Documente Academic
Documente Profesional
Documente Cultură
Deprivation and
Child Pedestrian
Road Casualties January 2010
Final Report
colinbuchanan.com
Deprivation and Child Pedestrian Road
Casualties
Final Report
46 Hill Street,
Belfast,
BT1 2LB
Telephone: 028 9023 4110
Fax: 028 9031 4182
Email : Belfast@cbuchanan.co.uk
____________________________________________ ____________________________________________
David McGuigan Stephen Wood
final_report_120110.doc
(C) Copyright Colin Buchanan and Partners Limited. All rights reserved.
This report has been prepared for the exclusive use of the commissioning party and unless otherwise agreed in writing by Colin
Buchanan and Partners Limited, no other party may copy, reproduce, distribute, make use of, or rely on the contents of the report.
No liability is accepted by Colin Buchanan and Partners Limited for any use of this report, other than for the purposes for which it
was originally prepared and provided.
Opinions and information provided in this report are on the basis of Colin Buchanan and Partners Limited using due skill, care and
diligence in the preparation of the same and no explicit warranty is provided as to their accuracy. It should be noted and is expressly
stated that no independent verification of any of the documents or information supplied to Colin Buchanan and Partners Limited has
been made
Deprivation and Child Pedestrian Road Casualties
Final Report
Contents
Summary 1
1 Introduction 3
1.1 Objectives of the study 3
1.2 Report structure 3
2 Literature Review 4
2.1 Purpose of the review 4
2.2 Context 4
2.3 Literature search 5
2.4 Literature review 6
3 Data sources 22
3.1 Road collision casualty data 22
3.2 Population data 22
3.3 Deprivation data 23
3.4 Urban Rural Classification 23
3.5 Road infrastructure data 23
3.6 People in employment 24
3.7 Workplace population 24
3.8 School pupil data 24
3.9 Car ownership data 25
3.10 Household tenure data 25
3.11 Weather data 25
3.12 Creation of composite variables 25
4 Ten Year Analysis 28
4.1 Datasets 28
4.2 Deprivation dataset 29
4.3 Population dataset 29
4.4 Casualty rate analysis 29
4.5 Location of collision as a proxy for location of residence 33
4.6 Analysis of collision factors 35
5 Statistical Modelling 39
5.1 Overview 39
5.2 Statistical package 39
5.3 Four stages 39
5.4 Stage 1: intercorrelation of dependent variables 40
5.5 Stage 2: Level 1 Modelling 44
5.6 Stage 3: Level 2 modelling 48
5.7 Stage 4: Level 3 modelling 56
6 Conclusions 62
6.1 Link between deprivation and child pedestrian casualties 62
6.2 Statistical modelling 62
6.3 Implications for road safety planning 63
Appendix 1
Appendix 2
Appendix 3
Appendix 4
Deprivation and Child Pedestrian Road Casualties
Final Report
Summary
The study was commissioned by the Northern Ireland Department of the Environment to review child
pedestrian road casualty data for the ten year period 1999-2008 and, in particular, to determine if
there is evidence that suggests a link between child pedestrian casualties and deprivation. The study
also included a comprehensive literature review of UK and Irish sources relevant to the study.
The study was based on demographic and infrastructure data for each of the 890 Census Super
Output Areas (SOAs) and road collision data for the period 1999-2008 supplied by the PSNI.
The road collision data comprised records for 64,651 collisions involving 109,385 casualties of which
8,829 were pedestrians. The number of child pedestrian casualties was 3,235.
The ten year data have been comprehensively tabulated according to a number of collision factors
recorded by the police.
In addition to the ten year dataset, a two year dataset was provided by the PSNI for the period April
2007 – March 2009 (inclusive). Postcode information relating to casualties has been recorded by the
PSNI since April 2007. This allowed an analysis to take place of how the location of collision and
location of residence of the casualty are distributed and whether or not – for the ten year data – the
location of collision could be used as a proxy for location of residence of the casualty. The conclusion
was that the use of location of collision as a proxy for location of residence of casualty was appropriate
in the context of this study. This view was supported by other research considered as part of the
literature review.
Tabulations from the ten year data provided an important quantitative insight into the relationship
between pedestrian casualty rates and area deprivation. The overall average child pedestrian
casualty rate is 0.81 casualties per year per 1,000 population with males in the in the 5 – 7 year age
group showing the highest rate of 3.34 casualties per year per 1,000 population. Across all child age
groups (ages 0 -15) it is shown that children resident in the most deprived areas are 4.8 times more
likely to be injured as a pedestrian in a road collision than those resident in the most affluent areas.
This figure varies by age and sex with females resident in the most deprived areas being 14.7 times
more likely to be injured as a pedestrian. These results were highly statistically significant.
The analysis was carried out in stages involving three levels of statistical modelling. In the Level 1
modelling it was firmly established that there were clear and unequivocal links between pedestrian
casualty rates and deprivation for children and adults and that the rates differed by age, sex of
casualty and an Urban-Rural Definition Group (URDG) classification. This finding was very much in
keeping with outcomes of studies elsewhere in the UK.
The Level 2 modelling looked in more detail at factors other than those used to establish the link in the
Level 1 modelling. These included roads infrastructure data (road lengths by road class and number
of junctions), Local Government District, workplace employment, population density and sunshine
hours. The Level 2 modelling established the key variables which are statistically significant in
explaining pedestrian casualty rates.
Finally, the Level 3 model included only those variables which had statistically “lasted the course” for
retention in the model. These variables related to:
The Multiple Deprivation Measure (MDM) score
MDM Crime domain score (a component of the MDM score)
MDM Proximity to Services domain score (a component of the MDM score)
‘Traffic proxy’ – as an alternative measure of relative traffic movements in SOAs
The number of school places
Sunshine hours
Annual Rainfall
1
Deprivation and Child Pedestrian Road Casualties
Final Report
URDG classification
The coefficients for the three MDM related variables, traffic proxy and school places are highly
statistically significant across most age groups and sexes (p generally <<0.001). The URDG
classification is highly significant for adult pedestrian casualties.
On the basis of the Level 3 model outputs some implications for road safety planning are also
ventured. The key implication relates to fact that deprivation remains a highly significant factor when a
number of other explanatory factors have been included in the model. This suggests that road safety
strategies should recognise the effect of area deprivation on child pedestrian casualty rates and that
interventions should be distributed with an appropriate focus on deprived areas. Specific interventions
might include:
a national “traffic club” scheme
road crossing training for children (Kerbcraft)
complementary speed enforcement devices and speed management programmes
improvements to infrastructure (including traffic calming and 20mph zones)
road safety inputs to the master planning of regeneration projects
local transport plans should include pedestrian reduction targets
a distribution of road safety measures which targets socio-economic environments
2
Deprivation and Child Pedestrian Road Casualties
Final Report
1 Introduction
carry out a 10 year analysis of available PSNI casualty data for child pedestrian
casualties aged 0-15 years.
- if there is no evidence of a link then outline the reasons why this might be
the case (i.e. any data issues or NI-specific factors which mean that the
deprivation effect is not as prevalent here as in the rest of the UK).
3
Deprivation and Child Pedestrian Road Casualties
Final Report
2 Literature Review
2.1.3 This literature review, therefore, has the purpose of identifying the body of existing
relevant research which will inform various aspects of the current commission, viz:
assess the likelihood that a link will be found
identify the key factors which will help to explain the existence of a link
inform the choice of statistical analyses
identify counter-measures which could help reduce the child pedestrian casualty
rate
2.2 Context
2.2.1 Up until the 1990s there was largely anecdotal evidence of the link between social
deprivation and road accident casualty rates. However, during the 1990s with a greater
emphasis being placed on social inclusion, this issue came to the fore in road safety. At
the time there was no specific formal data source for road accident casualty addresses
which could be used in research.
2.2.2 In Great Britain (GB), the Department for Transport undertakes a five year (quinquennial)
review of the STATS19 system for collecting and collating road collision data. In this
review the values of existing data items are reviewed together with assessments of new
data items which might add value to the data collected.
2.2.3 In 1997 Quinquennial Review it was recommended, with effect from January 1999, that
the postcode of the person's home address should be included as part of the statistical
records for each of the drivers and the casualties who were involved in an injury road
accident. It was recognised that the driver and casualty postcodes (and the use of the
special code for non-UK residents) are potentially of great value in the analysis of the
road accident statistics data including the analyses of "socio-economic" background
information about the kinds of areas in which people involved in road accidents live
(SGLGRAS (2000)).
2.2.4 The new PSNI road traffic collision data collection system implemented in NI on 1 April
2007 included for the first time in Northern Ireland details for drivers and casualties.
2.2.5 In GB there has now been just over 10 years worth of postcode data and this has allowed
detailed research to be undertaken for the first time on a national data set incorporating
links to place of residence and social deprivation indices. It is, therefore, this more recent
research which will have the most relevance to the present study. Our search has
4
Deprivation and Child Pedestrian Road Casualties
Final Report
focussed on literature since 1999 but includes relevant earlier work which is of value to
the current study.
2.3.5 This search revealed several hundred different papers and these were examined to
identify relevant material. The majority of the texts identified were either not based on
British or Irish research or on subjects not relevant to the present study (eg child
casualties in deprived areas caused by fire).
2.3.6 The citations in the relevant literature reviewed were also used to identify further
references. Finally, a Google search was also carried out to find items not found on the
databases searched. The Google search used the same keywords as used to search the
Web of Science.
2.3.7 All relevant items were read and where the content was considered to be of particular
interest to the present study it has been distilled and set out in the review.
1
See http://isiwebofknowledge.com/
2
a question mark (?) is used in a search term to indicate a wildcard, matching on zero or more characters in
place of the ?.
5
Deprivation and Child Pedestrian Road Casualties
Final Report
2.3.8 The review is based on a cross-section of literature with a variety of approaches and
conclusions using a number of datasets from across the UK.
2.3.9 Items included in the review are set out in the references section. Items read but not
used are listed in a short bibliography.
2.4.4 Kendrick (1993) noted that existing work on road traffic accidents was based on
analysing accident clusters despite evidence that accidents tended to be more dispersed.
Her study analysed 573 child pedestrian casualties (ages 0 -11) in Nottingham by
deprivation score for the years 1988-90. Casualty rates were mapped against deprivation
score and tabulations were presented. The analysis showed a highly significant
relationship between casualty rates and deprivation score (Chi square tests with p<0.001
and correlation coefficients (rs). Figure 2.1 shows Figure 2 from the report which shows
casualty rates mapped against deprivation score for children aged 5 – 11 years.
6
Deprivation and Child Pedestrian Road Casualties
Final Report
2.4.5 McGuigan, Bull and Gorman (1996) in a paper reviewing links between casualty rates
and deprivation based on over 16,000 casualty records in Lothian Region, Scotland
3
tabulated (their Figure 4) pedestrian casualty rates by social deprivation category which
showed a clear relationship between casualty rates and deprivation (see Figure 2.2).
3
In Figure 2.2 the x axis relates to the Carstairs and Morris Index of Deprivation (called” depcat”) which was
widely used in Scotland as a measure of quantifying relative socioeconomic deprivation or affluence in different
localities across the country.
7
Deprivation and Child Pedestrian Road Casualties
Final Report
Figure 2.2: Figure 4 reproduced from McGuigan, Bull and Gorman (1996)
2.4.6 Abdalla et al (1997) using a three year data set (1990 – 1992) comprising 8,560 casualty
records for Lothian Region, Scotland conducted a wide ranging study into the
relationships between casualty rates and social deprivation indicators for the casualties’
zone of residence. The study identified a statistically significant link between child
pedestrian casualty rates and deprivation.
2.4.7 Table 7 from their paper is reproduced below (Figure 2.3) where pedestrian casualty
rates are stratified by age. It is clear that the disparity between the most deprived and
most affluent groups differs by age. Of relevance to the present study, for children aged
0-4, 5-11 and 12-16 the ratios of the rates (15% most deprived / 15% most affluent) are
4
5.2, 7.6 and 2.4 respectively. Also the SCR values are over 1 for the most deprived
areas and under 1 for the most affluent areas. The difference in rates is highly significant
(p<0.01) based on a statistical test of the difference between the number of casualties per
resident population in the 15% most deprived and 15% most affluent zones.
4
SCR or standardised casualty ratio was a concept adopted by Abdalla et al in their paper. It is calculated as
the observed casualties ÷ expected casualties.
8
Deprivation and Child Pedestrian Road Casualties
Final Report
2.4.8 In a similar study, Chichester et al (1998), analysed 1,300 attendance records at the
Accident and Emergency unit at Monklands/Bellshill Hospital in North Lanarkshire,
Scotland, to seek to understand more about associations between road traffic accidents
and deprivation. Of the 1,300 records, 1,020 were linked to socio-economic data. The
study found significant associations indicating increasing casualty rates for all causes of
injury as deprivation increased. Of relevance to the present study, it was also noted that
analyses on child casualties showed particularly strong significance (Chi square test,
p<0.005).
2.4.9 Building on earlier work in Lothian Region, Scotland, White and Raeside (2001) sought to
understand the underlying reasons why children from socially excluded backgrounds are
more at risk. The authors undertook pupil interviews in schools in the both the most
deprived and most affluent areas in total 128 interviews were conducted at four schools
(two in an affluent area and two in a deprived area). The interviews focused on how
children travelled to/from school, how they played and participation in after school
activities. The conclusions of this study were:
the risk of child involvement in road accidents is highly class related
children from disadvantaged areas tend to suffer more serious injury
children from disadvantaged areas have greater exposure to risk relating to where
they play and lack of parental supervision when travelling to/from school
that there may also be cognitive factors at work because children up to about nine
years of age do not recognise dangerous situations
children in poor areas primarily walk to school, however, children from affluent
background cross more busy roads than their poorer counterparts – probably
reflecting greater travel distances for the affluent group.
there is less accompaniment by adults and more accompaniment by other children
for those in the poorest area compared to those in the most affluent areas
it is speculated that children from poorer areas will be more exposed to risk as they
are less likely to be driven than their affluent counterparts and are more likely to
play in the street
2.4.10 These findings suggest different behaviour patterns by children related to deprivation and
suggest that there may be more complex factors at work than deprivation itself.
2.4.11 Grayling et al (2002) point out that Britain’s good record in reducing road accident
casualties is marred by a relatively poor record on child pedestrian safety. This study for
9
Deprivation and Child Pedestrian Road Casualties
Final Report
the Institute of Public Policy Research mapped pedestrian casualties for the first time in
England and tested the links with deprivation and other area characteristics including:
number of children
traffic generating potential
residential and employment density
length and type of roads
weather and other specific local factors
2.4.12 In total there were more than 16,000 child pedestrian casualties in England in the
dataset. For some of the analysis the data were stratified by deprivation decile. The key
findings and aspects of the study were:
the use of deciles as the basis for analysis
children in the most deprived decile were more than three times as likely to be
pedestrian casualties than children in the least deprived decile – this result was
highly statistically significant
there was a considerable amount of unexplained variation between areas with
similar deprivation scores
the central conclusion of the study was that having controlled for other (non-
deprivation) factors there was still a distinct deprivation effect
population density had little impact on child pedestrian casualty rates
higher numbers of road junctions and A road length are linked to higher child
pedestrian casualty rates
higher annual rainfall and sunshine hours are linked to higher child pedestrian
casualty rates
the components of the deprivation measure were tested separately and it was
found that different components had different effects: in summary the components
relating to income, health, education and housing deprivation were associated with
higher pedestrian casualty rates whilst the components of employment and
accessibility were associated with lower rates
the effect of deprivation varied by region across England; for example, in London
the deprivation effect is particularly strong for children but less so for adults (with
respect to the average for the study)
the report ends with a discussion of possible remedial measures to reduce the
effects of deprivation on the rate of child pedestrian casualties
2.4.13 Figure 2.4 reproduces Figure 1.9 of the report which shows casualty rates by ward decile
and demonstrates how the casualty rate rises with increasing deprivation.
10
Deprivation and Child Pedestrian Road Casualties
Final Report
2.4.15 Graham et al (2004) report on a study into child and adult pedestrian casualties in
England focusing on the influence of socio-economic deprivation on casualty rates. The
study was based on STATS19 data for the years 1999 and 2000 provided by the
Department for Transport (DfT). The dataset comprised 28,228 child pedestrian casualty
records in 8,414 wards in England.
2.4.16 The report authors begin by stating that it is easy to see in the “raw” data that there is an
apparent strong relationship between casualty numbers and measures of deprivation.
They warn that one should not jump to the conclusion that there is a simple relationship
between deprivation and casualties which is unrelated to other factors. The authors
identified the possibility that deprivation is more commonly found in dense urban areas
and that collisions occur more frequently in dense urban areas and that this could have
more to do the relationship than deprivation. The work described in the paper sought to
disentangle the effects of area type from the influence of deprivation.
2.4.18 Regression modelling was used to link the independent variables to the casualty rates.
The key finding was that in addition to the area type variables used in the statistical
models, deprivation provides an additional explanation of the pedestrian casualty rates.
This outcome was rigorously tested and finding was confirmed as highly statistically
significant (t-statistic = 14.7 (p<< 0.001)) . The central conclusion was that “not only is
deprivation associated with an increase in the probability of a pedestrian casualty taking
place, but also with an increased probability of incidents involving a serious injury or a
fatality”. This deprivation effect was present for both adults and children but to a much
larger extent for children.
11
Deprivation and Child Pedestrian Road Casualties
Final Report
2.4.19 In Wales, a wide ranging report on deprivation and health by the National Public Health
Service (NPHS (2004)) sets out child pedestrian casualty rates by quintiles for child
pedestrian casualties reported to the Police and also for hospital inpatient episodes.
These are reproduced in Table 2.2. The data relate to slightly different time periods, age
grouping and casualty severity. Also, for the STATS19 data it is the location of accident
that provides the deprivation measure in contrast to the hospital inpatient episode
database which assigns the deprivation measure on the basis of place of residence.
Nevertheless, both datasets show a clear link between deprivation and child pedestrian
casualty rates. On the basis of the STATS19 data, children in the most deprived areas
were found to be two and a half times more likely to be injured in a road collision as a
pedestrian.
1 2
Reported To Police Hospital Inpatient Episodes
3 3
Deprivation fifth Children aged 4−16 years Children aged 5−14 years
1 (least deprived) 82.6 18.3
2 119.3 24.8
3 161.6 26.3
4 178.3 31.1
5 (most deprived) 209.3 39.3
Overall 159.8 29.6
Rate Ratio 5:1 2.52 2.15
1 STATS19 (1995-2000)
2 Patient Episode Database foe Wales (PEDW) (1997-2002)
3 Rate per 100,000 persons per year
2.4.20 The value of this report to the present study is that it complements the findings of the
Scottish and English research reviewed above on the basis of an independent dataset for
Wales.
2.4.21 Noland and Quddus (2004) pointed out that land use characteristics of an area had not
featured in earlier analyses and that deprived neighbourhoods may be subject to higher
traffic volumes than more affluent areas. In a detailed and wide ranging study based on
1999 casualty data for England casualty rates were modelled against a number of land
use, road network, traffic and demographic characteristics. The data comprised:
an undisclosed number of casualty records for 1999 across 8,414 wards
land use data
road characteristic data
demographic characteristic data
proxy for traffic flow data
2.4.22 The study did not look specifically at child pedestrian casualty rates. The main
conclusions were:
lower population densities were linked to higher casualty rates
higher employment densities were linked to higher casualty rates
larger levels of urbanisation were associated with fewer casualties
road lengths are associated with increases in serious injuries
deprivation is linked to higher casualty rates
higher exposure to traffic was linked to higher casualty rates
12
Deprivation and Child Pedestrian Road Casualties
Final Report
2.4.23 Silversides et al (2005) recognised that injuries in childhood represent a major public
health concern particularly in North and West Belfast where there is a high rate of
emergency department attendance due to injury and in where social deprivation is
widespread. In a study of nearly 479 children (aged 0 – 12) presenting to emergency
departments in North and West Belfast, Silversides et al (2005) identified inter alia a
highly significant correlation (p < 0.001) between economic deprivation (measured by the
Noble Index) and injuries due to collisions on the road. This was a small and prospective
study but it does add a Northern Ireland dimension to the link between deprivation and
road collisions.
2.4.24 In 2006 a wide-ranging report to the London Road Safety Unit based on STATS19 data
looked at the links between road collision casualty rates and deprivation (Edwards et al
(2006)). The purpose of the study was to examine the strength of the relationships
between deprivation and child pedestrian rates to help develop road safety policies and
interventions. The casualty data comprised 5,823 child pedestrian casualty records
(ages 0 – 15) for the six year period 1999 to 2004. The study used an adjusted Index of
Multiple Deprivation (IMD) to measure levels of deprivation in each of 4,765 Super Output
Areas (SOAs) in London. The adjustment was to remove an element of the
environmental component of the IMD which itself includes an indicator that relates to road
traffic injuries to pedestrians and cyclists. This adjusted IMD was called IMD2. This
problem will not arise with the NI data as the multiple deprivation measure (MDM)
adopted for Northern Ireland does not include an element for pedestrian and cyclist
casualties (NISRA (2005)).
2.4.25 The research methodology involved ranking the SOAs from 1 (least deprived) to 10 (most
deprived) and then basing the statistical analysis on this stratification. These were
referred to as deprivation deciles. The study also considered the effect that road network
variables had on the relationship.
2.4.26 The key aspects and findings of the London study of relevance to the NI study are:
the use of deciles as the basis for analysis
the pedestrian injury rate for children living in the most deprived decile were found
to 2.93 times as high as that for children resident in the least deprived decile
road length per hectare and number of junctions/km of road were included in the
multivariable regression analysis carried out
the adjustment to create IMD2 did not offer any material benefits in the analysis
over using the unadjusted IMD
for certain casualty types – especially child pedestrians and cyclist casualties – it is
acceptable to conduct an analysis linking casualties to the resident population of
the area in which the collision occurred (but not other casualty types)
2.4.27 The London study concluded that it is not necessarily the deprivation of an area per se
that causes the increased risk. For example, the risk of being involved in a road collision
increases with the time spent travelling and people resident in deprived areas may be
more likely to be injured as pedestrians because they don’t have a car and are more
likely to be walking.
2.4.28 London data were also used to monitor the relationship between injury and deprivation
over time. The outcome of this was an observed downward shift in overall child casualty
rates over time but that there had been little change in the relative distribution of
casualties by IMD. This result is broadly confirmed by a similar analysis of data for the
City of Edinburgh where child pedestrian casualty rates for 1991-94 were compared with
equivalent data for 2000-03 (McGuigan et al (2006)). Figure 2.5 clearly demonstrates the
overall fall in casualty rates between the two periods but note that the slope of the trend
line for 2000-03 is not as steep as that for 1991-94. This suggests that the overall
casualty rates are falling and that the disparity between rates for the most affluent and
13
Deprivation and Child Pedestrian Road Casualties
Final Report
most deprived are also falling. In 1991-94 the child pedestrian casualty rate for children
living in the most deprived areas (depcat = 7) was 4.58 times higher than children living in
the least deprived areas (depcat = 1). The equivalent figure for the 2000-03 data was
4.05).
5
Figure 2.5: Casualty rates by deprivation category (depcat ) for pedestrians in
age group 5-14 for 1991-94 and 2000-03 (City of Edinburgh)
2.4.30 These parts of the report may be of value to Road Safety Division in any policy
development work that may be required following consideration of the final report for this
study.
2.4.31 Later work by Graham and Stephens (2008) looked specifically at the impact of
deprivation on child pedestrian casualties in England. The STATS19 data for an
undisclosed number of casualties for the five year period 1998 to 2002 across 7,925
small areas were used in the study. An objective of the study was to decomposed the
deprivation effect by considering six separate dimensions related to:
income
employment
health
education
housing and services
crime
14
Deprivation and Child Pedestrian Road Casualties
Final Report
2.4.33 Area subsets were also coded to provide a relative measure of urbanisation according to
10 factors (ranging from Central London to rural) based on Department for Transport
(DfT) classifications. Separately generalised linear regression models were used. .
2.4.34 The findings were complex but consistent with previous research.
2.4.36 A study of fatal casualties in England by Clarke et al (2008) looked only at vehicle
occupant fatalities. This study, however, does have some relevance to the present study
because for every pedestrian road casualty there will be a vehicle driver involved: both
parties are likely to have contributed to the cause of the collision. [Earlier unpublished
work by McGuigan did show that there was a relationship between casualty and driver
indexes of deprivation. As most pedestrian casualties tend to be injured close to home it
is likely that the distribution of driver home addresses will reflect to an extent the area in
which the collision occurred – hence the relationship]. In the study by Clarke et al it was
found that there was a relationship between the IMD scores for drivers killed in road
collisions and deliberate risk taking. This may suggest that drivers resident in deprived
areas are more likely than drivers in less deprived areas to take more risks and drive less
responsibly and – as a consequence – are more likely to be over-represented in
pedestrian/vehicle collisions.
2.4.37 Edwards et al (2008) in a study to examine the variation in serious injuries in children by
the level of deprivation in an area and by type of settlement in England. Hospital
admission rates for serious injury to children aged 0-15 years (7,840 No) in census super
output areas in England during the 5-year period 1 April 1999 to 31 March 2004 were
analysed
15
Deprivation and Child Pedestrian Road Casualties
Final Report
reproduced in Figure 2.6. In this table it can be seen that a quarter of children in the 5-11
age group are injured as pedestrians at a distance of over 4.5km from their zone of
residence.
Pedestrian
Lowest Second Middle Fourth Highest
age
0-4 0.009 0.055 0.102 0.436 31.844
5-11 0.004 0.097 0.211 4.505 35.810
12-16 0.011 0.280 0.674 2.742 39.810
17-24 0.021 0.599 2.231 4.852 40.500
25-59 0.023 0.360 1.378 3.980 42.679
Elderly (≥ 60) 0.009 0.150 0.457 2.304 32.286
2.4.41 Petch and Henson (2000) in a study of factors relating to child accidents in Salford,
England looked at a number of factors to explain the distribution of casualties and
casualty rates. Of interest to the present study is a tabulation of the distance between
accident location and place of residence. Figures for child pedestrian casualties were set
out in the report and are reproduced here in Table 2.3. It can be seen that 78% of child
pedestrian casualties in Salford lived less than 1km away from the location of the
accident with 22% who lived more than 1km away. The proportion of casualties occurring
in collisions in excess of 2km from home is 11.4%
Table 2.3: Distance between accident location and place of residence (Petch
and Henson (2000))
2.4.42 These previous studies do not present direct comparisons of casualty residence and
collision location but they do suggest that up to one in four children may be injured in a
zone outside their zone of residence. It should noted, however, that this does not
necessarily mean that the deprivation score assigned on the basis of the accident
location will be different from the deprivation score of their home zone.
2.4.43 Hewson (2004) studied child pedestrian casualties in Devon. This study confirmed that
there were links between deprivation and increased child pedestrian casualty risk.
Hewson specifically looked at the question of whether it is better to aggregate the
casualty data on the basis of the ward in which the collision occurred or on the basis of
the ward in which the casualty was resident. It was found that where the data had been
16
Deprivation and Child Pedestrian Road Casualties
Final Report
aggregated by the ward in which the collision occurred there is evidence that that urban
wards have a higher casualty rate than rural wards. The same cannot be said where the
data were aggregated by ward of residence. This suggests that those collisions which
involve child pedestrians in areas in which they are not resident tend to be urban in
nature. Indeed, the author raises the question of whether or not the casualty rate is more
likely to be associated with the deprivation of the zone in which the collision occurred
than with that of the home address of the child (the author warns - given the limitations of
the underlying data - that conclusions have to be drawn with great caution).
2.4.44 A key finding of the London Study (Edwards et al (2006)) was that for certain casualty
types – especially child pedestrians and cyclist casualties – it is acceptable to conduct an
analysis linking casualties to the resident population of the area in which the collision
occurred. This needs to be compared with the finding by Abdalla et al (1997) that a
quarter of children in the 5-11 age group are injured as pedestrians at a distance of over
4.5km from their zone of residence (Figure 2.6 refers). Indeed, McGuigan et al (2006)
identified the proportion of “home” casualties by deprivation category. In this case, the
area level adopted was postcode sector and a “home” casualty is defined as a casualty
who was both injured and resident in the same postcode sector. Figure 2.7 shows that on
average about 60% of child casualties in Edinburgh were “home” casualties. The
average number of “home” casualties fell by a small amount between 1991-94 and 2000-
03. It should be noted the Edinburgh data are for all child casualties.
6
See footnote relating to Figure 2.2.
17
Deprivation and Child Pedestrian Road Casualties
Final Report
to what measures can be put in place to reduce the rate deficit between children living
affluent and deprived neighbourhoods.
2.4.46 Perhaps the first intervention was undertaken in Lothian Region, Scotland when on 1
April 1994 Lothian Regional Council and Lothian Health Board initiated the first region-
wide universally free Children's Traffic Club (CTC) in the UK. This intervention was
introduced directly as a result of the casualty rate deficit between children living in more
affluent and less affluent areas. Take-up rates for the first six month period showed a
clear gradation of take-up rates ranging from 62% in the most affluent areas to 39% in the
most deprived areas. Parallel measures were put in place to encourage membership of
the club and these included:
free distribution of CTC materials to nurseries and pre-school establishments
health visits to children included encouragement to join
mail drops
2.4.47 On the back of Lothian’s initiative, the then Scottish Office introduced a national CTCS
(Children’s Traffic Club for Scotland) in November 1995. This was supported by a
television advertising campaign. This initiative has recently been reviewed and a report
of consultation has been published (Road Safety Scotland (2009)). This report concludes
that there were many positive aspects of the resource but that it was some concern about
its general appropriateness. It was felt that the resource was too “busy” and “wordy” for
the age group it is aimed at. There was a strong sense that the CTCS was a worthwhile
but – after 13 years - a dated resource which needed refreshing. This report is currently
with Road Safety Scotland for consideration and future action.
2.4.48 The link between problem behaviour and social values is confirmed by the Department
for Transport DfT (1999) in a road safety research report which looked at problem
behaviour in children and accident risk. The report included a focused literature review of
previous research on accident risk and behavioural issues together with some original
research based on child and parent interviews. The conclusion is that social values
explained the link between problem behaviour and traffic collisions but not other types of
accidents. The study looked at the possibility that problem behaviour led to greater time
spent on the street exposed to risk and this was controlled for in the analysis. Taking this
into consideration, the report concluded with a high level of confidence that there is a link
between problem behaviour and traffic collision rates in children and that this “was not
simply a result of delinquent children spending more time in the traffic environment”. The
report also concluded that tackling this problem will require quite intensive and prolonged
intervention to establish significant improvements. The report, unfortunately, does not
indicate what these interventions might be.
2.4.49 Edwards et al (2008) in a wider study of serious injuries to children and deprivation
concluded that the major policy need related to safer road environments for pedestrians
and cyclists involving speed reduction. It is also suggested that, in the interests of
developing a “critical mass” effect, more walking and cycling should be encouraged as
this may well reduce the overall risks for these modes.
2.4.50 In an earlier report for the London Road Safety Unit, Edwards et al (2006) looked at a
range of potential remedial measures. They spelt out a general strategy in which they
recommended:
promotion of walking and cycling and discouragement of motor vehicle use
car free planning and designing particular urban areas for minimal motor vehicle
use
pedestrian-orientated commercial streets
public parks that encourage or require non-automotive access
18
Deprivation and Child Pedestrian Road Casualties
Final Report
2.4.51 They also highlighted the need for well targeted traffic calming and speed enforcement
action (eg radar, speed cameras etc). Education, training and publicity also have a place
in a balanced and holistic response to high casualty rates.
2.4.52 Grayling et al (2002) indicate that there are some highly cost-effective ways of reducing
pedestrian casualties including:
‘spot’ measures such as installing pedestrian crossings
more 20mph zones
local transport plans should include pedestrian reduction targets
all local transport plans should include a speed management strategy
2.4.53 Since this report there is evidence that more local authorities are looking closely at risks
in areas of deprivation and are taking wider action to reduce the impact of deprivation
socially and also specifically in terms of road safety. More 20mph zones (either
mandatory or advisory) have also been introduced.
2.4.56 Whelan et al (2008) in a comprehensive report look at the Kerbcraft initiative which has
been introduced in a number of deprived areas in Great Britain. The initiative is shown to
improve children’s crossing skills and behaviour and the report looks at ways in which
Kerbcraft can be improved and rolled out to other areas. This initiative may well be
suitable for areas of deprivation in Northern Ireland. It relies heavily on volunteers and
requires a strong commitment from officials to make it work successfully and sustainably.
19
Deprivation and Child Pedestrian Road Casualties
Final Report
2.4.58 It is perhaps worth noting that much of the research reviewed which relates to England
although independently carried out will have been based, in part, on the same data. The
sources for the casualty data would have been the STATS19 and the other data sources
for deprivation measures, population, road infrastructure, weather etc will have been
based on General Register Office, Met Office, Department for Communities and Local
Government, DfT etc. Given this, the findings are not based on entirely different datasets
and it would be surprising if interpretations and conclusions differed significantly. The
complementary findings by other researchers using different data sources are, therefore,
of particular relevance in supporting the overall conclusions [Scotland - McGuigan, Bull
and Gorman (1996), Abdalla et al (1997) and Chichester et al (1998),: Wales - NPHS
(2004): Northern Ireland - Silversides et al (2005)].
20
Deprivation and Child Pedestrian Road Casualties
Final Report
21
Deprivation and Child Pedestrian Road Casualties
Final Report
3 Data sources
3.1.3 The ten year dataset January 1999 to December 2008 (inclusive) comprises:
64,651 collision records
121,260 vehicle records
109,385 casualty records
3.1.4 The distribution of casualties by casualty class and age group is presented in Table 3.1.
Table 3.1: Casualties by casualty class and age group (January 1999 to
December 2008)
Age group
16
and All
1 2
Road user group 0 1-4 5-7 8-11 12-15 over ages
1. Driver 20 10 60 67 145 55,652 55,954
2. Pillion passenger 1 10 38 210 259
3. Vehicle passenger - Front 25 263 257 625 1,128 19,449 21,747
4. Vehicle passenger - Rear 35 1,506 1,198 1641 1,894 10,699 16973
5. Pedestrian 11 499 672 960 1,104 5,583 8,829
6. Motorcyclist 1 9 44 3,871 3,925
7. Pedal cyclist 25 150 264 262 997 1,698
All road users 92 2,304 2,337 3,576 4,615 96,461 109,385
1
age 0 is used to record an age which is unknown
2
age 1 is used to record casualties of age 0 (ie before first birthday) and age 1
See also paragraph 4.1.8
3.1.5 The collision data include all collisions in NI in which there was an injury to at least one
road user. The severity of injury is recorded as killed, serious or slight where
Killed is assigned if the casualty dies at the scene of the collision or within 30 days
as a result of the injuries sustained in the collision
Serious is assigned if the casualty is detained in hospital as an ‘in-patient’, or if any
of the following injuries whether or not the casualty is detained in hospital:
- fractures
- concussion
- internal injuries
22
Deprivation and Child Pedestrian Road Casualties
Final Report
- crushings
- burns
- severe cuts and lacerations
- severe general shock requiring medical treatment
Slight is assigned if the injury is of a minor character such as a sprain, bruise or cut
not judged to be severe, or slight shock requiring roadside attention.
