New National Connectivity Statistics are Welcome but Updated National Accessibility Statistics Should be a Far Higher Priority
In late 2025 the Department for Transport (DfT) finally published its proposed its new connectivity metric for England. When explaining the new metric, DfT present a partial history of their statistics covering the period from 2014 to 2019, but excluding the period between 2003 and 2014 when much of the learning took place about how to calculate and achieve effective use of national accessibility and connectivity statistics. DfT note that “these are our first estimates of connectivity and we welcome feedback to inform our future developments and refinement”. Given the apparent lack of corporate memory in DfT about the connectivity and accessibility statistics prior to 2014, observed by others when commenting on the new metric, I have focused my feedback to DfT below on the period prior to 2014. There are many lessons on how to improve the latest metric from the widespread use of the national statistics over the last 25 years, including comparisons with the approaches taken to the statistics in Scotland and Wales.
Prior to the recent review of the accessibility and connectivity statistics that led to the development of the new 2025 metric I was asked by DfT in 2019 to write a brief review with recommendations. This commentary does not repeat the contents of that earlier paper which can be viewed here.
Appropriate Application of Accessibility, Connectivity and Travel Demand metrics
A more collaborative approach to transport was promised in the Labour Party manifesto published nearly 30 years ago in 1996, seeking to make better connections between transport goals and wider policy using accessibility and connectivity metrics. The 1997 general election therefore gave the government a mandate to deliver these new approaches. After researching how this could best be achieved, the 2002 Cabinet Office policy paper “Making the Connections” explained how it was envisaged that better connectivity could be delivered by collaborative accessibility planning approaches.
Adding accessibility and connectivity to traditional transport planning metrics for movement and travel demand has sought to join up transport with everything else, to ensure that transport systems better meet the needs of the economy and society. The concepts of accessibility and connectivity are closely linked. In simple terms transport providers talk about connectedness to reflect a transport supplier or provider perspective, and transport users talk about accessibility to opportunities to reflect a user perspective.
It is therefore logical that DfT will use connectivity measures for internal discussions about transport supply where the solutions can be wholly or largely secured by government without needing participation by others. However inputs from other government departments and popular support from the travelling public are often also needed, particularly on more controversial behaviour change or tariff design challenges, so transparent measures of access to opportunity are often better suited to these needs.
DfT highlights the benefits of connectivity metrics in parallel with aggregate travel demand metrics to offer a more heterogenic approach to the economic assessment of transport policies and projects. This is extremely welcome, finally implementing the recommendations of the research from over 25 years ago proposing the strengthening of economic appraisal in this way. The DfT guidance notes which describe the new connectivity metric describe why logsum connectivity measures have been selected since they allow travel for different purposes (e.g. work, shopping, leisure, etc) to be combined in a single connectivity score. The connectivity score and the parallel travel demand based economic assessments can then be used to offer two different lenses on transport scheme benefits.
However, the selected connectivity metrics reflect the ‘utility of the transport system in supporting better access’, not necessarily the ‘value of access’ itself. The distinction between the value of the ‘transport system’ and the value of ‘accessibility using the transport system’ can often be important since people and businesses do not really care whether that access is delivered by a more useful transport system or by some other means like building a new hospital closer to where people live. DfT say “a normalisation method was applied by dividing raw scores by their theoretical maximum values, allowing fair comparison between sub-purposes with different diminishing returns scales” but this process can result in a misrepresentation of the value of access. This is not necessarily a problem if the only use of the metric is by DfT as a consistent measure of the value of transport connections, but it does severely limit the potential uses of the new metric.
Using the DfT metrics, land use changes such as adding a school or hospital close to a housing area, will not necessarily show as a positive benefit for connectivity. Sometimes less transport is better for the economy but the connectivity metric will not necessarily measure these benefits. This should be made far clearer when describing how the new metric can be used. For example the statement made by the transport minister on 11 December 2025 to promote the connectivity metric is very misleading “This landmark platform will serve as the new national metric of connectivity, transforming how we plan for new development and the transport infrastructure needed to support it, ensuring new homes and services can be easily accessed by sustainable modes of transport, helping kickstart economic growth, and delivering the government’s house building targets.” The minister refers to “ensuring new homes and services can be easily accessed” but the new connectivity measure is not a measure of access but of transport connection, and cannot be used as a reliable proxy for access.
‘Making the Connections’ Policies Require Accessibility Metrics to Make Connections
Attempts to use connectivity metrics have also proved to be disappointing in practical cross sector working over the last 30 years with most users finding them hard to trust. For this reason the DfT’s 2004 accessibility planning guidance recommended the use of the accessibility metrics that had been most successful in practice. Transport utility measures for better connectivity have not proved to be as successful as accessibility metrics in supporting better joint working between transport and other sectors towards a better connected economy and society[1].
