Puffin Pedestrian Crossings in London: Collisions, User Behaviour, and Pedestrian Perceptions

London Road Safety Unit March 2006 Research Summary No 8

Overview Puffin (Pedestrian User-Friendly Pedestrian behaviour observations Intelligent) crossings were designed to indicated that the level and nature of reduce delays to vehicles and improve road user conflict was similar at both pedestrians’ feelings of safety while Puffin and Pelican crossings. However, crossing the road. at Pelican crossings pedestrians are more likely to start crossing the road The potential road safety benefits of during the flashing green man period Puffin crossings (compared with Pelican than during the all red period at Puffin crossings), according to the DETR crossings. The expected reduction in Traffic Advisory Leaflet 1/01(1) are: delay to vehicles seems to be 1. Reduced harassment experienced by unfounded as mean delays to the first pedestrians as a result of the vehicle in the queue were higher at withdrawal of the flashing pedestrian Puffin (12.6 seconds) than at Pelican phase. crossings (8.8 seconds). 2. Reduced frequency of shunt Pedestrian perceptions of Puffin collisions as unnecessary signal crossings indicate that the majority changes are avoided. perceive Puffin crossings as safer than 3. Reduced confusion for pedestrians the facility they replaced. However, over as the blackout and flashing a quarter of pedestrians at Puffin pedestrian signals are not used. crossings report that their view of the 4. Increased convenience and reduced pedestrian signal is sometimes confusion associated with the wider obstructed by other pedestrians. use of pedestrian stages at signalised junctions. The use of detectors has resulted in pedestrians perceiving Collision analysis over all site groups that they have more time to cross at revealed non-significant reductions of Puffin than at Pelican crossings. 15% in total collisions and 26% in pedestrian collisions.

1

Objectives pedestrian crossing facilities such as Zebra and Pelican crossings. The overall aim of these studies was to Until now, only limited evaluation of assess all aspects of safety at stand- Puffin crossings has been undertaken. alone Puffin crossings (as opposed to at One early study of sites in London found junctions). little difference in collision rates at Collisions Pelican and Puffin crossings (2). However, this study examined only five • To compare total collisions and sites and found that some collision types pedestrian collisions before and after declined while others increased. installation of Puffin crossings. Another study of two pilot Puffin sites • To determine whether the effect incorporated both behavioural and varies with the previous crossing attitudinal measures but did not assess facility. the collision data (3). There was Road User Behaviour evidently a need for further evaluation of the safety record of Puffin crossings. • To compare pedestrian and driver behaviour at Pelican and Puffin The studies for TfL summarised here crossings and identify any unsafe focus on Puffin crossings as they are behaviour associated with Puffin functioning in London. crossings. • To compare the levels and nature of Method pedestrian conflict with other traffic at Puffin and Pelican crossings. Three complementary methods were employed: Pedestrian Perceptions 1. A before and after collision analysis • To determine the level of pedestrian at a selection of 23 sites where Puffin user understanding of Puffin and crossings have been installed. Pelican crossings and any difficulties 2. Video observation of road user they report in using them. behaviour and road user conflicts at • To determine pedestrian perceptions a selection of 5 Puffin and 5 Pelican of safety at Puffin and Pelican sites. crossings and levels of preference for 3. On-street pedestrian interviews at 14 each crossing type. Puffin and 10 Pelican crossings. • To identify any differences in pedestrian perceptions between more Results or less experienced users of Puffin crossings and those of different age- 1. Collision Analysis groups. Total and Pedestrian collision rates fall A full report for each study is available. relative to controls when Puffin crossings replace other crossings but Background not significantly. There are currently 111 stand alone Table 1 gives the percentage change in Puffin crossings in service in Greater collisions for each site type relative to London. The DfT intention is to borough control data. Over all the sites, gradually replace Pelicans with Puffins. following the introduction of a Puffin However, there is still some uncertainty crossing there were reductions of 15% in about the road safety implications of total collisions and 26% in pedestrian Puffins in comparison with other collisions.

