Using published statistics to support a needs case

Published national statistics are the least expensive credible evidence a planning submission can carry, and they are also the easiest to misuse. Office for National Statistics data, the census and the published projections can support a needs case, a transport chapter or a social infrastructure argument, but only if the geography matches the catchment you are talking about, the release is the current one, and the citation is precise enough that an officer can check it. A number without a dataset name, a geography and a release date carries no weight at all, and a case officer is entitled to ignore it.

Start from the question you have to answer, then choose the dataset. Doing it the other way round produces a chapter full of numbers that prove nothing.

Which datasets are usable as evidence

Four families do most of the work in planning submissions.

Census data gives you the most detailed cross-tabulated picture available: household composition, tenure, occupancy, car and van availability, method of travel to work, economic activity, health and disability, ethnicity, and qualifications, down to small-area geographies. It is a full count rather than a sample, which is why it stands up to scrutiny.

Population and household projections give a trend-based view forward. Household projections in particular sit under the national method for assessing housing need, and their base year matters because the method specifies which set to use.

Labour market and business data covers employment by industry, workplace population and business counts, which is what an employment land or economic argument runs on.

Small-area indices, including the index of multiple deprivation published by the department, give a ranked picture of relative deprivation by domain, useful in health and social infrastructure chapters.

Geography levels and choosing the right one

This is where most submissions go wrong. Statistical geographies are nested, and the choice changes the answer.

Two rules. Use the smallest geography that is statistically reliable for the point you are making, and always give the comparator: the same statistic for the district, the region and England. A figure with no comparator invites the question you did not answer.

Beware boundary changes. Small-area boundaries were revised between censuses, and ward boundaries change with electoral reviews, so a 2011 figure and a 2021 figure for the same named area may not describe the same land. Say which boundary version you used.

Household projections and their caveats

Projections are trend-based, not policy-based. They extrapolate recent demographic behaviour and do not account for a local plan, a new employer, a university expansion or a change in migration policy. That makes them a legitimate starting point and a poor conclusion.

State three things whenever you use them: the base year of the projection, the variant if you have used anything other than the principal projection, and the period you have taken. Then say what the projection cannot capture in your case. A consultant who names the limitation is more persuasive than one who presents a projection as a forecast, and inspectors notice the difference.

Where housing need is the issue, the national method sets out which datasets to use and how, so follow the method and show the arithmetic rather than substituting your own dataset choice. Our note on finding housing land supply and delivery test figures covers the supply side of the same argument.

Travel to work data for transport evidence

Census travel to work data supports mode share assumptions, and it is the dataset most often quoted without its caveat. The 2021 census was taken on a day when working patterns were disrupted, which affected recorded method of travel to work and the workplace population. Using it for mode share without acknowledging that is a weakness an opposing consultant will find.

The defensible approach is to use it alongside something else: local count data, the trip database used for trip generation, or an operator's published patronage figures. Where you rely on census mode share, state the release, the geography and the caveat in the same sentence.

Workplace population data has a related use in Transport Assessment work, for sense-checking whether a site's catchment behaves like the district as a whole. It rarely does.

Deprivation and health data in a needs case

Deprivation indices are ranked, not measured, so the correct statement is that an area falls within a given decile or percentile nationally, by domain. Saying an area is "deprived" without the domain and the rank is not evidence.

For a Health Impact Assessment or a social infrastructure chapter, combine the ranked deprivation position with census health and disability data for the same small areas, then relate both to what the scheme provides. The argument that works is specific: this catchment sits in the most deprived decile for the health domain, the census records a higher proportion of residents with a long-term condition than the district average, and the scheme delivers the following facilities within a stated walking distance.

Do not claim a health outcome. Planning evidence supports provision, not results.

Citing a dataset precisely

Every statistic needs five things: the publisher, the dataset or table name, the release or edition date, the geography with its code, and the figure with its unit. Table references are worth including, because that is what allows an officer to reproduce the number.

• publisher, for example the Office for National Statistics or the relevant department
• dataset and table name as published
• release date or edition, since revisions happen
• geography name and its statistical code
• the figure, with the unit and the base it is a proportion of

Put the citations in a table at the end of the chapter rather than in footnotes scattered through it. It makes the evidence base checkable, which is the whole point.

Where a statistic proves less than it seems

Three traps worth naming.

Small numbers. In small-area data, a percentage derived from a small base moves wildly and is sometimes suppressed for disclosure control. Quoting a 40 per cent figure derived from a base of fifteen households is not evidence of anything.

Correlation offered as need. A statistic showing an ageing population does not by itself establish a need for the particular housing product you are proposing. The chain from statistic to conclusion has to be written out.

Currency. A dataset that was current when the report was drafted may not be when the committee meets, particularly across a long determination. Date the evidence base and refresh the headline figures before committee.

Used properly, published statistics cost nothing and make a case look prepared. Used as decoration, they invite an officer to test them, and in a finely balanced case that is not the test you want. Draft the narrative sections in minutes on PlanScope, then build the evidence table yourself so every figure has a source.

This is general guidance, not legal advice; check the position with your local planning authority or a qualified professional before relying on it. PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval.