to ‘provide a welcoming environment’ through the use of LEDs. The use of light to create visual coherence has been to extent ‘corporatised’ in many high-income countries (see e.g. Jones27 on London’s South Bank and lighting), but for many local authorities in England and Wales, there was a still a keen sense that good lighting was a duty of local government, part of the obligation of public servants to the well-being of local communities. One lighting professional is particularly eloquent on the wide-ranging impacts that good-quality lighting can have across the range of environmental, economic and social aspects of localities:

A lot, a lot safer to do it, you know, even, even simple things as you, yourself and a neighbour going to play a game of badminton twice a week . . . and stopping for a glass of wine and a slice of, or a chicken salad on the way home . . . And it’s going to serve local people and feed back into the local economy [and on] inclusion, civic pride . . . We deliver a lighting installation but which can contribute towards our Transport Department’s drive for sustainable forms of transportation, on our education sector who are looking for safer routes to school to encourage parents to walk . . .

S8

For most lighting professionals, providing high-quality lighting for residents was, then, framed in terms of their obligations to provide a good ‘service’ and a responsive one which made communities feel that the local authority cared about them. Residents who discussed negative well-being implications of reduced light at night typically referenced a perceived failure in this regard. As one resident said: ‘What does disturb me is they are starting to turn lights off to save money. Because you never know when you’re going to be out at that time of night’ (I20 Wakefield). A feeling that security was being compromised by cost saving contributed to a sense of neglect and lack of trust that the local authority was acting in ‘good faith’:

Why do local authorities think it’s OK to light up the shopping areas – they seem more concerned at shops and businesses than they do the local residents and people’s own homes. We’re a bit of an island – when the lights are off, on this road, we’re cut off.

G2 Hertfordshire

For some, it was not, then, necessarily the darkness in and of itself that was disturbing, but rather the loss of a public good that had been enjoyed.

How much difference does part-night lighting make to well-being?

The household survey in one area of Shropshire provided some insight into both how typical the views outlined thus far were, and whether or not reductions in street lighting did make a difference to feelings about the local area, walking and driving after dark, fear of crime, sleep disturbance from street lights and seeing the night sky (Table 2).

The survey included a population that are generally in line with national views on how far they trust their local authority, with around 6% of people in streets affected and unaffected by part-night lighting reported that they ‘never’ trust their local council to do its best for their neighbourhood, compared with the national Citizenship Survey 2009–10, where the proportion was 7.9%. 28 Although respondents were significantly more likely to report that the number or brightness of street lights had been reduced within the last few years in the affected streets, only half had noticed the change. The only item where those in streets affected by part-night lighting made significantly different responses was ‘Thinking about the spring and autumn periods, how safe do you feel walking alone in your neighbourhood after dark?’, with fewer of those in affected streets more likely to report feeling ‘very safe’ (15.6% vs. 26.2%; p = 0.004). In the national Citizenship Survey 2009–10, the proportion reporting feeling ‘very safe’ was 31.8%. 28 There were no significant differences, however, in reporting walking alone after dark (29% vs. 34%; p = 0.27). In the Survey of English Housing 2004–5, the proportion reporting ever walking alone at night was 46.4%. 29

A small proportion (3%) of people reported being ‘very worried’ about having their car stolen or broken into after dark and this was similar in the affected and unaffected streets. Compared with national data, this is low: in the British Crime Survey 2007–8,30 10.7% reported being very worried about having their car stolen and 9.1% reported being very worried about having things stolen from their car.

In general, then, in a random sample of the population in affected and non-affected streets, there was little evidence that the introduction of part-night lighting had made significant differences to well-being, except with regard to residents’ feelings of personal security.

Pathways linking lighting reductions to well-being

We took the findings from the rapid appraisal, in the light of the literature on the impact of artificial light on health, to map out the possible pathways through which lighting reductions might theoretically impact on the public health. In summary, reductions in street lighting at night can affect well-being through a number of pathways. First, there are pathways from the direct effects of darker streets at night, which have both positive effects (improved sleep, existential capital from being able to see the night sky) and negative effects (anxiety from fear of crime, constraints on mobility at night). Although negative ones are emphasised in public accounts, deliberative public views balance these implications for personal well-being against broader determinants of health, such as carbon emission reduction. There are also pathways which are mediated by feelings about place and the governance of place. These relate to expectations that urban settings will be well lit, to the meaning of street lighting reductions in terms of trust in local government to make decisions in the interests of the neighbourhood and also more wide-ranging concerns about the implications of financial imperatives to ‘switch off’ lights. Finally, claims about well-being impacts are shaped by social attributes (such as age and gender) which potentially change the meaning of dark streets (in, for instance, making it more or less likely that one would be in them), but also the possibilities of expressing concern about them. Although highlighted in the scientific literature, concerns about the potential impacts of increases in LEDs on cancer or other physical health outcomes were not raised in any settings, public or private. These pathways are illustrated in Figure 1, with those of concern to the public in shaded boxes.

FIGURE 1.

A model of potential pathways linking reductions in street lighting to health and well-being. Health outcomes are on the right, with those raised in the rapid appraisal in shaded boxes. GHG, greenhouse gases.

