Using census travel to work data in a transport statement

Census method of travel to work data is the standard baseline for mode share in a transport submission, and it is the dataset most often quoted badly. It tells you how residents of a small area travelled to work on one day, for commuting only, at a geography you choose. It does not tell you how people will travel to a new supermarket, and the 2021 figures carry a caveat large enough to swallow the whole analysis if you ignore it.

Used properly it still does a job nothing else does: it gives a defensible, locally specific, freely available starting point that a highway officer already trusts.

What the census gives you and at what geography

The Office for National Statistics publishes method of travel to workplace as a census topic summary, counting residents aged 16 and over in employment by the mode they used for the longest part of their commute. Categories cover driving a car or van, passenger in a car or van, bus, train, underground and light rail, bicycle, on foot, motorcycle, taxi, other, and working mainly at or from home.

The data comes in two framings and mixing them up is the commonest error. Residence-based tables count people who live in an area, which is what you want for a residential scheme. Workplace-based tables count people who work in an area, which is what you want for an employment or commercial site. Origin and destination flow data, showing where people travel from and to, is published at coarser geographies than the headline tables.

Geographies run from output area, the smallest building block, up through lower layer and middle layer super output areas, wards, built-up areas and local authority districts.

Choosing the level to use

Smaller is not automatically better. An output area contains a few hundred people, so proportions move sharply on small counts and disclosure control adds noise. A single output area showing 8 per cent cycling may reflect a dozen people.

The approach that survives scrutiny is to present two or three levels together: the immediate area, a wider local area, and the district. Where the site's small area figure differs from the district, you then have to explain why, and the explanation is usually a real feature such as a railway station, a university, a hill or an absence of bus service.

For a site on the edge of a settlement, be careful which small areas you take. Output area boundaries follow population, not catchments, and an area that stretches into open countryside will mix two very different travel patterns.

Mode share as a baseline for assumptions

Census mode share does not generate trips. Trip rates come from surveyed sites in a trip rate database, and the census tells you how to split those trips between modes for the local context.

The usual sequence in a Transport Statement is: establish total person trips from the trip rate database, apply a locally derived mode split, convert car driver trips into vehicle movements, then test the network effect. Where the scheme is large enough to need a full Transport Assessment, the same baseline feeds distribution and assignment as well.

Three limitations to state openly rather than hide:

• The census covers journeys to work only, and commuting is a minority of all trips
• It records the mode used for the longest part of the journey, so a walk to a station is counted as rail
• It is a snapshot of one week in March, and seasonal variation is invisible

State them in the report. A highway officer who spots an unstated limitation stops trusting the rest of the analysis.

The age of the data and how to handle it

The census runs every ten years, so by the middle of a decade the baseline is old. For most purposes that is manageable, because relative mode shares between neighbouring areas move slowly.

The 2021 census is different. It was taken in March 2021, when working from home was widespread under public health restrictions, and the share recorded as working mainly at or from home rose sharply compared with 2011. The Office for National Statistics published guidance on comparability for exactly this reason.

That leaves three sensible options. Use the 2021 data and state the caveat plainly. Use 2011 data for the structure of mode share among those travelling, with 2021 for context. Or reprofile by excluding home workers from the denominator, so you are comparing mode shares among people who actually travelled, and say that this is what you have done.

Whichever you choose, do not present a 2021 car driver share as a forecast of car use without adjustment. It will understate it, and a highway authority that notices will discount your whole traffic case.

Home working and its effect on the figures

Home working is not a temporary artefact of one census. Patterns have settled somewhere between the 2011 and 2021 positions, and the honest answer is that nobody has a locally specific dataset that captures where.

For a residential scheme that generally works in the applicant's favour, and it is tempting to lean on it. Resist. Claiming a reduction in peak hour trips because of home working needs evidence for that location and that housing type, and the National Travel Survey or a local authority's own monitoring data is a better source than a census table taken during a lockdown.

Where the scheme includes home working space as a design feature, say so and let the design carry the argument rather than the statistics.

Combining census data with site surveys

Census data establishes context. Site-specific surveys establish behaviour. The strong submissions use both.

1. Manual classified counts at the site access and the nearest junctions, on neutral days outside school holidays.
2. A walk and cycle audit recording actual infrastructure quality within 400 m, 800 m and 2 km.
3. Public transport service data taken from current timetables, with walk distances to stops measured on the ground.
4. Parking beat surveys where on-street capacity is contested.
5. Comparison of the census mode share with any monitoring data from an existing travel plan nearby.

Where the census and the site survey disagree, explain the disagreement. That paragraph is often the most useful thing in the document, and it is where a Travel Plan target should come from rather than from a round number chosen for its optimism.

Citing the dataset correctly

Give the dataset name and table reference, the geography and its code, the census year, the source as the Office for National Statistics, and the date you downloaded it. Put the extracted table in an appendix rather than only a chart, so the officer can check your arithmetic.

Census outputs are published under the Open Government Licence, which permits reuse with attribution, and the same conditions apply to most public datasets used in planning submissions. The practicalities of reusing planning data under the Open Government Licence are worth knowing before you lift a map or a table into a report.

Keep the download in the project file. When a consultee queries a figure eighteen months later, the version you used matters more than the version currently online.

This is general guidance rather than legal advice, and transport evidence requirements are set by individual highway authorities, so check the position with the authority or a qualified transport planner before relying on it.

PlanScope drafts transport documents structured to UK planning policy, with survey data and trip rates flagged where they are missing. PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval.