Log in

Australian Political Debate
How to vote for what's best for YOU 
24th-Nov-2007 07:24 am

This is a nonpartisan 20 question questionaire which helps you to match your own priorities to those of your local candidates.

"How it works
We've asked every candidate in each electorate to answer twenty questions that cover a wide range of issues that are of interest to voters. Now you can answer the same quiz and compare your views to the people you are considering voting for."

This is a serious guide, not a fun meme, and it may well help you to break away from kneejerk partisan politics and direct your vote. I strongly urge everyone to do it before heading off to your polling booths today.
24th-Nov-2007 02:32 am (UTC)
Whereas I strongly urge everyone to completely disregard the results of this quiz and only consider it like another meme for the reasons I have explained here and here.

Yes, they had a professional pollster helped them craft the questions and the method of matching the responses to those of the candidates. Unfortunately the technical implementation (the code and the data sources it uses for the candidates who did not respond) remains under a cloud. They may not have intended to be misleading or deceptive in their practices, but that appears to be the result of what they've produced.

Contrary to the site's claim, the process is not transparent and the implementation of ranking non-responsive candidates is a possible breach of section 329 of the Commonwealth Electoral Act 1918.

Yes, I have raised this with the site maintainers, along with the AEC (see my second post above for details).

Additionally, there is an interesting discussion of the nature of the (very loaded) questions asked in the quiz in a friend's journal.

I'm fairly confident, following my own testing of the site, that all results, regardless of the electorate selected, lead to either a vote for the ALP or preferences being directed towards the ALP. If that's what people want, then fine, go for it; but don't think for a moment that the quiz will actually result in any alternative flow of preferences.
24th-Nov-2007 05:16 am (UTC)
You don't seem to have any actual concerns about the test other than you don't like it, because Liberal doesn't like it.

Every single candidate was sent the same set of questions. It is not the fault of Getup that almost no Liberal or National candidaters bothered to reply. Their lack of motivation is very unlikely due to percieved bias, Fred Nile's group aren't exactly treehuggers and they're very well represented on the database. The lack of motivation may well be explained by Liberal's opinion of the people hosting the questionaire however.

The results of the test not only give the highest ranking candidate to your answers, but to the percentage they scored. This means that even if your local conservative didn't bother, you can still note the low score Labor came in with and avoid them

I suggest your own biases have blinded you here. But you can keep telling yourslef it's because you're so much more clever than everyone else if you prefer.
24th-Nov-2007 05:48 am (UTC)
Interesting assumption there. I'm neither a Liberal nor a Labor voter (although I am in a safe Labor seat, so that point is fairly moot).

My objection is that in order to rank the candidates who did not respond their system must be assigning them some form of arbitrary value, even though their own FAQ states that to supply the non-responsive candidates with answers would be illegal (section 328 of the Electoral Act). Meanwhile the mechanics of the quiz requires a value of some kind in order to rank those candidates and create a formal ballot. So either they're assigning arbitrary values to those candidates or not (in which case it would be impossible to generate a formal ballot).

Those candidates and parties (the ALP supplied a policy document to cover all their candidates) which did supply responses had those responses converted to the data set (most likely a numerical value assigned to each of the five possible responses to each question, barring the last which is to be used for tie-breakers).

If that's the case, then the value assigned to non-responsive candidates should be either null or zero (ideally the former, to differentiate it from a comparable numerical value) for all "responses" to each question. However, if this null value had been correctly assigned to each of the non-responsive candidates it should be impossible to obtain a ranking of the non-responsive candidates, even amongst themselves.

Yet every completion of the quiz assigns these candidates a preferential rank after those candidates who supplied answers. Ergo, in spite of the statement that HSIV cannot legally supply answers for those candidates, they (or their web developers) have done precisely that behind the scenes and concealed it by running all the code on the server end (it's not possible for a site visitor to observe what that code is doing or what data source it uses, or even if that data source matches the candidate answers for those who did respond).

Some of what can be observed, as I've sketched above and on my own LJ, indicates some of what is actually taking place on the server, but not all of it and certainly not the specifics of the data set used for the non-responsive candidates. There's enough evidence to show that there are serious discrepancies between what they claim the quiz is able to do (legally) and what it actually does. Which, of course, is what prompted my emails to GetUp requesting access to the code and date sets for review. They have yet to supply any of this, although they did respond to the second email late yesterday afternoon, albeit only after I copied the AEC in on the email.

Anyway, that's beside the point. The issue isn't that some responded and some didn't, the issue is how they're handling the non-responses in order to produce a dynamically generated formal ballot for each visitor and whether that is being done appropriately.

Obviously I don't think it has been handled appropriately, although I suspect it will end up being a case of the quiz design methodology being valid, but the implementation of it not being able to handle that methodology correctly.

It may be that the data set for the non-responsive candidates was assigned by a website developer who was not aware of the legal ramifications of what s/he had done. Not that that's any kind of defense, though, since it indicates that no one who was aware of the relevent legislation inspected the technical implementation and they're still authorising the results of that implementation, so they've accepted responsibility for it.
This page was loaded Feb 24th 2017, 2:50 pm GMT.