Background/rationale

As described in Chapter 2, many UK local authorities responsible for street lighting have reduced, or are planning to reduce, some street lighting at night in order to reduce costs and greenhouse gas emissions. Local authorities are able to reduce street lighting at night using a range of interventions, including switching off lights permanently (switch-off), reducing the number of hours that lights are switched on (part-night lighting), replacing sodium lamps with white lights or LEDs and dimming lamps through centrally managed systems. Reductions to street lighting at night have attracted considerable public and media concern, centred on crime and road safety.

A previous systematic review35 of the effects of increasing street lighting on crime found a reduction in crime for those studies that examined changes over the course of the entire day (i.e. during hours of both daylight and darkness). However, for those that examined crime that occurred during the night-time alone (four studies), there was no evidence of an impact of changes to street lighting on crime. A major limitation of this research was the use of inadequate control areas (only two studies used multiple control areas, and in some studies the control area was adjacent to the treatment area) in the primary studies on which the review was based.

Previous research into the effects of the introduction of street lighting on road traffic injury found some evidence for improved road safety with increased street lighting [rate ratio (RR) 0.78, 95% CI 0.63 to 0.97]. 36 This systematic review included 17 controlled before-and-after studies; seven used a designated control site, while the other 10 studies collected data at one site only and used daytime data as the control. The pooled results of studies that used daytime data as the control also provided evidence for a stronger protective effect (RR 0.68, 95% CI 0.61 to 0.77). However, the methodological quality of the included studies was considered to be poor, with a high risk of bias.

We identified no previous research into effects of reduced street lighting on crime or road safety. By working together in collaboration with the local authorities of England and Wales, the LANTERNS project aimed to answer the question of whether or not reducing night-time street light for environmental and energy reasons has any impact on road traffic injuries and crime.

Objectives

The objectives of this component of the study were to collate information from local authorities in England and Wales on street light reduction and energy-saving schemes and to examine whether or not these schemes were associated with any changes to the rates of road traffic injuries or crime.

Study design

The study was a controlled interrupted time series analysis. It was based on national data on changes made to street lighting provision at night. Intervention data included the nature of changes to lighting made (e.g. switch-off, part-night lighting, dimming and white light), plus the dates of change and geographic location of street light columns.

To control for all of the important sources of confounding, we analysed street-level and area-level data as a ‘panel’ study, such that any factors that were constant over time in an area (e.g. road design) contributed no information to the analysis. The analysis was, therefore, of any association between changes to street lighting provision within streets/areas over time and any changes in counts of crime and traffic injuries.

Setting

We began in 2013 by establishing a clear project identity and framework for collaborating with the local authorities of England and Wales. This included a study acronym, ‘LANTERNS’, and a website (http://lanterns.lshtm.ac.uk/) that emphasised the collaborative nature of our project working with local authorities. We aimed to invite all local authorities that had made changes to street light provision at night to participate in the project. We developed a project database containing contact details of chief executives and street lighting managers, and summaries of lighting reduction schemes in the public domain (populated through searches of the web and other resources). In addition to local authority web pages, useful sources of information identified include previous freedom of information requests (e.g. www.whatdotheyknow.com), public sector database (www.public-sector.co.uk), street lighting Private Finance Initiative (PFI) websites and websites of interested organisations (e.g. www.stargazerslounge.com). The database was populated by April 2013 and, with NIHR approval for the website, project launch e-mails were sent to all contacts in the database on 10 April 2013.

Local authorities who responded to the first e-mail, or who registered interest in the project via the project website, were contacted by telephone. This personalised approach was intended to encourage participation in the project, to ascertain whether or not street lighting reduction schemes had been implemented, to discuss data extraction, to identify individuals who might offer expert lighting advice and to identify local authorities for the rapid appraisal. Further e-mails were sent in mid-April 2013 to follow up non-responding local authorities and to confirm the contact details of responders. A personalised letter was also sent from the principal investigator to all chief executive officers of England and Wales at the end of April 2013 to ask for their support of the project. In May 2013 the LANTERNS project was invited to partner with the ILP and the London Lighting Engineers Group. These organisations provided technical support and assisted in making further contact with non-responding local authorities.

A data request was discussed with the ILP and piloted with several local authorities. In June 2013 a formal letter and data request was sent to the 75 local authorities within which a contact person had been nominated, or with whom the LANTERNS project team had made telephone contact. In November 2013 a project newsletter was sent to all chief executives and primary street lighting contacts in the local authorities and made available on the LANTERNS website. This generated further interest and responses from the local authorities. By May 2014 we had direct contact with 140 (80%) of the 174 local authorities in England and Wales, resulting in useable data sets from 62 local authorities. Two local authorities additionally provided data via their unmetered supply operators (energy suppliers).

Participants

The study participants were, effectively, all people injured in road traffic collisions and all victims of crime who reported offences to the police in the local authorities in which street lighting reduction had been implemented and for which data were available (as described above). We used two routine data sources (described below), covering all local authorities and police forces in England and Wales, for the ascertainment of cases of road traffic injury and crime.

Variables

The exposure (intervention) and outcome variables used in the analysis were as follows.

Data sources/measurement

As described above, every local authority in England and Wales was approached in 2013 with a request for the specific locations of all street light columns where changes to street lighting either had been implemented or were planned, together with the month and year that changes were introduced.

For road traffic injuries, STATS19 data were obtained for the period 2000 to 2013. These data include the date, time of day, location (easting and northing of the location of the road traffic collision), severity (slight injury, serious injury or fatal injury) and type of casualty (pedestrian, cyclist, car occupant or powered two-wheeler) for all reported road collisions.

