The Number of Rough Sleepers Hits A New High - But How Are Homeless People Counted?

By James O Malley on at

Pic: Steven Lilley on Flickr.

New figures released today reveal that the number of people sleeping rough in Britain has risen to 4000 per night - a staggering 16% increase on last year, and terrible news.

The BBC, which reported the story, quotes Jon Sparkes, the chief executive of homeless charity Crisis slamming the new figures. He says "Behind these statistics are thousands of desperate people, sleeping in doorways, bin shelters, stations and parks - anywhere they can find to stay safe and escape the elements.”

I couldn’t agree more with his analysis - it’s pretty damning that our society is configured in such a way that people are forced to sleep in the street.

But this got me thinking: How are the numbers of people sleeping rough actually counted? They can’t actually know there are 4000 people sleeping rough, can they? Rough sleepers are almost by definition hard to track as they are living outside of many of the societal institutions and structures that might normally be used as proxies when counting people. If the government wants to collect data on the number of people living in an area, they can count the census data, or ask the local council for its list of people paying council tax, and so on. But surely rough sleepers would fall through these cracks? How do you count people living off the grid?

The answer is surprisingly interesting - and reveals an added layer of noise and complexity - which appear to add some rather large error bars to the shocking headline figures.

Counts and Estimates

For the Department of Communities and Local Government’s figures, since 2010, the government has required each local authority in England to submit a single number of the number of rough sleepers in each area - these are then totalled up.

But this number can be collected in one of two ways: Either a “count”, or an “estimate”. And while I certainly can’t think of a better system for doing it, it also appears to lack a little of scientific rigour you might implicitly expect.

Counts are what they sound like: Literally sending staff out to count people sleeping rough. According to the official guidance, this should be done between midnight and 5am - but adjusted to be later if need be, such as in city centres. Counters should also aim for areas where rough sleepers are known to be - so there’s no attempt at taking a random sample. And to be an official count, it must take place on a night between 1st October and 30th November, on a night with no unusual local factors - such as a festivals - taking place.

Who to count? Counters must look for people are “bedded down” or appear to be about to bed down, who have for example a sleeping bag or a tent. This latter criteria is actually important - apparently the guidance has been specifically broadened to include people who look like they’re about to bed down, so they don’t accidentally low-ball the figures.


Alternatively, local authorities can choose to submit an estimate of the number of rough sleepers - again, with the goal being to get a number that is a ‘snapshot’ of a night in October or November.

To get it, local authorities should host a meeting with groups who might have “intelligence” on rough sleepers - charities, faith groups, police, park rangers, health services and so on. Somehow, they’re all supposed to come up with a number they all agree on.

In London this process has been entirely formalised into a network known as a “CHAIN” - the Combined Homeless and Information Network. It’s administered by homeless charity St Mungo’s and shared with a number of these sorts of groups and organisations. Apparently it is built on the Salesforce platform - the CRM system which is more commonly used by companies to manage customer data.

So what does this mean? I’m not attempting to criticise and minimise the homeless figures by pointing out the methodology - but I am a little sceptical of how much meaning we can extract from this one headline figure.

Given that we’re talking about a relatively small number of people (4000 people across all of England), it strikes me that noise in the guesswork could conceivably contribute to such a large swing. The overall figure is essentially an aggregated set of educated guesses.

But even if we are sceptical, the 4000 number is also not entirely meaningless. As there are indeed rough sleepers out there who we should be doing more to help. Counting them - however spuriously - is surely a good first step towards recognising that there is a problem that needs to be solved.

You can donate to Crisis on its website.