p2w in bus lanes safety study v3 5 - bad science

65
London Road Safety Unit 1 DRAFT 22/02/2007 P2W in bus lanes safety study (V3_5) Written by: John Devenport Reviewed by: Chris Lines London Road Safety Unit Head of LRSU Road Network Performance Cleared by: Nick Morris Transport for London Street Management Director of RNP February 2007 Summary The study investigates the effect on collisions as part of a trial to permitting powered two- wheeled vehicles (P2Ws) to use bus lanes along three routes along the TLRN (A13, A23 & A41), using 36 month before and after collision data. The study forms part of a wider review of the trial by TfL based on traffic and user data. Analyses of collision types within the length of the routes containing bus lanes showed that overall, relative to control data, there was a 4% decrease in total collisions during bus lane operating hours. This could have occurred due to chance. For collisions involving a P2W during operating hours, there was a 1% increase, although again not statistically significant. For P2W collisions during operating hours, those resulting in injury to pedestrians increased by 3% and those resulting in injury to the P2W user increased by 6%, although neither increase was statistically significant. There were very few collisions involving a P2W and a pedal cyclist, but these showed a decrease. Most of the other collision types analysed showed small changes (some increases and some decreases) although almost all of them were not significant. Generally, however, it was found that there was significant variation between the trial routes. Detailed analysis of individual collisions involving P2Ws along the trial routes showed several changes in the conflict types, particularly involving right turning vehicles in conflict with the P2W during operating hours. Those involving the vehicle turning right across the P2W from the opposite direction increased whilst those turning right from a side road decreased. Increases in collisions were found for both those involving the P2W overtaking on the nearside and those reported as actually occurring in the bus lane. Overall, from this study, there was no clear evidence to suggest that permitting P2Ws in bus lanes provides safety benefits. A study of collisions before and after permitting use of bus lanes by powered two-wheelers on three trial routes on the Transport for London Road Network

Upload: others

Post on 30-May-2022

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 1

DRAFT 22/02/2007 P2W in bus lanes safety study (V3_5)

Written by: John Devenport Reviewed by: Chris Lines London Road Safety Unit Head of LRSU Road Network Performance Cleared by: Nick Morris Transport for London Street Management Director of RNP

February 2007

Summary The study investigates the effect on collisions as part of a trial to permitting powered two-wheeled vehicles (P2Ws) to use bus lanes along three routes along the TLRN (A13, A23 & A41), using 36 month before and after collision data. The study forms part of a wider review of the trial by TfL based on traffic and user data.

Analyses of collision types within the length of the routes containing bus lanes showed that overall, relative to control data, there was a 4% decrease in total collisions during bus lane operating hours. This could have occurred due to chance. For collisions involving a P2W during operating hours, there was a 1% increase, although again not statistically significant. For P2W collisions during operating hours, those resulting in injury to pedestrians increased by 3% and those resulting in injury to the P2W user increased by 6%, although neither increase was statistically significant. There were very few collisions involving a P2W and a pedal cyclist, but these showed a decrease. Most of the other collision types analysed showed small changes (some increases and some decreases) although almost all of them were not significant. Generally, however, it was found that there was significant variation between the trial routes.

Detailed analysis of individual collisions involving P2Ws along the trial routes showed several changes in the conflict types, particularly involving right turning vehicles in conflict with the P2W during operating hours. Those involving the vehicle turning right across the P2W from the opposite direction increased whilst those turning right from a side road decreased. Increases in collisions were found for both those involving the P2W overtaking on the nearside and those reported as actually occurring in the bus lane.

Overall, from this study, there was no clear evidence to suggest that permitting P2Ws in bus lanes provides safety benefits.

A study of collisions before and after permitting use of bus lanes by powered two-wheelers on three trial routes on the Transport for London Road Network

Page 2: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 2

Page 3: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 3

Contents 1. Introduction ..................................................................................................5

1.1 Background .........................................................................................5

2. Aims and Objectives ....................................................................................6

3. Method.........................................................................................................7

3.1 Site selection.......................................................................................7

3.2 Source of collision and casualty data ..................................................7

3.3 Study route definitions.........................................................................7

3.4 Study periods ......................................................................................8

3.5 Statistical analysis ...............................................................................8

3.6 Collision and casualty data analysed ..................................................9

3.7 Control data.......................................................................................10

3.8 Detailed collision conflict investigation ..............................................12

4. Results and commentary ........................................................................13

4.1 Changes in collisions.........................................................................13

4.1.1 Collisions involving any casualties ....................................................14

4.1.2 Collisions involving pedestrian casualties .........................................16

4.1.3 Collisions involving pedal cyclist casualties.......................................18

4.1.4 Collisions involving P2W casualties ..................................................20

4.1.5 Collisions involving bus occupant casualties.....................................22

4.1.6 Collisions Involving car occupant casualties .....................................24

4.1.7 Collisions involving killed or seriously injured casualties ...................26

4.2 Powered two-wheeler collision rates on trial routes...........................28

4.2.1 Calculation of P2W collision rates .....................................................28

4.2.2 Changes in powered two wheeler collision rates...............................28

4.3 Detailed conflict analysis ...................................................................31

4.3.1 Conflicts in P2W collisions during bus lane hours of operation .........31

4.3.2 Conflicts in P2W collisions outside of bus lane hours of operation....36

4.3.3 Collisions reported as being in bus lane............................................40

4.3.4 Collisions with P2W reported as overtaking on nearside...................41

Page 4: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 4

5. Conclusions ...............................................................................................42

5.2 Detailed before and after analysis .....................................................42

5.3 P2W collision rates............................................................................43

5.4 Detailed conflict analysis ...................................................................43

6. References.................................................................................................45

Appendices ....................................................................................................46

Appendix 1: Study route location maps .......................................................47

Appendix 2: Collision data summary table ...................................................52

Appendix 3: Casualty data summary table...................................................56

Appendix 4: Comparative test and control collision data summary..............59

Appendix 5: A solution to the problem of combining accident frequencies from

a number of sites...............................................................................63

Page 5: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 5

1. Introduction 1.1 Background This report presents an analysis of the effect on road safety of permitting powered two wheelers (P2Ws) in bus lanes on a trial basis along three routes on the Transport for London Road Network (TLRN). It forms part of a comprehensive review by TfL of the trial looking at changes in usage, speeds, bus journey times and road user attitudes.

The Mayor’s Transport Strategy (Proposal 4G.1) committed TfL to consider trials allowing P2Ws into bus lanes as a means of potentially reducing the exposure of P2W riders to general traffic in order to improve safety. P2W killed or seriously injured casualties have been the road user category that has shown least progress towards the 2010 casualty reduction targets, although it is recognised that this has been against a background of considerable growth in ownership and usage. Arguments against permitting P2Ws in bus lanes centre on the principle of private motorised vehicles being allowed access to bus lanes, additional vehicles in bus lanes having a negative impact on other users (notably buses and cyclists), and a potential increase in road safety conflicts particularly between P2Ws and cyclists, and P2Ws and pedestrians. Arguments for permitting P2Ws in bus lanes include encouragement of motorcycling and improved safety.

In September and October 2002, TfL introduced three pilot schemes on TLRN whereby P2Ws (motorcycles, mopeds and scooters) were permitted to use bus lanes along the three routes during the hours of operation.

The three pilot routes implemented were:

• the A13 between East India Dock Road (between Leamouth Road and Butcher Row), with hours of operation 7am to 7pm, Monday to Friday.

• the A23 Streatham High Road / Brixton Hill / Brixton Road (between Streatham Common and Camberwell New Road), with hours of operation 7 to 10am and 4 to 7pm, Monday to Friday.

• the A41 Finchley Road (between Marlborough Place and Platt’s Lane) with hours of operation 7 to 10am and 4 to 7pm, Monday to Friday.

All three routes have with-flow bus lanes in each direction along most of their length. Maps of the location of each route are shown in Appendix 1.

Page 6: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 6

2. Aims and Objectives The objectives of this study were:

• to provide an assessment of the changes in collisions and casualties along the parts of the three TLRN routes containing bus lanes, before and after implementation of the bus lane trial schemes, for all collisions and for those involving P2Ws, taking account of background trends by comparing with suitable control data; and

• to provide a detailed before and after analysis of some of the main conflict patterns for collisions involving P2Ws in the sections of routes containing bus lanes, to help identify any change in the type of conflicts with other road users as a result of the trials

The study provides a final report on the full three-year before and after assessment now that the full three years after data are now available.

Page 7: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 7

3. Method 3.1 Site selection The study investigated three routes on the TLRN chosen for the P2W in bus lanes trial. Details of the routes are given in Table 1 and also in Appendix 1.

Table 1: Trial routes for P2W in bus lanes

Route Name between Borough(s) Hours of operation Implemented

A13 East India Dock Road

Leamouth Road & Butcher Row

Tower Hamlets

7am to 7pm Mon-Fri

October 2002

A23 Streatham High Road / Brixton Road

Streatham Common & Camberwell New Road

Lambeth 7-10am & 4-7pm Mon-Fri

October 2002

A41 Finchley Road

Marlborough Place & Platt’s Lane

Camden & Westminster

7-10am & 4-7pm Mon-Fri

September 2002

3.2 Source of collision and casualty data The collision and casualty data used in this study is the Stats19 data held by TfL London Road Safety Unit for the Greater London area. This data is for collisions occurring on the public highway reported to the police, involving one or more vehicle and resulting in injury to one or more casualties. It is collected in London by the Metropolitan and City Police Forces and passed on to TfL for use by the highway and traffic authorities in London.

3.3 Study route definitions For this study, each of the routes was defined in terms of the nodes and links on the LAAU representation of the classified road network. The nodes are at the junctions of classified roads and the links are the classified roads between the nodes. The routes included the length of the routes described in which there was a bus lane, and were used as the basis of collision and casualty extraction for the analysis. Note that the definitions used in this study were slightly different to those used in previous interim reports on the trial, and were amended as a result of comments and suggestions made on the original analysis presented at a stakeholder seminar held by TfL in September 2006. It is important to recognise that the collisions and casualties analysed in the study are for those along the whole of the route sections selected and not just those that have been recorded as being explicitly associated with a bus lane in the Stats19 data, of which there are relatively few.

Page 8: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 8

3.4 Study periods The study periods for the analysis before and after implementation of the trial schemes were 36 months for each route, the usual length of time for this type of study. The month of implementation for each route was excluded from the analysis, to reflect the period when the changes were being introduced.

3.5 Statistical analysis The Tanner test was performed on a range of collision categories from the three trial sites (See 3.6). This aims to determine the effect of a given measure by combining the data over several sites. To conduct this test, appropriate control data were required (See 3.7).

The Tanner test uses two parameters to determine the statistical significance of differences in collision data before and after implementation of the P2W in bus lane schemes;

1. “t” is used to test the overall effect of the P2W in bus lane trial over all the sites combined, relative to control data;

2. “χ2” is used to test whether the effect of the P2W in bus lane trial varied between the sites groups.

In this study the 5% significance level was used as a threshold. Where the number of collisions both before and after introduction was zero, the zeros were substituted by 0.1 to allow the Tanner test to run correctly.

Full details of the Tanner test are given in Appendix 5

Note that only the collision data was subject to the rigorous statistical analysis, which is appropriate for events (collisions) rather than the outcomes (casualties). However, casualty data is presented for a number of categories to illustrate the scale of the changes relative to the control.