For crime, data from the Police.uk website were obtained from December 2010 (the earliest date for which the data are available) to December 2013. This large data set was downloaded and stored on a server to be interrogated as required. These data comprise the month, the name of roads where incidents occurred, the approximate geographic co-ordinates and the type of crime. One disadvantage of this publicly available data set is that time of day is not included. Another is that the spatial accuracy of the data is unknown as the data are obfuscated to preserve the anonymity of victims. However, as part of this project we assessed the reliability of these data by comparing them with detailed crime data (including the time and location of offences) from one police force. 39

Using a geographic information system (GIS) the data on crime and road traffic injuries were linked to a road segment database that includes the characteristics of all classified and unclassified roads. The road network used to link the data sets was the current version (at the time of analysis) of the Ordnance Survey Mastermap Integrated Transport Network. 40 We initially linked data to roads for two local authorities to test the concept (using Camden borough as an urban area and Norfolk as a rural location).

Street lighting data from the local authorities were formatted and imported into the GIS. Each road segment was classified according to the type of street lighting reduction scheme implemented (e.g. part-night lighting, dimming, part-night lighting and dimming, etc.) and by the census LSOA within which it was located. LSOAs contain around 670 households and represent the smallest census areas for which some data used in the analysis (indices of deprivation) are available.

From the combined data set, counts of crime and road traffic collisions for each road segment were generated by month and by year.

Bias

For optimal control of biases our analysis compared changes in counts of crimes and traffic collisions on roads before and after street lighting was changed, relative to trends on other roads. The estimated effect is, therefore, specific to roads with decreased street lighting compared with other roads.

We examined potential bias due to ‘regression to the mean’. This bias might arise if, for example, unusually low numbers of collisions or crime were factors influencing local decisions to reduce lighting in some areas. An increase in collisions or crime in those areas may then have been expected as a result of random variation over time, rather than owing to the lighting intervention. To test for this bias, we repeated our analysis excluding data for periods of time prior to the introduction of street lighting changes.

To examine whether or not potential under-reporting of events to police might bias the estimates of effect, we analysed separately the most reliably reported events (i.e. killed or seriously injured casualties, and crime for which insurance claims require a crime report, e.g. burglary of dwelling or theft of vehicle).

Study size

Sample size (power) calculations assumed that street lighting changes had been implemented on streets for which only 1% of pre-intervention traffic injuries and crime occurred.

For road traffic injuries, statistical power was maximised by using data for a period of up to 10 years before lighting changes were implemented. Assuming that 150 night-time injuries per year might be expected on intervention roads, this was expected to give 1500 night-time injuries on intervention roads during the 10 years before lighting reduction was implemented and 150 injuries 1 year after the reduction was implemented, providing 90% power to detect an increase of 32% above pre-intervention injury levels. This magnitude of effect is consistent with the reduction in injuries estimated in a systematic review of the introduction of street lighting. 36

For crime, based on data from the crime survey of England and Wales30 and police-recorded crime data, we assumed that around 20,000 daytime and night-time offences would be expected per year on intervention roads, such that the study would have 90% power to detect a 5% increase in crime above pre-intervention levels (a much smaller increase than the estimated decrease of 22% associated with increased lighting reported in Welsh and Farrington35), and for major crime subcategories (e.g. violence against the person) it would have 90% power to detect increases of about 10% in crime.

Quantitative variables

We aimed to estimate the association between changes to street lighting provision over time and changes in counts of collisions at night, and counts of criminal offences that commonly occur at night: burglary, theft of (or from) a vehicle, robbery, antisocial behaviour, criminal damage and violence including sexual assault. We defined ‘night-time’ as 1 hour after sunset and 1 hour before sunrise. For the analysis of part-night lighting, ‘night-time’ was defined as 12 a.m. to 6 a.m., a common period used by local authorities when switching lights off for part of the night.

Statistical methods

To control for all of the important sources of confounding we analysed street-level and area-level data using a panel study design with fixed effects. This approach uses data within (not between) the streets/areas (panels) to estimate any association between lighting reduction and changes in crime/traffic injuries. Any factors that are constant in a street/area over time (e.g. road design, presence of trees and buildings, etc.), therefore, contribute no information and cannot confound the primary association of interest. Other types of analyses (e.g. random effects for street/area) draw some information from how interventions correlate with overall outcome rates across streets/areas, but at the cost of potential confounding by other things that vary over space (but not time).

Participants

Local authorities in England and Wales were generally very positive about participating in the LANTERNS project. Although some of the local authorities were keen to participate, they were unable to provide the data required within the time scale. Figure 2 shows the flow of local authorities participating in the project and Figure 3 shows which local authorities were included in the analysis. Eight local authorities were unable to extract the required data from their asset management systems; three authorities confirmed that they had made no relevant changes to street lighting in their areas. In the case of 32 local authorities, despite an initial interest in the project and several follow-up e-mails and telephone calls, we could obtain no further response. Agreement to participate was obtained from 82 local authorities, of which 71 were able to provide data within the time scale. Of the 71 data sets received, 62 spanned sufficient periods of time to contribute to the time series analysis (see Figure 2).

FIGURE 2.

Flow of local authority participants in the LANTERNS project.

FIGURE 3.

