How we measure travel determines what we find – are we using the right metrics to address today’s and tomorrow’s challenges?

DATA AVAILABILITY is a central theme in transport analysis and policy-making nowadays, with rapidly increasing volumes and ever widening sources of data threatening to swamp our ability to process it – adopting AI is seen as the solution for the transport sector.

But, aside from issues about the quantity and quality of data, a less considered question is about whether we are measuring the right – or, most useful – things? This, of course, partly depends on the policy perspective. From the 1950s, the focus was on catering for the rapid growth in motor vehicles and so most attention was paid to vehicle trips and maximising vehicle movements (e.g. at junctions); with the gradual switch in policy focus from the 1970s onwards to encouraging modal shift from car, then most statistics were instead quoted in terms of person trips and maximising person capacity. And things have not moved on much, since then.

Since the early days of household travel surveys and four-stage modelling, the focus has been on measuring person trips and trip rates – the number of travel episodes in a day. This provides a fairly good indication of the frequency of ‘movement’ (particularly that which is mechanised) and has the merit of treating each episode as being of equal value, regardless of length or time (e.g. a short trip to a local shop is given equal ‘weight’ as a very long business trip). It is a very powerful metric, but has come to dominate analysis – for example, the very insightful annual Travel in London reports present most analysis of travel behaviour in terms of trips and trip rates - and is used in contexts where an alternative metric might be more informative.

A previous Deep Thinking contribution by Kristine Beuret and Terence Bendixson highlighted the differences in interpretation resulting from looking at walking stages instead of walking trips, with walking becoming the dominant mode when using the former metric.

For example, consider transport carbon reduction and modal switching. Local authorities have been encouraged to set targets for 50% of trips within their area to be made by active travel, and this has been trumpeted as making a major contribution to reducing CO2 emissions – 50% sounds a lot. However, emissions are much more closely related to mileage, from which perspective active travel makes a much smaller contribution – albeit that it has other major benefits, such as producing no air pollution and promoting physical and mental health. This distinction can be seen when looking at published data from the National Travel Survey for England (NTS, Table 0303). Nationally, in 2023, walking and cycling account for 25.9% of all trips, but only 3.8% of mileage (conversely, car drivers account for 33.8% of trips and 50.1% of mileage). So major carbon savings will not come from mode shift from car to active travel – unless this is achieved through switching to local destinations and a major reduction in trip lengths.

Trip lengths/distances provide a more relevant metric than trip rates in other areas too. Most evidently, for measuring things like vehicle emissions, fuel efficiency and for assessing vehicle reliability, plus measures of overall road traffic volumes – things mainly related to vehicle movement. But not in all cases – for example, when investigating road casualties.

Road casualties are usually reported either as absolute numbers (X KSIs per annum), or in relation to mileage (e.g. casualties per billion vehicle miles). It is claimed by Wes Marshall (‘Killed by a Traffic Engineer’) that the widespread use of mileage as the denominator in the USA came about as a result of lobbying by the motor car industry, to present the casualty statistics in a more favourable light. But risk is more related to exposure, which can better be represented by travel time. Other things being equal, a pedestrian travelling a mile is exposed to a greater risk than a car driver travelling a mile, as for the former this will involve around 20 minutes spent in a potentially hostile environment, compared to only 1-5 minutes for a car driver and passenger.

A different type of issue concerns the common denominator used for reporting travel metrics, which is conventionally the numbers, length or time spent per person. This is entirely logical but may not be fully informative – particularly in a post-pandemic world. This can be seen in London from the LTDS data. In 2010/11, 18% of people interviewed reported no trips on their travel day; this rose to 28% in 2022/23. Over this period, average trip rates per person dropped by a substantial 17%, from 2.48 to 2.07 per person per day. But, in terms of trip rates per traveller, the drop was less than 5%: from 3.02 to 2.88. This refines the narrative. Instead of the general conclusion that ‘on average, people are travelling less’, we can see that most of this is due to ‘fewer people are travelling’ on an average day. Increased working from home is likely to be a major factor, as well as an ageing population.

Finally, one issue that is hardly discussed is the usefulness/appropriateness of all this data on travel behaviour for policy making, in terms of identifying problems and determining solutions. Clearly, for practitioners providing specific transport services, such information is crucial in planning service patterns and capacity; but, more generally, what does it tell us about planning for the future? Not as much as we might think, I would contend.

For most phenomenon, it is clear whether an increase or decrease in the value of an indicator is ‘better’: fewer not more collisions, cleaner not dirtier air, less not more carbon emissions – these are all to be preferred. But, what about trip rates, trip lengths and time spent travelling? In each case, it would be possible to come up with plausible arguments in favour of either more or less as the preferred outcome. Thus a higher number of person trips could be seen as being positive if they are made ‘by choice’ and reflect a more activity-rich out-of-home lifestyle; but, could be viewed as a negative if they are ‘forced’ car driver chauffeuring trips due to lack of mode availability for other household members. Conversely, a lower trip rate could reflect a choice outcome – a preference for working from home and making extensive use of on-line services, or a more efficient trip-chaining travel pattern – or a negative outcome if a person is home-bound, due to the lack of accessible modes. Similarly, does a higher mileage indicate the enjoyment of an enlarged choice set, or a need to travel further due to the closure of local shops, hospitals, etc? And so on. In the background are little considered questions about the nature of suppressed and generated travel.

So, in summary, there is clearly no single metric of travel behaviour best suited to all purposes; we probably need to make greater use of distance- and time-based measures, and consider whether basing these on people or travellers will give the greater insight to a particular issue: it is ‘horses for courses’. Travel diaries and other data collection methods enable us to calculate trip distances and times, as well as trip rates, and we should use these metrics in appropriate contexts. But, they don’t help answer more fundamental questions as to whether current transport networks and land use patterns are enabling people’s needs and aspirations to be fulfilled.

If the standard travel behaviour metrics are not very informative for assessing whether current behaviour is in some sense ‘optimal’ and fulfilling peoples’ needs, what might be? This can be addressed from three perspectives.

First, from a trip/activity viewpoint, it is important to understand how we measure the number and types of primary out-of-home activities. This is not the same as counting trips. The number of trips required to carry out three non-home activities would be six if each involved a separate journey to/from home, four if the three activities were chained on one outing from home, and only two trips if all three activities were carried out at the same non-home destination.

This might provide one set of measures but ideally this would need to be augmented by additional information on the extent of in-home activity substitution (e.g. home working) and on the home as a trip attraction (e.g. receiving home deliveries and social visits). Such information could be added to conventional household travel surveys – probably by combining them with a time use survey format.

The second perspective would involve collecting more information on perceptions and attitudes: levels of satisfaction with mobility provision, ability to carry out desired activities out of home (taking into account mode availability, cost, comfort, etc,), measures of quality of life, etc. And the third perspective would switch the focus from mobility to one based on accessibility: can people meet their daily needs with minimum ‘friction’, either through local trips or digitally? Referring back to travel metrics, this inverts the significance of distance: implicitly, from a mobility viewpoint longer trips are often seen as more ‘valuable’ (in terms of revealed preference), whereas from an accessibility perspective, shorter trips are seen as being more ‘efficient’.

This whole area of transport metrics and policy interpretation would benefit from considerably more debate.

Peter Jones is professor of transport and sustainable development in the Centre for Transport Studies at UCL. Previously he was Director of the Transport Studies Group at the University of Westminster, and Deputy Director of the Transport Studies Unit at Oxford University.

This article was first published in LTT magazine, LTT914, 1 May 2025.