Page 9: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 9

3.6 Collision and casualty data analysed For each of the three trial routes the following collision data were extracted for the appropriate before and after periods and were used for the subsequent statistical analysis (Table 3)

Table 3: Collision categories analysed

Collision category

1. All collisions, all severities, all hours

2. All collisions, all severities, bus lane hours of operation

3. P2W collision, all severities, all hours

4. P2W collision, all severities, bus lane hours of operation

5. All collisions, involving pedestrian casualty, all severities, all hours

6. All collisions, involving pedestrian casualty, all severities, bus lane hours of operation

7. P2W collision, involving pedestrian casualty, all severities, all hours

8. P2W collision, involving pedestrian casualty, all severities, bus lane hours of operation

9. All collisions, involving pedal cyclist casualty, all severities, all hours

10. All collisions, involving pedal cyclist casualty, all severities, bus lane hours of operation

11. P2W collision, involving pedal cyclist casualty, all severities, all hours

12. P2W collision, involving pedal cyclist casualty, all severities, bus lane hours of operation

13. All collisions, involving P2W user casualty, all severities, all hours

14. All collisions, involving P2W user casualty, all severities, bus lane hours of operation

15. P2W collision, involving P2W user casualty, all severities, all hours

16. P2W collision, involving P2W user casualty, all severities, bus lane hours of operation

17. All collisions, involving bus occupant casualty, all severities, all hours

18. All collisions, involving bus occupant casualty, all severities, bus lane hours of operation

19. P2W collision, involving bus occupant casualty, all severities, all hours

20. P2W collision, involving bus occupant casualty, all severities, bus lane hours of operation

21. All collisions, involving car occupant casualty, all severities, all hours

22. All collisions, involving car occupant casualty, all severities, bus lane hours of operation

23. P2W collision, involving car occupant casualty, all severities, all hours

24. P2W collision, involving car occupant casualty, all severities, bus lane hours of operation

25. All collisions, involving KSI casualty severities, all hours

26. All collisions, involving KSI casualty severities, bus lane hours of operation

27. P2W collision, involving KSI casualty severities, all hours

28. P2W collision, involving KSI casualty severities, bus lane hours of operation

Page 10: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 10

In addition, data for the following casualty categories were extracted and presented, although not subjected to the same rigorous statistical analysis, Casualties, which are the outcome of events (collisions), may not be independent from each other and therefore not suitable for statistical analysis. (Table 4)

Table 4: Casualty categories analysed

Casualty category

1. All collisions, all casualties, all severities, all hours

2. All collisions, all casualties, all severities, operating hours

3. All collisions, P2W casualties, all severities, all hours

4. All collisions, P2W casualties, all severities, operating hours

5. All collisions, pedal cyclist casualties, all severities, all hours

6. All collisions, pedal cyclist casualties, all severities, operating hours

7. All collisions, pedestrian casualties, all severities, all hours

8. All collisions, pedestrian casualties, all severities, operating hours

9. P2W collisions, all casualties, all severities, all hours

10. P2W collisions, all casualties, all severities, operating hours

11. P2W collisions, P2W casualties, all severities, all hours

12. P2W collisions, P2W casualties, all severities, operating hours

13. P2W collisions, pedal cyclist casualties, all severities, all hours

14. P2W collisions, pedal cyclist casualties, all severities, operating hours

15. P2W collisions, pedestrian casualties, all severities, all hours

16. P2W collisions, pedestrian casualties, all severities, operating hours

17. All collisions, all casualties, KSI severity, all hours

18. All collisions, all casualties, KSI severity, operating hours

19. P2W collisions, all casualties, KSI severity, all hours

20. P2W collisions, all casualties, KSI severity, operating hours

21. P2W collisions, P2W casualties, KSI severity, all hours

22. P2W collisions, P2W casualties, KSI severity, operating hours

3.7 Control data There are limitations on the type of control data that can be used in the Tanner test. Ideally the control data should be at least 10 times as large as the trial data and the two sets of data should be mutually exclusive. Therefore, to enable the use of statistical tests, collision figures for the TLRN were used as the basis of the control. This was sufficiently large compared with the trial data and also representative of the types of roads on which the trial routes were located. It

Page 11: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 11

was recognised that the Congestion Charging scheme would have influenced the collision levels in the Central London area, so collisions and casualties on the sections of the TLRN within the Congestion Charging Zone (CCZ) and the Inner Ring Road itself were excluded from the control data.

To ensure that each trial site had mutually exclusive control data, the data for the whole TLRN (minus the CCZ) was divided into three parts, each comprising the boroughs in the vicinity of the three trial routes. For the statistical analysis, the test data for the whole lengths of the three trial sites were subtracted from the TLRN minus CCZ control data to give a Net value. For the purposes of this report the ‘TLRN minus CCZ’ will be referred to as the ‘TLRN’ for the rest of the report.

The collision data on the TLRN from the following boroughs were used as control data.

Table 5 Groupings of boroughs for TLRN control data

Route A13 A23 A41 Control boroughs

City of London Islington Hackney Tower Hamlets Waltham Forest Redbridge Havering Barking & Dagenham Newham

Greenwich Lewisham Southwark Lambeth Wandsworth Bexley Bromley Croydon Sutton Merton Kingston-upon Thames Richmond-upon Thames

Westminster Camden Hammersmith & Fulham Kensington & Chelsea Hounslow Hillingdon Ealing Brent Barnet Haringey Enfield

NB. No TLRN in Harrow

TLRN control data were extracted for the equivalent collision categories analysed, e.g. for all times, for operating day/times etc.

In earlier reports providing interim results on these P2W in bus lane trials, ‘control’ data for Greater London and two other bus lane route sections (on parts of the A5 and A10) had been shown for comparison purposes. Figures for these areas/routes are shown for comparison in Appendix 4 but have not been used as actual control data in the statistical testing. Collision data for the A5 and A10 were far too small to use as controls in the Tanner test, particularly for some of the smaller collision categories analysed, and there is considerable fluctuation in

Page 12: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 12

the data for these two routes compared with the changes observed for the larger control data for the TLRN and also the whole Greater London area.

To test the robustness of the TLRN control data, statistical analysis of a sample of collision categories was carried out using Greater London collision data as a control, to compare results using the TLRN controls. The same borough groupings were used as shown in Table 5. Overall, the results obtained with Greater London control data were very similar to those using the TLRN data, so it was decided that all testing would use the TLRN control data, which it was considered most robust and representative of the types of routes included in the trial.

3.8 Detailed collision conflict investigation In addition to the statistical analysis of the collision data along the lengths of route with bus lanes along the study routes, a detailed investigation was carried out of the conflicts occurring for collisions involving P2Ws before and after the start of the trials. This was to gain a greater understanding of how conflict patterns might have changed. This involved reconstructing the conflict occurring in each P2W collision using the full Stats19 data and then grouping these conflicts into a number of categories to determine what changes (if any) had occurred in each collision category. Although the overall number of P2W collisions is relatively large, the number of collisions in most of the conflict categories is generally very small, so that it was not appropriate to carry out rigorous statistical analysis for this part of the study.

In addition, as part of this analysis, note was made of any collisions that were directly associated with the bus lane, recorded either as part of the ‘Vehicle location – restricted lanes’ field (Stats19 field 2.9) or mentioned explicitly in the text description of the collision circumstances. In addition to the direct association with the bus lane, note was also made of any P2W collisions where the P2W manoeuvre was recorded as ‘Overtaking on the nearside’.

The findings from this part of the analysis are presented in Section 4.2.

Page 13: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 13

4. Results and commentary 4.1 Changes in collisions Table 6 gives the overall basic summary of the percentage changes occurring across all of the sites combined together with an indication of the significance level of the changes for all of the cases investigated, using collisions on the TLRN as a control. The table also gives an indication of whether there was any variation between the trial routes. Table 6: Overall summary table

Collision type Casualty type Severity Hours Overall Significance Significant % change of overall variationrelative to change between

control routes?

1 All collisions All All All -2% NSD Yes

2 All collisions All All Operating hours -4% NSD Yes

3 P2W collisions All All All -3% NSD Yes

4 P2W collisions All All Operating hours +1% NSD Yes

5 All collisions Pedestrian All All -2% NSD Yes

6 All collisions Pedestrian All Operating hours -7% NSD Yes

7 P2W collisions Pedestrian All All -13% NSD Yes

8 P2W collisions Pedestrian All Operating hours +3% NSD Yes

9 All collisions Pedal cyclist All All -12% NSD Yes

10 All collisions Pedal cyclist All Operating hours +7% NSD Yes

11 P2W collisions Pedal cyclist All All -14% NSD Yes

12 P2W collisions Pedal cyclist All Operating hours -58% NSD Yes

13 All collisions P2W user All All +2% NSD Yes

14 All collisions P2W user All Operating hours +6% NSD Yes

15 P2W collisions P2W user All All +2% NSD Yes

16 P2W collisions P2W user All Operating hours +6% NSD Yes

17 All collisions Bus occupants All All -4% NSD Yes

18 All collisions Bus occupants All Operating hours +24% NSD Yes

19 P2W collisions Bus occupants All All ∞ NSD Yes

20 P2W collisions Bus occupants All Operating hours ∞ NSD Yes

21 All collisions Car occupants All All +1% NSD Yes

22 All collisions Car occupants All Operating hours -28% 5% level Yes

23 P2W collisions Car occupants All All +77% NSD No

24 P2W collisions Car occupants All Operating hours +47% NSD Yes

25 All collisions All KSI collision All +7% NSD Yes

26 All collisions All KSI collision Operating hours -3% NSD Yes

27 P2W collisions All KSI collision All +15% NSD Yes

28 P2W collisions All KSI collision Operating hours +29% NSD Yes

NSD=No significant difference It is important to be aware that changes in some categories are based on very small collision numbers which are shown in the subsequent tables and also in Appendix 2. Summary tables for the casualty categories analysed are presented in Appendix 3, although they are not subject to the rigorous statistical analysis. (See Section 3.5). Most of the casualty categories have an almost equivalent collision category that has been subject to robust statistical analysis, and shows a similar picture for casualties as would be expected.

Page 14: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 14

4.1.1 Collisions involving any casualties

Table 7

Cas

eC

ollis

ion

type

Cas

ualty

type

Seve

rity

Hou

rsR

oute

TEST

DAT

A - B

US

LAN

E O

NLY

CO

NTR

OL

DAT

A (N

ET o

f Tes

t + C

CZ)

Site

Si

te

Ove

rall

Sign

ifica

nce

Sign

ifica

nt%

cha

nge

% c

hang

e K

-val

ue%

cha

nge

% c

hang

eof

ove

rall

varia

tion

Bef

ore

Afte

r(s

impl

e)C

ontr

olB

efor

eAf

ter

(sim

ple)

rela

tive

to

rela

tive

toch

ange

betw

een

cont

rol

cont

rol

rout

es?

1Al

l col

lisio

nsAl

lAl

lAl

lA1

316

314

5-1

1%TL

RN

6,27

75,

027

-20%

1.11

11%

A23

653

493

-25%

9,64

58,

090

-16%

0.90

-10%

A41

199

177

-11%

7,93

96,

237

-21%

1.13

13%

All

1015

815

-20%

23,8

6119

,354

-19%

-2%

NSD

Yes

2Al

l col

lisio

nsAl

lAl

lO

pera

ting

hour

sA1

391

81-1

1%TL

RN

3,68

03,

039

-17%

1.08

8%A2

322

717

6-2

2%3,

325

2,89

2-1

3%0.

89-1

1%A4

175

61-1

9%2,

493

1,97

4-2

1%1.

033%

All

393

318

-19%

9,49

87,

905

-17%

-4%

NSD

Yes

3P2

W c

ollis

ions

All

All

All

A13

5250

-4%

TLR

N1,

506

1,15

1-2

4%1.

2626

%A2

319

913

9-3

0%2,

932

2,42

3-1

7%0.

85-1

5%A4

158

51-1

2%1,

711

1,29

9-2

4%1.

1616

%Al

l30

924

0-2

2%6,

149

4,87

3-2

1%-3

%N

SDYe

s4

P2W

col

lisio

nsAl

lAl

lO

pera

ting

hour

sA1

336

398%

TLR

N1,

019

798

-22%

1.38

38%

A23

9068

-24%

1,35

41,

183

-13%

0.86

-14%

A41

2823

-18%

757

594

-22%

1.05

5%Al

l15

413

0-1

6%3,

130

2,57

5-1

8%+1

%N

SDYe

s

Page 15: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 15

Table 7 shows that at all times for the trial sites combined, there was a 2% decrease in all collisions relative to the control, but this change was not statistically significant and could be due to chance. During the bus lane hours of operation all collisions decreased by 4% relative to the control data, but again this was not significant.

With regards to collisions involving a powered two-wheeler, for all times there was a 3% decrease relative to the control data, but again not significant. During bus lane hours of operation, collisions involving a P2W along the routes increased by 1% but again this change was not significant.

It was also found however, that for most of these categories there was significant variation in terms of changes in collisions between the trial routes, which indicates that any changes might not be repeatable at other locations.

Page 16: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 16

4.1.2 Collisions involving pedestrian casualties

Table 8

Cas

eC

ollis

ion

type

Cas

ualty

type

Seve

rity

Hou

rsR

oute

TEST

DAT

A - B

US

LAN

E O

NLY

CO

NTR

OL

DAT

A (N

ET o

f Tes

t + C

CZ)

Site

Si

te

Ove

rall

Sign

ifica

nce

Sign

ifica

nt%

cha

nge

% c

hang

e K

-val

ue%

cha

nge

% c

hang

eof

ove

rall

varia

tion

Bef

ore

Afte

r(s

impl

e)C

ontr

olB

efor

eAf

ter

(sim

ple)

rela

tive

to

rela

tive

toch

ange

betw

een

cont

rol

cont

rol

rout

es?