Local authorities included in LANTERNS analysis. Contains National Statistics data © Crown copyright and database right 2013; contains Ordnance Survey data © Crown copyright and database right 2013.

Descriptive data

The street lighting interventions of primary interest were switch-off, part-night lighting, dimming, trimming and white light. However, many local authorities had introduced these interventions in combination with others, for example changing lamps to white light and using part-night lighting. Before proceeding with the analysis, we specified which combinations would be considered in addition to the single interventions. Trimming is where the time period during which lights are switched on is slightly reduced. This is usually achieved using photo cells which are set to switch lights on/off when it is darker (e.g. at 35 lx rather than 70 lx). Thus, the period of time whereby a trimming intervention can have an effect on crime or road safety is very marginal compared with the other light-reduction interventions, which last for several hours. For the purposes of analysis we therefore merged all combinations of lighting interventions that included trimming with the same combinations that excluded trimming. The street lighting interventions analysed were, therefore, switch-off, part-night lighting (including part-night lighting with trimming), white light (including white light with trimming) and dimming (including dimming with trimming).

The introduction of street lighting adaptation strategies increased steadily from 2009 (Figure 4). By December 2013, the local authorities participating in this study had implemented white light on a total of 15,833 km of road (7% of total road km in the 62 participating local authorities); part-night lighting on 12,101 km of road (5%); dimming on 10,519 km of road (4%); and switch-off on 946 km of road (0.4%). The proportions of total km of road in each MSOA that had lights switched off ranged from 0% to 60% [median 0.03%, interquartile range (IQR) 0.01–0.05%]. For part-night lighting the proportions ranged from 0% to 84% (median 0.2%, IQR 0.1–18%); for dimming the proportions ranged from 0% to 93% (median 0.14%, IQR 0.07–5%); and for white light the proportions ranged from 0% to 81% (median 1%, IQR 0.15–11%).

FIGURE 4.

Road km with lighting adaptation strategies implemented. PNL, part-night lighting.

Outcome data

Main results

Road traffic injuries

We found little evidence for any associations between street lighting adaptation strategies and day-adjusted night-time collision rates (Figure 5).

FIGURE 5.

Associations between street lighting adaptation strategies and night-time road traffic collisions. (a) Switch-off; (b) part-night lighting; (c) dimming; and (d) white light. PNL, part-night lighting.

There was no evidence from any of the regional models or the overall estimates for an association between switch-off (RR 0.97, 95% CI 0.82 to 1.15), part-night lighting (RR 0.95, 95% CI 0.84 to 1.07) or dimming (RR 1.00, 95% CI 0.91 to 1.10).

There was weak evidence in London for an association between the introduction of white light and increased night-time collisions (RR 1.30, 95% CI 1.03 to 1.65); however, the overall national estimate provides no evidence for an association (RR 1.01, 95% CI 0.93 to 1.09).

When models were restricted to collisions where casualties had suffered a serious injury or were killed, the estimated associations with street lighting adaptation strategies were unchanged and estimates were less precise (Figure 6).

FIGURE 6.

Associations between street lighting adaptation strategies and night-time collisions (casualties killed or seriously injured). (a) Switch-off; (b) part-night lighting; (c) dimming; and (d) white light. PNL, part-night lighting.

Crime

Figure 7 shows the estimated mean associations between changes in the proportion of total km of road in each MSOA that had introduced each street lighting adaptation intervention and counts of each offence across England and Wales. The RRs indicate the expected change in crime if 100% of total km of road in an area were to receive the lighting intervention.

FIGURE 7.

Associations between street lighting adaptation strategies and crime. (a) Switch-off; (b) part-night lighting; (c) dimming; and (d) white light. PNL, part-night lighting.

There was no evidence from the overall estimates for an association between the aggregate count of crime and switch off (RR 0.11, 95% CI 0.01 to 2.75) or part-night lighting (RR 0.96, 95% CI 0.86 to 1.06).

There was weak evidence for a reduction in the aggregate count of crime and dimming (RR 0.84, 95% CI 0.70 to 1.02) and white light (RR 0.89, 95% CI 0.77 to 1.03).

These RRs estimate change in outcomes associated with 100% implementation of the street lighting intervention in an area. To assist with interpretation, if dimming was applied to street lights on 10% of total road km in an area, the estimated change in crime in the area would be equivalent to 0.84^0.10 = 0.98 (i.e. a 2% decrease in crime) with a 95% CI from 0.70^0.10 = 0.96 to 1.02^0.10 = 1.00 (i.e. between a 4% decrease and no change in crime). If applied to street lights on 50% of roads, the change would be 0.84^0.50 = 0.92 (i.e. an 8% decrease in crime) with a 95% CI from 0.84 to 1.01 (i.e. between a 6% decrease and a 1% increase in crime).

We found significant heterogeneity between the estimates at police force level (see Appendix 6). There was strong evidence for an association between part-night lighting, dimming and white light and a decrease in crimes in some police forces, and strong evidence for an association between part-night lighting, dimming and white light and an increase in crimes in others.

When specific crime offences were considered, the estimates provide suggestive evidence that part-night lighting may be associated with an increase in robbery (RR 1.48, 95% CI 0.99 to 2.21) and that dimming may be associated with a decrease in violence (RR 0.78, 95% CI 0.60 to 1.01).