(http://www.nisranew.nisra.gov.uk/census/Excel/commissioned_output/EXT20060206.xls)
3.2.2 The table presents these data by single year of age and by sex.
3.3.2 The Multiple Deprivation Measure for 2005 (MDM 2005) was constructed by combining
the seven transformed domain scores, using the following weights:
Income (25%)
Employment (25%)
Health Deprivation and Disability (15%)
Education, Skills and Training (15%)
Proximity to Services (10%)
Crime and Disorder (5%)
Living Environment (5%)
3.3.3 In addition the table provides two further deprivation measures:
the Income Deprivation Affecting Children (IDAC)
the Income Deprivation Affecting Older People (IDAOP)
3.3.4 The distribution of the MDM (organised by decile) is presented in Figure 4.1 in Appendix
4.
http://www.ninis.nisra.gov.uk/mapxtreme_deprivation2005/Viewdata/UrbanRuralClassific
ation2005.xls
23
Deprivation and Child Pedestrian Road Casualties
Final Report
number of junctions (i.e. nodes on the digitised network with three or more
attached road links)
3.5.2 From these a junction density can be calculated as:
3.6.2 The data are disaggregated according to the following occupation groups:
Managers and senior officials
Professional occupations
Associate professional and technical occupations
Administrative and secretarial occupations
Skilled trades occupations
Personal service occupations
Sales and customer service occupations
Process, plant and machine operatives
Elementary occupations
24
Deprivation and Child Pedestrian Road Casualties
Final Report
3.12.2 Traffic proxy: this variable is designed to provide a proxy for travel activity levels in each
SOA. This variable was described by Graham et al (2005). It is a function of the two
variables:
people in employment (i.e. the number of people in employment resident in the
SOA)
workplace population (i.e. the number of people with jobs within the SOA)
taking the form:
3.12.3 Road length per hectare: this variable is designed to provide a measure for road density
in each SOA. This variable was described by Noland and Quddus (2004). It is a function
of the two variables:
total road length
SOA area (hectares)
taking the form:
3.12.4 Junctions per road km: this variable is designed to provide a measure for junction
density in terms of road length in each SOA. It is a function of the two variables:
number of junctions
25
Deprivation and Child Pedestrian Road Casualties
Final Report
3.12.5 Junctions per hectare: this variable is designed to provide a measure for junction
density in each SOA. This variable was described by Graham et al (2005) and Noland
and Quddus (2004). It is a function of the two variables:
number of junctions
SOA area (hectares)
taking the form:
3.12.6 Traffic density per road km: this variable is designed to provide a relative measure for
traffic density levels in each SOA in terms of road length. It is a function of the two
variables:
traffic proxy
total road length
taking the form:
3.12.7 Traffic density per junction: this variable is designed to provide a relative measure for
traffic density levels in each SOA in terms of the number of road junctions. It is a function
of the two variables:
traffic proxy
number of junctions
taking the form:
3.12.8 Resident population per road junction: this variable is designed to provide a relative
measure for population exposure at junctions in each SOA. This variable was described
by Graham et al (2005). It is a function of the two variables:
total resident population
number of junctions
taking the form:
3.12.9 Employed residents per road junction: this variable is designed to provide a relative
measure for employed resident exposure at junctions in each SOA in terms of the
number of road junctions. This variable was described by Graham et al (2005). It is a
function of the two variables:
employed residents
number of junctions
taking the form:
3.12.10 Workplace population per road junction: this variable is designed to provide a relative
measure for workplace population at junctions in each SOA in terms of the number of
26
Deprivation and Child Pedestrian Road Casualties
Final Report
27
Deprivation and Child Pedestrian Road Casualties
Final Report
4.1 Datasets
Data Sources
4.1.2 The data used in the study have been either downloaded from the Northern Ireland
Statistics and Research Agency (NISRA) website or have been provided directly by the
appropriate agency. For a detailed description see Chapter 3.
4.1.4 The main dataset comprises data for the ten year period covering the calendar years
1999 to 2008. This dataset will be used for the main purpose of the study to examine if,
in Northern Ireland, a relationship exists between deprivation and the child pedestrian
casualty rate.
4.1.5 The other dataset comprises data for the two year period from April 2007 to March 2009
(inclusive). This data set will be used to test the assumption that the location of collision
may be used as a proxy for location of casualty residence. It is important to investigate
this because information regarding people’s deprivation status is usually related to their
home address. In Northern Ireland (NI), casualty postcodes have only recently been
assigned to the collision data (from April 2007 onwards) and it is not, therefore, possible
to use the socio-economic status of the home address for the ten year dataset. Using the
two year dataset, however, it is possible to compare the assignment of deprivation to
individual casualties based on location of collision and location of residence to determine
the differences that exist and draw conclusions about the use of location of collision as a
proxy for location of residence.
4.1.8 The PSNI reserve the use of an age of 0 for casualties of an unknown age and the
dataset includes 92 such casualties. Accordingly, these 92 casualties over the 10 year
period have been excluded from the analysis. Casualties who are aged under one year
are coded as age “1”.
4.1.9 As can be seen in Table 2 there are four age groups for children accounting for 3,235
child pedestrian casualties.
7
All Tables and Figures referred to in this Chapter are presented in Appendix 2.
28
Deprivation and Child Pedestrian Road Casualties
Final Report
4.1.12 As can be seen in Table 4 there are four age groups for children totalling 476 child
pedestrian casualties.
4.2.2 The SOAs have been sorted by MDM values and split into ten groups ranging from 1
(most affluent) to 10 (most deprived). These groupings are referred to as deciles.
4.4.3 The corresponding populations stratified by MDM deciles, age group and sex are set out
in Table 6.
4.4.4 The casualty rates for these stratifications are set out in Table 7. The casualty rate is
expressed as the annual average number of casualties per 1,000 population, viz:
4.4.5 It can be seen in Table 7 that for each age group for both males and females, the
casualty rates show an upward trend as the MDM decile rises. In some cases the trend
is clear (e.g. for males aged 8 to 11) in others it is less clear (e.g. for females aged 12-
15). In all cases, however, the relationships are highly statistically significant at least at
the 2.5% level, with most of the relationships being statistically significant at the 0.5%
level (Table 8 refers).
4.4.6 Table 9 shows deprivation disparity ratios (DDRs) which are the ratios of rates for decile
10 (most deprived) to rates for decile 1 (most affluent). The DDR is a measure of the
additional risk of being involved as a pedestrian road collision casualty between children
resident in the most affluent and most deprived areas. It is noted that the overall DDR for
females is higher than the value for males. It would, however, be unwise to draw specific
conclusions from this result because in terms of casualty numbers the data for decile 1
29
Deprivation and Child Pedestrian Road Casualties
Final Report
(most affluent) are the most sparsely populated area of the data. There are 46 female
casualties and 87 male casualties recorded for decile 1 (Table 5 refers).
4.4.7 The rates set out in Table 7 are shown in graphical form in Figures 1 – 8.
Child pedestrian casualty rates by decile and sex (all age groups)
4.4.10 Figure 2 shows the casualty rates for all child age groups grouped by sex.
30
Deprivation and Child Pedestrian Road Casualties
Final Report
4.4.22 The corresponding populations stratified by MDM deciles, age group and sex are set out
in Table 11.
4.4.23 The casualty rates for these stratifications are set out in Table 12.
4.4.24 It can be seen in Table 13 that for each age group for both males and females, the
casualty rates show an upward trend. In all cases the relationships are highly statistically
significant at the 0.5% level.
4.4.25 Table 14 shows deprivation disparity ratios (DDRs) which are the ratios of rates for decile
10 (most deprived) to rates for decile 1 (most affluent). Here the DDR is a measure of
the additional risk of being involved as a pedestrian road collision casualty between
adults resident in the most affluent and most deprived areas.
4.4.26 The rates set out in Table 12 are shown in graphical form in Figures 9 – 11.
31
Deprivation and Child Pedestrian Road Casualties
Final Report
Young adult pedestrian casualty rates by decile and sex (ages 16 – 24)
4.4.27 Figure 9 shows the casualty rates for young adults aged 16-24 grouped by sex.
Adult pedestrian casualty rates by decile and sex (ages 25 and over)
4.4.29 Figure 10 shows the casualty rates for young adults aged 25 and over grouped by sex.
Pedestrian casualty rates for all ages by decile and sex (ages 0 and over)
4.4.31 Figure 11 shows the casualty rates for all ages (0 and over) grouped by sex.
32
Deprivation and Child Pedestrian Road Casualties
Final Report
Overview
4.5.2 In this section the two year – April 2007 to March 2009 (inclusive) - dataset is used for
the analysis.
4.5.3 During this period there were 476 child pedestrian casualties. All but 28 (6%) of these
casualties were reported with a postcode which has been assigned to an SOA. The
location of each collision has also been assigned to an individual SOA and we describe a
“home” casualty as a person who was injured in a collision in the same SOA in which
he/she is resident. An “away” casualty is a person injured in an SOA outside the SOA of
his/her residence. Table 15 summarises these.
Analysis
4.5.4 Information regarding the SOA of residence has been recorded in Northern Ireland for
each collision since 1 April 2007. This means that the analysis of the ten year data set
cannot use the casualty “home” SOA information as the link does not exist for all but 21
months. There is a concern that using the location of collision as a proxy may impair the
analysis. This concern is not, however, generally, borne out by the literature review. The
purpose of this analysis is to examine the issue using data local to Northern Ireland.
4.5.5 Of the casualties for which a postcode was provided, 54% were injured in collisions
outside their home SOA with the remaining 46% injured within their home SOA. Although
the “away” collisions may be in any other SOA in the province, it is likely that the majority
will occur more close to home (perhaps in an adjacent SOA) and possibly in an SOA with
the same or similar MDM and deprivation decile.
4.5.6 The literature review includes a study by Petch and Henson (2000) which examined
factors relating to child accidents in Salford, England. In this study the authors reported
on a tabulation of the distance between accident location and place of residence. Figures
for child pedestrian casualties were set out in the report and are summarised in the first
two columns of Table 16. The third column of the table shows the equivalent data for NI.
The NI distances were provided by the PSNI for 447 of the 448 child pedestrian
casualties in the dataset. The data for NI show that whist 68.5% of casualties are injured
within 1km of their home address, the corollary is that 31.5% are injured at distances of
more than 1km with 21.4% at distances in excess of 2km.
4.5.7 The NI child pedestrian casualties tend to be involved in collisions further from their home
than is reported for Salford. Over one in five (21.4%) of NI casualties occur more than
2km from the home address compared with an equivalent figure for Salford of 11.4%.
The Salford data would be expected to show a different distribution for longer distances
because the dataset was limited to casualties resident in Salford and who were injured in
2
Salford. Salford has an area of 96.4 km which is very much smaller than the area of NI
2 9
which is 13,576.5 km (excluding major inland waterways) . It is interesting to note that
the proportions of casualties injured within .25km of home are almost identical for both
Salford and NI.
4.5.8 For two reasons this need not suggest that the location of the collision is necessarily a
poor proxy for the location of residence of casualties. These reasons are:
some longer distances – particularly in rural areas – may occur within an SOA
which covers a wide area
9
http://www.nisranew.nisra.gov.uk/census/area_measurement.html
33
Deprivation and Child Pedestrian Road Casualties
Final Report
where the SOA for the collision is different from the SOA of residence of the
casualty, the SOAs may be adjacent and may share the same (or similar)
deprivation score.
4.5.9 Table 17 summarises the analysis by decile and crosstabulates the decile for the location
of the collision for “away” casualties. Columns 2 and 3 identify the number of “home” and
“away” casualties. The next ten columns tabulate the deprivation deciles for the location
of collision for “away” casualties. Although some of the numbers are small, it is evident
that there is some clustering of “away” collision deciles around the MDM decile value of
the casualty’s home SOA.
4.5.10 The shaded cells in Table 17 highlight the “away” deciles that are within two deciles of
the “home” decile. Two deciles have been selected as an indicative “buffer” to identify
broadly similar levels of deprivation. So, for example, where the decile of residence is 7,
the “buffer” area includes away deciles ranging from 5 to 9. The second last column
sums the number of casualties in the “buffer” and the final column adds these to the
decile of residence to create a “buffer total” value. This, therefore, creates a value for:
4.5.11 The buffer total for all the data sums to 362 which comprises 81% of the total number of
child pedestrian casualties in the dataset. This means that for 81% of all child pedestrian
casualties the use of location of collision provides the same or similar (within 2 deciles)
deprivation score that would be assigned from location of residence. Or, to put it another
way, fewer than 20% of the casualties are injured in away SOAs with widely differing
MDM scores.
4.5.12 Figure 14 shows pedestrian casualty rates where the deprivation decile is based on
location of residence. Figure 15 shows pedestrian casualty rates where the deprivation
decile is based on location of collision. Figure 16 shows the difference in rates by
deprivation decile: here the difference is:
4.5.13 Where the difference is positive it means that the collision based rate is higher and when
it is negative the opposite is true: the residence based rate is higher.
4.5.14 Figure 16 has deliberately been kept to the same scale as Figures 14 and 15 for
comparison. The key points to note are:
the value of the differences are generally low
there is no particular pattern or trend to the differences in relation to the decile
values
Conclusion
4.5.15 The analysis shows that:
for 81% of all child pedestrian casualties the use of location of collision provides
the same or similar (within 2 deciles) deprivation score that would be assigned
from location of residence
the difference between casualty rates based on location of collision and location of
residence are:
- generally low
- not biased in relation to deprivation decile
34
Deprivation and Child Pedestrian Road Casualties
Final Report
4.5.16 The conclusion is that the use of location of collision as a proxy for location of residence
is very unlikely to skew results to an extent that would substantially impact on the
robustness of the findings of the study.
10
The figure for all casualties includes casualties whose severity is defined as slight. The difference between
KSI casualties and all casualties represents slights casualties.
11
The age groups 0-4, 8-11 and 12-15 cover periods of four years. The age group 5-8 covers three years. To
control for this difference, casualty totals in the former periods have been divided by 4 and the casualty totals for
the 5-7 age group have been divided by 3 to provide average values for each age group by year of age.
35
Deprivation and Child Pedestrian Road Casualties
Final Report
4.6.5 From Table 18 it can be seen that for weekdays casualty numbers peak for the three
hour travel home from school period 15.00 (3pm) to 18.00 (6pm). There is a smaller
peak in the morning during the travel from home to school hour beginning 08.00 (8am).
These four hours account for 50% of all child pedestrian casualties during weekdays. In
terms of monthly distribution it can be seen that child pedestrian casualties tend to peak
during spring (April – June) and the autumn (September – October) although this is not
very pronounced.
4.6.6 For weekends (Table 19 refers) the picture is rather different with casualty numbers
peaking throughout the afternoon and into the evening over the nine hour period 12.00
(noon) until 21.00 (9pm). These nine hours account for 87% of all child pedestrian
casualties at weekends. As is the case for weekday casualties, in terms of monthly
distribution it can be seen that child pedestrian casualties tend to peak during spring
(April – June) and the autumn (September – October) although this is not very
pronounced.
Speed limit
4.6.7 The distribution of child pedestrian casualties by speed limit is shown in Table 20. The
data have been stratified by urban rural classification, sex and age group.
4.6.8 It is evident that in rural areas that the proportion of casualties injured on roads subject to
a 30mph speed limit falls with increasing age with a corresponding rise in the casualties
on unrestricted roads where the speed limit is 60mph. This is observed for both males
and females.
4.6.9 In urban areas there is similar but much less marked situation where the proportion of
casualties injured on roads subject to a 30mph speed limit falls with increasing age with a
counterbalancing rise in the casualties on roads where the speed limit is 40 to 60mph.
This is observed for both males and females.
Weather
4.6.10 The distribution of child pedestrian casualties by speed limit is shown in Table 21. The
data have been stratified by urban rural classification, sex and age group.
4.6.11 The significant majority of casualties are injured in collisions when fine weather prevails.
The urban rural classification has no clear effect. For females and males it is evident that
in urban areas the proportion of casualties injured during fine weather falls with
increasing age with a corresponding rise in the casualties injured when it is raining. The
same pattern is not evident for rural areas.
4.6.13 In total, 80% of casualties are involved in collisions during daylight hours; this figure rises
to over 95% in the summer months (May – July) and falls to under 65% in the winter
months (November – January).
4.6.14 Conversely, in total, 20% of casualties are involved in collisions during darkness hours;
this figure rises to 50% in December and falls to under 5% in the summer months (May –
July).
36
Deprivation and Child Pedestrian Road Casualties
Final Report
Carriageway type
4.6.18 The carriageway type associated with child pedestrian casualties is set out in Table 26
stratified by urban rural classification. Collisions on single carriageway roads account for
over 89% of all child pedestrian casualties. The urban rural classification does not
significantly influence the overall distribution of carriageway types associated with child
pedestrian casualties.
37
Deprivation and Child Pedestrian Road Casualties
Final Report
casualties. The urban rural classification does not significantly influence the overall
distribution of road surface conditions associated with child pedestrian casualties.
Carriageway hazards
4.6.21 Table 29 sets out the carriageway hazards present for collisions involving child
pedestrian casualties. 99% of the collisions are reported to have no carriageway hazards
with 1% relating to other object in carriageway.
4.6.23 Two thirds (67%) of vehicles were recorded as “going ahead other” with 9% passing a
stationary vehicle on the off side.
38
Deprivation and Child Pedestrian Road Casualties
Final Report
5 Statistical Modelling
5.1 Overview
5.1.1 Statistical modelling is a process which requires a great deal of care and consideration to
avoid misuse and consequent misinterpretations of the model outputs. In the context of
this study there are three considerations which need to be taken into account.
1. In general terms it is widely accepted as good practice that there is a virtue in the
parsimonious use of independent variables to keep models practical and simple.
2. Where there is strong intercorrelation between independent variables a statistical
model does not need both independent variables in order to explain the dependent
variable. It can happen that one of the independent variables will be eliminated
from the regression and it can often be that it is the wrong variable. This is usually
called "nonsense-elimination". Another problem with intercorrelation is that the
confidence intervals on the regression coefficients will be very wide and may even
include zero, which means you can’t even be confident whether an increase in the
X value is associated with an increase, or a decrease, in Y. Where the confidence
intervals are so wide, excluding a variable (or adding a new one) can change the
coefficients dramatically – and may even change their signs.
3. Identify and manage any interactions there may be between variables. In simple
situations the effect of each independent variable is completely separate from the
other independent variables. Whatever the effects of the different independent
variables are, they just add up or accumulate in a simple way. These are called
additive effects (i.e. there is no interaction). In more complicated situations the
effect of one independent variable depends on another independent variable(s). In
terms of road safety males may take more risks than females but the level of
additional risk may be linked to age or possibly deprivation. This effect is
interaction.
5.1.2 These considerations will be germane as the analysis progresses.
39
Deprivation and Child Pedestrian Road Casualties
Final Report
5.4.4 For each of the 890 SOAs values for the seven domains have been determined and
these have been combined to generate an MDM for each SOA.
5.4.5 The intercorrelations between these domains (across all 890 SOAs) are shown in Table
5.1.
40
Deprivation and Child Pedestrian Road Casualties
Final Report
2
Table 5.1: Intercorrelations (R values) between multiple deprivation measure,
IDAC and constituent domains
Environment
Proximity To
Employment
Education
Services
Income
Health
Living
Crime
IDAC
MDM
MDM domain
MDM -
Income .932 -
Employment .910 .878 -
Health .669 .551 .567 -
Education .796 .693 .666 .468 -
Proximity to Services .105 .100 .109 .271 .086 -
Crime .217 .182 .159 .342 .117 .440 -
Living environment .604 .521 .475 .436 .459 .066 .152 -
IDAC .874 .934 .789 .556 .689 .149 .228 .517 -
2
5.4.6 The R values effectively measure the proportion of variation in one variable explained by
the other. So, for example, the variation in the Income factor explains some 93% of the
variation in the MDM. At the other end of the scale, the variation in the Proximity to
Services factor explains just over 10% of the variation in the MDM. Although all the
values for R2 are highly statistically significant at least at the 0.1% level, many of them do
not explain very large proportions of variation in other variables.