2

Table 1: Percentage Change in Total and Pedestrian Collisions for each Site Type

All Collisions Pedestrian collisions Change in Change in Site Type No Probability* Probability* collisionsa collisionsa All site groups 16 -15% 30% -26% 22% Previously no formal crossing 6 8% 82% -35% 48% Previously 6 -14% 62% -8% 80% Previously Pelican Crossing 6 -39% 7% -30% 48%

No Number of site groups, a Change in collisions at Puffin crossings compared to control sites, * Probability that result is due to chance However, the absolute number of 2. Road User Behaviour and Puffin collisions was relatively small and these Crossing Operation reductions were not statistically Collision analysis was supplemented significant at the 5% level. Where there with information about road user had previously been no formal crossing, behaviour and conflicts collected at five total collisions rose slightly (although this Puffin and five Pelican crossing sites in result was not statistically significant). London. There were differences in Shunt collisions have been predicted to pedestrian flows between the Puffin and decrease where Puffin crossings are Pelican sites at two of the site pairs installed because of the expected which affected interpretation of the cancellation of unnecessary pedestrian findings. demands. However, only a small Pedestrian Behaviour reduction in the average monthly rate per site of shunt collisions from 0.007 in The proportion of pedestrians who use the before period to 0.006 in the after the pedestrian demand button varies period was found (which equated to an widely between sites. actual reduction from 6 to 5 collisions overall) this difference was not Over all the sites 20% of users of statistically significant. Over all borough Pelican and 28% of users of Puffin site groups, the monthly total collision crossings did not use the pedestrian rate fell between the before and after demand button. However, this periods from 0.205 to 0.149. However, proportion varied from 2% to 49% the rate rose at five of the 23 sites, two between sites. of which had previously had no formal Pedestrians are more likely to start crossing facility, two of which had been crossing during the Pelican flashing Zebra crossings and one had been green man period than in the Puffin all converted from a Pelican crossing. red phase period. Over all sites, the monthly average pedestrian collision rate per site fell from While 0.1% of pedestrians started 0.071 to 0.043. The monthly rate rose at crossing in the all-red period at Puffin seven out of 23 sites, four previously crossings, this figure was 1% at Pelican had no formal crossing. Two had been a crossings. This difference is statistically Zebra crossing and one had previously significant at the 1% level. This been a Pelican crossing. suggests that pedestrians may be more cautious when faced with the red man

3

Table 2: Vehicle and pedestrian conflicts at 5 Pelican and 5 Puffin crossings (0700-1900)

Violations Encounters Conflicts Collisions Puffin crossings 15 22 1 0 Pelican crossings 3 18 1 1

than a flashing green man. Pedestrians • The all-red extension periods were were also more likely to cross the road frequently activated at Puffin outside the pedestrian green phase at crossings. The proportion of sites operating under Urban Traffic pedestrian phases using the Control (UTC) than sites on local control extensions ranged from 20% to 91% for both types of crossings. at different Puffin sites. Pedestrian vehicle conflicts • The call cancel facility (where pedestrian demands are cancelled at Frequency and types of conflict are Puffin crossings if no longer required similar at Puffin and Pelican crossings. because a pedestrian crosses before the pedestrian phase) was used only Conflicts between pedestrian and once out of 500 crossing events and vehicles were observed on video (0700- there were only 8 instances observed 1900) and categorised in order of when a call should have been descending severity: cancelled but was not. 1. Collisions • Checks of the crossing equipment 2. Conflicts (where two road users were revealed that 4 out of 5 Puffin on such a course that one of the two crossings had faults with detectors had to take sudden aversive action to which affected operation of the avoid a collision) crossing. The videos also suggested 3. Encounters (less serious interactions that pedestrian on-crossing detectors than conflicts) failed for 1-3% of crossing events. 4. Violations (of the Highway Code) • Finally, Puffin timings may have been Table 2 shows the frequency of each of set conservatively in relation to the these at Puffin and Pelican crossings. most recent guidance which has been The frequency and types of conflicts published since the data were were similar at Puffin and Pelican collected. crossings. Most occurred when To test the relative contribution of these pedestrians crossed against the lights or two last factors in increased vehicle in platoons or when vehicles violated the delay, further data were collected at the red light. There were too few conflicts to Puffin site which had the longest mean determine any significant differences in vehicle red time (21.1 seconds). By frequency between Puffin and Pelican making adjustments to detector crossings. alignment and signal timings (to conform Puffin Signal Operation to recent guidance), the subsequent effect on vehicle delay was calculated. Delay to the first vehicle in the queue is There were significant reductions to the longer at Puffin crossings. mean delay of the first vehicle in the queue when detectors were adjusted (to The mean delay to the first vehicle in the 12.8 seconds Standard Deviation=5.8) queue was 12.6 seconds at Puffin and timings altered (to 10.29 seconds, crossings and 8.8 seconds at pelican SD= 4.45). crossings. Reasons for this may be:

4

3. Pedestrian Interviews Understanding

On-street interviews were completed Over one quarter of pedestrians show with 232 pedestrians using 10 Pelican little awareness of crossing design. crossings and 309 pedestrians using 14 Puffin crossings. To enable Nearly a quarter of respondents (27% at investigation of possible changes in Pelicans and 22% at Puffins) could not perceptions over time, six ‘new’ Puffin correctly recall which side of the road crossings (implemented within one year the green pedestrian signal had been on of the study) and eight ‘old’ Puffin the pedestrian crossing they had just crossings (implemented for more than a used. A higher proportion of users of year) were studied. older Puffin crossings recalled this. The sites were selected to have About 20% of all respondents reported comparable vehicle flows, road widths, that, according to the Highway Code, the land use and crossing control and flashing green man at a Pelican crossing spread over inner and outer London. A meant “it’s ok to start crossing the road – short observation period at each site but slower walkers might prefer to wait suggested that sites were well matched until the next green man” whereas the in terms of pedestrian flows while hourly Highway code advises that “When the vehicle flows were slightly higher at green figure begins to flash you should Pelican crossings (547 compared to 500 not start to cross”. vehicles). The observations also showed 12% of Pelican users and 17% of Puffin that 80% of pedestrians crossed on the users agreed that they are sometimes green man at all crossing sites. confused about when to start crossing Interviews were conducted over four Only 9% of users hours including one peak period at each recalled ever having seen a leaflet or site during July and August 2005. other publicity about Puffin crossings. Pedestrians were approached after they had crossed the road and the interviews Respondents were asked to rate the included a mixture of open and closed crossing they had used for ease of use. questions which aimed to determine The majority of users report that both pedestrians’ levels of understanding of types of crossing were “very” or “fairly” Pelican and Puffin crossings, their easy to use as shown in Table 3. perceptions of safety and their attitudes towards Puffin crossings.

Table 3: Respondents ratings of ease of using Puffin and Pelican crossings

Pelicans Puffins

Very easy 92 % 84 %

Fairly easy 4 % 10 %

Neither easy nor difficult 3 % 2 %

Fairly difficult 1 % 4 %

Very difficult 0 % *

Total easy 96 % 94 %

Total difficult 1 % 4 %

5

Problems experienced Over a quarter of Puffin users reported At both crossing types, the main that their view of the pedestrian signal problems for pedestrians are perceived was sometimes obscured. as cars not stopping. Respondents were asked to rate their agreement with a number of statements The 5% of respondents who had reported about the crossing they had used any difficulties were asked about the (shown in Table 5). At Pelican nature of these. Waiting too long for the crossings users were more likely to pedestrian phase and the shortness of agree that cars did not give them the pedestrian phase was mentioned for enough time to cross the road. More both Pelican and Puffin crossings. Puffin users (29%) agreed that their A few pedestrians reported insufficient view of the pedestrian signal was time to cross at Pelicans and confusion at sometimes obstructed by other people Puffin crossings. (compared to 8% at Pelican crossings). In addition, 13% of pedestrians at both Perceptions of Safety now and before Pelican and Puffin crossings reported Pedestrians generally perceive Puffin having had problems using the crossing crossings as safer than other types of at some time in the past. Pedestrians crossings. who used the crossing more frequently (3 or more days per week) were most likely At Pelican crossings 81% of to report having experienced problems in pedestrians reported feeling safe the past than infrequent users (15% compared with 91% at Puffin crossings. compared to 8%). Cars not stopping at This proportion is similar for both old the red light was a problem mentioned at and new Puffin crossings. However, both types of crossing. Some Pelican some differences in responses may crossing users also mentioned not having result from the local characteristics of enough time to cross before traffic moved each crossing such as speeding drivers off while a few Puffin crossing users or topography. This difference in mentioned equipment faults. perceived safety is wider for older pedestrians as shown in Table 4. Table 4: Overall Feelings of Safety, by age “I feel safe using this crossing to Age Group get across the road” 14-40 years 41-60 years 61+years All ages