We found consistency in the direction of the estimated associations between crime and dimming, with all point estimates indicating reductions in crime. There was similar consistency in the estimated associations with white light, all suggesting reductions in crime.

Area deprivation (England only)

When the analysis was repeated in each of three strata of areas based on the IMD,43 the stratum-specific estimates were not materially different from the overall estimates, and each was imprecise (Table 3). There were insufficient data for switch-off.

There was no strong evidence for any associations between any street lighting changes and road traffic injuries within any IMD strata (Table 4).

Regression to the mean

Had we observed any evidence for an increase in collisions or crime following reduction of street lighting we had planned to assess whether or not the associations may be, in part, due to bias from regression to the mean. We found no evidence for increases in either outcome and so this analysis was not conducted. In addition, there were insufficient months of crime data available prior to street lighting interventions being implemented for an assessment of potential bias due to regression to the mean to be meaningful.

Key results

This study has provided detailed evidence on the impact of street lighting energy-saving schemes on two important public health outcomes, namely crime and road traffic injury. We obtained data from 62 local authorities in England and Wales, which collectively account for a total of over 30,000 road km affected by lighting changes by 2013.

In summary, the key results are:

• Switch-off (permanently turning street lights off) was not associated with an increase in night-time traffic collisions or crime; however, the results are imprecise because of the small number of areas in which switch-off was implemented, and so should be treated with caution.

• Part-night lighting (e.g. street lights are switched off between 12 a.m. and 6 a.m.) was not associated with an increase in night-time traffic collisions or crime.

• Replacing conventional yellow lighting with white light was not associated with an increase in night-time traffic collisions and was associated with a reduction in crime, though estimates were imprecise.

• Dimming of conventional yellow light or white light was not associated with an increase in night-time traffic collisions and was associated with a reduction in crime, though estimates were imprecise.

These results were not materially changed by the sensitivity analyses undertaken.

Limitations

The study used routine data sources on road traffic injuries (police STATS19 data) and crime (Police.uk data). These data sets have several limitations; in particular, they may be incomplete owing to under-reporting. However, for under-reporting to affect the results of our analysis, there would have had to have been differential changes over time in the recording of crime and road traffic injuries in the streets where lighting has been changed, compared with streets without changes to street lighting. This seems unlikely.

Before making crime data publicly available, the spatial locations of reported crimes in the Police.uk data are altered to protect the anonymity of victims. This introduces the potential for misclassifying crime to roads adjacent to those on which they occurred, which could lead to them being incorrectly allocated to roads on which lighting reduction interventions have been introduced (or vice versa). To address this limitation, we conducted a reliability study that compared these data with detailed police-recorded crime data, and to identify the spatial level at which counts of crime using both sources are in good agreement. Analyses were then conducted at this level of spatial resolution. Thus, while we are reasonably confident in the direction of the associations between street lighting reductions and crime, we must remain cautious about their relative magnitudes as the effects are measured at the area, rather than street segment, level.

To address the potential for under-reporting to bias estimates of effect on collisions, we analysed separately those collisions where casualties were fatally or seriously injured, as these collisions are more likely to be reported to the police. These analyses found no evidence for associations between street lighting changes and night-time collisions. We also analysed crimes that are more likely to be reported to the police (i.e. burglary and vehicle theft reports are required for insurance claims). The estimated effects were not consistently larger for these two types of crime. Our analysis did not, however, include any measures of fear of crime, which might be expected to have increased in those areas with reduced lighting.

We could not take into account the potential impact of other road safety or crime prevention initiatives, such as improved road markings, policing interventions or CCTV, for example. If such measures were introduced more often in streets where lighting has been changed than elsewhere, it is possible that some of the associated changes in crime in areas where lighting has been changed may be attributable to these other measures. It is likely that such confounding will account for part of the observed associations with crime within the areas where street lighting was changed, compared with other areas.

We found no convincing evidence for associations between street lighting reductions and road traffic injuries. In designing the study we had estimated the hypothesised effect of reducing street lighting on casualties (a 32% increase) by using the inverse of the relative risk estimated in the Cochrane systematic review. 36 We had assumed there would be 1500 night-time injuries on intervention roads during 10 years before lighting reduction and 150 injuries 1 year after, to achieve 90% power to detect this increase of 32% above pre-intervention injury levels. For our analysis of switch-off, we were able to include over 1780 night-time casualties on roads before street lighting was reduced (i.e. during 2000–10) and 298 night-time casualties after street lighting was reduced (i.e. during 2011–13). It is possible that we did not have sufficient statistical power to detect smaller, plausible associations. However, for other street lighting interventions, such as part-night lighting and dimming, the numbers of night-time collisions included in the analysis were much larger.

It is also possible that the numbers of pedestrians, cyclists, motorcyclists and car drivers who use and travel within the streets and areas where street lighting was reduced declined at the same time as the lighting was reduced, resulting in fewer road casualties in those streets and areas. If so, any increase in hazard to pedestrians and cyclists due to lower lighting conditions might have been obscured by a reduction in numbers of people exposed to road injury risk at night.