5.4.7 At this point it was decided to drop the Income, Employment, Health, Education, Living
environment domains and IDAC from further analysis and take forward the MDM, with the
Crime and Proximity to Services domains to the detailed modelling stage. It was also
observed that there was a strong relationship between car (and van) ownership levels
and the MDM: accordingly it was decided to eliminate the level of car ownership from the
detailed modelling stage (see Figure 5.1). This was also the case with housing tenure
variables and they too were eliminated from the detailed modelling stage (see Figure 5.2)
41
Deprivation and Child Pedestrian Road Casualties
Final Report
Figure 5.1: Plot of MDM Score with proportion of households with no car
80
Proportion of households with no car(%)
60
40
20
0
0 20 40 60 80
MDM Score
80
60
40
20
0
0 20 40 60 80
MDM Score
5.4.8 At this stage it was decided to keep the composite variables generated above in the
dataset and deal with any intercorrelation effects as the modelling progressed. The
42
Deprivation and Child Pedestrian Road Casualties
Final Report
reason for this was to keep in the dataset a number of variables that reported in the
literature review.
43
Deprivation and Child Pedestrian Road Casualties
Final Report
Model form
5.5.2 If data conform to a Poisson distribution, the key assumption is that mean is equal to the
variance. In the case of the NI data, the distribution of the casualty totals by SOA show
variances in excess of the mean.
5.5.3 The average number of pedestrian casualties for all children (ages 0-15) is 3.6 and the
variance is 15.9. This indicates that the data are overdispersed and that it would be
inappropriate to use a Poisson model to analyse the data. The appropriate model would
be one with a negative binomial error structure.
5.5.4 Negative binomial regression models can be used to model both counts and rates.
Examples of each would be:
The year by year number of collisions at a particular junction
The number of casualties in collisions in a given time period occurring in SOAs
5.5.5 The number of collisions at a junction is purely a count outcome and can be successfully
modelled using standard regression techniques.
5.5.6 But how do the models handle rates? In the current study, the rate is just a count per
head of population. If there is a variation in the population this will affect the number of
casualties observed. Population, here, can be considered as a measure of exposure.
5.5.7 Statistical models can manage exposure variables by allowing users to define such
variables as offset variables. The final model structure adopted for the study was a
negative binomial model with log(population) defined as an offset variable. The R
package provides fully for this model structure.
Contingency table
5.5.8 In this stage a multi-way classification in the form of a contingency table was prepared
where the pedestrian casualty and population data were classified according to the key
characteristics of deprivation index, sex of casualty, age of casualty and URDG
Classification each of which is expressed as a non-numeric factor. The columns of the
contingency table were:
MDM decile (10 classifications/factors)
Sex of casualty (2)
Age of casualty (6)
URDG Classification (2)
The average number of casualties by year of year of age
The average population by year of age
5.5.9 The casualty numbers and population data have been standardised by dividing by the
number of years of age covered by the age group. The means that the modelled
coefficient values for age are relative. If this standardisation had not been undertaken the
coefficient values for age would have biased by the number of years in the age group
particularly in the case of the adult age group.
5.5.10 The table comprised 240 rows (ie 10 x 2 x 6 x 2). Table 5.3 shows the structure or the
data for the first 6 and last 6 rows in the table.
44
Deprivation and Child Pedestrian Road Casualties
Final Report
5.5.11 To demonstrate how the inclusion of variables affects the model and goodness of fit, the
following analysis outlines the effect of adding variables one by one and then looking at
interaction affects.
5.5.12 Tables A 1 – A 6 set out in Appendix 3 show the coefficient estimates for the intercept
value and each factor used in the model. Note that the coefficients for the first level of
each factor are aliased. Each table has six columns:
Column 1 Factor: identifies the factor and factor level (e.g. mdmdec3, indicates
the third decile for MDM decile)
Column 2 Estimate: provides the value for the coefficient relating to the factor
Column 3 Std, Error: is the standard error for the coefficient
Column 4 Z-value: is the estimate divided by the standard error
Column 5 Pr((>|Z|): is the probability of the value of Z occurring by chance - the
lower the number the more significant is the value for the estimate
Column 6 Significance code: is purely a visual presentation of the significance
(the more stars the more significant the estimate value is)
5.5.13 Figures A 1 – A 6 in Appendix 3 show graphically the goodness of fit associated with
these models.
5.5.15 It is evident that deprivation is a highly significant factor with increasing deprivation and
higher pedestrian casualty rates.
5.5.16 Figure A 1 shows the relationship in graphical form where it can be seen that although
there is a strong relationship between MDM decile and pedestrian casualty rates, the
overall fit is poor. This would not be a good model to use for predictive purposes.
45
Deprivation and Child Pedestrian Road Casualties
Final Report
5.5.18 It is evident that both deprivation and age are highly significant factors with:
increasing deprivation predicting higher casualty rates
casualty rates varying by age group
5.5.19 Figure A 2 shows the relationship in graphical form where it can be seen that the fit is
better than it is for MDM decile, there are a number of outliers particularly where the
observed number of casualties is low.
5.5.21 It is evident that the independent variables are highly significant factors with:
increasing deprivation predicting higher casualty rates
casualty rates varying by age group
males giving rise to higher casualty rates than females
5.5.22 Figure A 3 shows the relationship in graphical form where it can be seen that although
sex is a significant factor, the fit is not observably better than it is for mdm decile and age.
There remain a number of outliers particularly where the observed number of casualties
is low.
5.5.24 It is evident that the independent variables are highly significant factors with:
increasing deprivation predicting higher casualty rates
casualty rates varying by age group
males giving rise to higher casualty rates than females
the urban classification giving rise to higher casualty rates than the rural
classification
5.5.25 Figure A 4 shows the relationship in graphical form where it can be seen that the
inclusion of the URDG classification factor improves the fit measurably.
5.5.27 With interaction effects the output it is more difficult to analyse the model output. It is
evident that the independent variables and the interaction between MDM decile and
URDG classification are highly significant factors with:
increasing deprivation predicting higher casualty rates
casualty rates varying by age group
males giving rise to higher casualty rates than females
46
Deprivation and Child Pedestrian Road Casualties
Final Report
the urban classification giving rise to higher casualty rates than the rural
classification
the effect of the URDG classification is uneven with rural casualty rates for MDM
deciles 9 and 10 being less than for deciles 2 to 8. The high estimate value for the
MDM decile 10/urban interaction suggests that casualty rates for this category are
significantly higher than for other DMD decile values.
5.5.28 This model needs to be treated with a degree of caution because the populations in rural
areas with MDM deciles 9 and 10 particularly in the lower age groups are low. The
modelled outputs may reflect this with some deceptive or unexpected estimates.
5.5.29 Table 5.4 shows the combined estimates for male child pedestrian casualties (by age
group) in urban areas where it can be seen that the combination of the estimates relating
to the MDM decile*URDG interaction begin to make sense with the combined estimates
generally rising with increasing deprivation.
5.5.30 Figure A 5 shows the relationship in graphical form where it can be seen that the level of
fit is not observably improved by the introduction of the interaction term.
Table 5.4: Combined estimates for child pedestrian casualties (age 0-15)
Age group
MDM decile 0-4 5-7 8 - 11 12 - 15
1 0.40 1.21 1.27 1.40
2 0.45 1.26 1.32 1.45
3 0.68 1.49 1.54 1.68
4 0.66 1.47 1.53 1.66
5 0.49 1.30 1.36 1.49
6 1.00 1.81 1.87 2.00
7 1.01 1.83 1.88 2.02
8 1.27 2.08 2.14 2.27
9 1.68 2.49 2.55 2.68
10 2.33 3.14 3.20 3.33
5.5.32 With interaction effects the output it is more difficult to analyse the model output. It is
evident that the independent variables and the interaction between MDM decile and
URDG classification are highly significant factors with:
increasing deprivation predicting higher casualty rates
casualty rates varying by age group
males giving rise to higher casualty rates than females
the urban classification giving rise to higher casualty rates than the rural
classification
the effect of the URDG classification is uneven is discussed above
the effect of the age/sex interaction shows (because of the negative and significant
values for the interaction terms ageA2-6sexM) that the male casualty rates for
47
Deprivation and Child Pedestrian Road Casualties
Final Report
ageA1 (ages 0-4) is higher than for the other age groups. The standard errors or
the interaction terms show that there are no significant differences between the
estimates calculated.
5.5.33 Figure A 6 shows the relationship in graphical form where it can be seen that the level of
fit is not observably improved by the introduction of the new interaction term.
5.5.34 This Level 1 model analysis demonstrates that there are clear and highly significant
relationships between casualty rate and deprivation, sex of casualty, age of casualty and
the URDG classification of the collision location. This is a key finding in the context of
this study as it leads on to the more detailed Stage 2 modelling.
Overview
5.6.2 In the high level (Level 1) model it was established that there are clear and highly
significant relationships between casualty rate and deprivation, sex of casualty, age of
casualty and the URDG classification of the collision location. In this more detailed part
of the analysis datasets for each casualty age group, casualty sex and
urban/classification were created with each set being the subject of a similarly designed
model. In this way it is possible to see how the independent variables representing
environmental and other factors behave in the models across, age, sex and URDG
classification stratifications.
Initial analysis
5.6.3 In Stage 1 it was established that there are clear and highly significant relationships
between casualty rate and deprivation, sex of casualty, age of casualty and the URDG
classification of the collision location. In this more detailed part of the analysis datasets
for each casualty age group, casualty sex and urban/classification were created with
each set being the subject of a similarly designed model. In this way it is possible to see
how the independent variables behave in the models across, age, sex and URDG
classification stratifications.
5.6.4 The variables relating to school places and free school meals were made relevant to the
age group being modelled. When younger children walk to school they will generally be
accompanied by a responsible adult and may also be in the company of younger siblings;
accordingly, these adults and the younger children will be exposed to risk as pedestrians.
Following an initial analysis it was decided to use values for all school places and
proportion of all pupils entitled to free school meals as variables for the pre-school age
group (ages 0-4) and for the young adult and adult age groups. The variables for school
places and free school meals entitlement included in the modelling are set out in Table
5.5.
48
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 5.5: Descriptions of School Places and Free School Meals variables
5.6.5 In the initial analysis, all variables set out in Table 5.2 were tested in the model. Many of
these variables were found either not to be significant in explaining casualty rates or were
found to not as effective in explaining casualty rates as other similar variables. Table 5.6
summarises the outcome of this initial analysis
49
Deprivation and Child Pedestrian Road Casualties
Final Report
5.6.6 The significance of the coefficients are set out in Table A 7 and Table A 8 for urban and
rural areas respectively. Also shown is the sign associated with the coefficient. A plus
sign (“+”) indicates a positive correlation and a minus sing (“-“) indicates an inverse
correlation for the variable with casualty rates. In a positive correlation, as one variable
gets larger the other gets larger. In an inverse correlation, as one gets larger, the other
gets smaller.
50
Deprivation and Child Pedestrian Road Casualties
Final Report
5.6.7 It should be noted that because of the smaller datasets for rural casualties a number of
the models did properly converge and the software indicated a warning message and
these are identified in the right hand column of Table A 8.
51
Deprivation and Child Pedestrian Road Casualties
Final Report
road length is found to be a significant correlated variable for half the models
examined – for most of these the correlation is inverse- the greater the road length,
the lower the casualty rate (in most instances)
the traffic proxy variable which provides a relative measure of travel activity in the
SOA is generally a highly significant positively correlated variable across all age
groups and sexes – the more activity there is in the SOA, the greater the casualty
rate
the number of school places is a highly significant positively correlated variable for
about half of the models – the more school places, the greater the casualty rate
the proportion of free school meals is seen to be a significant factor for about half
the models. Where is it significant the relationship is positive - the higher the
proportion of pupils with free school meals, the greater the casualty rate
sunshine hours is seen to be a highly significant generally positively correlated
variable in the models for most child age groups – the greater the number of
sunshine hours, the higher the casualty rate
like sunshine hours, annual rainfall is seen to be a highly significant generally
positively correlated variable in the models for most child age groups – the greater
rainfall, the higher the casualty rate
5.6.10 There are some interesting differences between the rural and urban models.
the population density is generally positively correlated with casualty rates for
urban areas and inversely for rural areas.
the length of road is generally positively correlated with casualty rates for urban
areas and inversely for rural areas.
sunshine hours is generally inversely correlated with casualty rates for urban areas
and positively for rural areas.
annual rainfall is generally inversely correlated with casualty rates for urban areas
and positively for rural areas.
Discussion
5.6.11 For the MDM score and the MDM Crime domain the relationships show positive
correlations and this is as intuition would suggest that the more deprived an area is the
higher the casualty rate.
5.6.12 The MDM Proximity to Services domain shows inverse correlations which may seem
counter-intuitive at first. The domain, however, is scored to give higher values to areas
which are further away from services and lower values to areas closer to services.
Higher values are, therefore, more deprived for this domain. This may suggest that the
closer the population is to services the more likely there will be pedestrian trips and
consequent increased exposure. Where services are located further away, fewer walking
trips may be made and there is less exposure with a consequent lower casualty rate.
5.6.13 Population density is generally positively correlated with casualty rates in urban areas
and mostly inversely correlated in rural areas. The reason for this may rely on the strong
relationship between population density and road length which is set out clearly in Figure
5.3 where it can be seen that higher population densities are associated with lower road
length and vice versa. This strong intercorrelation indicates why there may be some
confusing model outputs regarding these two variables and suggests that only one of
them need be in the model. Given that population density includes population as its
numerator and that road length is a wholly independent variable, it is road length that
should survive with population density being eliminated.
52
Deprivation and Child Pedestrian Road Casualties
Final Report
100
50
0
0 50 100 150
5.6.14 Total road length is generally positively correlated with casualty rates in urban areas and
generally inversely correlated in rural areas. There may be a parallel here with the MDM
Proximity to Services domain where in rural areas road length and distance to services
may be correlated. Indeed, this is borne out by a plot of road length against MDM
Proximity to Services which is shown in Figure 5.4. Road length itself not a measure of
distance required to be travelled whereas the MDM Proximity to Services domain is
certainly a social measure relating to distances that may need to be travelled as part of
day-to-day living. Accordingly road length was eliminated from the model in preference to
MDM Proximity to Services.
53
Deprivation and Child Pedestrian Road Casualties
Final Report
1
0
-1
-2
0 50 100 150
5.6.15 Traffic proxy is highly significant and positively correlated with casualty rates. This result
follows intuition. Traffic proxy is a relative measure of the amount of activity in an area
both in terms of vehicular and pedestrian movements. High values for traffic proxy
suggest greater exposure to risk for pedestrians and consequent higher casualty rates:
this is borne out by the model outputs.
5.6.16 The number of school places in an area is highly significant and positively correlated with
casualty rates in urban areas. The models for rural areas show less conclusive outputs
but they do show some highly significant and positively correlated coefficients. This
result is in keeping with intuition. For the avoidance of misinterpretation of this result, it is
to be noted that the values for the variables used to measure school places relate to the
number of school places at schools in an SOA (not to the number of school children
resident in an SOA. Therefore, the relationship is most likely to be the result of collisions
around schools occurring on journeys to/from school. The relationships hold good for
models including children and adults. The fact that the relationships are much stronger
and more consistent in urban areas may be explained by the higher proportion of children
in rural areas who will be entitled to (and will use) free school transport. This would
reduce exposure as pedestrians on journeys to/from school for the pupils, parents and
younger siblings alike.
5.6.17 The proportion of children with free school meals does not demonstrate a consistent
pattern and it would be unwise to place emphasis on the model outputs for this variable.
Some coefficients are significant but show no definable pattern. Some of the correlations
are positive and others inverse. The reason for this may lie in the fact that there is
correlation between the proportion of free school meals and the MDM score and this is
likely to give rise to unreliable coefficient estimates. Figure 5.5 shows a plot of the
proportion of all pupils entitled to free school meals against MDM Score which shows a
clear correlation between the two variables. Note that most SOAs do not have schools
54
Deprivation and Child Pedestrian Road Casualties
Final Report
within their boundaries and that it these SOAs which form the vertical column of data at
point “0” on the x-axis of the chart.
Figure 5.5: Plot of proportion of all pupils entitled to free school meals against
MDM Score
40
20
0
0 20 40 60
5.6.18 Both sunshine hours and annual rainfall are generally positively correlated with casualty
rates in rural areas and inversely correlated in urban areas. The exception to this relates
to the young adult age group (16-24) where in urban areas both sunshine hours and
annual rainfall are positively correlated with casualty rates. A possible reason for these
differences is discussed below.
55
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 5.7: Local Government Districts which impact on casualty rates (Level 2
model)
5.6.20 It is interesting to note that sunshine hours and rainfall are both correlated with Local
Government District. This being the case makes it difficult to judge which variable is
dominant. If Local Government District is the dominant factor it may explain why the
model outputs relating to sunshine hours and annual rainfall show differences between
rural and urban areas. It is difficult within the scope of this study to measure objectively
the differences between Local Government Districts which may impact on pedestrian
casualty rates; examples of such differences may relate to:
the level of pedestrianisation in the district
the proportion of children in the district entitled to and using free school transport
local policies and design standards relating to facilities for pedestrians
implementation of urban design standards at a district level which minimise
pedestrian/vehicle conflict
5.6.21 In the absence of a clear reason to believe that Local Government District is a dominant
factor it was decided to retain sunshine hours and annual rainfall in the model.
56
Deprivation and Child Pedestrian Road Casualties
Final Report
5.7.3 Table A 9 and Table A 10 set out the significance and signs for the coefficients for urban
and rural areas respectively.
57
Deprivation and Child Pedestrian Road Casualties
Final Report
5.7.6 There remain differences between the rural and urban models.
Sunshine hours is generally inversely correlated with casualty rates for urban areas
and positively for rural areas.
Annual rainfall is generally inversely correlated with casualty rates for urban areas
and positively for rural areas.