Pedestrians at Pelicans agree 84 % 79 % 75 % 81 %

Pedestrians at Pelicans disagree 10 % 13 % 10 % 11 %

Pedestrians at Puffins agree 91 % 86 % 98 % 91 %

Pedestrians at Puffins disagree 4 % 4 % 0 % 3 %

Number of Pedestrians at Pelicans 136 56 40 232 Number of Pedestrians at Puffins 173 85 51 309

6

143 of the 309 (46%) respondents at Puffin crossings is usually associated Puffin crossings said they remembered with reductions in collision rates relative the previous crossing facility. Of these, to controls and to other formal crossing 29% felt that there was no difference in types. However, these findings are safety before and after implementation of based on only 23 Puffin crossings and, the Puffin crossing, 59% felt the Puffin since the start of the collision analysis crossing was safer and 20% had felt many more Puffin crossings have been safer on the previous crossing. implemented. Extending the collision analysis to more sites would test the Feelings of safety had increased most at robustness of this finding. Puffin sites which had replaced a Zebra or no formal crossing facility. Even at The expected reduction in shunt type Puffin crossings that had replaced collisions at Puffin crossings was not Pelican crossings, 48% of the 88 found, perhaps because there was not respondents thought the Puffin crossing enough collision data but perhaps also was safer while 30% saw no difference. because the call cancel facility was found to be rarely used. The reasons given for preferring the previous crossing included difficulties in The observational and pedestrian understanding the Puffin crossing and in interviews suggest that crossing non- seeing the green man. Reasons for compliance is no higher at Puffin preferring the Puffin crossing included crossings than at Pelican crossings. In introduction of signals and having more fact, the proportion of pedestrians who time to cross. Some respondents start crossing during the all red period at mentioned the audible signals Puffins is lower than the proportion who suggesting that, in the absence of a start crossing during the flashing green farside pedestrian signal, some man phase at Pelican sites. However, pedestrians may relay more heavily on this effect may diminish as public these. understanding of Puffin crossings and extension periods increases as Pedestrians at each type of crossing pedestrians may become less cautious were asked how often they crossed at during the all red phase. At present only that crossing against the green man. a minority of pedestrians have read Levels of self-reported non-compliance information about puffin crossings. with the crossing were similar at both Pelican (39%) and Puffin crossings The intended goal of reducing pedestrian (41%). harassment by vehicle drivers seems to have been achieved with the introduction of Puffin crossings. Pedestrians, Discussion especially older pedestrians, report These three studies provide a detailed having more time to cross the road snapshot of puffin crossings as they are before the traffic starts. This benefit for being used and operated across London pedestrians comes at a cost to vehicle today. They offer reassurance on some drivers as the first vehicle in the queue is of safety issues which have concerned delayed longer at Puffin than Pelican practitioners, while highlighting other crossings. However, these increased potential areas of concern. delays may be partly a result of misaligned detectors and conservative The available data on collisions and signal timings. conflicts suggest there is no significant difference in safety between Puffin and The subsequent data collection showed Pelican crossings and that installation of that, with fine tuning, these delays can