Interpretation

In the context of reported concern about street lights going out at night, this study found little evidence of harmful effects of switch-off, part-night lighting, dimming or changes to white light/LEDs on road collisions or crime in England and Wales. Given previous evidence that introducing street lighting is associated with a reduction in road traffic collisions, it is notable that we found no evidence for an increase in collisions where street lighting was reduced at night. We did not, however, have direct measures of exposure, and so we do not know whether or not there were changes in cars or pedestrians using the affected areas. Neither do we have measures of risk compensation, which might shed light on whether or not road users take more care in darker streets. Certainly, the mechanisms linking the reduction of street lighting to impacts on crime and road safety may be different from those linking improvements to those impacts. First, given potential pathways that operate through feelings about neighbourhood or trust in local authority, removal of a public good may have very different effects from maintaining or improving it. There are, then, potential explanations that relate to both the motivations for the intervention on the part of local authorities and the ways in which those interventions are understood by those living in, and using, affected streets. As Chapter 2 outlined, concerns about dark streets are overlaid to some extent by concerns about the fact that a local authority has removed a public good: this may play out differently in different areas of the country. Indeed, we found significant evidence for heterogeneity in the impacts estimated at police force level, indicating the importance of context. In some forces there were statistically significant reductions in crime, whereas in others there were significant increases.

Permanently switching lights off in an area may signal decline and a lack of governance. 35 If switch-off is taken as a sign of disengagement with a community, mechanisms such as that described by the ‘broken window’ thesis, whereby disorder generates further disorder,44,45 might operate to escalate crime incidence. Dark streets might also encourage some forms of criminal activity. For instance, thieves may feel less conspicuous and less concerned about being seen or identified by either passers-by or CCTV operators. Further research is necessary to uncover the mechanism(s) through which changes (increases or decreases) to street lighting might bring about their effects on crime. To date, though, it appears that the decisions made by those local authorities that have shared data with us have not increased the risks of road traffic injuries or crime to their residents.

Generalisability

This study used data from a range of settings in the UK, including urban, rural, deprived and affluent areas, and has shown that the effect of changes to street lighting on crime (but not road traffic injuries) appears to vary across contexts. Our stratified analyses must be treated with a degree of caution, but they suggest that switch-off was associated with an increase in burglary and vehicle crime in more deprived areas and city/town centres, whereas part-night lighting was associated with decreases in burglary and vehicle crime in less deprived areas. The extent to which these results may be applied to a single area or street remains a matter for further enquiry.

Costs of interventions

Although determining the costs of interventions is necessary in any economic evaluation, costing large-scale public health interventions can have particular challenges. Costs of large-scale public health interventions may be incurred by a range of agencies, making estimates of total costs tricky. For example, the costs of introducing free bus travel for young people in London, an intervention with many potential public health implications,57,58 were incurred by two different agencies. The costs of manufacturing and posting bus passes to young people were incurred by Transport for London (i.e. administrative costs of the scheme). The costs of providing increased bus capacity were incurred by the bus network operators within London (i.e. operating costs of the scheme). The marginal cost of an additional concessionary journey can be estimated using a variety of methods, which can lead to different results, further complicating cost estimation. In addition to having costs shared by multiple agencies, large-scale public health interventions may also shift costs (for instance, from public service authorities to individual households) making total cost calculations even more problematic.

‘Wider’ effects of policies

Traditionally, public health research has focused on the direct impacts of interventions on the health of individuals. This ‘micro-level’ focus often excludes important outcomes that indirectly effect health, such as social capital, community empowerment, etc. 48 Most research argues that these types of outcomes should ideally be included in evaluations of large-scale policies and interventions. Large-scale interventions can also affect non-health outcomes47 and conceptual decisions need to be made about whether or not these outcomes are included in the scope of the analyses. Identifying the wider health effects (and non-health effects if they are to be included) of large-scale policies is not straightforward. Causal pathways are complex and often unknown. Complicating matters is the fact that large-scale public policies are often implemented within what are called ‘complex systems’ with their own emergent properties that can have important health effects. 46

Attribution of health and wider effects to large-scale interventions is also problematic. Large-scale interventions can occur at the same time as other changes, making it difficult to assess which specific changes in health outcomes result from the intervention. In complex environments there are often no obvious comparison or ‘control’ areas to help contextualise changes. Randomised controlled trial (RCT) designs are often not feasible (or necessarily appropriate) in evaluations of large-scale policies and so evaluations may need to rely on robust observational methods to attribute health and wider effects of policies.

Monetising the effects of interventions

Valuing large-scale public health interventions also begets a number of methodological concerns. There are two main ways of valuing the benefits of interventions: revealed preference techniques and stated preference techniques. Revealed preference techniques use market information to place a monetary value on particular benefits. For example, economists can use the difference in compensation for different types of jobs to monetise the ‘risk’ of different types of workplace injuries. However, there are a number of fundamental challenges in identifying the trade-offs that individuals make (particularly the difficulty of controlling for other influences on decisions). Revealed preference techniques are very rarely used in monetising interventions of public goods: the non-excludable nature of public goods ensures that there are fewer opportunities for trade-offs to be identifiable. Therefore, research aiming to monetise benefits of large-scale public health interventions tends to concentrate on stated preference techniques. Stated preference techniques ask individuals to place a value on changes in their health and welfare, usually through a process called contingent valuation. Broadly, contingent valuation processes ask individuals hypothetically about the maximum amount of money they would be willing to pay for the benefits associated with the introduction of a service, or about the minimum amount of money they would be willing to accept as compensation for the removal of a service.