For young adults in urban areas are an exception where sunshine hours and
annual rainfall are positively correlated with casualty rates
5.7.7 The reasons for these differences relating to sunshine hours and annual rainfall are not
clear and were discussed earlier in paragraph 5.6.20.
5.7.9 For the combined urban and rural models it is observed that:
The MDM score is a consistent and highly significant positively correlated variable
across all age groups and sexes – the more deprived the SOA, the greater the
casualty rate
The MDM Crime domain score is a consistent and highly significant positively
correlated variable across all age groups and sexes – the more crime related to the
SOA the greater the casualty rate
The MDM Proximity to Services domain score is a consistent and highly significant
inversely correlated variable across all age groups and sexes – the more deprived
the SOA is in terms of proximity to services related to the SOA the greater the
casualty rate.
The traffic proxy variable which provides a relative measure of travel activity in the
SOA is a consistent and highly significant positively correlated variable across all
age groups and sexes – the more activity there is in the SOA, the greater the
casualty rate
The number of school places is a consistent and highly significant positively
correlated variable for about half of the models – the more school places, the
greater the casualty rate
Sunshine hours is seen to be a highly significant generally positively correlated
variable in the models for most child age groups – the greater the number of
sunshine hours, the higher the casualty rate
Like sunshine hours, annual rainfall is seen to be a highly significant generally
positively correlated variable in the models for most child age groups – the greater
rainfall, the higher the casualty rate
URDG classification is generally positively correlated with child casualty rates and
generally inversely correlated with adult casualty rates
58
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 5.8: Local Government Districts which impact on casualty rates (Level 3
model)
Impact on Level of
Age group Local Government District casualty rates significance
Children Ballymena Positive 0.05
(0 – 15) Belfast Positive 0.05
Carrickfergus Positive 0.001
Castlereagh Positive 0.05
Derry Positive 0.05
Larne Positive 0.001
Limavady Positive 0.01
Lisburn Positive 0.01
Magherafelt Inverse 0.05
North Down Inverse 0.001
Newry and Mourne Inverse 0.05
Newtownabbey Positive 0.05
Young Adults Antrim Inverse 0.001
(16 – 24) Ards Inverse 0.01
Craigavon Positive 0.001
Dungannon Positive 0.001
Magherafelt Positive 0.05
North Down Inverse 0.05
Adults Derry Inverse 0.001
(25 and over) Dungannon Positive 0.01
Moyle Positive 0.001
Strabane Inverse 0.05
5.7.11 Figure 5.6 shows a Level 3 model plot for all child pedestrian casualties (age 0 -15).
Whilst there is a clear correlation between the modelled and observed values there is a
quite a lot of statistical “white noise”. The “white noise” was not so marked with the Level
1 model which is an entirely expected outcome. The Level 1 model is an aggregation of
data and the individual cell values in the contingency table represent average values
which will to a large extent have removed a significant proportion of variation which will
give rise to the “white noise”.
59
Deprivation and Child Pedestrian Road Casualties
Final Report
Figure 5.6: Observed and modelled casualty totals for all child pedestrian
casualties (ages 0 – 15) (Level 3 model)
25
20
15
10
5
0
0 5 10 15 20 25 30 35
Modelled value
(Level 3 model)
5.7.13 Table A 12 in Appendix 3 tabulates the 50 poorest performing SOAs and Table A 13
tabulates the 50 best performing SOAs. A review of the poorest performing SOAs may
shed some light on the reasons why particular they have been listed and may assist in
developing remedial action. On the other hand, a review of the best performing SOAs
may highlight some examples of good practice which could be transferred to other areas
of Northern Ireland. Care, however, should be exercised in undertaking this work as the
data relate to casualty records over a ten year period and there may have been
significant demographic and infrastructure changes over this period.
5.7.14 Figure 4.1 in Appendix 4 maps the distribution of performance (as deciles) across
Northern Ireland. This map can be compared with the equivalent map for MDM decile.
The comparison shows that whilst deprivation tends to be clustered, the performance
shows less clustering. This could be investigated further using the processes outlined in
the previous two paragraphs.
60
Deprivation and Child Pedestrian Road Casualties
Final Report
5.7.16 The ratio of the two values (second five year period : first five year period) is 0.590.
Therefore, an estimate of the number of casualties expected in an SOA for the second
five year period (2004 - 2008) would be the number of casualties for the preceding five
years (1999 – 2003) times 0.590.
5.7.17 These estimated casualty totals can be compared with the observed casualty totals to
determine the change in the number of casualties. Here the value is:
5.7.18 Where the number is positive this represents an increase in casualty totals. Conversely,
where the number is negative this represents a decrease in casualty totals.
5.7.19 Table A 14 in Appendix 3 tabulates the 50 SOAs with the greatest increases in casualty
totals and Table A 14 tabulates the 50 SOAs with the largest decreases. A review of
these tabulations may be helpful in identifying trends over the past ten years which may
have been occasioned by changes in demography, traffic patterns, school locations or
specific road safety interventions.
61
Deprivation and Child Pedestrian Road Casualties
Final Report
6 Conclusions
6.1.2 Overall children resident in SOAs comprising the most deprived 10% of SOAs are 4.78
times more likely to be injured as a pedestrian in a road collision as children resident in
the least deprived 10% of SOAs. This value is defined as the deprivation disparity ratio
(DDR). This overall figure of 4.78 masks different risks for children of different ages sexes
which are shown in Table 9 and reproduced below in Table 6.1
Table 6.1: Deprivation disparity ratio of child pedestrian casualty rates1 by sex,
Multiple Deprivation Measure (MDM) decile and age group (1999 –
13
2008)
6.1.3 This finding fulfils one of the key objectives of the study. It is clear from this table that
there a clear differences in the DDR values by age and sex. For example:
Males in the 5-7 age group are 5.7 more likely than males in the 12-15 age group
to be injured as a pedestrian
Males in the 8-11 age group are 1.4 times more likely than females in the same
group to be injured as a pedestrian
Females in the 12-15 age group are 1.6 times more likely than males in the same
age group to be injured as a pedestrian
6.2.2 The statistical modelling carried out has shown that key variables relating to how busy an
area is, proximity to services, crime and number of school places are linked to the child
pedestrian casualty rates. Other variables - weather conditions, URDG classification and
local government district are also linked but less robustly to casualty rates.
6.2.3 The reliability of the models appears to be good with very high levels of consistency
across the age groups and sexes for the key variables with moderate consistency for the
other variables.
13
For notes on the data see Table 9.
62
Deprivation and Child Pedestrian Road Casualties
Final Report
6.3.2 The key finding is that there is a clear link between deprivation and child pedestrian
casualty rates in Northern Ireland. From a road safety planning point of view this would
strongly suggest that road safety initiatives should target or at least be prioritised in part
on areas of deprivation
6.3.3 The number of school places in an SOA (as a proxy for the number of children travelling
to and from a school in that area as opposed to the number of children living in the area)
is a consistent and highly significant variable. This suggests that the presence of schools
adds significant additional risk for children. This perhaps not surprising as most children
will attend school just under 200 days a year and many will travel as pedestrians either
alone or accompanied by an adult. The model, however, shows quantitatively that there
is a highly significant additional risk. This finding would support interventions which make
travel to school safer including through improved infrastructure and possible pedestrian
skills training. This effect is additional to impact deprivation has on casualty rates and
suggests that priorities for interventions of this type be given to deprived areas. Part time
20 mph zones may also be an effective counter measure.
6.3.4 The traffic proxy variable has also been found to be a consistent and highly significant
variable. This, too, is perhaps not a surprising conclusion but the modelling process has
quantitatively shown a strong and robust link between traffic and child pedestrian casualty
rates. This suggests that busier areas should be a focus for an intervention strategy
perhaps involving speed reduction strategies.
6.3.5 The model can be used to identify areas of concern (child pedestrian casualty blackspots)
which might respond to site specific intervention. The production of a list of areas of
concern for investigation could prove helpful in implementing low cost remedial
measures. Table A 12 in Appendix 3 sets out an indicative list based on areas with more
casualties than expected based on the model outputs. This list is indicative and care
needs to exercised to avoid an “over-interpretation” of the data (see paragraph 5.7.13).
6.3.6 The models could also be used to identify good practice by preparing a list of areas
where there are fewer casualties than predicted by the model outputs. These areas
could be the subject of a review to determine if there are any underlying reasons why
casualty rates are lower than expected. Table A 13 in Appendix 3 sets out an indicative
list of such areas.
6.3.7 Child pedestrian casualty rates are modelled to be higher in urban areas than in rural
areas. Whilst the traffic proxy variable is a measure of traffic in an area it is not a
measure of the traffic density on roads which will be higher in urban than in rural areas.
Also in urban areas it is likely that individual walking trips will involve more street
crossings than in rural areas. Consequently, some priority should be given to
interventions relating to urban areas.
6.3.8 Finally, proximity to services is seen to be linked to casualty rates where the closer
children live to services the higher the casualty rates. This is perhaps more widely
related to planning issues and suggests that road safety ought to be a key consideration
in the design of regeneration and major development schemes.
63
Deprivation and Child Pedestrian Road Casualties
Final Report
64
Deprivation and Child Pedestrian Road Casualties
Final Report
Appendix 1
1
Deprivation and Child Pedestrian Road Casualties
Final Report
References
Abdalla IM, Raeside R, Barker D and McGuigan DRD. An investigation into the relationships between
area social characteristics and road accident casualties. Accident Analysis and Prevention Vol 29,
pp583-593, 1997.
http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6V5S-3SX1G85-
4&_user=10&_rdoc=1&_fmt=&_orig=search&_sort=d&_docanchor=&view=c&_acct=C000050221&_ve
rsion=1&_urlVersion=0&_userid=10&md5=15f57e30af5f3942421da81c871a28d2
Chichester BM, Gregan JA, Anderson DP and Kerr JM. Associations between road traffic accidents
and socio-economic deprivation in Scotland’s west coast. Scottish Medical Journal Vol 43 pp135-
1398, 1998.
Not available on the web.
Childhood accidents update: Promoting young people’s health using the Health Behaviour in School-
aged Children (HBSC) study and the Health Development Agency’s Evidence Base. NHS Health
Development Agency, undated.
http://www.wiredforhealth.gov.uk/Word/HBSCBriefingAccidentalInjury.doc.
Clarke DD, Ward P, Truman W and Bartle C (School of Psychology University of Nottingham). A poor
way to die: social deprivation and road traffic fatalities. Department for Transport 2008.
http://www.dft.gov.uk/pgr/roadsafety/research/rsrr/theme5/poorwaytodie.pdf.
Cran GW. Social deprivation and childhood injuries in North and West Belfast. The Ulster Medical
Journal, Volume 74 No. 1, pp. 22-28, May 2005.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2475484/pdf/ulstermedj00001-0025.pdf
Department for Transport (DfT). Childhood accidents and their relationship with problem behaviour
(No.07). DfT, July 1999.
http://www.dft.gov.uk/pgr/roadsafety/research/rsrr/theme1/childhoodaccidentsandtheirre4729?page=5
Edwards P, Green J, Lachowycz K, Grundy C and Roberts I. Serious Injuries in children: variation by
area deprivation and settlement type. Archive of Disease in Childhood Vol 93 pp485-489 2008.
http://www.ncbi.nlm.nih.gov/pubmed/18381345?dopt=AbstractPlus
Edwards P, Green J, Roberts I, Grundy C and Lachowycz K. Deprivation and Road Safety in London:
A report to the London Road Safety Unit. London: London School of Hygiene and Tropical Medicine,
2006.
http://www.tfl.gov.uk/assets/downloads/deprivation-and-road-safety.pdf
Graham D, Glaister S and Anderson R. The effects of deprivation on the incidence of child and adult
pedestrian casualties in England. Accident Analysis and Prevention Vol 37, pp125-135, 2005.
http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6V5S-4DKKDWP-
1&_user=10&_coverDate=01%2F31%2F2005&_alid=1018631002&_rdoc=1&_fmt=high&_orig=search
&_cdi=5794&_sort=r&_docanchor=&view=c&_ct=1&_acct=C000050221&_version=1&_urlVersion=0&
_userid=10&md5=e5afa54340109afb8f405df78f60134f
2
Deprivation and Child Pedestrian Road Casualties
Final Report
Graham DJ and Stephens DA. Decomposing the impact of deprivation on child pedestrian casualties
in England. Accident Analysis & Prevention Vol 40, Issue 4, pp1351-1364, 2008.
http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6V5S-4S26HCC-
1&_user=10&_coverDate=07%2F31%2F2008&_alid=1018627151&_rdoc=3&_fmt=high&_orig=search
&_cdi=5794&_sort=r&_docanchor=&view=c&_ct=59&_acct=C000050221&_version=1&_urlVersion=0
&_userid=10&md5=3ff2102fff5d9c73a950258f644af710
Grayling T, Hallam K, Graham D, Anderson R and Glaister S. Streets ahead safe and liveable streets
for children. Institute for Public Policy Research, 2002.
http://www.ippr.org.uk/members/download.asp?f=%2Fecomm%2Ffiles%2Fstreets%5Fahead%2Epdf
Kendrick D. Prevention of pedestrian accidents. Archives of Disease in Childhood 1993; 68: 669-672
http://adc.bmj.com/cgi/reprint/68/5/669
McGuigan D, Bull M and Gorman D. Connecting social deprivation and involvement in road accidents:
results and implications. 24th PTRC European Transport Forum. Brunel University, England.
Proceedings of Seminar H. Traffic management and road safety, 2-6 Sept 1996. PTRC Education and
Research Services Ltd.
http://www.etcproceedings.org/paper/connecting-social-deprivation-and-involvement-in-road-
accidents-results-and-im
McGuigan D, Fraser D and McBride A. Road accidents and social deprivation. Scottish Transport
Applications and Research (STAR) Conference, Glasgow, 2006.
Not available on the web
National Public Health Service for Wales (NPHS). Deprivation and Health - A report by the National
Public Health Service for Wales. NPHS, 2004.
http://www2.nphs.wales.nhs.uk:8080/hiatdocs.nsf/1f8687d8da97650980256fa30051b0be/83086791e1
fb86538025726f0043322c/$FILE/Deprivationreport10Dec04.pdf
Noland RB and Quddus MA. A spatially disaggregate analysis of road casualties in England. Accident
Analysis and Prevention Vol 36 pp973-984, 2004.
http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6V5S-4BM6B16-
1&_user=10&_coverDate=11%2F30%2F2004&_alid=1018618644&_rdoc=8&_fmt=high&_orig=search
&_cdi=5794&_sort=r&_docanchor=&view=c&_ct=8&_acct=C000050221&_version=1&_urlVersion=0&
_userid=10&md5=f342bc500a40767afb4f8ff554f9224e
Northern Ireland Statistics & Research Agency (NISRA). Northern Ireland Multiple Deprivation
Measure 2005. NISRA, May 2005.
http://www.nisra.gov.uk/deprivation/archive/NIMDM2005FullReport.pdf
3
Deprivation and Child Pedestrian Road Casualties
Final Report
Petch RO and Henson RR. Child road safety in the urban environment. Journal of Transport
Geography Vol 8 pp197-211, 2000.
http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6VG8-40JFYM1-
4&_user=10&_coverDate=09%2F30%2F2000&_alid=1018625738&_rdoc=1&_fmt=high&_orig=search
&_cdi=6032&_sort=r&_docanchor=&view=c&_ct=1&_acct=C000050221&_version=1&_urlVersion=0&
_userid=10&md5=37240e1564772964d94ff6a8a574a96b
Road Safety Scotland (RSS). Road Safety Early Years Resource Consultation: Analysis of
Responses. RSS, April 2009.
http://www.road-safety.org.uk/Images/EarlyYearsAnalysisFinalReport,ODS,April2009_tcm4-
549089.pdf
SGLGRAS. Paper LGRAS(00)6: Driver and casualty postcodes, and their use in a research project on
tourist road accidents in rural Scotland. Scottish Government Liaison Group on Road Accident
Statistics (SGLGRAS) 27th June 2000.
http://cci.scot.nhs.uk/Topics/Statistics/scotstat/transport/roadaccidentstats/paper-lgras-00-6
Silversides JA, Gibson A, Glasgow JFT, Mercer R and Cran GW. Social deprivation and childhood
injuries in North and West Belfast. The Ulster Medical Journal, Volume 74 No. 1, pp. 22-28, May 2005.
http://www.pubmedcentral.nih.gov/picrender.fcgi?artid=2475484&blobtype=pdf
Whelan K, Towner E, Errington G and Powell J (Centre for Child and Adolescent Health, University of
the West of England, Bristol). Evaluation of the National Child Pedestrian Training Pilot Projects. Road
Safety Research Report No. 82. Department for Transport., 2008.
http://www.dft.gov.uk/pgr/roadsafety/research/rsrr/theme1/childpedestrianprojects/kerbcraftevaluation
main.pdf
4
Deprivation and Child Pedestrian Road Casualties
Final Report
Bibliography
Abdalla IM, Barker D and Raeside R. Road accident characteristics and socio-economic deprivation.
Traffic Engineering and Control Vol 38 pp672-676, December 1997.
http://www.tecmagazine.com/index.cfm?fuseaction=magazine.Article&ArticleID=327
Cattan M, Green J, Newell C, Ayrton R and Walker J. Child–Parent Interaction in Relation to Road
Safety Education: Part 1 – A Critical Literature Review. Road Safety Research Report 101,
Department for Transport, December 2008.
http://www.dft.gov.uk/pgr/roadsafety/research/rsrr/theme1/no101roadsafetyreport.pdf
Dowswell T and Towner E. Social deprivation and the prevention of unintentional injury in childhood: a
systematic review. Health Education Research, Vol. 17, No. 2, 221-237, April 2002.
http://her.oxfordjournals.org/cgi/content/full/17/2/221?maxtoshow=&HITS=10&hits=10&RESULTFOR
MAT=1&author1=dowswell&author2=towner&andorexacttitle=and&andorexacttitleabs=and&andorexa
ctfulltext=and&searchid=1&FIRSTINDEX=0&sortspec=relevance&resourcetype=HWCIT
Haynes R, Jones A, Harvey I, Tony Jewell T and Lea D. Geographical distribution of road traffic
deaths in England and Wales: place of accident compared with place of residence. Journal of Public
Health VoI. 27, No. 1, pp. 107–111, February 2005.
http://jpubhealth.oxfordjournals.org/cgi/reprint/27/1/107.pdf
Hewson PJ. Epidemiology of child pedestrian casualty rates: Can we assume spatial independence?
