Posts by George Darroch
Last ←Newer Page 1 2 3 4 5 Older→ First
-
Hard News: The Language of Climate, in reply to Che Tibby,
just don’t be part of the problem. do all the things that you want everyone to do to save the world, so if they ever start to catch up you’re already showing the way. and if need be you can help them
Agreed. But it's okay to drive to an anti-oil protest.
-
I don't talk about climate change anymore. I'm inarticulate with rage.
-
Hard News: Gower Speaks, in reply to Henry Barnard,
Especially as NZF and Green are, as Ben puts it, `positively correlated’?
They are correlated, but that actually makes their voters different to each other. When the vote of one goes up, so does the vote of the other - that's because their vote reflects either National or Labour weakness, and they tend to gain from their disaffection.
-
Furthermore, the panel provided "no documentation for the method used by the panel to moderate the proposal evaluation results".
Which is, to say the least, very interesting. The rest of the article outlines irregularities and possible irregularities. There isn't much robustness in this process, and I wonder what they were expecting the response of the sector to be.
A proposed merger between the helpline services is being pushed through, and I hope the process has a much greater degree of integrity than this one.
-
Download the code and play with the numbers. It’s a lot of fun.
I would start making other assumptions in that case. Firstly, that National would take a hit of a few percent, made disproportionately from young women and ‘aspirational liberals’, and that the Conservative Party would gain votes, some from NZFirst and others from National. Labour and the Greens would also be beneficiaries.
However, none of this moves things sufficiently towards Labour, or away from National. NZ First has enough catchment in the middle at the moment to be the main beneficiaries. Organic growth of 4-5% is necessary for Labour to start looking like they’re in a position to win with the Greens outright.
Peter Green’s model responsibly assumes that the Green Party's ability to turn out polled ‘Green-voters’ is similar to past elections. I’m in the Green Party and we are working towards both higher support and higher turnout, and hopeful that we can achieve that – but it isn’t a given, and I wouldn’t expect anyone to assume it.
-
More playing… taking away an Epsom ACT electorate win takes National’s chances of an outright victory down from 40% to around 33%. That’s maths.
-
I decided to run the simulation with the figures produced by Peter Green’s poll aggregator (which uses a generalised additive model(GitHub)), and made assumptions about voter enrolment (3,150,000; slighly below average increase) and turnout (77%; slightly below the average of recent turnout). Just for kicks I set the loop to 100,000 simulations, and ran it once.
national_led
0.39955nzf_decides
0.60045This may be an abuse of both models, and an extrapolation too far. However, it speaks to a significantly different narrative than the one being communicated, one which adheres to a reality in which:
“you’ve got to stick with the information that you’re given. And you’ve got to stick with it from month to month and pretend that there is nothing else out there, for the sanctity of that information.
I don’t judge him. A man must have a code, and Gower sticks to his.
-
But most New Zealanders expect Labour to do something so utterly compellingly stupid before the election that they will fail to win
This is a way to explain it.
I suspect it’s actually more that most New Zealanders have been conditioned by the polls and reportage thereof to believe that National’s victory on the day is assured.
As is this.
It's also possible that A is the cause of B, or B is the cause of A.
-
Snap.
Quite happy to help if for any reason things are difficult (95% of the time it's as easy as pie, but sometimes it isn't).
-
Hard News: Gower Speaks, in reply to danduran,
This might be a dumb question, but how does one run this at home? I guess I have to install a language or something?
Go to http://www.r-project.org/ and install R. This is pretty easy, and can be done in about ten to fifteen minutes without specialist knowledge. I'd also go to https://www.rstudio.com/ and install R-Studio, because it makes this work nicer and easier if you're going to do more of it, but it's not at all necessary.
Then it's simply a matter of cutting and pasting the code at https://github.com/thoughtfulbloke/OneVoteTwoVoteRedVoteBlueVote/blob/master/oneVoteTwoVoteRedVoteBlueVote.R into your R session and pressing enter. Because it is the first time you're using it, it will require the installation of packages, but this is fairly automatic. Just choose the Auckland server and they'll install directly.