Although contingent valuation methods are the dominant valuation techniques used in public health, they involve a number of methodological choices which may impact on their validity and reliability. Briefly, these issues include how best to construct the hypothetical ‘scenario’ presented to individuals, the choice of ‘payment vehicle’ (the means by which payment will hypothetically be made), the time period for valuation and whether individuals are valuing the intervention under conditions of certainty or risk. 59 The choice of methods may have implications for the extent of hypothetical and strategic bias present in the valuation. Large-scale interventions that involve characteristics of public goods present even greater challenges for stated preference valuations. Evidence for ‘free rider’ hypotheses suggests that public goods are often undervalued and principles of ‘loss aversion’, the ‘endowment effect’ and the well-known willingness-to-pay/willingness-to-accept gap suggest that it is particularly difficult to value removal of a public good.

We use street lighting as a case study to illustrate some of the complexities of CBA of large-scale public health interventions with public good characteristics.

Costs of intervention

In England and Wales, street lighting is delivered by different agencies depending on the location of the street. The Highways Authority is responsible for providing street lighting on motorways. For the most part, local authorities deliver street lighting on roads within their boundaries. However, in major metropolitan areas, street lighting on ‘trunk roads’ (usually major roads that span different local authorities) is provided by metropolitan agencies.

Interventions to reduce street lighting have taken a number of forms. Some local authorities have introduced part-night lighting, where street lights are turned off between certain hours of the night (typically between 12 a.m. and 6 a.m., although this can vary). Other local authorities have introduced dimming, where lights stay on all night but emit a reduced amount of light. Trimming, where street lights are switched on slightly later and switched off slightly earlier, is another popular intervention. Finally, local authorities have also switched to different (usually more energy efficient) types of lighting. Many local authorities introduce multiple interventions involving the same street light.

If we take a societal perspective in the CBA framework, calculating the costs of these different interventions to different agencies poses a number of problems. Local authorities incur four major costs associated with street lighting reduction interventions, energy costs, maintenance costs, lighting adaptation costs and consultation costs, each of which may be difficult to tease out. Not all local authorities consult residents about changes to street lighting provision; among those that do, consultations on reduced street lighting can take place within a broader discussion of local issues, which means making some assumptions when attributing costs to the street lighting intervention in particular. In terms of energy costs, energy tariffs differ not only by company but by time of day, and energy companies have sometimes changed their prices in response to reduced street lighting. Maintenance costs are also problematic. Reduced street lighting is often part of a larger programme of works to update street lighting generally, so, again, some assumptions need to be made to attribute part of these costs to the street lighting intervention.

‘Wider effects’ of street lighting

There are many potential effects of reduced street lighting on public health and many of these may be mediated through related changes that increase and/or decrease the likelihood of particular outcomes (see Figure 1). For instance, one theory suggests that reduced street lighting will lead to poorer visibility on the streets at night. This decrease in visibility will, in turn, lead to an increase in road traffic collisions and particular types of crime. However, reduced visibility at night may also reduce residents’ mobility, if they are less likely to go out at night because of fear of crime or injury. Reduced mobility may lead to a decrease in the number of crimes and road traffic injuries that take place in affected areas.

In addition, there are a number of non-health related outcomes that may be affected by street light interventions. For example, if reduced visibility leads individuals to travel more slowly or to avoid particular areas then this has implications for the cost of travel. A reduction in mobility due to the reduced street lighting schemes may also have implications for the night-time economy near affected areas. Conceptual decisions need to be made about whether or not to include these non-health outcomes.

In a real-world context, it may be difficult to attribute any change in health outcomes to the street lighting intervention. Local authorities choose to reduce street lighting in particular areas, for different reasons, but many choose to reduce street lighting in what they consider to be ‘low-risk’ areas. It may be difficult to identify matching comparison areas, there are a number of potential confounders, and other changes in the streets may have occurred around the same time (e.g. increased police presence; traffic-calming measures), making attribution of causation difficult. For instance, in response to concerns about crime, policing policies may have changed to increase presence of police in those areas affected by the interventions. Additionally, companies or individuals may have invested privately in their own lighting in response to the intervention, which may mitigate any effects of the intervention. Depending on the scope of an economic evaluation these private investments may need to be considered in the costs of the intervention.

Difficult valuation issues

Street lighting reduction offers a complex case for stated preference approaches to valuation for a number of reasons. Research into the value of introducing street lighting has identified a few technical challenges. First, in a context where there is a lack of transparency in local government spending,62 the public may lack trust in the agency responsible for street lighting provision. This presents challenges in choosing a method of eliciting willingness to pay/willingness to accept. Using repeated valuations may anger and surprise the respondent as a result of expectations about the first valuation representing actual cost of the intervention; respondents may wish to send a message about keeping prices low to the untrusted authority; and values may be contingent on other changes in local taxes that have occurred recently. 61 Second, individuals are likely to base their preferences for reduced street lighting on the direct effects of the intervention (such as crime, fear of crime and road safety concerns) and are unlikely to consider other more minor implications such as light pollution. 60

Overcoming challenges

A number of recent articles have made recommendations for the conduct of CBA in the face of these challenges. Broadly, these include questionnaires or consultations with key stakeholders to inform cost calculations, conceptual modelling and quasi-experimental methods to elucidate the wider benefits of an intervention, and utilising guidance and literature from a range of disciplines on valuation.