5Al

l col

lisio

nsPe

dest

rian

All

All

A13

4532

-29%

TLR

N96

774

9-2

3%0.

92-8

%A2

316

512

3-2

5%1,

676

1,39

8-1

7%0.

89-1

1%A4

141

4715

%97

180

2-1

7%1.

3939

%Al

l25

120

2-2

0%3,

614

2,94

9-1

8%-2

%N

SDYe

s6

All c

ollis

ions

Pede

stria

nAl

lO

pera

ting

hour

sA1

327

19-3

0%TL

RN

547

424

-22%

0.91

-9%

A23

6236

-42%

568

468

-18%

0.70

-30%

A41

1623

44%

296

231

-22%

1.84

84%

All

105

78-2

6%1,

411

1,12

3-2

0%-7

%N

SDYe

s7

P2W

col

lisio

nsPe

dest

rian

All

All

A13

107

-30%

TLR

N15

811

3-2

8%0.

98-2

%A2

328

13-5

4%25

923

6-9

%0.

51-4

9%A4

15

1112

0%13

410

2-2

4%2.

8918

9%Al

l43

31-2

8%55

145

1-1

8%-1

3%N

SDYe

s8

P2W

col

lisio

nsPe

dest

rian

All

Ope

ratin

g ho

urs

A13

77

0%TL

RN

116

83-2

8%1.

4040

%A2

314

7-5

0%13

711

9-1

3%0.

58-4

2%A4

13

610

0%54

45-1

7%2.

4014

0%Al

l24

20-1

7%30

724

7-2

0%+3

%N

SDYe

s

Page 17: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 17

London has a large number of vulnerable road users therefore it is important to consider any change in collisions involving vulnerable road users, particularly those involving pedestrians, cyclists and P2W users. For all collisions resulting in injury to pedestrians, Table 8 shows that there were decreases relative to the control data of 2% and 7% at all times and during bus lane hours of operation respectively, although neither of these changes were statistically significant. For collisions involving a P2W resulting in injury to pedestrians there was a 13% decrease relative to the control data at all times, but during bus lane hours of operation an increase of 3% relative to control was observed. However, neither of these changes was significant, and this may partly be a reflection of the relatively small numbers of pedestrians injured in collisions with P2Ws. For all of these categories there was significant variation between the three routes.

Page 18: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 18

4.1.3 Collisions involving pedal cyclist casualties

Table 9

Cas

eC

ollis

ion

type

Cas

ualty

type

Seve

rity

Hou

rsR

oute

TEST

DAT

A - B

US

LAN

E O

NLY

CO

NTR

OL

DAT

A (N

ET o

f Tes

t + C

CZ)

Site

Si

te

Ove

rall

Sign

ifica

nce

Sign

ifica

nt%

cha

nge

% c

hang

e K

-val

ue%

cha

nge

% c

hang

eof

ove

rall

varia

tion

Bef

ore

Afte

r(s

impl

e)C

ontr

olB

efor

eAf

ter

(sim

ple)

rela

tive

to

rela

tive

toch

ange

betw

een

cont

rol

cont

rol

rout

es?

9Al

l col

lisio

nsPe

dal c

yclis

tAl

lAl

lA1

34

40%

TLR

N45

642

5-7

%1.

077%

A23

7355

-25%

871

833

-4%

0.79

-21%

A41

2021

5%41

335

7-1

4%1.

2121

%Al

l97

80-1

8%1,

740

1,61

5-7

%-1

2%N

SDYe

s10

All c

ollis

ions

Peda

l cyc

list

All

Ope

ratin

g ho

urs

A13

23

50%

TLR

N31

926

7-1

6%1.

7979

%A2

333

32-3

%44

041

2-6

%1.

044%

A41

1312

-8%

191

168

-12%

1.05

5%Al

l48

47-2

%95

084

7-1

1%+7

%N

SDYe

s11

P2W

col

lisio

nsPe

dal c

yclis

tAl

lAl

lA1

30.

10.

10%

TLR

N14

2043

%0.

70-3

0%A2

31

0-1

00%

2928

-3%

0.00

-100

%A4

13

30%

1311

-15%

1.18

18%

All

4.1

3.1

-24%

5659

5%-1

4%N

SDYe

s12

P2W

col

lisio

nsPe

dal c

yclis

tAl

lO

pera

ting

hour

sA1

30.

10.

10%

TLR

N10

1551

%0.

66-3

4%A2

30.

10.

10%

1716

-6%

1.06

6%A4

13

1-6

7%7

6-1

4%0.

39-6

1%Al

l3.

21.

2-6

3%34

379%

-58%

NSD

No

Page 19: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 19

Considering collisions involving injury to another key group of vulnerable road users, pedal cyclists, Table 9 shows that there was a decrease relative to control in pedal cyclists of 12% at all times, but a 7% increase relative to control during bus lane hours of operation. Once again, neither of these changes was statistically significant.

For collisions involving a P2W resulting in injury to pedal cyclists, the numbers were extremely small, but there was an decrease relative to control of 14% at all times and a decrease of 58% relative to control during bus lane hours of operation. None of the changes were significant.

For all four of these categories, there was significant variation between the three routes in the trial.

Page 20: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 20

4.1.4 Collisions involving P2W casualties

Table 10

Cas

eC

ollis

ion

type

Cas

ualty

type

Seve

rity

Hou

rsR

oute

TEST

DAT

A - B

US

LAN

E O

NLY

CO

NTR

OL

DAT

A (N

ET o

f Tes

t + C

CZ)

Site

Si

te

Ove

rall

Sign

ifica

nce

Sign

ifica

nt%

cha

nge

% c

hang

e K

-val

ue%

cha

nge

% c

hang

eof

ove

rall

varia

tion

Bef

ore

Afte

r(s

impl

e)C

ontr

olB

efor

eAf

ter

(sim

ple)

rela

tive

to

rela

tive

toch

ange

betw

een

cont

rol

cont

rol

rout

es?

13Al

l col

lisio

nsP2

W u

ser

All

All

A13

4446

5%TL

RN

1,42

21,

064

-25%

1.40

40%

A23

180

135

-25%

2,78

62,

289

-18%

0.91

-9%

A41

5647

-16%

1,63

01,

256

-23%

1.09

9%Al

l28

022

8-1

9%5,

838

4,60

9-2

1%+2

%N

SDYe

s14

All c

ollis

ions

P2W

use

rAl

lO

pera

ting

hour

sA1

331

3410

%TL

RN

948

726

-23%

1.43

43%

A23

8167

-17%

1,26

91,

106

-13%

0.95

-5%

A41

2620

-23%

719

573

-20%

0.97

-3%

All

138

121

-12%

2,93

62,

405

-18%

+6%

NSD

Yes

15P2

W c

ollis

ions

P2W

use

rAl

lAl

lA1

344

465%

TLR

N1,

422

1,06

4-2

5%1.

4040

%A2

318

013

5-2

5%2,

786

2,28

9-1

8%0.

91-9

%A4

156

47-1

6%1,

630

1,25

6-2

3%1.

099%

All

280

228

-19%

5,83

84,

609

-21%

+2%

NSD

Yes

16P2

W c

ollis

ions

P2W

use

rAl

lO

pera

ting

hour

sA1

331

3410

%TL

RN

948

726

-23%

1.43

43%

A23

8167

-17%

1,26

91,

106

-13%

0.95

-5%

A41

2620

-23%

719

571

-21%

0.97

-3%

All

138

121

-12%

2,93

62,

403

-18%

+6%

NSD

Yes

Page 21: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 21

One of the major collision categories of interest in this evaluation are those involving injury to a P2W users.

Table 10 shows that for all collisions involving a P2W and resulting injury to a P2W user, at all times there was a 2% increase relative to control, whereas during the bus lane hours of operation there was an 6% increase.

Neither of these changes was statistically significant but again, once again there was significant variation between routes.

Page 22: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 22

4.1.5 Collisions involving bus occupant casualties

Table 11

Cas

eC

ollis

ion

type

Cas

ualty

type

Seve

rity

Hou

rsR

oute

TEST

DAT

A - B

US

LAN

E O

NLY

CO

NTR

OL

DAT

A (N

ET o

f Tes

t + C

CZ)

Site

Si

te

Ove

rall

Sign

ifica

nce

Sign

ifica

nt%

cha

nge

% c

hang

e K

-val

ue%

cha

nge

% c

hang

eof

ove

rall

varia

tion

Bef

ore

Afte

r(s

impl

e)C

ontr

olB

efor

eAf

ter

(sim

ple)

rela

tive

to

rela

tive

toch

ange

betw

een

cont

rol

cont

rol

rout

es?

17Al

l col

lisio

nsBu

s oc

cAl

lAl

lA1

316

176%

TLR

N37

730

8-1

8%1.

3030

%A2

385

73-1

4%72

364

4-1

1%0.

96-4

%A4

120

17-1

5%26

430

415

%0.

74-2

6%Al

l12

110

7-1

2%1,

364

1,25

6-8

%-4

%N

SDYe

s18

All c

ollis

ions

Bus

occ

All

Ope

ratin

g ho

urs

A13

717

143%

TLR

N26

421

8-1

7%2.

9419

4%A2

321

18-1

4%18

018

63%

0.83

-17%

A41

67

17%

6178

28%

0.91

-9%

All

3442

24%

505

482

-5%

+24%

NSD

Yes

19P2

W c

ollis

ions

Bus

occ

All

All

A13

0.1

2.1

2000

%TL

RN

00

-200

%-2

1.00

-220

0%A2

30.

10.

10%

22

0%1.

000%

A41

0.1

0.1

0%1

0-1

11%

-9.0

0-1

000%

All

0.3

2.3

667%

32

-30%

∞N

SDYe

s20

P2W

col

lisio

nsBu

s oc

cAl

lO

pera

ting

hour

sA1

30.

12.

120

00%

TLR

N0

00%

21.0

020

00%

A23

0.1

0.1

0%1

10%

1.00

0%A4

10.

10.

10%

10

-90%

10.0

090

0%Al

l0.

32.

366

7%2

1-5

0%∞

NSD

Yes

Page 23: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 23

Collisions resulting in injuries to bus occupants are clearly of interest in a study on bus lanes.

For all collisions in which a bus occupant was injured, Table 11 shows that at all times there was an increase of 2% relative to control, but this change was not significant.

For all collisions in which a bus occupant was injured during bus lane operating hours there was a 41% increase relative to control, although not significant , but this does not appear to be directly associated with P2W involvement, as described in the following paragraph.

For P2W collisions resulting in injury to a bus occupant, there were no collisions before introduction of the scheme, and only a very small number after (two), so that there was no significant change. Both of these occurred during bus lane hours of operation on the A13.

For all of these cases there was significant variation between the routes.

Page 24: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 24

4.1.6 Collisions Involving car occupant casualties

Table 12

Cas

eC

ollis

ion

type

Cas

ualty

type

Seve

rity

Hou

rsR

oute

TEST

DAT

A - B

US

LAN

E O

NLY

CO

NTR

OL

DAT

A (N

ET o

f Tes

t + C

CZ)

Site

Si

te

Ove

rall

Sign

ifica

nce

Sign

ifica

nt%

cha

nge

% c

hang

e K

-val

ue%

cha

nge

% c

hang

eof

ove

rall

varia

tion

Bef

ore

Afte

r(s

impl

e)C

ontr

olB

efor

eAf

ter

(sim

ple)

rela

tive

to

rela

tive

toch

ange

betw

een

cont

rol

cont

rol

rout

es?

21Al

l col

lisio

nsC

ar o

ccAl

lAl

lA1

394

80-1

5%TL

RN

4,32

63,

387

-22%

1.09

9%A2

322

717

1-2

5%5,

304

4,31

6-1

9%0.

93-7

%A4

182

70-1

5%6,

449

4,76

8-2

6%1.

1515

%Al

l40

332

1-2

0%16

,079

12,4

71-2

2%+1

%N

SDYe

s22

All c

ollis

ions

Car

occ

All

Ope

ratin

g ho

urs

A13

3827

-29%

TLR

N1,

888

1,49

5-2

1%0.

90-1

0%A2

352

30-4

2%1,

267

1,06

4-1

6%0.

69-3

1%A4

121

8-6

2%1,

541

1,17

1-2

4%0.

50-5

0%Al

l11

165

-41%

4,69

63,

730

-21%

-28%

5% le

vel

Yes

23P2

W c

ollis

ions

Car

occ

All

All

A13

22

0%TL

RN

5549

-11%

1.12

12%

A23

35

67%

120

89-2

6%2.