7

be reduced. The call cancel facility was • Only to install the call cancel facility if originally intended to offset the increased the crossing is not running pre-timed delay to the first vehicles by reducing the maximum and pedestrian flows on the delay to vehicles overall. However, footway are low. observations showed that this was rarely • Consider reviewing on-crossing used. detector positioning to establish A small minority of pedestrians did pedestrians on the crossing. mention problems with confusion • If installed, ensure kerbside detectors because of the nearside pedestrian function properly and allow signal and signal failure, the latter of cancellation of pedestrian demand. which was borne out in the observational • Distribute information and publicity on study. The few pedestrians who find the using Puffin crossing more widely to nearside signals confusing may rely pedestrians, especially at new sites. more heavily on the audible signal. • Ensure audible signals are functioning correctly as pedestrians may rely on It is interesting to note that pedestrian these more at Puffin crossings. perceptions of safety don’t always match reality. One of the situations where pedestrians perceived the greatest Next steps increases in safety (i.e. where a Puffin The studies summarised here uncovered crossing had replaced a Zebra crossing) a number of issues which could be actually shows the smallest reduction in investigated further: pedestrian collision rates. • The collision analysis provides only It may be surprising that pedestrians feel preliminary results and further analysis safer at Puffin crossings than at other should include more sites and perhaps types of crossings given anecdotal should focus on selection of relevant evidence to the contrary. This finding collisions. may be a result of the study design • Conduct interviews with non-users of which made it difficult to ask pedestrians Puffin crossings (i.e. those who for explicit comparisons between Puffin choose to cross the road away from and Pelican crossings. A pedestrian the crossing because they find it too survey comparing perceptions of safety confusing). before and after conversion of a Pelican • Conduct before and after pedestrian to a Puffin crossing which confirmed interviews at Pelican to Puffin these findings would provide added conversion sites. reassurance that this is a robust result. • Studies of atypical Puffin crossing sites could be carried out. For example puffin crossings with central Conclusions island (4). Given the evidence available from Puffin • The research presented here did not sites in London, there seems no safety assess changes in network capacity. argument against the installation of Future work could investigate whether Puffin crossings where appropriate. the provision of Puffin crossings has However, the following recommend- any impact on network capacity. dations should be considered: Also, a DfT study currently underway on Puffin crossings will give information on Puffin technology when it works perfectly and the subsequent potential benefits to road users.

8

Table 5: Levels of agreement with statements about crossing used by pedestrians Pelican Puffin Statement Rating Users Users “It is very easy to see the Strongly agree 68 % 46 % green and red men at this Tend to agree 20 % 36 % crossing” Neither agree/disagree 5 % 2 % Tend to disagree 4 % 7 % Strongly disagree 3 % 8 %

“Sometimes I can’t see the Strongly agree 1 % 8 % red and green men Tend to agree 6 % 22 % because there are people Neither agree/disagree 3 % 7 % in the way” Tend to disagree 23 % 24 % Strongly disagree 66 % 38 %

“You have to wait too long Strongly agree 12 % 10 % for the green man to Tend to agree 19 % 21 % appear” Neither agree/disagree 7 % 8 % Tend to disagree 21 % 25 % Strongly disagree 40 % 34 %

“If I start crossing when the Strongly agree 34 % 41 % green man is showing, I Tend to agree 34 % 47 % have enough time to get Neither agree/disagree 12 % 2 % across the road before the Tend to disagree 9 % 4 % traffic starts” Strongly disagree 9 % 5 %

“Sometimes the cars start Strongly agree 28 % 18 % moving off here before Tend to agree 34 % 35 % everybody has got across Neither agree/disagree 9 % 7 % the road” Tend to disagree 15 % 23 % Strongly disagree 11 % 13 %

“I feel safe using this Strongly agree 36 % 40 % crossing to get across the Tend to agree 45 % 50 % road” Neither agree/disagree 8 % 5 % Tend to disagree 6 % 2 % Strongly disagree 5 % 1 %

Number of Pedestrians interviewed (232) (309)

9

Studies summarised Outlook research (2005): Puffin and Pelican crossings. Views of pedestrian users. Research report prepared for the London Road Safety Unit. Walker, R., Winnett, M., Martin, A., Kennedy, J. (2005): Puffin crossing operation and behaviour study. Unpublished TRL report for the London Road Safety Unit. Webster, N. (2006): The effect of newly installed Puffin crossings on collisions. London Road Safety Unit research report.

Selected references (1) Department for Environment, Transport and the Regions, (2001): Puffin Pedestrian Crossings. Traffic Advisory Leaflet 1/01, DETR. (2) Highways Agency (1997): Puffin Crossing Study; An Analysis of Collisions at Five Locations where Pelican Crossing were converted to Puffin Method Control. London Research Centre. (3) Davies, H.E.H (1992): The Puffin Pedestrian Crossing: Experience with the First Experimental Sites. TRL Research report 364. Transport Research Laboratory, Crowthorne: Berks. (4) York. I. & Gutteridge, S. (2006): Assessment of Puffin Control Strategy. Unpublished TRL report for the London Borough of Lambeth.

10