Accident Analysis & Prevention, Vol 37, Issue 4, Pages 651-659, July 2005.
http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6V5S-4FWKMDM-
3&_user=10&_coverDate=07%2F31%2F2005&_alid=1018621186&_rdoc=2&_fmt=high&_orig=search
&_cdi=5794&_sort=r&_docanchor=&view=c&_ct=2&_acct=C000050221&_version=1&_urlVersion=0&
_userid=10&md5=a8f759fe2931f2dc77b7a7a856497973
Sanjay Rana. Investigation of geographical dependence of road traffic casualties in highly deprived
areas of Greater Manchester. Neighbourhood Road Safety Initiative (NRSI) 2006.
http://eprints.ucl.ac.uk/1731/1/CTS_Talk_Presentation.pdf
ODPM. Making the connections: final report on transport and social exclusion. ODPM Social
Exclusion Unit, 2003.
http://www.cabinetoffice.gov.uk/media/cabinetoffice/social_exclusion_task_force/assets/publications_1
997_to_2006/making_transport_2003.pdf
White D and Raeside R. The socio-economic influences on the risk of child involvement in road traffic
accidents. The Royal Society for the Prevention of Accidents Road Safety Congress 12, 14 March
2001.
http://www.rospa.org/roadsafety/conferences/congress2001/proceedings/white.pdf
White D, Raeside R and Barker D (Napier University). Road accidents and children living in
disadvantaged areas: a literature review. The Scottish Executive Central Research Unit, 2000.
http://www.scotland.gov.uk/Resource/Doc/156570/0042052.pdf
5
Deprivation and Child Pedestrian Road Casualties
Final Report
Appendix 2
1
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 1: Casualties by casualty class and age group (January 1999 to December 2008)
Age group
14
Casualty Class 0 0-4 5-7 8-11 12-15 16 and over All ages
1. Driver 20 10 60 67 145 55,652 55,954
2. Pillion passenger 1 10 38 210 259
3. Vehicle passenger - Front 25 263 257 625 1,128 19,449 21,,747
4. Vehicle passenger - Rear 35 1,506 1,198 1641 1,894 10,699 16973
5. Pedestrian 11 499 672 960 1,104 5,583 8,829
6. Motorcyclist 1 9 44 3,871 3,925
7. Pedal cyclist 25 150 264 262 997 1,698
All casualty classes 92 2,304 2,337 3,576 4,615 96,461 109,385
Casualties
Age Group Male Female Both sexes
0-4 Pre-school 344 155 499
5-7 Infant (lower primary) 407 265 672
8 - 11 Upper primary 590 370 960
12 - 15 Secondary 638 466 1,104
All age groups 1979 1256 3,235
Table 3: Casualties by casualty class and age group (April 2007 to March 2009)
Age group
Casualty Class 0 0-4 5-7 8-11 12-15 16 and over All ages
1. Driver 15 1 16 9,920 9,952
2. Pillion passenger 7 38 45
3. Vehicle passenger - Front 20 24 36 98 175 3,155 3,508
4. Vehicle passenger - Rear 29 266 170 273 299 1,641 2,678
5. Pedestrian 11 87 71 138 180 1,187 1674
6. Motorcyclist 1 7 905 913
7. Pedal cyclist 1 5 20 57 58 308 449
All casualty classes 77 382 297 567 742 1,7154 19,219
Table 4: Child pedestrian casualties by age group (April 2007 to March 2009)
Casualties
Age Group Male Female Both sexes
0-4 Pre-school 54 33 87
5-7 Infant (lower primary 44 27 71
8 - 11 Upper primary 78 60 138
12 - 15 Secondary 98 82 180
All age groups 274 202 476
14
Age “0” is reserved by the PSNI for casualties of unknown age. Casualties of 0 (ie aged under 1 year) are
coded as “1”.
2
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 5: Child pedestrian casualty numbers by sex, Multiple Deprivation Measure (MDM)
decile and age group (1999 - 2008)
3
Deprivation and Child Pedestrian Road Casualties
Final Report
1
Table 6: Child population by sex, Multiple Deprivation Measure (MDM) decile and age
group (1999 - 2008)
4
Deprivation and Child Pedestrian Road Casualties
Final Report
1
Table 7: Child pedestrian casualty rates by sex, Multiple Deprivation Measure (MDM) decile
and age group (1999 - 2008)
5
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 8: Spearman’s ranked correlation coefficients and significance levels for the
relationship between deprivation decile and child pedestrian casualty rates by sex
and age group (1999 - 2008)
1
The ρ values corresponding to probability levels are:
Probability 0.05 (5%) 0.025 (2.5%) 0.01 (1%) 0.005 (0.5%)
ρ .564 .648 .745 .794
This means, example, where the significance level is 0.5% there is less than 1 in 200 probability that
the relationship could have occurred by chance.
1
Table 9: Deprivation disparity ratio of child pedestrian casualty rates by sex, Multiple
Deprivation Measure (MDM) decile and age group (1999 – 2008)
6
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 10: Child and adult pedestrian casualty numbers by sex, Multiple Deprivation Measure
(MDM) decile and age group (1999 - 2008)
7
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 11: Child and adult population by sex, Multiple Deprivation Measure (MDM) decile and
age group (1999 - 2008)
8
Deprivation and Child Pedestrian Road Casualties
Final Report
1
Table 12: Child and adult pedestrian casualty rates by sex, Multiple Deprivation Measure
(MDM) decile and age group (1999 - 2008)
9
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 13: Spearman’s ranked correlation coefficients and significance levels for the
relationship between deprivation decile and child and adult pedestrian casualty
rates by sex and age group (1999 - 2008)
1
The ρ values corresponding to probability levels are:
Probability 0.05 (5%) 0.025 (2.5%) 0.01 (1%) 0.005 (0.5%)
ρ .564 .648 .745 .794
This means, example, where the significance level is 0.5% there is less than 1 in 200 probability that
the relationship could have occurred by chance.
1
Table 14: Deprivation disparity ratio of child and adult pedestrian casualty rates by sex,
Multiple Deprivation Measure (MDM) decile and age group (1999 – 2008)
10
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 15: Number of child pedestrian casualties by location of collision (April 2007 – March
2009
Table 16: Distance between collision location and home address for child pedestrian
casualties in Salford and Northern Ireland
11
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 17: Decile of residence by decile of location of collision for child pedestrian casualties (April 2007 to March 2009)
Number of
child Home +
pedestrian Number of Number of
Decile of casualties away deciles away deciles
residence with known Decile of location of away collision within two within two
(home home Home Away deciles of deciles of
decile) postcode Casualties Casualties 1 2 3 4 5 6 7 8 9 10 home decile home decile
1 23 4 19 7 2 2 2 1 1 1 2 1 11 15
2 23 12 11 1 2 1 1 2 1 2 1 4 16
3 37 14 23 3 1 3 1 1 5 4 2 3 8 22
4 30 9 21 1 1 3 1 6 1 2 2 3 1 12 21
5 39 21 18 2 1 5 4 1 1 1 3 11 32
6 42 20 22 2 1 2 4 3 2 1 4 3 12 32
7 40 19 21 1 3 1 6 2 1 2 4 1 15 34
8 46 29 17 2 1 3 4 1 3 1 2 11 40
9 59 27 32 2 2 1 2 2 3 3 7 3 7 20 47
10 109 51 58 3 3 3 12 37 52 103
All deciles 448 206 242 19 11 17 15 23 21 20 26 35 55 156 362
12
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 18: Child pedestrian casualties by month and time of day (weekdays)
13
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 19: Child pedestrian casualties by month and time of day (weekends)
14
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 20: Child pedestrian casualties by speed limit and age group for males and females
(1999 - 2008)
Speed limit
All
Urban rural Age speed
1
classification group <30 30 40 50/55 60 70 limits
A: Males
number
0-4 0 34 1 0 8 0 43
5-7 0 41 3 0 8 0 52
Rural
8 - 11 0 53 4 2 28 0 87
12 - 15 0 63 5 1 55 0 124
0-4 0 294 5 1 1 0 301
5-7 0 348 5 1 1 0 355
Urban
8 - 11 0 466 25 5 7 0 503
12 - 15 1 454 34 9 15 1 514
row percentage
0-4 79% 2% 0% 19% 100%
5-7 79% 6% 0% 15% 100%
Rural
8 - 11 61% 5% 2% 32% 100%
12 - 15 51% 4% 1% 44% 100%
0-4 98% 2% 0% 0% 100%
5-7 98% 1% 0% 0% 100%
Urban
8 - 11 93% 5% 1% 1% 100%
12 - 15 0% 88% 7% 2% 3% 0% 100%
B: Females
number
0-4 1 22 1 0 2 0 26
5-7 0 22 1 0 8 0 31
Rural
8 - 11 0 39 2 0 15 0 56
12 - 15 1 38 2 0 32 1 74
0-4 1 125 2 0 1 0 129
5-7 0 224 9 0 1 0 234
Urban
8 - 11 1 298 9 2 4 0 314
12 - 15 0 357 19 10 6 0 392
row percentage
0-4 4% 85% 4% 8% 100%
5-7 71% 3% 26% 100%
Rural
8 - 11 70% 4% 27% 100%
12 - 15 1% 51% 3% 43% 1% 100%
0-4 1% 97% 2% 1% 100%
5-7 96% 4% 0% 100%
Urban
8 - 11 0% 95% 3% 1% 1% 100%
12 - 15 91% 5% 3% 2% 100%
1
One speed limit was recorded at 55mph.
15
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 21: Child pedestrian casualties by weather condition by age group for males and
females (1999 - 2008)
number
0-4 21 4 0 0 0 1 26
5-7 24 7 0 0 0 0 31
Rural
8 - 11 43 7 1 2 3 0 56
12 - 15 60 9 1 2 0 2 74
0-4 115 13 0 0 0 1 129
5-7 205 20 0 4 1 4 234
Urban
8 - 11 257 50 1 0 3 3 314
12 - 15 309 66 2 2 7 6 392
row percentage
0-4 81% 15% 4% 100%
5-7 77% 23% 0% 100%
Rural
8 - 11 77% 13% 2% 4% 5% 0% 100%
12 - 15 81% 12% 1% 3% 3% 100%
0-4 89% 10% 1% 100%
5-7 88% 9% 2% 0% 2% 100%
Urban
8 - 11 82% 16% 0% 1% 1% 100%
12 - 15 79% 17% 1% 1% 2% 2% 100%
1
Weather conditions include fine, rain and snow both with and without high winds. The number of
weather conditions reported with high winds was 80 (fine (38), rain (46) and snow (2)). The table
combines with and without high winds for fine, rain and snow.
16
Deprivation and Child Pedestrian Road Casualties
Final Report
17
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 23: Child pedestrian casualties by principle causation factor (1999 – 2008)
19
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 24: Child pedestrian casualties by junction detail and junction control (1999 – 2008)
Junction Control
20
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 25: Child pedestrian casualties by pedestrian crossing facility and age
group (1999 – 2008)
Age group
All age
Pedestrian crossing facility 0-4 5-7 8-11 12-15 groups
number
None within 50 metres 439 570 747 821 2,577
Pelican 13 30 77 103 223
Other light controlled crossing 12 15 45 44 116
Pedestrian Phase at traffic signal junction 13 16 32 48 109
Zebra crossing 13 18 25 37 93
Central refuge - no control 6 7 19 26 58
Site controlled by crossing patrol 2 10 9 13 34
Zebra controlled by School Crossing Patrol 5 3 6 14
Footbridge or subway 1 1 2 4 8
Site controlled by other authorised person 1 2 3
All 499 672 960 1,104 3,235
column percentage
None within 50 metres 88% 85% 78% 74% 80%
Pelican 3% 4% 8% 9% 7%
Other light controlled crossing 2% 2% 5% 4% 4%
Pedestrian Phase at traffic signal junction 3% 2% 3% 4% 3%
Zebra crossing 3% 3% 3% 3% 3%
Central refuge - no control 1% 1% 2% 2% 2%
Site controlled by crossing patrol 0% 1% 1% 1% 1%
Zebra controlled by School Crossing Patrol 0% 1% 0% 1% 0%
Footbridge or subway 0% 0% 0% 0%
Site controlled by other authorised person 0% 0% 0%
All 100% 100% 100% 100% 100%
21
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 26: Child pedestrian casualties by carriageway type and urban rural classification
(1999 – 2008)
Number %ages
Table 27: Child pedestrian casualties by road surface condition and urban rural
classification (1999 – 2008)
Number %ages
Road surface condition Rural Urban All Rural Urban All
Dry 354 2,016 2,370 72% 74% 73%
Wet / Damp 127 660 787 26% 24% 24%
Other 1 25 26 0% 1% 1%
Frost / Ice 2 18 20 0% 1% 1%
Slippery (after dry Spell) 4 8 12 1% 0% 0%
Snow 2 9 11 0% 0% 0%
Mud 1 3 4 0% 0% 0%
Flood 2 1 3 0% 0% 0%
Oil 1 1 0% 0%
Leaves 1 1 0% 0%
All 493 2,742 3,235 100% 100% 100%
22
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 28: Child pedestrian casualties by special condition at site and urban rural
classification (1999 – 2008)
Table 29: Child pedestrian casualties by carriageway hazard and urban rural
classification (1999 – 2008)
23
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 30: Child pedestrian casualties by vehicle manoeuvre and vehicle location at time of impact (1999 – 2008)
Vehicle location at time of impact
Pedestrian
and On lay- Leaving Entering
On Other Entering Leaving vehicle by or lay-by / Bus lay-by /
On main minor public main Footpath main shared hard hard lane / hard
Vehicle manoeuvre road road place road (pavement) road precinct shoulder shoulder busway shoulder All %age
1
Going Ahead other 1,431 666 16 12 17 7 7 3 2 4 2,166 67%
Passing Stationary Vehicle on o/s 176 115 1 1 1 1 295 9%
Reversing 20 42 20 11 16 5 9 2 2 1 128 4%
Other / not known 47 41 7 2 11 1 3 112 3%
. About to Go Ahead / Held up 51 17 1 1 1 71 2%
Turning right 19 22 2 6 2 18 1 70 2%
Stopping 36 19 4 1 1 2 1 1 1 66 2%
Starting 27 22 4 5 1 1 4 64 2%
Going ahead left hand bend 39 20 3 1 63 2%
Going ahead left hand bend 26 23 2 5 2 58 2%
Passing on Nearside 30 26 56 2%
Turning left 13 11 18 10 1 53 2%
Parked 7 5 1 2 15 0%
Overtaking Moving Vehicle on o/s 6 3 1 10 0%
Waiting to turn left 1 2 1 4 0%
Changing Lane to Right 2 1 3 0%
U turn 1 1 0%
All 1,931 1036 58 56 54 46 22 13 9 6 3 3,235 100%
%age 60% 32% 2% 2% 2% 1% 1% 0% 0% 0% 0% 100%
1
Includes one miscoded value for vehicle location at time of impact.