Methods

In our cost–benefit model, we have simulated forward-looking estimates of the costs and benefits of adapting a proportion of street lights in a typical local authority area over 5-, 10- and 15-year time horizons. All prices are reported in 2015 GBP.

Results

Results from the initial model suggest that after 5 years the cost of street lighting provision in the part-night lighting scenario are greater than in the ‘do-nothing’ scenario by £74,754 (Table 7). Over a 5-year time horizon, therefore, a reduction in energy consumption in the part-night lighting scenario is outweighed by the capital costs of implementing part-night lighting through a CMS. The benefits of the part-night lighting scenario, however, are also greater than the benefits of the ‘do-nothing’ scenario by £2,029,519 (range –£6,596,018 to £10,117,131). In other words, it appears that the disbenefits of an estimated increase in robbery and violence associated with the introduction of part-night lighting are outweighed by the benefits of the estimated decrease in collisions, burglary and vehicle crime and carbon reductions. It is important to note that the estimates of effect of part-night lighting on collisions and crimes are imprecise and so lead to big differences in estimates of benefits.

If we use the upper bounds from the CIs for estimated associations with part-night lighting, collisions and all crime types are estimated to increase with part-night lighting. In this case, the model indicates that after 5 years the disbenefits of estimated increases in collisions and crimes would outweigh the benefits from reductions in carbon emissions by £6,596,018.

Alternatively, if we use the lower bound of CIs around estimates of association between part-night lighting and collisions and crimes, all crimes and collisions would decrease after the implementation of part-night lighting. In this case the benefits of reductions in collisions and crimes in addition to benefits from a reduction in carbon emissions would total £10,117,131. Comparing the benefits to the costs of the part-night lighting scenario with the ‘do-nothing’ scenario results in a NPV of £1,954,765 (range –£6,670,772 to £10,042,377). Implementing part-night lighting on 5% of roads in this hypothetical local authority might, therefore, result in a net benefit to society of nearly £2M after 5 years.

Looking at 10- and 15-year time horizons, the costs of street lighting provision under the part-night lighting scenario are lower than the costs under the ‘do-nothing’ scenario: that is, the reduction in spend on energy exceeds the capital cost of changing to a CMS over these longer time horizons. The benefits implementing part-night lighting driven by reductions in collisions, burglary, vehicle crime and carbon emissions remain large at £3,640,780 (range –£11,293,681 to £17,618,549) after 10 years and £4,911,520 (range –£14,715,109 to £23,255,609) after 15 years. Combined with cost savings these benefits drive a large overall benefit to society (NPVs) after 10 years of £3,657,574 (range –£11,276,887 to £17,635,343) and after 15 years of £5,039,696 (range –£14,586,933 to £23,383,786).

In the following sensitivity analyses we calculate the results under different assumptions about increases in energy prices. Table 8 presents results of the CBA model assuming energy prices rise with inflation only.

Under this assumption, our CBA model estimates that implementing part-night lighting on 5% of road km in an area will result in marginally smaller benefits to society after 5, 10, and 15 years than with larger energy price rises.

Table 9 presents results of the CBA model using the assumption that energy prices will rise to 4% above inflation. Under this assumption, the costs savings from introducing part-night lighting are even greater than the initial model, contributing to an even larger net benefit to society from the intervention.

Discussion

We conducted a CBA to compare the societal costs of street lighting reduction with its societal benefits. The scope of our CBA included infrastructure, maintenance and energy consumption costs, and any associated changes in crime and road traffic collisions (using estimates from Chapter 3 of this report). Our CBA framework considered two scenarios over different time horizons: one where street lighting is reduced and one where it is maintained. Specifically, we considered the economic argument for part-night lighting of a proportion of street lights in a typical local authority. We found that part-night lighting regimes are likely to represent a large net benefit to the local authority, although it is important to note the considerable uncertainty around the estimates.

Our results need to be interpreted with caution owing to a number of limitations. Our CBA model did not include other outcomes that may be hypothesised to be associated with part-night lighting regimes (see Figure 1). We might reasonably expect that most of these outcomes, such as the potential increase in fear of crime and decrease in mobility, if incorporated, would serve to reduce the net benefit to the local authority of part-night lighting regimes. This suggests that our CBA model is likely to have underestimated overall disbenefits from part-night lighting regimes. We cannot know whether or not these disbenefits, if incorporated, would outweigh benefits from reduced collisions, burglaries and vehicle crimes.

Another key limitation is the number of simplifying assumptions made within the model, such as those for energy prices, inflation rates and lighting maintenance costs. Our CBA was able to incorporate uncertainty around the estimates of the effect of part-night lighting on collisions and crime, leading to a wide range of estimates of NPVs of part-night lighting for local authorities. We did not consider uncertainty around other elements in the model, such as costs or time trends. If we were to incorporate these uncertainties, it is likely that the range of estimates of the NPV of part-night lighting regimes would widen. Furthermore, this CBA did not incorporate other public health outcomes, such as sleep patterns and neighbourhood morale, which might change when lighting reduction strategies are implemented. Reductions in energy use, CO₂ emissions and crime, however, suggest that the economic case for street lighting reduction strategies may be promising.

Local authorities in England and Wales may wish to extend or adapt our CBA model to their localities, using energy costs, maintenance costs and lighting adaptation costs specific to their areas (the spreadsheet is available from the authors on request).