2512

5%A4

10.

10.

10%

7752

-33%

1.48

48%

All

5.1

7.1

39%

252

190

-25%

+77%

NSD

Yes

24P2

W c

ollis

ions

Car

occ

All

Ope

ratin

g ho

urs

A13

22

0%TL

RN

3534

-3%

1.03

3%A2

32

350

%55

42-2

4%1.

9696

%A4

10.

10.

10%

3021

-30%

1.43

43%

All

4.1

5.1

24%

120

97-1

9%+4

7%N

SDYe

s

Page 25: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 25

Collisions resulting in injuries to car occupants are the largest group generally, so it is important to investigate how they might have changed along the trial route.

For all collisions resulting in injury to car occupants, there was an increase of 1% relative control at all times. During bus lane hours of operation there was a decrease of 28% relative to control, which was found to be significant at the 5% level and indicative of a real change. However, this is not associated with P2W involvement, as described in the paragraph below.

For P2W collisions resulting in injury to car occupants, there were only extremely small numbers in both the before and after periods, although they did result in large percentage increases relative to control, 77% at all times and 47% during operating hours. However, none of these changes were significant.

Again, for all of these categories, significant variation between routes was found.

Page 26: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 26

4.1.7 Collisions involving killed or seriously injured casualties

Table 13

Cas

eC

ollis

ion

type

Cas

ualty

type

Seve

rity

Hou

rsR

oute

TEST

DAT

A - B

US

LAN

E O

NLY

CO

NTR

OL

DAT

A (N

ET o

f Tes

t + C

CZ)

Site

Si

te

Ove

rall

Sign

ifica

nce

Sign

ifica

nt%

cha

nge

% c

hang

e K

-val

ue%

cha

nge

% c

hang

eof

ove

rall

varia

tion

Bef

ore

Afte

r(s

impl

e)C

ontr

olB

efor

eAf

ter

(sim

ple)

rela

tive

to

rela

tive

toch

ange

betw

een

cont

rol

cont

rol

rout

es?

25Al

l col

lisio

nsKS

I col

lAl

lA1

324

23-4

%TL

RN

947

675

-29%

1.34

34%

A23

9770

-28%

1,48

01,

133

-23%

0.94

-6%

A41

2625

-4%

1,16

185

5-2

6%1.

3131

%Al

l14

711

8-2

0%3,

588

2,66

3-2

6%+7

%N

SDYe

s26

All c

ollis

ions

KSI c

oll

Ope

ratin

g ho

urs

A13

1212

0%TL

RN

449

335

-25%

1.34

34%

A23

3725

-32%

445

357

-20%

0.84

-16%

A41

139

-31%

293

199

-32%

1.02

2%Al

l62

46-2

6%1,

187

891

-25%

-3%

NSD

Yes

27P2

W c

ollis

ions

KSI c

oll

All

A13

912

33%

TLR

N26

619

7-2

6%1.

8080

%A2

331

26-1

6%48

243

9-9

%0.

92-8

%A4

111

110%

307

233

-24%

1.32

32%

All

5149

-4%

1,05

586

9-1

8%+1

5%N

SDYe

s28

P2W

col

lisio

nsKS

I col

lO

pera

ting

hour

sA1

35

860

%TL

RN

144

123

-15%

1.87

87%

A23

1311

-15%

215

191

-11%

0.95

-5%

A41

55

0%13

178

-40%

1.68

68%

All

2324

4%49

039

2-2

0%+2

9%N

SDYe

s

Page 27: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 27

Collision severity is an important consideration in a study of this type, particularly as most of the casualty reduction targets are based on the killed or seriously injured casualty numbers.

For all collisions resulting in killed or seriously injured casualties (to all modes) there was an increase relative to control of 7% at all hours, but during bus lane operating hours there was a decrease of 3% relative to control. Neither of these changes was statistically significant.

For collisions involving a P2W and resulting in killed or seriously injured casualties there were increases relative to control of 15% at all hours and of 29% during bus lane hours of operation, although neither change was statistically significant.

As in many of the other cases examined, there was significant variation between the three trial routes.

As the overall KSI collision changes were not significant, analysis into KSI collisions for different modes was not taken further as the numbers in any of the sub-categories would be very small.

Page 28: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 28

4.2 Powered two-wheeler collision rates on trial routes 4.2.1 Calculation of P2W collision rates None of the collision analysis presented so far has taken account of any changes in the usage of powered two-wheel vehicles along the trial routes and also the overall background trend for P2W collision rates in London over the same period. Collision rates, which take account of both the number of collisions together and a measure of the exposure, help us to understand such changes.

For collision rate calculations, the numerator was the number of collisions involving powered two-wheel vehicles in the study route lengths containing the bus lane, and the denominator was the number of vehicle kilometres travelled by such vehicles along the length of the trial routes.

For the trial routes, the information used for the denominator was taken from the Department for Transport’s Annual Road Traffic Census Counts, which provide the Annual Average Daily Traffic Flow Estimates by year. These were factored from daily to annual figures to give the P2W flow over the three years before and after the start of the trial. Along each of the trial routes were a number of count sites, from which the powered two-wheeler flows were obtained and an average flow for each trial route calculated. These average flows were based on six sites on the A13, on ten sites on the A23 and on three sites on the A41.

Note that the rates calculated were for ‘all periods of the day (as opposed to when the bus lanes were operating) as the DfT traffic volume counts were not available for just the bus lanes operating hours.

For the Greater London comparison, the denominator information was based on the DfT National Road Traffic Surveys, which gives the annual volume of traffic for the Greater London area.

4.2.2 Changes in powered two wheeler collision rates P2W collision rates were calculated for the three trial routes separately, for the three trial routes combined, and for Greater London for a three year period before and after the start of the trial (Table 14).

Page 29: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 29

Table 14: P2W collision rates over three years before and after the start of the trial

Route 3 years P2W km travelled (million km)

P2W collisions in three years

P2W collision rate per 100million vehicle

kilometres

Before After Before After Before After

A13 4.388 5.361 52 50 1,185 933

A23 6.683 8.044 199 139 2,978 1,728

A41 2.956 4.328 58 51 1,962 1,178

Trial routes combined 14.027 17.733 309 240 2,203 1,353

Greater London (net) 2,222.246 2,499.433 23,920 18,249 1,076 730

The data indicate that P2W kilometres travelled have risen (Table 14 columns 2/3), and P2W collisions have fallen (Table 14, columns 4/5) both on the trial routes and in Greater London since the start of the trial. This is why for both trial routes and Greater London as a whole, the P2W collision rate has fallen since the trial was started.

On the trial routes, the P2W collision rate fell from 2,203 to 1,353 P2W collisions per 100 million vehicle km travelled and in Greater London the P2W collision rate fell from 1,076 to 730 P2W collisions per 100 million vehicle km travelled (Table 14 columns 6/7).

To investigate whether the collision rates significantly changed since the start of the trial, rate ratios together with their confidence intervals were calculated comparing the three year collision rates before with the three year collision rates after the start of the trial (Table 15).

Table 15: Rate Ratios comparing P2W collision rates before and after the start of the trial

Route P2W collision rate per

100million vehicle kilometres

Before After Rate Ratio

(after/before)

Lower Confidence

Interval (95%)

Upper Confidence

Interval (95%)

P-Value (95%)

A13 1,185 933 0.79 0.53 1.16 0.227

A23 2,978 1,728 0.58 0.47 0.72 <0.001*

A41 1,962 1,178 0.60 0.41 0.88 0.008*

Trial routes combined 2,203 1,353 0.61 0.52 0.73 <0.001*

Greater London 1,076 730 0.68 0.67 0.69 <0.001*

* Very strong evidence indicating that change over time is significant at 95% level. The findings from the analysis presented in Table 15 indicate that the collision rate on the trial routes combined has fallen by 39% since the start of the trial

Page 30: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 30

(significant at 95% confidence level). Similarly, the collision rate in Greater London has fallen by 32% since the start of the trial (significant at 95% confidence level).

Because the confidence intervals of the trial routes (0.52 to 0.73) overlap with those of Greater London (0.67 to 0.69), the size of the reduction (minus 7%) between the trial routes and Greater London is not statistically significant.

This means that the trial has not resulted in a significant reduction in the P2W collision rate on the trial routes in comparison with the rest of London. This broadly confirms the findings from the collision analysis earlier in the report.

Page 31: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 31

4.3 Detailed conflict analysis Tables 16 and 17 present a summary of the types of conflict that occurred for all collisions involving P2Ws within the sections of the study routes containing bus lanes. Table 16 shows P2W collisions occurring during the bus lane hours of operation and Table 17 shows those occurring outside of the bus lane hours of operation. Each table shows the number of collisions in the 36 month before and after periods for each route together with the numeric difference. The overall before and after figures are shown for all sites, together with both the numeric and percentage changes, but care must be taken for interpretation as the numbers in many of the conflict types are extremely small, so that some small numeric changes give large percentage changes.

In the tables, the paths of the vehicles are shown in the conflict diagram with ‘sticks’; P2Ws are represented by a dashed line and the other vehicle represented by a full line and pedestrians are represented by a line with a semi-circle on the end. A brief description of each conflict type is given. Thus, for example, the conflict for a vehicle turning right into the path of the P2W from the opposite direction is represented by the following conflict.

It should be noted that the conflict reference number is simply a working number and not a reflection of any ranking.

4.3.1 Conflicts in P2W collisions during bus lane hours of operation Table 16 shows a very wide range of conflicts occurring during bus lane hours of operation, although many had very small numbers in both the before and after periods. However several types of conflicts had reasonable numbers associated with them, some of which may have been influenced by the introduction of the bus lane scheme.

The examination of the conflicts indicates that of particular concern in both the before and after periods were conflicts that involved P2Ws in conflict with turning vehicles, particularly those making right turns, but there was some evidence of a change in the type of conflict. For example, the largest conflict category was the vehicle turning right across the path of the P2W from the opposite direction (Type 1), which increased by 21% overall. However here was a 60% decrease in vehicles turning right across the P2W from a side road on the nearside (Type 35). There was also a 50% reduction in vehicles turning right

Page 32: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 32

across the P2W from the same direction (Type 2). A possible explanation for these changes is that the P2Ws might be generally following a riding line nearer the kerb, so are less visible to vehicles turning right from the opposite direction (with visibility of the P2W possibly obscured by stationary or slow moving traffic alongside the bus lane), but more visible to vehicles turning right into their path from the nearside. If P2Ws are riding closer to the kerb in the bus lane schemes, there may be less opportunity physically for other vehicles to turn right across their path from the same direction.

Regarding left turning vehicles in conflict with P2Ws, there was a 63% increase in the number of collisions involving a vehicle turning left across the path of the P2W (Type 20) and a small increase in vehicles turning left from the side road into the main road, although the numbers were very small (Type 4).

Collisions involving P2Ws and pedestrians increased by over 75% for those occurring at formal pedestrian crossings, with pedestrians crossing from the nearside (Type 23). This might be due to pedestrians not expecting P2Ws to be in the bus lanes. There was a 20% decrease in the number of collisions involving a pedestrian crossing from the offside, although the numbers were small. (Type 24).

P2W collisions with pedestrians away from formal crossings (from both nearside and offside) showed 43% decreases in the after period compared with before (Types 25 and 26).

Several other conflict types showed changes but for some it was not obvious how they related to the P2W in bus lanes. Shunt type collisions in which the P2W was hit from the rear by the other vehicle (Type 16) increased by 100%, but shunt collisions where the P2W ran into the rear of the other vehicle (Type 8) increased by 50%. In addition, vehicles making U-turns across the path of the P2W remained at similar levels. (Types 5 and 6).