24
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 31: Child pedestrian casualties by school pupil casualty by age group for week and
weekend days (1999 – 2008)
25
Deprivation and Child Pedestrian Road Casualties
Final Report
Table 32: Child pedestrian casualties by pedestrian location and pedestrian movement (1999 – 2008)
Pedestrian movement
Crossing Crossing
from from
drivers drivers In c'way
Crossing Crossing nearside offside Walking In c'way Walking stationary
from from masked masked in c'way stationary in c'way - not
drivers drivers by static by static Unknown/ back to - not facing crossing -
Pedestrian location nearside offside vehicle vehicle other traffic crossing traffic masked All %age
In carriageway crossing elsewhere 627 373 375 286 31 8 3 4 3 1,710 53%
In centre of carriageway 94 140 71 66 12 6 25 1 1 416 13%
In carriageway crossing elsewhere
94 64 44 54 3 1 2 1 263 8%
within 50 m of crossing
On footway or verge 6 1 1 1 125 49 1 63 247 8%
In carriageway, not crossing 23 11 22 6 65 21 48 13 9 218 7%
In carriageway on pedestrian crossing 88 73 23 17 2 1 1 205 6%
Unknown or other 28 15 6 8 85 4 4 1 3 154 5%
On refuge, central island or central
2 9 1 1 13 0%
reservation
In carriageway, within zigzag lines at
3 2 1 6 0%
crossing approach
In carriageway within zigzag lines at
1 2 3 0%
crossing exit
All 966 688 545 439 323 89 84 82 19 3,235 100%
%age 30% 21% 17% 14% 10% 3% 3% 3% 1% 100%
26
Deprivation and Child Pedestrian Road Casualties
Final Report
1.80
1.60
Child pedestrian casualty rate
1.40
1.20
1.00
0.80
0.60
0.40
0.20
0.00
1 2 3 4 5 6 7 8 9 10
Deprivation decile
2.50
Child pedestrian casualty rate
2.00
1.50
Male
Female
1.00
0.50
0.00
1 2 3 4 5 6 7 8 9 10
Deprivation decile
27
Deprivation and Child Pedestrian Road Casualties
Final Report
1.60
Child pedestrian casualty rate
1.40
1.20
1.00
0.80
0.60
0.40
0.20
0.00
1 2 3 4 5 6 7 8 9 10
Deprivation decile
3.50
Child pedestrian casualty rate
3.00
2.50
2.00
Male
1.50 Female
1.00
0.50
0.00
1 2 3 4 5 6 7 8 9 10
Deprivation decile
28
Deprivation and Child Pedestrian Road Casualties
Final Report
2.00
1.50
Male
1.00 Female
0.50
0.00
1 2 3 4 5 6 7 8 9 10
Deprivation decile
1.80
Child pedestrian casualty rate
1.60
1.40
1.20
1.00 Male
0.80 Female
0.60
0.40
0.20
0.00
1 2 3 4 5 6 7 8 9 10
Deprivation decile
29
Deprivation and Child Pedestrian Road Casualties
Final Report
4
3.5
3
Casualty rate
1 - 4 years
2.5
5 - 7 years
2
8 - 11 years
1.5
12 -15 years
1
0.5
0
1 2 3 4 5 6 7 8 9 10
Deprivation decile
30
Deprivation and Child Pedestrian Road Casualties
Final Report
2.5
2
Casualty rate
1 - 4 years
1.5
5 - 7 years
8 - 11 years
1
12 -15 years
0.5
0
1 2 3 4 5 6 7 8 9 10
Deprivation decile
31
Deprivation and Child Pedestrian Road Casualties
Final Report
1.80
Child pedestrian casualty rate
1.60
1.40
1.20
1.00 Male
0.80 Female
0.60
0.40
0.20
0.00
1 2 3 4 5 6 7 8 9 10
Deprivation decile
1.40
Child pedestrian casualty rate
1.20
1.00
0.80 Male
Female
0.60
0.40
0.20
0.00
1 2 3 4 5 6 7 8 9 10
Deprivation decile
32
Deprivation and Child Pedestrian Road Casualties
Final Report
1.80
1.60
Child pedestrian casualty rate
1.40
1.20
1.00
Male
0.80
Female
0.60
0.40
0.20
0.00
1 2 3 4 5 6 7 8 9 10
Deprivation decile
Figure 12: Pedestrian casualty rates by deprivation decile and age group
for all children and adults (males) (1999 - 2008)
2.50
2.00
Casualty rate
1.50 0 - 15 years
16 - 24 years
1.00 25 and over
0.50
0.00
1 2 3 4 5 6 7 8 9 10
Deprivation decile
33
Deprivation and Child Pedestrian Road Casualties
Final Report
Figure 13: Pedestrian casualty rates by deprivation decile and age group
for all children and adults (females) (1999 - 2008)
1.60
1.40
1.20
Casualty rate
1.00 0 - 15 years
0.80 16 - 24 years
0.60 25 and over
0.40
0.20
0.00
1 2 3 4 5 6 7 8 9 10
Deprivation decile
1.40
1.20
Child pedestrian casualty rate
1.00
0.80
0.60
0.40
0.20
0.00
1 2 3 4 5 6 7 8 9 10
Deprivation decile
34
Deprivation and Child Pedestrian Road Casualties
Final Report
1.40
1.20
Child pedestrian casualty rate
1.00
0.80
0.60
0.40
0.20
0.00
1 2 3 4 5 6 7 8 9 10
Deprivation decile
rate difference
Difference in child pedestrian casualty rate
0.50
0.30
0.10
-0.10 1 2 3 4 5 6 7 8 9 10
-0.30
-0.50
-0.70
Deprivation decile
35
Deprivation and Child Pedestrian Road Casualties
Final Report
Figure 17: KSI child pedestrian casualties by year and sex (1999 -
2008)
100
90
Number of casualties
80
70
60
Male
50
Female
40
30
20
10
0
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
Year
Figure 18: All child pedestrian casualties by year and sex (1999 -
2008)
350
300
Number of casualties
250
200 Male
150 Female
100
50
0
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
Year
36
Deprivation and Child Pedestrian Road Casualties
Final Report
Figure 19: Average male child pedestrian casualties by year of age and
age group (1999 - 2008)
Number of casualties
30
25
1-4
20
5-7
15
8-11
10
12-15
5
0
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
Year
20
15 1-4
5-7
10
8-11
5 12-15
0
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
Year
37
Deprivation and Child Pedestrian Road Casualties
Final Report
Appendix 3
1
Deprivation and Child Pedestrian Road Casualties
Final Report
Significance
Factor Estimate Std. Error Z Value Pr(>|Z|) code
(Intercept) 0.76548 0.09195 8.325 <2e-16 ***
mdmdecM2 0.11573 0.12564 0.921 0.35702
mdmdecM3 0.32195 0.12334 2.610 0.00905 **
mdmdecM4 0.34826 0.12281 2.836 0.00457 **
mdmdecM5 0.31371 0.12311 2.548 0.01083 *
mdmdecM6 0.69341 0.12002 5.777 7.59e-09 ***
mdmdecM7 0.71597 0.11995 5.969 2.39e-09 ***
mdmdecM8 0.90485 0.11893 7.608 2.78e-14 ***
mdmdecM9 1.21956 0.11884 10.262 <2e-16 ***
mdmdecM10 1.98809 0.11933 16.660 <2e-16 ***
Significance codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Table A 2: Level 1 model output with MDM decile and age as independent variables
Significance
Coefficient Estimate Std. Error Z value Pr(>|z|) code
(Intercept) 0.06881 0.10296 0.668 0.503948
mdmdecM2 0.13670 0.11585 1.180 0.238005
mdmdecM3 0.32541 0.11340 2.869 0.004111 **
mdmdecM4 0.39946 0.11235 3.556 0.000377 ***
mdmdecM5 0.34637 0.11288 3.069 0.002151 **
mdmdecM6 0.72863 0.10927 6.668 2.59e-11 ***
mdmdecM7 0.75236 0.10915 6.893 5.47e-12 ***
mdmdecM8 0.96688 0.10774 8.974 <2e-16 ***
mdmdecM9 1.27366 0.10714 11.887 <2e-16 ***
mdmdecM10 2.05901 0.10640 19.351 <2e-16 ***
ageA2 0.77367 0.08016 9.652 <2e-16 ***
ageA3 0.96016 0.07906 12.145 <2e-16 ***
ageA4 1.15707 0.07810 14.815 <2e-16 ***
ageA5 0.65797 0.08093 8.131 4.27e-16 ***
ageA6 -0.56574 0.09620 -5.881 4.08e-09 ***
Significance codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
2
Deprivation and Child Pedestrian Road Casualties
Final Report
Table A 3: Level 1 model output with MDM decile, age and sex as independent variables
Significance
Coefficient Estimate Std. Error Z Value Pr(>|Z|) Code
(Intercept) -0.20944 0.10420 -2.010 0.044433 *
mdmdecM2 0.14618 0.11335 1.290 0.197177
mdmdecM3 0.33244 0.11088 2.998 0.002717 **
mdmdecM4 0.38775 0.11001 3.525 0.000424 ***
mdmdecM5 0.33174 0.11060 2.999 0.002705 **
mdmdecM6 0.75155 0.10654 7.054 1.74e-12 ***
mdmdecM7 0.74415 0.10666 6.977 3.01e-12 ***
mdmdecM8 0.97624 0.10507 9.292 <2e-16 ***
mdmdecM9 1.28022 0.10438 12.265 <2e-16 ***
mdmdecM10 2.06573 0.10341 19.976 <2e-16 ***
ageA2 0.79233 0.07824 10.127 <2e-16 ***
ageA3 0.97366 0.07714 12.622 <2e-16 ***
ageA4 1.18034 0.07612 15.507 <2e-16 ***
ageA5 0.67209 0.07905 8.502 <2e-16 ***
ageA6 -0.53863 0.09439 -5.706 1.16e-08 ***
sexM 0.46494 0.04368 10.645 <2e-16 ***
Significance codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Table A 4: Level 1 model output with MDM decile, age, sex and URDG as independent
variables
Significance
Coefficient Estimate Std. Error Z value Pr(>|z|) code
Intercept) -1.20367 0.07714 -15.604 <2e-16 ***
mdmdecM2 0.21272 0.07939 2.679 0.00737 **
mdmdecM3 0.43138 0.07594 5.680 1.35e-08 ***
mdmdecM4 0.46758 0.07539 6.202 5.57e-10 ***
mdmdecM5 0.45862 0.07588 6.044 1.51e-09 ***
mdmdecM6 0.83590 0.07099 11.776 <2e-16 ***
mdmdecM7 0.83913 0.07079 11.855 <2e-16 ***
mdmdecM8 1.01032 0.06896 14.650 <2e-16 ***
mdmdecM9 1.24538 0.06691 18.614 <2e-16 ***
mdmdecM10 1.91408 0.06339 30.196 <2e-16 ***
ageA2 0.80782 0.04698 17.195 <2e-16 ***
ageA3 0.87007 0.04647 18.722 <2e-16 ***
ageA4 1.00702 0.04556 22.103 <2e-16 ***
ageA5 0.56376 0.04889 11.532 <2e-16 ***
ageA6 -0.64881 0.06832 -9.496 <2e-16 ***
sexM 0.43388 0.02454 17.681 <2e-16 ***
urdgU 1.50940 0.03349 45.064 <2e-16 ***
Significance codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
3
Deprivation and Child Pedestrian Road Casualties
Final Report
Table A 5: Level 1 model output with MDM decile, age, sex and URDG as independent
variables (where MDM decile and URDG are modelled with interaction)
Std. Significance
Coefficient Estimate Error Z Value Pr(>|Z|) Code
(Intercept) -2.26149 0.28385 -7.967 1.62e-15 ***
mdmdecM2 1.57541 0.30226 5.212 1.87e-07 ***
mdmdecM3 1.75154 0.29841 5.870 4.37e-09 ***
mdmdecM4 1.93064 0.29525 6.539 6.20e-11 ***
mdmdecM5 2.22592 0.29130 7.641 2.15e-14 ***
mdmdecM6 2.35219 0.29008 8.109 5.12e-16 ***
mdmdecM7 2.34555 0.29040 8.077 6.63e-16 ***
mdmdecM8 2.26973 0.29142 7.788 6.79e-15 ***
mdmdecM9 0.60789 0.34164 1.779 0.0752 .
mdmdecM10 -1.57153 0.73763 -2.131 0.0331 *
urdgU 2.66163 0.28741 9.261 <2e-16 ***
ageA2 0.81135 0.04637 17.497 <2e-16 ***
ageA3 0.86772 0.04591 18.902 <2e-16 ***
ageA4 1.00162 0.04501 22.254 <2e-16 ***
ageA5 0.56139 0.04835 11.611 <2e-16 ***
ageA6 -0.65403 0.06792 -9.629 <2e-16 ***
sexM 0.43230 0.02416 17.897 <2e-16 ***
mdmdecM2:urdgU -1.52785 0.31394 -4.867 1.13e-06 ***
mdmdecM3:urdgU -1.47565 0.30911 -4.774 1.81e-06 ***
mdmdecM4:urdgU -1.66892 0.30614 -5.451 5.00e-08 ***
mdmdecM5:urdgU -2.13537 0.30347 -7.037 1.97e-12 ***
mdmdecM6:urdgU -1.75458 0.29985 -5.851 4.87e-09 ***
mdmdecM7:urdgU -1.73094 0.30001 -5.770 7.95e-09 ***
mdmdecM8:urdgU -1.39892 0.30016 -4.661 3.15e-06 ***
mdmdecM9:urdgU 0.67255 0.34828 1.931 0.0535 .
mdmdecM10:urdgU 3.50347 0.74040 4.732 2.22e-06 ***
Significance codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
4
Deprivation and Child Pedestrian Road Casualties
Final Report
Table A 6: Level 1 model output with MDM decile, age, sex and URDG as independent
variables (where MDM decile*URDG and age*sex are modelled with interaction)
Std. Significance
Coefficient Estimate Error Z value Pr(>|z|) code
(Intercept) -2.50359 0.28934 -8.653 <2e-16 ***
mdmdecM2 1.57537 0.30226 5.212 1.87e-07 ***
mdmdecM3 1.75136 0.29841 5.869 4.38e-09 ***
mdmdecM4 1.93054 0.29525 6.539 6.21e-11 ***
mdmdecM5 2.22586 0.29130 7.641 2.15e-14 ***
mdmdecM6 2.35200 0.29008 8.108 5.14e-16 ***
mdmdecM7 2.34539 0.29040 8.076 6.67e-16 ***
mdmdecM8 2.26958 0.29142 7.788 6.81e-15 ***
mdmdecM9 0.60784 0.34164 1.779 0.075208 .
mdmdecM10 -1.57174 0.73763 -2.131 0.033106 *
urdgU 2.66145 0.28741 9.260 <2e-16 ***
ageA2 1.05568 0.08081 13.063 <2e-16 ***
ageA3 1.09170 0.08032 13.591 <2e-16 ***
ageA4 1.31480 0.07825 16.803 <2e-16 ***
ageA5 0.79623 0.08376 9.506 <2e-16 ***
ageA6 -0.29110 0.10857 -2.681 0.007337 **
sexM 0.80430 0.08371 9.609 <2e-16 ***
mdmdecM2:urdgU -1.52786 0.31394 -4.867 1.13e-06 ***
mdmdecM3:urdgU -1.47551 0.30911 -4.773 1.81e-06 ***
mdmdecM4:urdgU -1.66889 0.30614 -5.451 5.00e-08 ***
mdmdecM5:urdgU -2.13537 0.30347 -7.037 1.97e-12 ***
mdmdecM6:urdgU -1.75446 0.29985 -5.851 4.89e-09 ***
mdmdecM7:urdgU -1.73079 0.30001 -5.769 7.97e-09 ***
mdmdecM8:urdgU -1.39876 0.30016 -4.660 3.16e-06 ***
mdmdecM9:urdgU 0.67259 0.34828 1.931 0.053463 .
mdmdecM10:urdgU 3.50367 0.74040 4.732 2.22e-06 ***
ageA2:sexM -0.37536 0.09884 -3.798 0.000146 ***
ageA3:sexM -0.34204 0.09802 -3.490 0.000484 ***
ageA4:sexM -0.49142 0.09591 -5.124 2.99e-07 ***
ageA5:sexM -0.35973 0.10276 -3.501 0.000464 ***
ageA6:sexM -0.58162 0.14007 -4.152 3.29e-05 ***
Significance codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
5
Deprivation and Child Pedestrian Road Casualties
Final Report
Table A 7: Significance of coefficients for pedestrian casualties in urban areas by age and sex for the Level 2 Model
Free
(Intercep MDM MMD MDM Population Road Traffic School school Sunshine Annual
Age Group Sex t) Score Crime Proximity density length proxy places meals hours Rainfall
Child Both 0.001(+) 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) 0.001(+) 0.001(+) 0.001(+) 0.001(-) 0.01(-)
casualties Female NS 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) 0.001(+) 0.001(+) 0.001(+) 0.001(-) 0.01(-)
(0-14) Male 0.05(+) 0.001(+) 0.001(+) 0.001(-) 0.001(+) NS 0.001(+) 0.001(+) NS 0.001(-) 0.05(-)
Child Both NS 0.001(+) 0.001(+) 0.01(-) 0.05(+) NS 0.001(+) 0.001(+) NS 0.001(-) 0.001(-)
casualties Female NS 0.001(+) NS 0.01(-) 0.01(+) 0.05(+) 0.05(+) 0.01(+) 0.001(+) 0.05(-) NS
(0-4) Male NS 0.001(+) 0.001(+) NS NS NS 0.05(+) 0.01(+) NS 0.01(-) 0.001(-)
Child Both NS 0.001(+) 0.001(+) NS 0.001(+) 0.05(-) 0.001(+) 0.001(+) NS 0.001(-) 0.01(-)
casualties Female NS 0.001(+) 0.001(+) 0.05(+) NS NS NS 0.001(+) NS 0.001(-) 0.01(-)
(5-7) Male NS 0.001(+) 0.05(+) NS 0.001(+) NS 0.001(+) 0.001(+) NS 0.001(-) NS
Child Both 0.05(-) 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) 0.001(+) 0.001(+) 0.01(+) NS NS
casualties Female 0.001(-) 0.001(+) 0.05(+) 0.001(-) NS 0.05(+) 0.01(+) 0.001(+) 0.01(+) NS NS
(8-11) Male NS 0.001(+) 0.001(+) 0.01(-) 0.001(+) 0.05(+) 0.001(+) 0.001(+) NS NS NS
Child Both NS 0.001(+) 0.001(+) 0.001(-) NS 0.001(+) 0.001(+) 0.001(+) NS 0.05(-) 0.01(-)
casualties Female NS 0.05(+) 0.001(+) 0.001(-) NS 0.001(+) 0.001(+) 0.001(+) NS 0.001(-) 0.001(-)
(12-15) Male NS 0.001(+) 0.001(+) 0.001(-) NS NS 0.001(+) 0.001(+) 0.001(-) NS NS
Young adult Both 0.001(-) 0.001(+) 0.001(+) 0.001(-) 0.01(+) 0.01(+) 0.001(+) 0.001(+) 0.001(-) 0.01(+) 0.001(+)
casualties Female 0.001(-) 0.001(+) 0.001(+) 0.001(-) 0.01(+) NS 0.001(+) 0.001(+) 0.01(-) 0.01(+) 0.001(+)
(16-24) Male 0.001(-) 0.001(+) 0.001(+) 0.001(-) NS 0.05(+) 0.001(+) 0.001(+) NS 0.01(+) 0.01(+)
Adult Both NS 0.001(+) 0.001(+) 0.001(-) NS 0.001(+) 0.001(+) 0.001(+) 0.001(-) NS NS
casualties Female 0.001(-) 0.001(+) 0.001(+) 0.001(-) NS 0.01(+) 0.001(+) 0.001(+) 0.001(-) NS NS
(25 and over) Male NS 0.001(+) 0.001(+) 0.001(-) NS 0.001(+) 0.001(+) 0.001(+) NS 0.05(-) 0.001(-)
Both 0.001(+) 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) 0.001(+) 0.001(+) NS NS NS
All Casualties Female NS 0.001(+) 0.001(+) 0.001(-) 0.01(+) 0.001(+) 0.001(+) 0.001(+) NS NS NS
Male 0.05(+) 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) 0.001(+) 0.001(+) NS NS NS
6
Deprivation and Child Pedestrian Road Casualties
Final Report
Table A 8: Significance of coefficients for pedestrian casualties in rural areas by age and sex for the Level 2 Model
Free
MDM MMD MDM Population Road Traffic School school Sunshine Annual
Age Group Sex (Intercept) Score Crime Proximity density length proxy places meals hours Rainfall
Child Both 0.01(-) 0.001(+) 0.01(+) 0.001(-) NS 0.05(-) 0.001(+) NS 0.001(+) 0.01(+) 0.001(+)
casualties (0- Female 0.01(-) 0.01(+) 0.05(+) 0.001(-) NS 0.01(-) 0.001(+) 0.05(+) 0.05(+) NS NS
14) Male 0.001(-) 0.001(+) 0.05(+) 0.001(-) NS NS 0.01(+) NS 0.01(+) 0.01(+) 0.001(+)
Child Both NS 0.001(+) NS 0.001(-) NS NS NS NS NS 0.01(-) NS W
casualties (0- Female NS 0.001(+) 0.05(+) 0.01(-) NS 0.05(+) 0.05(-) NS 0.001(-) 0.01(-) NS W
4) Male NS 0.001(+) NS 0.001(-) 0.05(-) NS 0.01(+) NS NS NS NS W
Child Both 0.01(-) 0.001(+) NS NS 0.05(+) 0.001(-) NS NS NS NS NS W
casualties (5- Female NS 0.001(+) NS NS NS 0.001(-) NS 0.001(+) NS NS NS W
7) Male 0.001(-) NS NS NS 0.05(+) NS NS NS NS 0.01(+) NS W
Child Both 0.01(-) NS 0.001(+) 0.05(-) NS 0.01(-) NS 0.001(+) 0.001(+) NS NS W
casualties (8- Female 0.01(-) NS 0.01(+) NS 0.05(-) 0.001(-) NS 0.001(+) 0.05(+) 0.05(+) NS W
11) Male 0.05(-) NS 0.001(+) 0.01(-) NS NS NS 0.001(+) 0.01(+) NS 0.05(+) W
Child Both 0.001(-) 0.001(+) NS 0.001(-) 0.01(-) NS 0.001(+) NS NS 0.01(+) 0.001(+)
casualties Female 0.001(-) 0.01(+) NS 0.001(-) 0.05(-) NS 0.001(+) NS NS NS 0.05(+)
(12-15) Male 0.001(-) 0.05(+) NS 0.01(-) NS NS 0.001(+) NS NS 0.05(+) 0.001(+) W
Young adult Both 0.001(-) 0.001(+) NS 0.05(-) NS NS 0.001(+) NS NS 0.001(+) 0.001(+)
casualties Female 0.001(-) NS NS NS 0.05(-) 0.001(-) 0.001(+) 0.001(+) NS 0.05(+) 0.01(+)
(16-24) Male 0.001(-) 0.001(+) NS 0.001(-) NS NS 0.001(+) NS NS 0.001(+) 0.001(+)
Adult Both NS 0.001(+) 0.001(+) 0.001(-) 0.01(-) NS 0.001(+) 0.01(+) NS NS 0.05(+)
casualties (25 Female 0.01(-) 0.001(+) 0.001(+) NS NS 0.001(-) 0.001(+) NS 0.001(+) NS 0.01(+)
and over) Male NS 0.001(+) 0.001(+) 0.001(-) 0.001(-) 0.01(+) 0.001(+) 0.001(+) 0.001(+) NS NS
Both NS 0.001(+) 0.001(+) 0.001(-) 0.01(-) NS 0.001(+) 0.01(+) NS 0.01(+) 0.001(+)
All Casualties Female 0.05(-) 0.01(+) 0.001(+) 0.01(-) NS 0.001(-) 0.001(+) 0.01(+) 0.010(+) NS 0.01(+)
Male 0.01(-) 0.001(+) 0.001(+) 0.001(-) 0.01(-) 0.01(+) 0.001(+) NS NS 0.001(+) 0.001(+)
A “W” in the right hand column means that the model did not fully converge and the outputs may not be robust
7
Deprivation and Child Pedestrian Road Casualties
Final Report
Table A 9: Significance of coefficients for pedestrian casualties in urban areas by age and sex for the Level 3 Model
MDM School Sunshine Annual
Age Group Sex (Intercept) MDM Score MMD Crime Proximity Traffic proxy places hours Rainfall
Child Both 0.001(+) 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) 0.001(-) 0.001(-)
casualties (0- Female 0.001(+) 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) 0.001(-) 0.001(-)
14) Male 0.001(+) 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) 0.001(-) 0.001(-)
Child Both NS 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) 0.001(-) 0.001(-)
casualties (0- Female NS 0.001(+) NS 0.001(-) 0.01(+) 0.001(+) 0.01(-) NS
4) Male NS 0.001(+) 0.001(+) 0.01(-) 0.01(+) 0.01(+) 0.01(-) 0.001(-)
Child Both 0.01(+) 0.001(+) 0.01(+) 0.001(-) NS 0.001(+) 0.001(-) 0.001(-)
casualties (5- Female NS 0.001(+) 0.01(+) NS NS 0.001(+) 0.001(-) 0.01(-)
7) Male NS 0.001(+) NS 0.001(-) 0.05(+) 0.001(+) 0.001(-) 0.05(-)
Child Both NS 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) NS NS
casualties (8- Female 0.001(-) 0.001(+) 0.01(+) 0.001(-) 0.001(+) 0.001(+) NS NS
11) Male NS 0.001(+) 0.01(+) 0.001(-) 0.001(+) 0.001(+) 0.01(-) 0.05(-)
Child Both NS 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) 0.05(-) 0.01(-)
casualties Female NS 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) 0.001(-) 0.001(-)
(12-15) Male 0.05(-) 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) NS NS
Young adult Both 0.001(-) 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) 0.01(+) 0.01(+)
casualties Female 0.001(-) 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.01(+) 0.01(+) 0.05(+)
(16-24) Male 0.001(-) 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) NS 0.01(+)
Adult Both NS 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) NS 0.01(-)
casualties (25 Female 0.001(-) 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) NS NS
and over) Male NS 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) NS 0.001(-)
Both 0.001(+) 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) 0.001(-) 0.01(-)
All Casualties Female NS 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) NS NS
Male 0.001(+) 0.001(+) 0.001(+) 0.001(-) 0.001(+) 0.001(+) 0.01(-) 0.01(-)
8
Deprivation and Child Pedestrian Road Casualties
Final Report
Table A 10: Significance of coefficients for pedestrian casualties in rural areas by age and sex for the Level 3 Model
MDM Sunshine Annual
Age Group Sex (Intercept) MDM Score MMD Crime Proximity Traffic proxy School places hours Rainfall
Child Both 0.01(-) 0.001(+) 0.01(+) 0.001(-) 0.001(+) 0.05(+) 0.01(+) 0.001(+)
casualties (0- Female 0.01(-) 0.001(+) 0.001(+) 0.001(-) 0.05(+) 0.01(+) NS NS
14) Male 0.01(-) 0.001(+) 0.001(+) 0.001(-) 0.05(+) 0.01(+) NS NS
Child Both NS 0.001(+) NS 0.001(-) NS NS 0.01(-) NS W
casualties (0- Female NS 0.001(+) NS 0.001(-) 0.05(-) NS 0.01(-) NS W
4) Male NS 0.001(+) NS 0.001(-) 0.01(+) NS NS NS W
Child Both 0.01(-) 0.001(+) NS 0.001(-) NS NS NS NS W
casualties (5- Female NS 0.001(+) NS 0.001(-) NS 0.01(+) NS NS W
7) Male 0.001(-) 0.05(+) NS NS NS NS 0.01(+) NS W
Child Both 0.01(-) 0.001(+) 0.001(+) 0.05(-) NS 0.001(+) NS NS W
casualties (8- Female 0.01(-) NS 0.001(+) 0.001(-) NS 0.01(+) 0.05(+) NS W
11) Male 0.05(-) 0.001(+) 0.001(+) 0.001(-) NS 0.001(+) NS 0.05(+) W
Child Both 0.001(-) 0.01(+) NS 0.001(-) 0.001(+) NS 0.01(+) 0.001(+)
casualties Female 0.001(-) 0.05(+) NS 0.001(-) 0.001(+) NS NS 0.05(+)
(12-15) Male 0.001(-) 0.05(+) NS 0.01(-) 0.001(+) NS 0.05(+) 0.001(+) W
Young adult Both 0.001(-) 0.001(+) NS 0.01(-) 0.001(+) NS 0.001(+) 0.001(+)
casualties Female 0.001(-) NS NS NS 0.01(+) 0.001(+) 0.01(+) 0.001(+)
(16-24) Male 0.001(-) 0.001(+) NS 0.001(-) 0.001(+) 0.01(-) 0.001(+) 0.001(+)
Adult Both NS 0.001(+) 0.001(+) 0.001(-) 0.001(+) NS NS 0.05(+)
casualties (25 Female 0.01(-) 0.001(+) 0.001(+) 0.001(-) 0.001(+) NS NS 0.01(+)
and over) Male NS 0.001(+) 0.001(+) NS 0.001(+) NS NS NS
Both NS 0.001(+) 0.001(+) 0.001(-) 0.001(+) NS 0.01(+) 0.001(+)
All Casualties Female 0.05(-) 0.01(+) 0.001(+) 0.001(-) 0.001(+) 0.01(+) NS 0.001(+)
Male 0.01(-) 0.001(+) 0.001(+) 0.001(-) 0.001(+) NS 0.01(+) 0.001(+)
A “W” in the right hand column means that the model did not fully converge and the outputs may not be robust
9
Deprivation and Child Pedestrian Road Casualties
Final Report
Table A 11: Significance of coefficients for pedestrian casualties in all areas (urban and rural combined) by age and sex for the Level 3 Model
10
Deprivation and Child Pedestrian Road Casualties
Final Report
Table A 12: 50 poorest performing SOAs in terms of child pedestrian casualties (Level 3 Model)
11
Deprivation and Child Pedestrian Road Casualties
Final Report
Table A 13: 50 best performing SOAs in terms of child pedestrian casualties (Level 3 Model)
12
Deprivation and Child Pedestrian Road Casualties
Final Report
Table A 14: 50 SOAs with highest casualty increases in most recent 5 year period (2004 – 08)
Local
Government Observed Expected Observed
SOA Code District SOA Name 1999 - 03 2004 - 08 2004 - 08 Increase
95XX05W1 North Down Bangor Castle 5 2.95 12 9.05
95PP20W1 Fermanagh Portora 4 2.36 8 5.64
95GG05S3 Belfast Ballynafeigh_3 1 0.59 6 5.41
95GG35S3 Belfast New Lodge_3 7 4.13 9 4.87
95PP17W1 Fermanagh Lisnaskea 2 1.18 6 4.82
95KK04W1 Cookstown Gortalowry 4 2.36 7 4.64
95GG21S2 Belfast Falls_2 6 3.54 8 4.46
95GG17S1 Belfast Cliftonville_1 3 1.77 6 4.23
95YY01W1 Omagh Beragh 0 0.00 4 4.00
95NN07S2 Down Cathedral_2 7 4.13 8 3.87
95GG37S2 Belfast Ravenhill_2 2 1.18 5 3.82
95JJ16W1 Coleraine Portstewart 2 1.18 5 3.82
95WW14W1 Newtownabbey Dunanney 2 1.18 5 3.82
95DD21W1 Ballymena Park 4 2.36 6 3.64
95WW12W1 Newtownabbey Coole 4 2.36 6 3.64
95MM09S1 Derry Claudy_1 1 0.59 4 3.41
95MM13S2 Derry Culmore_2 1 0.59 4 3.41
95DD08W1 Ballymena Castle Demesne 8 4.72 8 3.28
95BB19S1 Ards Movilla_1 3 1.77 5 3.23
95GG16S1 Belfast Chichester Park_1 5 2.95 6 3.05
95GG07S3 Belfast Beechmount_3 0 0.00 3 3.00
95II15S2 Castlereagh Knockbracken_2 0 0.00 3 3.00
95II20W1 Castlereagh Newtownbreda 0 0.00 3 3.00
95LL02S1 Craigavon Annagh_1 0 0.00 3 3.00
95MM12S2 Derry Crevagh_2 0 0.00 3 3.00
95MM19S2 Derry Kilfennan_2 0 0.00 3 3.00
95NN19S1 Down Saintfield_1 0 0.00 3 3.00
95WW02S1 Newtownabbey Ballyclare North_1 0 0.00 3 3.00
95YY15S1 Omagh Lisanelly_1 0 0.00 3 3.00
95JJ22W1 Coleraine Waterside 2 1.18 4 2.82
95KK16W1 Cookstown Tullagh 2 1.18 4 2.82
95NN17W1 Down Murlough 2 1.18 4 2.82
95SS21S1 Lisburn Maghaberry_1 2 1.18 4 2.82
95GG50S3 Belfast Woodstock_3 4 2.36 5 2.64
95LL08S2 Craigavon Corcrain_2 4 2.36 5 2.64
95SS18S2 Lisburn Lagan Valley_2 4 2.36 5 2.64
95SS28W1 Lisburn Tonagh 4 2.36 5 2.64
95GG31S3 Belfast Ladybrook_3 6 3.54 6 2.46
95II05S2 Castlereagh Carryduff East_2 6 3.54 6 2.46
95GG32S1 Belfast Legoniel_1 1 0.59 3 2.41
95LL07W1 Craigavon Church 1 0.59 3 2.41
95LL09S2 Craigavon Court_2 1 0.59 3 2.41
95MM01S2 Derry Altnagelvin_2 1 0.59 3 2.41
95MM20S1 Derry Lisnagelvin_1 1 0.59 3 2.41
95OO21W1 Dungannon Mullaghmore 1 0.59 3 2.41
Florence Court
95PP12W1 Fermanagh and Kinawley 1 0.59 3 2.41
95RR03W1 Limavady Coolessan 1 0.59 3 2.41
95WW08S1 Newtownabbey Burnthill_1 1 0.59 3 2.41
95WW22S1 Newtownabbey Mossley_1 1 0.59 3 2.41
95GG39S2 Belfast Shaftesbury_2 8 4.72 7 2.28
13
Deprivation and Child Pedestrian Road Casualties
Final Report
Table A 15: 50 SOAs with highest casualty reductions in most recent 5 year period (2004 – 08)
Local
Government Observed Expected Observed
SOA Code District SOA Name 1999 - 03 2004 - 08 2004 - 08 Reduction
95SS06S1 Lisburn Collin Glen_1 11 6.50 1 -5.50
95GG40S1 Belfast Shankill_1 10 5.90 1 -4.90
95GG04S2 Belfast Ballymacarrett_2 8 4.72 0 -4.72
95GG21S1 Belfast Falls_1 13 7.68 3 -4.68
95BB06W1 Ards Central_Ards 18 10.63 6 -4.63
95ZZ06W1 Strabane East 11 6.50 2 -4.50
95GG07S1 Belfast Beechmount_1 9 5.31 1 -4.31
95QQ07W1 Larne Central_Larne 10 5.90 2 -3.90
95GG47S1 Belfast Water Works_1 15 8.86 5 -3.86
95GG22S3 Belfast Falls Park_3 8 4.72 1 -3.72
95MM10S1 Derry Creggan Central_1 8 4.72 1 -3.72
95HH11W1 Carrickfergus Love Lane 6 3.54 0 -3.54
95II16W1 Castlereagh Lisnasharragh 6 3.54 0 -3.54
95ZZ12W1 Strabane Sion Mills 6 3.54 0 -3.54
95GG16S2 Belfast Chichester Park_2 9 5.31 2 -3.31
95MM25S1 Derry Shantallow West_1 9 5.31 2 -3.31
95VV10S2 Newry and Mourne Daisy hill_2 9 5.31 2 -3.31
95GG31S1 Belfast Ladybrook_1 7 4.13 1 -3.13
95GG38S3 Belfast Rosetta_3 7 4.13 1 -3.13
95GG17S3 Belfast Cliftonville_3 5 2.95 0 -2.95
95GG27S3 Belfast Glencolin_3 5 2.95 0 -2.95
95GG43S3 Belfast Sydenham_3 5 2.95 0 -2.95
95HH02S1 Carrickfergus Bluefield_1 5 2.95 0 -2.95
95OO05W1 Dungannon Ballysaggart 5 2.95 0 -2.95
95SS07S2 Lisburn Derryaghy_2 5 2.95 0 -2.95
95SS26S1 Lisburn Poleglass_1 5 2.95 0 -2.95
95VV07S1 Newry and Mourne Clonallan_1 5 2.95 0 -2.95
95GG47S2 Belfast Water Works_2 10 5.90 3 -2.90
95GG40S2 Belfast Shankill_2 8 4.72 2 -2.72
95AA19W1 Antrim Toome 6 3.54 1 -2.54
95MM01S1 Derry Altnagelvin_1 6 3.54 1 -2.54
95MM20S2 Derry Lisnagelvin_2 6 3.54 1 -2.54
95SS13S1 Lisburn Hilden_1 6 3.54 1 -2.54
95GG48S1 Belfast Whiterock_1 11 6.50 4 -2.50
95GG18S2 Belfast Clonard_2 16 9.45 7 -2.45
95AA16W1 Antrim Steeple 4 2.36 0 -2.36
95CC07W1 Armagh Derrynoose 4 2.36 0 -2.36
95CC14W1 Armagh Laurelvale 4 2.36 0 -2.36
95EE07W1 Ballymoney Fairhill 4 2.36 0 -2.36
95FF06S1 Banbridge Dromore South_1 4 2.36 0 -2.36
95GG05S2 Belfast Ballynafeigh_2 4 2.36 0 -2.36
95GG23S3 Belfast Finaghy_3 4 2.36 0 -2.36
95GG32S3 Belfast Legoniel_3 4 2.36 0 -2.36
95HH07W1 Carrickfergus Gortalee 4 2.36 0 -2.36
95JJ07W1 Coleraine Cross Glebe 4 2.36 0 -2.36
95JJ20S2 Coleraine The Cuts_2 4 2.36 0 -2.36
95JJ21W1 Coleraine University 4 2.36 0 -2.36
95KK09W1 Cookstown Newbuildings 4 2.36 0 -2.36
95LL14S2 Craigavon Drumnamoe_2 4 2.36 0 -2.36
95MM21S1 Derry New Buildings_1 4 2.36 0 -2.36
14
Deprivation and Child Pedestrian Road Casualties
Final Report
Figure A 1: Graph of observed and modelled pedestrian casualty totals with MDM decile as
independent variable for the Level 1 Model
Figure A 2: Graph of observed and modelled pedestrian casualty totals with MDM decile and
age as independent variables for the Level 1 Model
15
Deprivation and Child Pedestrian Road Casualties
Final Report
Figure A 3: Graph of observed and modelled pedestrian casualty totals with MDM decile, age
and sex as independent variables for the Level 1 Model
Figure A 4: Graph of observed and modelled pedestrian casualty totals with MDM decile, age,
sex and URDG as independent variables for the Level 1 Model
16
Deprivation and Child Pedestrian Road Casualties
Final Report
Figure A 5: Graph of observed and modelled pedestrian casualty totals with MDM decile, age,
sex and URDG as independent variables (where MDM decile and URDG are modelled with
interaction) for the Level 1 Model
Figure A 6: Graph of observed and modelled pedestrian casualty totals with MDM decile, age,
sex and URDG as independent variables (where MDM decile*URDG and age*sex are modelled
with interaction) for the Level 1 Model
17
Deprivation and Child Pedestrian Road Casualties
Final Report
Appendix 4
Mapping
0
Deprivation and Child Pedestrian Road Casualties
Final Report
1
Deprivation and Child Pedestrian Road Casualties
Final Report
2
Deprivation and Child Pedestrian Road Casualties
Final Report