Idea: A re-usable framework for “Who Should I Vote For?” tests

Those “Who Should I Vote For?” sites are, I think, a great idea. Many of them fail though. They;

  • often miss out huge policy areas (e.g. Europe)
  • often just copy/paste from manifestos making it easy to see which party is which
  • over simplify questions or provide limited answer choices that don’t always match views
  • become out of date and are slow to be updated

I’d like to propose we (anyone with the inclination to help me) build a framework for creating such tests that we can use again and again.  I have an idea how we can do it.

It needs to be;

  • Open source and collaborative (obviously!)
  • Easy for anyone (i.e. non-developers) to contribute to improving the policy questions

Here’s what I propose;

1. Agree/disagree questions

In order to determine the political leanings of the user without revealing actual policies we should use questions to which the answer has to be one of five choices;

  • Strongly Agree
  • Agree
  • Neither
  • Disagree
  • Strongly Disagree

Each political party will then be attached to one of those five answers. For example, consider the proposition “The UK should leave the EU” to which the matching would probably be;

  • Strongly Agree – UKIP
  • Agree – Conservatives
  • Neither
  • Disagree – Labour
  • Strongly Disagree – Liberal Democrats

For each matching answer, a “point” will be given to that party, and at the end of the survey those points will be used to generate percentage “matches” to each party. E.g. if the user was Nick Clegg you’d hope he achieved a 100% match to the Liberal Democrats, and lesser percentage matches to some other parties too.

2. Markdown for question and answer generation

So that anyone can help to contribute, the questions and the matching of parties to answers should use Markdown. The application would then parse the markdown to generate the actual questions and calculate the points based on the answers given.

This is an example of how that Markdown might look.

It’s not great for parsing, but should be straightforward for contributing too. I’m very open to ideas on a better format, in any case.


So, what do you think? Is it a good idea? Is it workable? Will you help me build it?



, ,




2 responses to “Idea: A re-usable framework for “Who Should I Vote For?” tests”

  1. Paul Robinson Avatar
    Paul Robinson

    Great idea – but you can see from your example why these sites/apps get into such difficulty. I wouldn’t describe the Conservative Party’s position on that statement as ‘Agree’. The Party is quite notorious for being split on this issue. There are a growing number of ‘rebels’ [ie they disagree with the party’s position] who would “Strongly Agree” with that statement. But the leadership’s official position is highly nuanced: wanting drastic reform, but *might* leave if it can’t get it. But then there are also others (ie Clarke or Heseltine) who may still favour UK entry into the Euro currency. Likewise I suspect behind the scenes the Labour leadership’s position on this is exactly as you say (ie “Disagree”. But Milliband’s speech on the subject last month was even more nuanced than the Conservatives – they are clearly pro-EU but too afraid to say so publicly. That is real challenge of listening or reading politicians opinions on different subjects: trying to convert their weasel words into a one-word summary. It’s why they so rarely respond to Paxman with a simple yes/no answer.

    1. philipjohn Avatar

      I thought someone might point that out 😉 My thinking was that “on balance” they are more out than in and are too scared to be as open about that as they could be, in the same way Labour are.

      Anyway, back to the idea, which I think has a solution of sorts in two ways…

      Given the setting of those party-answer associations would be based on contributions, we’d probably need a similar system to the OpenPolitics Manifesto where they have to be voted through. Debates would ensue and consensus reached.

      Secondly, you’ll notice I’ve also added references into the MD format. This is designed to be the “evidence please” of this project. Changes to answers would require supporting evidence. These may well come from analysis of MP voting records for example.

Leave a Reply

Your email address will not be published. Required fields are marked *