Alternative Vote Pie visualisation
A project intended to help people visualise how alternative voting would work in their constituency (or in general).
The visualisation starts off by showing you:
- An inner pie chart, which breaks down the proportions of first-choice votes
- A second outer ring of the pie chart shows how the second-choice votes break down for voters of each party
The predictions at present are based on opinion poll data together with the results of the 2010 election and a simple 'swing' model. The breakdown of second-choice votes are predicted based on national data (eg "20% of labour voters will put lib deb as their second choice").
The tool then walks you through each round or 'runoff' as may be necessary under alternative vote.
- When a party is eliminated, the second choice votes of voters from that party are moved from the outer ring of the pie chart, into the inner pie, showing visually how this process works. (NB: the plan was to animate this in a nice visual way, but there was insufficient time on the day; see below)
Ideas for taking this project forward
As stated the plan was to visualise the redistribution of second-choice votes at every stage via a nice animation where the outer "second choice" slices of the pie slide inwards and become part of the inner pie slice. This will take a bit more time though.
The predictive model used for both first and second-choice votes could be improved significantly with more time and/or data; the tool could also use third-party predictions if people like the BBC produce them. The opinion poll data used was also rather old, I just used what I was able to find in a convenient format on the day but more recent data should be available.
The UI could also be improved to explain more about the visualisation and the AV process, better usability for selecting constituencies and for moving between rounds of voting, better rollover tooltips on the various chunks of the pie, etc.
Also, the visualisation currently lumps together all minority parties as "Other"; this could potentially be changed if it bugs anyone.
Estimated costs for taking this project forward
To develop a significantly improved predictive model for UK election results under alternative vote: you'd probably need to pay a reasonable daily rate for a consulting statistician, for anything from a day or two up to a few weeks depending on how deep you want them to get into it, how much and what kind of data is available to be used. If you (not sure who exactly I'm pitching to here?) can source good data and analysis from a third party or you have stats people on staff this could save on the costs though.
Personally speaking I'd be happy to have a crack at either if anyone's interested. In the former case, while I have some knowledge of the subject I'm not a professional statistician so might take a bit longer and/or not charge as much as a pro.
I might also make some further improvements for my own amusement if I get time, although no guarantees there :)
About the data used for this project
Data was sourced from http://refusingthedefault.blogspot.com/p/alternative-vote-swingometer.html and included:
2010 election results by constituency
Data on share of first and second-choice votes derived from 3 opinion polls from later in 2010 (Populous/Times 23 June, Yougov/Channel 4 28 June, Yougov/Spectator 6 July).
Add photos by tagging them with rewiredstate:project="votegeek-alternative-vote-visualisation"