Page 33: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 33

Table 16 (Part 1 of 3): P2W collision conflict categories - during bus lane operating hours

A13 A23 A41 All routes

Conflict Description Before After Change Before After Change Before After Change Before After Change % change

1

Other vehicle turns right into path of P2W (Opposite directions)

6 10 4 15 16 1 7 8 1 28 34 6 21%

2Other vehicle turns right into path of P2W (Same direction)

0 1 1 10 4 -6 0 10 5 -5 -50%

3

Other vehicle disobeys junction control and turns right into path of P2W

0 1 1 3 1 -2 0 3 2 -1 -33%

4

Other vehicle disobeys junction control and turns left into path of P2W

1 1 0 0 2 2 0 1 3 2 200%

5

Other vehicle u-turns into path of P2W (From opposite direction)

0 1 1 0 0 1 1 0 0%

6Other vehicle u-turns into path of P2W (From same direction)

1 2 1 3 0 -3 0 1 1 4 3 -1 -25%

8P2W runs into rear of other vehicle

2 2 0 1 6 5 3 1 -2 6 9 3 50%

9P2W loses control (and may hit other vehicle)

1 1 0 2 2 0 0 3 3 0 0%

10Other vehicle changes lane from n/s across the path of P2W

5 0 -5 4 1 -3 1 0 -1 10 1 -9 -90%

11P2W brakes and/or swerves to avoid collision

0 2 0 -2 0 2 0 -2 -100%

12

Other veh fails to give way or disobeys junction control and collides with P2W

6 1 -5 0 2 2 1 1 0 7 4 -3 -43%

14

Head on collision between P2W and other vehicle

0 2 0 -2 0 2 0 -2 -100%

*

*

*

Page 34: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 34

Table 16 (Part 2 of 3): P2W collision conflict categories - during bus lane operating hours

A13 A23 A41 All routes

Conflict Description Before After Change Before After Change Before After Change Before After Change % change

15P2W loses control -and hits kerb, barrier or wall etc.

0 0 1 0 -1 1 0 -1 -100%

16Other vehicle runs into rear of P2W

0 4 4 3 2 -1 0 3 6 3 100%

17

P2W fails to give way at junction control and collides with other vehicle

1 0 -1 1 0 -1 3 0 -3 5 0 -5 -100%

20Other vehicle turns left across the path of P2W user

1 3 2 4 8 4 3 2 -1 8 13 5 63%

22P2W turns right into path of other vehicle .

0 0 1 1 0 0 1 1 ∞

23

P2W strikes pedestrian at or within 50m of a formal pedestrian crossing (from nearside)

2 3 1 2 3 1 0 1 1 4 7 3 75%

24

P2W strikes pedestrian at or within 50m of a formal pedestrian crossing (from offside)

3 3 0 1 -1 1 1 0 5 4 -1 -20%

25

P2W strikes pedestrian crossing from nearside (not at crossing)

0 7 3 -4 0 1 1 7 4 -3 -43%

26

P2W strikes pedestrian crossing from offside (not at crossing)

2 1 -1 4 1 -3 1 2 1 7 4 -3 -43%

27 Other pedestrian collision 0 0 1 0 -1 1 0 -1 -100%

29Other vehicle starts off or pulls out into path of P2W

0 1 1 0 0 1 1 0 0%

30

P2W hits open door / swerves to avoid open door of other vehicle.

1 0 -1 0 2 2 1 1 0 2 3 1 50%

*

Page 35: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 35

Table 16 (Part 3 of 3): P2W collision conflict categories - during bus lane operating hours

A13 A23 A41 All routes

Conflict Description Before After Change Before After Change Before After Change Before After Change % change

32P2W starts off or pulls out into path of other vehicle

0 1 1 0 1 1 0 0 2 2 ∞

34 No details No details 0 1 0 -1 0 1 0 -1 -100%

35Other vehicle turns right from side road across path of P2W

2 1 -1 17 7 -10 1 0 -1 20 8 -12 -60%

36Other vehicle changes lane from o/s across path of P2W

0 4 4 3 2 -1 3 2 -1 6 8 2 33%

37P2W changes lane across path of other vehicle from o/s.

1 0 -1 1 0 -1 1 1 0 3 1 -2 -67%

38P2W makes U-turn across path of other vehicle

0 0 0 1 1 0 1 1 ∞

39P2W turnd left into path of other vehicle

1 0 -1 0 0 1 0 -1 -100%

40

P2W turns right across path of other vehicle from opposite direction.

0 2 2 0 0 2 2 0 0%

Totals 36 39 90 68 28 23 154 130

*

Page 36: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 36

4.3.2 Conflicts in P2W collisions outside of bus lane hours of operation Table 17 also shows a very wide range of conflicts occurring for collisions occurring outside of the bus lane hours of operation. Again, many categories had very small numbers in both the before and after periods.

It is important to investigate if there are any notable changes in the main collision types outside of the hours of operation as the riding line along the whole route for P2Ws at these times might be have changed and been influenced by the ability to use the bus lane during hours of operation.

For vehicles turning right across the path of the P2W there was a general reduction, with a 32% decrease in collisions with the vehicle from the opposite direction (Type 1) and a 67% decrease in vehicles turning across the P2W from the same direction (Type 2).There was also a reduction of 19% of vehicles turning right from a side road across the path of the P2W (Type 35).

For left turning conflicts, there was a 67% decrease in vehicles turning left from a side road into the path of the P2W (Type 4), but a 33% increase in vehicles turning left across the path of the P2W from the same direction (Type 20).

For U-turning conflicts, there was a 57% reduction in collisions with the other vehicle turning across the P2W from the same direction (Type 6).

For shunt type collisions, quite different changes were noted. For collisions with the P2W running into the rear of the other vehicle (Type 8) there was a reduction of 83% but for collisions with the other vehicle running into the rear of the P2W (Type 16) there was an increase of 100%.

Page 37: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 37

Table 17 (Part 1 of 3): P2W collision conflict categories - outside of bus lane operating hours

A13 A23 A41 All routes

Conflict Description Before After Change Before After Change Before After Change Before After Change % change

1

Other vehicle turns right into path of P2W (Opposite directions)

1 2 1 19 12 -7 5 3 -2 25 17 -8 -32%

2Other vehicle turns right into path of P2W (Same direction)

0 12 4 -8 0 12 4 -8 -67%

3

Other vehicle disobeys junction control and turns right into path of P2W

0 5 3 -2 0 5 3 -2 -40%

4

Other vehicle disobeys junction control and turns left into path of P2W

0 3 1 -2 0 3 1 -2 -67%

5

Other vehicle u-turns into path of P2W (From opposite direction)

0 2 2 0 1 1 0 3 3 0 0%

6Other vehicle u-turns into path of P2W (From same direction)

3 0 -3 7 4 -3 4 2 -2 14 6 -8 -57%

8P2W runs into rear of other vehicle

0 1 1 7 0 -7 5 1 -4 12 2 -10 -83%

9P2W loses control (and may hit other vehicle)

0 1 1 5 3 -2 0 2 2 5 6 1 20%

10Other vehicle changes lane from n/s across the path of P2W

3 1 -2 1 3 2 1 1 0 5 5 0 0%

11P2W brakes and/or swerves to avoid collision

0 2 0 -2 0 2 0 -2 -100%

12

Other veh fails to give way or disobeys junction control and collides with P2W

2 1 -1 0 1 1 0 1 1 2 3 1 50%

13 P2W hits parked vehicle 0 1 0 -1 0 1 1 1 1 0 0%

14

Head on collision between P2W and other vehicle

0 2 1 -1 0 2 1 -1 -50%

*

*

*

Page 38: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 38

Table 17 (Part 2 of 3): P2W collision conflict categories - outside of bus lane operating hours

A13 A23 A41 All routes

Conflict Description Before After Change Before After Change Before After Change Before After Change % change

15P2W loses control -and hits kerb, barrier or wall etc.

0 1 1 1 0 -1 0 1 1 0 0%

16Other vehicle runs into rear of P2W

0 1 1 2 7 5 2 -2 4 8 4 100%

17

P2W fails to give way at junction control and collides with other vehicle

0 0 2 2 0 1 1 0 3 3 ∞

18

P2W loses control -and may hit other vehicle- (road surface condition)

0 1 0 -1 0 1 1 1 1 0 0%

20Other vehicle turns left across the path of P2W user

1 1 0 0 2 2 2 1 -1 3 4 1 33%

23

P2W strikes pedestrian at or within 50m of a formal pedestrian crossing (from nearside)

2 0 -2 6 1 -5 1 2 1 9 3 -6 -67%

24

P2W strikes pedestrian at or within 50m of a formal pedestrian crossing (from offside)

0 1 1 0 1 0 -1 2 1 -1 -50%

25

P2W strikes pedestrian crossing from nearside (not at crossing)

1 0 -1 4 0 -4 1 1 0 6 1 -5 -83%

26

P2W strikes pedestrian crossing from offside (not at crossing)

0 0 1 1 0 1 1 0 2 2 ∞

27 Other pedestrian collision 0 1 4 3 0 1 1 1 5 4 400%

29Other vehicle starts off or pulls out into path of P2W

0 0 0 1 1 0 1 1 ∞

30

P2W hits open door / swerves to avoid open door of other vehicle.

0 0 1 1 2 1 -1 2 2 0 0%

*

Page 39: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 39

Table 17 (Part 3 of 3): P2W collision conflict categories - outside of bus lane operating hours

A13 A23 A41 All routes

Conflict Description Before After Change Before After Change Before After Change Before After Change % change

34 No details No details 1 0 -1 1 0 -1 0 1 1 2 1 -1 -50%

35Other vehicle turns right from side road across path of P2W

1 0 -1 14 12 -2 1 1 0 16 13 -3 -19%

36Other vehicle changes lane from o/s across path of P2W

1 1 0 9 5 -4 4 3 -1 14 9 -5 -36%

37P2W changes lane across path of other vehicle from o/s.

0 2 0 -2 0 2 0 -2 -100%

38P2W makes U-turn across path of other vehicle

0 0 0 1 1 0 1 1 ∞

40

P2W turns right across path of other vehicle from opposite direction.

0 1 1 1 0 -1 0 1 1 0 0%

41

P2W turns right across path of other vehicle from same direction.

0 0 1 1 0 0 1 1 ∞

Totals 16 11 109 71 30 28 155 110

Page 40: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 40

4.3.3 Collisions reported as being in bus lane As part of the Stats19 data, the location of vehicles at the time of collision is recorded, with regards to whether they are in a ‘Restricted lane’, with one such category being a bus lane (Field 2.9). This is a relatively new item of information to be recorded by police officers, but there is uncertainty as to how completely or reliably it is reported. In addition, the bus lane may be mentioned in the text description of the collisions circumstances. Both of these are obviously of interest and relevance for this study so the number of occurrences recorded for each of the three routes in the before and after periods is shown in Table 18.

Table 18: Collisions involving a P2W reported as occurring actually in a bus lane

Period Bus lane Collisions Before After Change % change

During hours of operation A13 1 4 3 300%A23 7 9 2 29%A41 1 1 0 0%Total 9 14 5 56%

Outside of hours of operation A13 0 1 1 ∞A23 4 6 2 50%A41 0 0 0 0%Total 4 7 3 75%

Table 18 shows that there was considerable variation between the routes with most of the P2W collisions reported as being actually in the bus lane being on the A23, both during hours of operation and outside hours of operation. Overall, although the numbers were very small, there was an increase of 56% during bus lane hours of operation and 75% outside of hours of operation.

Page 41: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 41

4.3.4 Collisions with P2W reported as overtaking on nearside Another collision category that may help to identify safety issues for P2W users in bus lanes is where the P2W is recorded as overtaking on the nearside. This might be recorded in either or both of the text description of the collision circumstances or the Stats19 ‘Vehicle manoeuvre’ field (Field 2.7).

Table 19: P2W collisions with P2W reported as overtaking nearside

Period Bus lane Collisions Before After Change % change

During hours of operation A13 3 7 4 133%A23 1 6 5 500%A41 0 2 2 ∞Total 4 15 11 275%

Outside of hours of operation A13 0 1 1 ∞A23 4 4 0 0%A41 1 1 0 0%Total 5 6 1 20%

Table 19 shows that although the frequency of such collisions was very low, the number of P2Ws recorded as overtaking on the nearside increased substantially during hours of operation (275%%) but by a relatively small amount outside of hour of operation (20%), following introduction of theP2W in bus lane trial.

Page 42: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 42

5. Conclusions 5.1.1 This study has investigated the effect on collisions as part of a trial to permit powered two wheeled vehicles (P2Ws) to use three bus lanes along important radial routes on the Transport for London Road Network (TLRN). The analysis is based on personal injury collision and casualty data within the lengths of the trial routes along the A13, A23 and A41containing bus lanes, for a period of three years before and after introduction of the schemes. The analysis considered the changes in collisions during bus lane operating hours and also at all times.

5.1.2 Rigorous statistical analysis was carried out on a wide range of collision categories, taking into account the changes observed against a control based on collisions on the rest of the TLRN (minus that part within the Congestion Charging Zone), which was considered to be representative of the types of roads on which the three trial routes were located. In addition, detailed analysis was undertaken on the conflicts occurring for collisions involving powered two wheelers, to ascertain if there were any noticeable changes in particular types occurring. This study forms one part of a wider review of other traffic and user issues associated with the P2W in bus lanes trials on the TLRN.

5.2 Detailed before and after analysis 5.2.1 All collisions on trial routes decreased by 2% relative to control lengths at all times and by 4% during hours of bus lane operation. Neither of these changes was statistically significant, i.e. they could have occurred by chance.

5.2.2 Collisions on trial routes involving powered two-wheeler decreased by 3% at all times relative to control lengths, whereas during hours of operation they increased by 1%, although neither of these changes was statistically significant.

5.2.3 For other vulnerable road users, collisions involving P2Ws in collision with a pedestrian on trial routes decreased by 13% relative to control lengths at all times but increased by 3% during hours of bus lane operation. Neither was statistically significant. For collisions involving a P2W and a pedal cyclist, the numbers were very small, with non- significant changes at any time.

5.2.4 P2W collisions which resulted in injury to the P2W users on trial routes, increased relative to control lengths by 2% at all times and 6% during bus lane operating times, although neither was statistically significant.

Page 43: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 43

5.2.5 One aspect of the statistical analysis is that for most (but not all) of the collision categories studied, there was significant variation between the trial routes, indicating that it is unlikely the results would be repeated if similar schemes were introduced generally.

5.2.6 Overall, it is concluded that generally there were small increases or decreases in most of the collision categories on the trial routes relative to the control lengths, although generally these were not statistically significant. Therefore there is no clear evidence which demonstrates that permitting P2Ws in bus lanes provides a real safety benefit.

5.3 P2W collision rates 5.3.1 With regards to P2W collision rate per 100million vehicle kilometres, there was a statistically significant reduction of 39%on the trial routes (combined) following introduction of the trial. However, for the same period across London, there was a reduction of 31% in the P2W collision rate per 100million vehicle kilometres. The difference between these two rate reductions was found not to be significant. This broadly confirms the findings from the detailed before and after collision analysis.

5.4 Detailed conflict analysis 5.4.1 With regards to the change in conflicts in collisions involving powered two wheelers on trial routes, there were several notable changes during hours of bus lane operation. However, it must be noted that the numbers of collisions in most of the conflict types are generally very small.

5.4.2 For right turning collisions, there was a change of type with a 21% increase in those involving the other vehicle turning right across the path of the P2W from the opposite direction. However there was a decrease of 60% in vehicles turning right from a nearside side road across the P2W and a 50% reduction in vehicles turning right across the P2W from the same direction. A possible explanation for these changes might be due to the P2Ws generally following a riding line nearer the kerb along the bus lane routes, making them more visible to vehicles on the nearside but less visible to vehicles turning from the opposite direction.

5.4.3 For collisions during operating hours involving left turning vehicles, there was a 63% increase in vehicles turning left across the path of the P2W from the same direction.

5.4.4 Collisions involving P2Ws injuring pedestrians in the vicinity of formal crossings from the nearside increased by over 75% during operating hours, whilst those crossing from the offside decreased by 20%.

Page 44: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 44

5.4.5 For those P2W collisions reported as actually occurring in a bus lane, the numbers were relatively small, but showed an increase of 56% during bus lane hours of operation and 75% outside of bus lane operating hours.

5.4.6 Collisions involving the P2W reported as overtaking on the nearside were another type of collision category that was investigated, and again although the numbers were very small, there were increases of 275% during hours of bus lane operation and 20% outside of bus lane operating times.

5.4.7 The detailed analysis of conflicts showed some notable changes, including some major categories that resulted in increases in collisions involving P2Ws during hours of bus lane operation. A major problem for P2Ws continued to be those involving right turning vehicles, although the nature of these appeared to change as a result of the trial. Problems were also identified for P2W collisions involving pedestrians during hours of operation, especially in the vicinity of formal pedestrian crossings.

Page 45: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 45

6. References Stats 20 -Instructions for the completion of road accident reports Department for Transport (October 2004) The Mayor’s Transport Strategy Greater London Authority (July 2001)

Page 46: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 46

Appendices

Page 47: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 47

Appendix 1: Study route location maps

Page 48: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 48

A23 Bus Lane Trial Route Other TLRN Routes

Reproducd by permission of Geograhers' A-Z Map Co Ltd. Crown Copyright 2006. All rights reserved. License Number 100017302.

Page 49: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 49

A41 Bus Lane Trial Route Other TLRN Routes

Reproducd by permission of Geograhers' A-Z Map Co Ltd. Crown Copyright 2006. All rights reserved. License Number 100017302.

Page 50: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 50

A5 Bus Lane Control Route Other TLRN Routes

Reproducd by permission of Geograhers' A-Z Map Co Ltd. Crown Copyright 2006. All rights reserved. License Number 100017302.

Page 51: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 51

A10 Bus Lane Control Route Other TLRN Routes

Reproducd by permission of Geograhers' A-Z Map Co Ltd. Crown Copyright 2006. All rights reserved. License Number 100017302.

Page 52: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 52

Appendix 2: Collision data summary table COLLISION test and TLRN control data for P2W in Bus Lane Study (36 months Before v. 36 months After)(Part 1 of 4) NSD=No significant difference

Case Collision type Casualty type Severity Hours Route TEST DATA - BUS LANE ONLY CONTROL DATA (NET of Test + CCZ) Site Site Overall Significance Significant% change % change K-value % change % change of overall variation

Before After (simple) Control Before After (simple) relative to relative to change betweencontrol control routes?

1 All collisions All All All A13 163 145 -11% TLRN 6,277 5,027 -20% 1.11 11%A23 653 493 -25% 9,645 8,090 -16% 0.90 -10%A41 199 177 -11% 7,939 6,237 -21% 1.13 13%All 1015 815 -20% 23,861 19,354 -19% -2% NSD Yes

2 All collisions All All Operating hours A13 91 81 -11% TLRN 3,680 3,039 -17% 1.08 8%A23 227 176 -22% 3,325 2,892 -13% 0.89 -11%A41 75 61 -19% 2,493 1,974 -21% 1.03 3%All 393 318 -19% 9,498 7,905 -17% -4% NSD Yes

3 P2W collisions All All All A13 52 50 -4% TLRN 1,506 1,151 -24% 1.26 26%A23 199 139 -30% 2,932 2,423 -17% 0.85 -15%A41 58 51 -12% 1,711 1,299 -24% 1.16 16%All 309 240 -22% 6,149 4,873 -21% -3% NSD Yes

4 P2W collisions All All Operating hours A13 36 39 8% TLRN 1,019 798 -22% 1.38 38%A23 90 68 -24% 1,354 1,183 -13% 0.86 -14%A41 28 23 -18% 757 594 -22% 1.05 5%All 154 130 -16% 3,130 2,575 -18% +1% NSD Yes

5 All collisions Pedestrian All All A13 45 32 -29% TLRN 967 749 -23% 0.92 -8%A23 165 123 -25% 1,676 1,398 -17% 0.89 -11%A41 41 47 15% 971 802 -17% 1.39 39%All 251 202 -20% 3,614 2,949 -18% -2% NSD Yes

6 All collisions Pedestrian All Operating hours A13 27 19 -30% TLRN 547 424 -22% 0.91 -9%A23 62 36 -42% 568 468 -18% 0.70 -30%A41 16 23 44% 296 231 -22% 1.84 84%All 105 78 -26% 1,411 1,123 -20% -7% NSD Yes

7 P2W collisions Pedestrian All All A13 10 7 -30% TLRN 158 113 -28% 0.98 -2%A23 28 13 -54% 259 236 -9% 0.51 -49%A41 5 11 120% 134 102 -24% 2.89 189%All 43 31 -28% 551 451 -18% -13% NSD Yes

8 P2W collisions Pedestrian All Operating hours A13 7 7 0% TLRN 116 83 -28% 1.40 40%A23 14 7 -50% 137 119 -13% 0.58 -42%A41 3 6 100% 54 45 -17% 2.40 140%All 24 20 -17% 307 247 -20% +3% NSD Yes

Page 53: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 53

COLLISION test and TLRN control data for P2W in Bus Lane Study (36 months Before v. 36 months After)(Part 2 of 4) NSD=No significant difference

Case Collision type Casualty type Severity Hours Route TEST DATA - BUS LANE ONLY CONTROL DATA (NET of Test + CCZ) Site Site Overall Significance SignificantK-value % change % change of overall variation

Control Before After relative to relative to change betweencontrol control routes?

9 All collisions Pedal cyclist All All A13 4 4 0% TLRN 456 425 -7% 1.07 7%A23 73 55 -25% 871 833 -4% 0.79 -21%A41 20 21 5% 413 357 -14% 1.21 21%All 97 80 -18% 1,740 1,615 -7% -12% NSD Yes

10 All collisions Pedal cyclist All Operating hours A13 2 3 50% TLRN 319 267 -16% 1.79 79%A23 33 32 -3% 440 412 -6% 1.04 4%A41 13 12 -8% 191 168 -12% 1.05 5%All 48 47 -2% 950 847 -11% +7% NSD Yes

11 P2W collisions Pedal cyclist All All A13 0.1 0.1 0% TLRN 14 20 43% 0.70 -30%A23 1 0 -100% 29 28 -3% 0.00 -100%A41 3 3 0% 13 11 -15% 1.18 18%All 4.1 3.1 -24% 56 59 5% -14% NSD Yes

12 P2W collisions Pedal cyclist All Operating hours A13 0.1 0.1 0% TLRN 10 15 51% 0.66 -34%A23 0.1 0.1 0% 17 16 -6% 1.06 6%A41 3 1 -67% 7 6 -14% 0.39 -61%All 3.2 1.2 -63% 34 37 9% -58% NSD No

13 All collisions P2W user All All A13 44 46 5% TLRN 1,422 1,064 -25% 1.40 40%A23 180 135 -25% 2,786 2,289 -18% 0.91 -9%A41 56 47 -16% 1,630 1,256 -23% 1.09 9%All 280 228 -19% 5,838 4,609 -21% +2% NSD Yes

14 All collisions P2W user All Operating hours A13 31 34 10% TLRN 948 726 -23% 1.43 43%A23 81 67 -17% 1,269 1,106 -13% 0.95 -5%A41 26 20 -23% 719 573 -20% 0.97 -3%All 138 121 -12% 2,936 2,405 -18% +6% NSD Yes

15 P2W collisions P2W user All All A13 44 46 5% TLRN 1,422 1,064 -25% 1.40 40%A23 180 135 -25% 2,786 2,289 -18% 0.91 -9%A41 56 47 -16% 1,630 1,256 -23% 1.09 9%All 280 228 -19% 5,838 4,609 -21% +2% NSD Yes

16 P2W collisions P2W user All Operating hours A13 31 34 10% TLRN 948 726 -23% 1.43 43%A23 81 67 -17% 1,269 1,106 -13% 0.95 -5%A41 26 20 -23% 719 571 -21% 0.97 -3%All 138 121 -12% 2,936 2,403 -18% +6% NSD Yes

Page 54: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 54

COLLISION test and TLRN control data for P2W in Bus Lane Study (36 months Before v. 36 months After)(Part 3 of 4) NSD=No significant difference

Case Collision type Casualty type Severity Hours Route TEST DATA - BUS LANE ONLY CONTROL DATA (NET of Test + CCZ) Site Site Overall Significance SignificantK-value % change % change of overall variation

Control Before After relative to relative to change betweencontrol control routes?

17 All collisions Bus occ All All A13 16 17 6% TLRN 377 308 -18% 1.30 30%A23 85 73 -14% 723 644 -11% 0.96 -4%A41 20 17 -15% 264 304 15% 0.74 -26%All 121 107 -12% 1,364 1,256 -8% -4% NSD Yes

18 All collisions Bus occ All Operating hours A13 7 17 143% TLRN 264 218 -17% 2.94 194%A23 21 18 -14% 180 186 3% 0.83 -17%A41 6 7 17% 61 78 28% 0.91 -9%All 34 42 24% 505 482 -5% +24% NSD Yes

19 P2W collisions Bus occ All All A13 0.1 2.1 2000% TLRN 0 0 -200% -21.00 -2200%A23 0.1 0.1 0% 2 2 0% 1.00 0%A41 0.1 0.1 0% 1 0 -111% -9.00 -1000%All 0.3 2.3 667% 3 2 -30% ∞ NSD Yes

20 P2W collisions Bus occ All Operating hours A13 0.1 2.1 2000% TLRN 0 0 0% 21.00 2000%A23 0.1 0.1 0% 1 1 0% 1.00 0%A41 0.1 0.1 0% 1 0 -90% 10.00 900%All 0.3 2.3 667% 2 1 -50% ∞ NSD Yes

21 All collisions Car occ All All A13 94 80 -15% TLRN 4,326 3,387 -22% 1.09 9%A23 227 171 -25% 5,304 4,316 -19% 0.93 -7%A41 82 70 -15% 6,449 4,768 -26% 1.15 15%All 403 321 -20% 16,079 12,471 -22% +1% NSD Yes

22 All collisions Car occ All Operating hours A13 38 27 -29% TLRN 1,888 1,495 -21% 0.90 -10%A23 52 30 -42% 1,267 1,064 -16% 0.69 -31%A41 21 8 -62% 1,541 1,171 -24% 0.50 -50%All 111 65 -41% 4,696 3,730 -21% -28% 5% level Yes

23 P2W collisions Car occ All All A13 2 2 0% TLRN 55 49 -11% 1.12 12%A23 3 5 67% 120 89 -26% 2.25 125%A41 0.1 0.1 0% 77 52 -33% 1.48 48%All 5.1 7.1 39% 252 190 -25% +77% NSD Yes

24 P2W collisions Car occ All Operating hours A13 2 2 0% TLRN 35 34 -3% 1.03 3%A23 2 3 50% 55 42 -24% 1.96 96%A41 0.1 0.1 0% 30 21 -30% 1.43 43%All 4.1 5.1 24% 120 97 -19% +47% NSD Yes

Page 55: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 55

COLLISION test and TLRN control data for P2W in Bus Lane Study (36 months Before v. 36 months After)(Part 4 of 4) NSD=No significant difference

Case Collision type Casualty type Severity Hours Route TEST DATA - BUS LANE ONLY CONTROL DATA (NET of Test + CCZ) Site Site Overall Significance SignificantK-value % change % change of overall variation

Control Before After relative to relative to change betweencontrol control routes?

25 All collisions KSI coll All A13 24 23 -4% TLRN 947 675 -29% 1.34 34%A23 97 70 -28% 1,480 1,133 -23% 0.94 -6%A41 26 25 -4% 1,161 855 -26% 1.31 31%All 147 118 -20% 3,588 2,663 -26% +7% NSD Yes

26 All collisions KSI coll Operating hours A13 12 12 0% TLRN 449 335 -25% 1.34 34%A23 37 25 -32% 445 357 -20% 0.84 -16%A41 13 9 -31% 293 199 -32% 1.02 2%All 62 46 -26% 1,187 891 -25% -3% NSD Yes

27 P2W collisions KSI coll All A13 9 12 33% TLRN 266 197 -26% 1.80 80%A23 31 26 -16% 482 439 -9% 0.92 -8%A41 11 11 0% 307 233 -24% 1.32 32%All 51 49 -4% 1,055 869 -18% +15% NSD Yes

28 P2W collisions KSI coll Operating hours A13 5 8 60% TLRN 144 123 -15% 1.87 87%A23 13 11 -15% 215 191 -11% 0.95 -5%A41 5 5 0% 131 78 -40% 1.68 68%All 23 24 4% 490 392 -20% +29% NSD Yes

Page 56: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 56

Appendix 3: Casualty data summary table CASUALTY test and TLRN control data (net of study route and CCZ Area) for P2W in Bus Lane Study (36 months Before v. 36 months After)(Part 1 of 3)

Case Collision type Casualty type Severity Hours Route TEST DATA - BUS LANE ONLY CONTROL DATA (NET of test and CCZ) Site K-value Site % change % change % change

Before After (simple) Control Before After (simple) relative to control

1 All collisions All All All A13 212 180 -15% TLRN 7,929 6,226 -21% 1.08 8%A23 743 582 -22% 11752 9788 -17% 0.94 -6%A41 240 177 -26% 10214 7874 -23% 0.96 -4%All 1195 939 -21% 29895 23888 -20%

2 All collisions All All Operating hours A13 111 101 -9% TLRN 5502 4310 -22% 1.16 16%A23 255 194 -24% 4776 4016 -16% 0.90 -10%A41 90 71 -21% 3730 2915 -22% 1.01 1%All 456 366 -20% 14008 11241 -20%

3 All collisions P2W cas All All A13 44 40 -9% TLRN 1422 1064 -25% 1.21 21%A23 180 135 -25% 2786 2289 -18% 0.91 -9%A41 56 47 -16% 1630 1256 -23% 1.09 9%All 280 222 -21% 5838 4609 -21%

4 All collisions P2W cas All Operating hours A13 31 34 10% TLRN 948 726 -23% 1.43 43%A23 81 67 -17% 1269 1106 -13% 0.95 -5%A41 26 20 -23% 719 571 -21% 0.97 -3%All 138 121 -12% 2936 2403 -18%

5 All collisions Pedal cyclist All All A13 4 4 0% TLRN 456 425 -7% 1.07 7%A23 73 55 -25% 871 835 -4% 0.79 -21%A41 20 21 5% 413 357 -14% 1.21 21%All 97 80 -18% 1740 1617 -7%

6 All collisions Pedal cyclist All Operating hours A13 2 3 50% TLRN 319 267 -16% 1.79 79%A23 33 32 -3% 440 412 -6% 1.04 4%A41 13 12 -8% 191 168 -12% 1.05 5%All 48 47 -2% 950 847 -11%

7 All collisions Pedestrian All All A13 45 32 -29% TLRN 967 749 -23% 0.92 -8%A23 165 123 -25% 1676 1398 -17% 0.89 -11%A41 41 47 15% 971 802 -17% 1.39 39%All 251 202 -20% 3614 2949 -18%

8 All collisions Pedestrian All Operating hours A13 27 19 -30% TLRN 547 424 -22% 0.91 -9%A23 62 36 -42% 568 469 -17% 0.70 -30%A41 16 23 44% 296 231 -22% 1.84 84%All 105 78 -26% 1411 1124 -20%

Page 57: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 57

CASUALTY test and TLRN control data (net of study route and CCZ Area) for P2W in Bus Lane Study (36 months Before v. 36 months After)(Part 2 of 3)

Case Collision type Casualty type Severity Hours Route TEST DATA - BUS LANE ONLY CONTROL DATA (NET of test and CCZ) Site K-value Site % change % change % change

Before After (simple) Control Before After (simple) relative to control

9 P2W collisions All All All A13 57 57 0% TLRN 1652 1250 -24% 1.32 32%A23 214 153 -29% 3217 2650 -18% 0.87 -13%A41 66 62 -6% 1857 1431 -23% 1.22 22%All 337 272 -19% 6726 5331 -21%

10 P2W collisions All All Operating hours A13 41 45 10% TLRN 1111 861 -23% 1.42 42%A23 98 77 -21% 4414 3780 -14% 0.92 -8%A41 34 28 -18% 3385 2649 -22% 1.05 5%All 173 150 -13% 8910 7290 -18%

11 P2W collisions P2W cas All All A13 44 46 5% TLRN 1422 1064 -25% 1.40 40%A23 180 135 -25% 2786 2289 -18% 0.91 -9%A41 56 47 -16% 1630 1256 -23% 1.09 9%All 280 228 -19% 5838 4609 -21%

12 P2W collisions P2W cas All Operating hours A13 31 34 10% TLRN 948 729 -23% 1.43 43%A23 81 67 -17% 1269 1106 -13% 0.95 -5%A41 26 20 -23% 719 571 -21% 0.97 -3%All 138 121 -12% 2936 2406 -18%

13 P2W collisions Pedal cyclist All All A13 0 0 0% TLRN 14 20 43% n/a n/aA23 1 0 -100% 29 28 -3% 0.00 -100%A41 3 3 0% 13 11 -15% 1.18 18%All 4 3 -25% 56 59 5%

14 P2W collisions Pedal cyclist All Operating hours A13 0 0 0% TLRN 10 15 50% n/a n/aA23 0 0 0% 17 16 -6% n/a n/aA41 3 1 -67% 7 6 -14% 0.39 -61%All 3 1 -67% 34 37 9%

15 P2W collisions Pedestrian All All A13 10 7 -30% TLRN 158 113 -28% 0.98 -2%A23 28 13 -54% 269 239 -11% 0.52 -48%A41 5 11 120% 134 102 -24% 2.89 189%All 43 31 -28% 561 454 -19%

16 P2W collisions Pedestrian All Operating hours A13 7 7 0% TLRN 116 83 -28% 1.40 40%A23 14 7 -50% 137 119 -13% 0.58 -42%A41 3 6 100% 54 45 -17% 2.40 140%All 24 20 -17% 307 247 -20%

Page 58: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 58

CASUALTY test and TLRN control data (net of study route and CCZ Area) for P2W in Bus Lane Study (36 months Before v. 36 months After)(Part 3 of 3)

Case Collision type Casualty type Severity Hours Route TEST DATA - BUS LANE ONLY CONTROL DATA (NET of test and CCZ) Site K-value Site % change % change % change

Before After (simple) Control Before After (simple) relative to control

17 All collisions All KSI All A13 26 25 -4% TLRN 1046 745 -29% 1.35 35%A23 105 73 -30% 1588 1224 -23% 0.90 -10%A41 32 26 -19% 1282 947 -26% 1.10 10%All 163 124 -24% 3916 2916 -26%

18 All collisions All KSI Operating hours A13 12 12 0% TLRN 488 359 -26% 1.36 36%A23 41 26 -37% 474 386 -19% 0.78 -22%A41 14 9 -36% 330 217 -34% 0.98 -2%All 67 47 -30% 1292 962 -26%

19 P2W collisions All KSI All A13 9 12 33% TLRN 279 202 -28% 1.84 84%A23 31 26 -16% 507 456 -10% 0.93 -7%A41 11 12 9% 312 241 -23% 1.41 41%All 51 50 -2% 1098 899 -18%

20 P2W collisions All KSI Operating hours A13 5 8 60% TLRN 211 144 -32% 2.34 134%A23 13 11 -15% 225 202 -10% 0.94 -6%A41 5 5 0% 137 80 -42% 1.71 71%All 23 24 4% 573 426 -26%

21 P2W collisions P2W cas KSI All A13 6 9 50% TLRN 242 175 -28% 2.07 107%A23 26 24 -8% 438 387 -12% 1.04 4%A41 10 11 10% 265 218 -18% 1.34 34%All 42 44 5% 945 780 -17%

22 P2W collisions P2W cas KSI Operating hours A13 4 5 25% TLRN 139 114 -18% 1.52 52%A23 13 11 -15% 194 173 -11% 0.95 -5%A41 5 5 0% 122 74 -39% 1.65 65%All 22 21 -5% 455 361 -21%

Page 59: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 59

Appendix 4: Comparative test and control collision data summary 36 months before and after Part 1 of 4

Case Collision type Casualty type Severity Hours Route / Area % change Before After (simple)

1 All collisions All All All A13 trial 163 145 -11%A23 trial 653 493 -25%A41 trial 199 177 -11%

All trial routes 1,015 815 -20%A5 183 171 -7%A10 66 60 -9%

A5 + A10 combined 249 231 -7%TLRN (Net of trial & CCZ) 23,861 19,354 -19%

Greater London 109,173 88,608 -19%2 All collisions All All Operating hours A13 trial 91 81 -11%

A23 trial 227 176 -22%A41 trial 75 61 -19%

All trial routes 393 318 -19%A5 93 97 4%A10 35 29 -17%

A5 + A10 combined 128 126 -2%TLRN (Net of trial & CCZ) 9,498 7,905 -17%

Greater London 62,584 50,847 -19%3 P2W collisions All All All A13 trial 52 50 -4%

A23 trial 199 139 -30%A41 trial 58 51 -12%

All trial routes 309 240 -22%A5 54 54 0%A10 10 5 -50%

A5 + A10 combined 64 59 -8%TLRN (Net of trial & CCZ) 6,149 4,873 -21%

Greater London 24,390 18,619 -24%4 P2W collisions All All Operating hours A13 trial 36 39 8%

A23 trial 90 68 -24%A41 trial 28 23 -18%

All trial routes 154 130 -16%A5 31 34 10%A10 5 1 -80%

A5 + A10 combined 36 35 -3%TLRN (Net of trial & CCZ) 3,130 2,575 -18%

Greater London 15,769 12,170 -23%5 All collisions Pedestrian All All A13 trial 45 32 -29%

A23 trial 165 123 -25%A41 trial 41 47 15%

All trial routes 251 202 -20%A5 50 44 -12%A10 16 9 -44%

A5 + A10 combined 66 53 -20%TLRN (Net of trial & CCZ) 3,614 2,949 -18%

Greater London 24,558 19,752 -20%6 All collisions Pedestrian All Operating hours A13 trial 27 19 -30%

A23 trial 62 36 -42%A41 trial 16 23 44%

All trial routes 105 78 -26%A5 23 24 4%A10 10 6 -40%

A5 + A10 combined 33 30 -9%TLRN (Net of trial & CCZ) 1,411 1,123 -20%

Greater London 14,929 11,877 -20%7 P2W collisions Pedestrian All All A13 trial 10 7 -30%

A23 trial 28 13 -54%A41 trial 5 11 120%

All trial routes 43 31 -28%A5 5 7 40%A10 5 6 20%

A5 + A10 combined 10 13 30%TLRN (Net of trial & CCZ) 551 451 -18%

Greater London 2,554 1,983 -22%8 P2W collisions Pedestrian All Operating hours A13 trial 10 7 -30%

A23 trial 14 7 -50%A41 trial 5 11 120%

All trial routes 29 25 -14%A5 3 5 67%A10 3 0 -100%

A5 + A10 combined 6 5 -17%TLRN (Net of trial & CCZ) 307 247 -20%

Greater London 1,870 1,438 -23%

Page 60: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 60

Part 2 of 49 All collisions Pedal cyclist All All A13 trial 4 4 0%

A23 trial 73 55 -25%A41 trial 20 21 5%

All trial routes 97 80 -18%A5 13 11 -15%A10 5 3 -40%

A5 + A10 combined 18 14 -22%TLRN (Net of trial & CCZ) 1,740 1,615 -7%

Greater London 10,039 8,913 -11%10 All collisions Pedal cyclist All Operating hours A13 trial 2 3 50%

A23 trial 33 32 -3%A41 trial 13 12 -8%

All trial routes 48 47 -2%A5 9 10 11%A10 3 2 -33%

A5 + A10 combined 12 12 0%TLRN (Net of trial & CCZ) 950 847 -11%

Greater London 6,778 5,979 -12%11 P2W collisions Pedal cyclist All All A13 trial 0 0 0%

A23 trial 1 0 -100%A41 trial 3 3 0%

All trial routes 4 3 -24%A5 0 1 ∞A10 0 0 0%

A5 + A10 combined 0 1 ∞TLRN (Net of trial & CCZ) 56 59 5%

Greater London 311 287 -8%12 P2W collisions Pedal cyclist All Operating hours A13 trial 0 0 0%

A23 trial 0 0 0%A41 trial 3 1 -67%

All trial routes 3 1 -67%A5 0 1 ∞A10 0 0 0%

A5 + A10 combined 0 1 ∞TLRN (Net of trial & CCZ) 34 37 9%

Greater London 235 207 -12%13 All collisions P2W user All All A13 trial 44 46 5%

A23 trial 180 135 -25%A41 trial 56 47 -16%

All trial routes 280 228 -19%A5 53 44 -17%A10 6 6 0%

A5 + A10 combined 59 50 -15%TLRN (Net of trial & CCZ) 5,838 4,609 -21%

Greater London 22,856 17,459 -24%14 All collisions P2W user All Operating hours A13 trial 31 34 10%

A23 trial 81 67 -17%A41 trial 26 20 -23%

All trial routes 138 121 -12%A5 28 28 0%A10 2 1 -50%

A5 + A10 combined 30 29 -3%TLRN (Net of trial & CCZ) 2,936 2,405 -18%

Greater London 14,476 11,069 -24%15 P2W collisions P2W user All All A13 trial 44 46 5%

A23 trial 180 135 -25%A41 trial 56 47 -16%

All trial routes 280 228 -19%A5 53 44 -17%A10 6 6 0%

A5 + A10 combined 59 50 -15%TLRN (Net of trial & CCZ) 5,838 4,609 -21%

Greater London 22,856 17,459 -24%16 P2W collisions P2W user All Operating hours A13 trial 31 34 10%

A23 trial 81 67 -17%A41 trial 26 20 -23%

All trial routes 138 121 -12%A5 28 28 0%A10 2 1 -50%

A5 + A10 combined 30 29 -3%TLRN (Net of trial & CCZ) 2,936 2,403 -18%

Greater London 14,476 11,069 -24%

Page 61: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 61

Part 3 of 417 All collisions Bus occ All All A13 trial 16 17 6%

A23 trial 85 73 -14%A41 trial 20 17 -15%

All trial routes 121 107 -12%A5 4 13 225%A10 4 9 125%

A5 + A10 combined 8 22 175%TLRN (Net of trial & CCZ) 1,364 1,256 -8%

Greater London 7,120 6,533 -8%18 All collisions Bus occ All Operating hours A13 trial 7 17 143%

A23 trial 21 18 -14%A41 trial 6 7 17%

All trial routes 34 42 24%A5 4 10 150%A10 3 8 167%

A5 + A10 combined 7 18 157%TLRN (Net of trial & CCZ) 505 482 -5%

Greater London 4,907 4,459 -9%19 P2W collisions Bus occ All All A13 trial 0 2 ∞

A23 trial 0 0 0%A41 trial 0 0 0%

All trial routes 0 2 ∞A5 0 1 ∞A10 0 0 0%

A5 + A10 combined 0 1 ∞TLRN (Net of trial & CCZ) 3 2 -30%

Greater London 19 37 95%20 P2W collisions Bus occ All Operating hours A13 trial 0 2 ∞

A23 trial 0 0 0%A41 trial 0 0 0%

All trial routes 0 2 ∞A5 0 1 ∞A10 0 0 0%

A5 + A10 combined 0 1 ∞TLRN (Net of trial & CCZ) 2 1 -50%

Greater London 11 25 127%21 All collisions Car occ All All A13 trial 94 80 -15%

A23 trial 227 171 -25%A41 trial 82 70 -15%

All trial routes 403 321 -20%A5 103 70 -32%A10 44 46 5%

A5 + A10 combined 147 116 -21%TLRN (Net of trial & CCZ) 16,079 12,471 -22%

Greater London 63,352 49,998 -21%22 All collisions Car occ All Operating hours A13 trial 38 27 -29%

A23 trial 52 30 -42%A41 trial 21 8 -62%

All trial routes 111 65 -41%A5 43 31 -28%A10 18 16 -11%

A5 + A10 combined 61 47 -23%TLRN (Net of trial & CCZ) 4,696 3,730 -21%

Greater London 29,376 23,393 -20%23 P2W collisions Car occ All All A13 trial 2 2 0%

A23 trial 4 5 25%A41 trial 0 0 0%

All trial routes 6 7 16%A5 3 1 -67%A10 0 0 0%

A5 + A10 combined 3 1 -67%TLRN (Net of trial & CCZ) 252 190 -25%

Greater London 993 765 -23%24 P2W collisions Car occ All Operating hours A13 trial 2 2 0%

A23 trial 2 3 50%A41 trial 0 0 0%

All trial routes 4 5 24%A5 2 1 -50%A10 0 0 0%

A5 + A10 combined 2 1 -50%TLRN (Net of trial & CCZ) 120 97 -19%

Greater London 547 435 -20%

Page 62: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 62

Part 4 of 425 All collisions KSI coll All A13 trial 24 23 -4%

A23 trial 97 70 -28%A41 trial 26 25 -4%

All trial routes 147 118 -20%A5 23 24 4%

A10 13 10 -23%A5 + A10 combined 36 34 -6%

TLRN (Net of trial & CCZ) 3,588 2,663 -26%Greater London 16,441 12,269 -25%

26 All collisions KSI coll Operating hours A13 trial 12 12 0%A23 trial 37 25 -32%A41 trial 13 9 -31%

All trial routes 62 46 -26%A5 12 11 -8%

A10 6 3 -50%A5 + A10 combined 18 14 -22%

TLRN (Net of trial & CCZ) 1,187 891 -25%Greater London 8,664 6,384 -26%

27 P2W collisions KSI coll All A13 trial 9 12 33%A23 trial 31 26 -16%A41 trial 11 11 0%

All trial routes 51 49 -4%A5 3 7 133%

A10 3 1 -67%A5 + A10 combined 6 8 33%

TLRN (Net of trial & CCZ) 1,055 869 -18%Greater London 4,183 3,295 -21%

28 P2W collisions KSI coll Operating hours A13 trial 5 8 60%A23 trial 13 11 -15%A41 trial 5 5 0%

All trial routes 23 24 4%A5 1 3 200%

A10 2 0 -100%A5 + A10 combined 3 3 0%

TLRN (Net of trial & CCZ) 490 392 -20%Greater London 2,469 1,924 -22%

Page 63: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 63

Appendix 5: A solution to the problem of combining accident frequencies from a number of sites The Tanner test is based on a method devised by J.C. Tanner of the Transport and Road Research Laboratory. It is used in situations where it is required to combine collision data from a number of sites where similar changes have taken place, such as introduction of a traffic management feature. The explanation of Tanner's method has been simplified in this note, but a detailed analysis is given in Biometrika, Volume 45, parts 3 & 4, pages 331-to 342 A problem in the combination of accident frequencies. The analysis performs two functions. Firstly it tests the overall effects for all sites combined, and secondly it tests for variation between the different sites. The theory of the analysis is as follows: Let N = number of sites from which data is to be combined

bi = Number of accidents in the before period at site i (where i = 1, 2, …..N)

ai = Number of accidents in the after period at site i

Ci = Ratio of accidents after to before the control area for site i (The control is assumed free from error. In practice, control figures should be at least 10 times greater than the test data)

ni = ai + bi

Then ki = ai/(biCi) is the measure of the apparent effect of the change at site i.

Firstly, it is necessary to calculate a value for K that will be the overall value of the apparent effect of the change at all sites. This is done by solving the equation:

___(1)

The right hand of the equation (1) is calculated for trial values of K until a sufficiently accurate solution is obtained, and can be done by iteration using a computer program.

∑∑ +=

i

ii KC1

nb

Page 64: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 64

Having calculated K it is then possible to calculate T, which indicates the level of significance of the overall effect of the changes.

___(2)

where The sampling variance of logeK is approximately given by:

___(3)

Where ___(4)

and ___(5)

Two values of T are calculated. The first value is calculated by putting K = 1 in equation (3). If T is found to be significant, a second value of T is calculated using the value of K from equation (1) to calculate the variance and hence the standard error. If the first value of T is not significantly different, the second is not calculated. The test is really a test of the significance of the departure of K from unity. The Χ2 value given by equation (5) indicates, if significant, that the effects of the change varied form site to site. The overall percentage change is determined from the value of K. A value of less than one indicates a decrease whereas a value above one indicates an increase. A K-value of 1 indicates no change.

For example, a K value of 0.75 indicates a percentage change of (K-1)x100%, i.e. a decrease of 25%.

K)Error(Log StandardKlogT

e

e=

K)ogVariance(l K)(Log Error Standard ee =

∑∑

+

⎟⎟⎠

⎞⎜⎜⎝

⎛+Φ+

=

2ii

ie

KCi)(1nKC

n21 )(1

Klog Variance

∑∑

⎟⎟⎠

⎞⎜⎜⎝

⎛−

−=Φ 2

i

2i

2

)n(nN

11N

X

∑ −=

ii

2iii2

nKC)Ckb(aX

Page 65: P2W in bus lanes safety study V3 5 - Bad Science

London Road Safety Unit 65

The values of T and Χ2 are interpreted as follows: T is interpreted by referring to a table of percentage points of the normal distribution. However the percentage value in the table has to be doubled to allow for the fact that Tanner's test is two-sided. Hence a T-value of 1.96 would be interpreted as 2.5% from the table but would in fact represent 5% significance level. It should be noted that T is unrestricted by degrees of freedom. (An alternative means of interpreting T is by referring to a table of percentage points of the t-distribution for infinite degrees of freedom. In this case it is not necessary to double the percentage value, as the t-distribution is two-sided). Χ2 is interpreted by reference to a table of percentage points for the Χ2 distribution, for N-1 degrees of freedom, where N is the number of sites being tested. If Χ2is significant, we may conclude that the effect of the treatment is not the same at all sites. This method of combining accident frequencies does have certain disadvantages, including the fact that it uses a control ratio rather than absolute figures. However, provided certain conditions are observed it can be used quite successfully to combine accident data. The conditions are: (i) The control data must be at least ten times greater than the test data. (ii) Wherever possible, each site should have its own mutually exclusive control data. However, it must be realised that this is not always possible. Also the dividing up of a large control figure into a number of mutually exclusive control figures does not always adequately reflect real differences. (iii) The element (1+ Φ) in equation (3) must be omitted when there is no evidence of significant variation between sites. Although in Tanner's analysis it stated that (1+ Φ) may be omitted when there is no firm evidence of variations between sites, experience has shown that (1+ Φ) must be omitted in such a case.