Each ward in London with our estimate of %age of voting-age population registered to vote. Red: <75%; Yellow: 75%-88%; Green: >88%.
London Voter Registration Prototype Meme Generator Thingy
This week, we built takepart.london. You pop in your postcode and we tell you how many people are registered to vote in your ward. Then, because we’ve checked it against the ONS population estimates, we give it to you as a percentage, and stick it on a nice local images from geograph.
The point was that folks who are already registered would use the site, share the image and become ambassadors to their friend who might not be registered. Their friend would find it in the first place, because it’s shared on facebook, not via a TV or radio channel that they’re not watching. Their friend would pay attention thanks to the locally relevant image (it has more salience than another old white guy talking about numbers). And then their friend would see the stat (typically something like 80% of people in this area are registered to vote) and would experience a social norming effect: they wouldn’t want to be left out. It’s fairly quick and dirty behavioural science, but it might have held up in trials had we had the time to run them.
The site’s had about 1,500 visitors, thanks particularly to a big helping hand from Sadiq Khan. It’s fair to say it’s not gone terribly viral. People seemed to genuinely enjoy the site, trying various postcodes, checking out the league table. But we were hoping people would share their unique local image, which would encourage more people to share theirs, etc. Perhaps the images weren’t attractive enough? See Joe’s area picture below. More time would have allowed us to manually check some of these to ensure we’d got the most compelling image, and applied the filters to make them unique and atypical. More time would have also allowed us to play with the UX: moving the share buttons and ensuring the twitter image share worked. Or perhaps we needed more cat photos.
So while the website got lots of use, it’s not clear whether — as was the point — we are having any social norming effect in local communities that might not be fully registered. It’s extremely hard to measure, of course: the people we are trying to reach aren’t the people logging onto the website, they are the unregistered friends of the people who share the images. And short of being Facebook, knowing how many people saw images shared on facebook is impossible. And knowing how many of them are unregistered is doubly impossible.
We may have ended up simply providing lots of interest to politics geeks looking at wards where fewer folks are registered. Which wasn’t the aim. But still useful. And if anyone’s out on a registration drive this weekend, perhaps the league table can guide their efforts.
Our London First partner reported lots of feedback that people would like this for the whole of the UK, which seems like a good plan. If councils would open up more data, this is the kind of thing we could do. And this kind of data might motivate people to work smarter and more efficiently to get more people involved. If I don’t know what the registration percentage is like in my area, I’m not going to be motivated to do anything to bring the numbers up. The more local the data, the more I’m likely to care: if it’s my street, estate or block of flats, then better still.
So hopefully this has been a useful experiment that shows what’s possible.
Ed Saperia, always willing to provide helpful critique towards these things, rightfully pointed out that people care about issues. And what we’ve done here is assume that people care about the issue of democratic engagement; that registering to vote is in itself an important ‘issue’ that we want to act on. Of course, Democracy Club folks believe that, and many others too, but it’s rare. It’s a bit of a meta-issue. Most people don’t get passionate about voter registration figures. They care about schools and hospitals and potholes. So a non-partisan approach like this may be doomed to fail unless we could have connected in a non-partisan way to local issues. Goes the theory. Which reminds Joe of his favourite ‘remember to vote’ advert.
Lastly, we’ll publish the research that led to this effort soon. Soooon.
More polling stations fun
There was much excitement when we realised that polling districts were included in the latest release of Ordnance Survey’s AddressBase (UK-wide mapping data). We figured that if the data was reliable, we could have crowdsourced the polling station locations (from the Notice of Polls that councils are legally required to produce), then we could have worked out which station went in each district, and bingo, we’d have magicked coverage of much of the UK.
But alas, it wasn’t to be. Poking away at the data, such as testing it against a friendly council’s information on their polling districts for 23 June, and a chat with Ordnance Survey (OS) revealed that the data just isn’t reliable enough. It’s only as good as OS were able to glean from councils and districts do change relatively regularly (at a minimum every five years). But this must be solvable in future.
Instead, then, we’re back to where we started. If we want to be able to swiftly, accurately tell users where their polling station is, we need to keep pleading and working with councils to produce up-to-date address-to-polling station lookup tables. The London boroughs all did this for the 5 May polling station finder, which has since been shuttered, so we think they’re in the best place to provide it, and with our London First partners we’re chasing all over London.
There were elections this week too
Seven council elections were held yesterday. We didn’t cover them on WhoCanIVoteFor, but not because we didn’t want to. We need to create the system and platform to create the database of all elections.
It’s not just about the satisfaction of achieving universal coverage, or the social impact of helping voters in those seven areas, but by-elections like these also represent the opportunity to test and experiment.
For example, does anything we do affect turnout? It would have been fun to take three of those seven elections at random, gotten the candidate data, mobilised volunteers in those areas to get photos of them, get all their social media details, email addresses, put up posters locally, maybe print some leaflets, and push the information as far as we could. Then we could have looked at turnout trends across the seven and see if our work made any difference. Far from a RCT, but could have been fun to try.
The only way we can do things like this is by knowing about all the elections in advance. Lots of enthusiasts do collect this kind of information, but we need it in a systematised and machine-readable format. We’re really looking forward to doing some work on this over the next three-to-six months, if we can raise funds for it.
On that note…
We’re currently plotting applications to MassChallenge, a curious non-equity accelerator thingy, and going back, with a stronger application to ODINE, a European open data grant funder. Hints, tips and thoughts very welcome!