Fullfact has been looking at various media reports on accident statistics in 20mph zones. In essence official statistics show that the number of accidents in 20mph zones has been rising. The conclusion drawn by some media is that this shows that 20mph zones don't work. In reality Fullfact finds that the figures take no account of the expansion in the use of 20mph limits. So an increase in the number of accidents tells us nothing about the effectiveness of 20 mph zones.
To make any real sense of the raw data we would need to know how many roads have a 20 mph limit, and how much traffic they carry. It seems that the DfT can't help, because they don't know.
I wondered if the OSM data could provide any more information.
OSM contributors have added speed limits to about 9% of the UK road network. As a result, about 3,500km of road currently show a speed limit of 20 mph. That represents just under 1% of the whole road network, and just under 3% of minor urban roads.
We must bear in mind that some 20 mph zones will not yet be added to the database. And OSM data isn't always accurate, so it's also possible that some roads have been marked incorrectly as having a limit of 20mph. We could make some assumptions about roads that don't yet carry speed limit information, but it doesn't help a lot here. It is normally safe to assume that OSM data is incomplete, but less likely that data is inaccurate. So we can make a reasonable assumption that the true extent of 20mph roads is going to be higher than the recorded figure. In other words, we can use the data to estimate that at least 1% of the UK road network now has a 20mph limit, and 20mph limits now represent almost 3% of minor urban roads.
Meanwhile, according to ONS/DfT data, 20mph zones account for just over 1% of all accidents, and 1.6% of accidents in built-up areas. In other words, the proportion of accidents in 20mph zones seems no higher than the proportion of 20mph roads. If we just look at built-up areas then the accident rate in 20mph zones is about half what we might otherwise expect.
Perhaps more importantly, the DfT data on its own shows that the mix of accidents is different under different speed limits. With a 40mph limit, one in 100 accidents involves a death; at 30mph it is one in 200, and at 20mph it better than one in 300. In 20mph zones the number of deaths on the road could be as little as a tenth of the level elsewhere.
All this doesn't get us to complete and reliable figures, it can take no account at all of traffic volumes, and there are surely some more complex causes and effects underlying the data. Nevertheless, if the limited numbers that we have tell us anything, it is that accidents in 20mph zones are in line with the obvious hypothesis that lower speed limits are associated with fewer, less serious, accidents.
More here - http://www.sustrans.org.uk/resources/in-the-news/are-20-streets-really-less-safe