DfT’s accessibility planning approaches have often been described as the most successful method yet identified for influencing travel demand, which is increasingly important for the UK to meet its carbon reduction aims[2]. The 2013 Parliamentary Inquiry also recommended that “the Department should focus more closely on improving accessibility with transport funding directed to ‘accessibility’-focused initiatives, which will have a swifter impact on people’s well-being than large infrastructure projects”.
Shifting the national metrics to a connectivity focus using the latest metric is inconsistent with the recommendation of the Parliamentary Inquiry for a stronger accessibility focus. Partners of transport authorities in health, education, environment, planning, social services, leisure and other areas rely on the transparency of accessibility measures with many case studies demonstrating how these have successfully been used.
Calculation Methodology
The notes on the methodology accompanying the new connectivity metric make extensive reference to DfT accessibility and connectivity statistics from 2014 to 2019 which I described in my 2019 paper for DfT as “at best misleading and at worst nonsensical”. It is therefore very encouraging that many of my suggestions from that paper have now been implemented, but many more improvements are also needed.
- Lack of clarity about the reasons for the choices of data and algorithms – DfT simply describe data sources and algorithms they have used rather than offer any explanation of why the particular dataset or algorithm was chosen e.g. “For cycling and walking, our graphs are constructed using OpenStreetMap data”. Much greater trust in the chosen approaches could be created if DfT sought to evidence the benefits of each approach such as the reasons why OS Mastermap was used for the road network but not for cycling and walking, or why different forms of routing algorithms were used for each mode of transport.
- Calibration to the real world – Anyone who has used two different navigation systems for a journey will know how different recommended journey times and routes can be when using different systems, even when detailed preferences have been set such as preferred maximum length of walk or how time, cost and reliability are traded when making a travel choice. As stated in my 2019 review, DfT appear to be prioritising replicability over usefulness, so the likely usage will be limited to those seeking government funding. Offering a replicable platform for appraisal could deliver consistent nonsense without better calibration to the real world. It would be far better to calculate journey times using a few different algorithms to understand the uncertainty in the conclusions being made, or use a routing engine already widely used for journey planning by the travelling public to calibrate the route choices with real world practice.
- Accounting for the variation in journey times throughout the day to each opportunity – In 2004, when the first pilot national statistics were published by DfT, many sensitivity tests were undertaken to identify how many times of day needed to be included in the analysis to reflect the variation in travel times to key destinations due to public transport schedules, road congestion and other factors. The new DfT metric uses four times of day to reflect the journey time variation on the road network, journey times at 10 minute intervals throughout the day for public transport trips, and a single journey time for walking and cycling. DfT do not explain why they have checked journey times by public transport at 10 minute intervals throughout the day, when 15 minute intervals had been successfully used in the statistical series in the past for many years [3]. Also road congestion varies throughout the day, particularly from 3pm to about 8pm when the evening peak is widely spread, so from 2010 to 2014 road journey times were calculated at 15 minute intervals in DfT’s national statistics matching the approach for the public transport times. The proposal for four times of day in the road connectivity metric is not likely to be sufficient to represent the travel times accurately.
- Spatial detail – The national DfT accessibility statistics from 2004 to 2016 were published at census output area level because this was the smallest area where information about the population was known such as age, gender, car ownership and other factors. There was no computational reason why a finer spatial detail could not be used. The choice of the 100 metre squares across the country in the new DfT connectivity metric offers no apparent advantages for computational accuracy, but in the 25 years since the design of the original DfT approach computing capabilities have improved substantially. The new metric could perhaps have learned from the experiences calculating the accessibility domain of the Welsh Index of Multiple Deprivation, where useful increases in accuracy were obtained by using house address point data as the chosen level of spatial detail, particularly when representing trips using demand responsive transport. Address point data can be associated with many other datasets such as house prices to represent the characteristics of the affected population. With modern computing capabilities using address point data would be easily feasible across England, not just for a smaller country like Wales.
- Willingness to travel – DfT say that in their connectivity metric “impedance functions are assumed to be the willingness to travel of the average person”. This seems to cut across the aim of the new approach for a more segmented understanding of travel choices so a more representative typology of the population would be more useful. In 2003, the willingness to travel by different population groups and trip purposes was reviewed in both Scotland and England for the accessibility/connectivity statistics, but with digital connections travel behaviour has changed markedly since then for each trip purpose. The 2003 impedance functions were derived using the National Travel Survey data, and the parallel Scottish analysis used Scottish Household Survey travel diary data. A key conclusion of this analysis was that different traveller groups had different behavioural characteristics. An updated analysis using not just the National Travel Survey but other travel diary data from household surveys could offer users of new accessibility and connectivity metrics a choice of deterrence parameters to suit the analysis being undertaken. Although ‘deterrence parameters’ in national statistics can be representative of some choices, it is always be important to check the sensitivity of the analysis for the target populations groups expected to benefit from the transport or land use change.
[1] To support the technical design of the new metrics in the late 1990s several projects were undertaken for government departments and referenced in the DfT guidance.
[2] For example as reported in the 2013 Parliamentary Inquiry
[3] The methodology note is no longer available at the DfT website so has been made available here.