spinofflive
simon bridgesmemefeat

OPINIONPoliticsJuly 28, 2020

Memebers of parliament: the week in politics, told in memes

simon bridgesmemefeat

Welcome to memebers of parliament, a political column for people who just want the memes. 

In 53 days there will be an election and already there’s too much news. Too many politicians doing too many things. Who’s bothering to keep up with politics? Well, technically we all should because we live in a democracy and an informed vote is a good vote. But who has the time when there are memes to look at? Now you can have both. Join me every Tuesday to look back at the past seven days and think Jesus was that huge scandal only last week?

The grading is simple: YES, NO, or HUH?

Did you oversee an important piece of legislation this week? That’s a YES from me.

Did you send sexually graphic material to a teenaged person this week? That’s a NO from me.

Did you [insert most things that politicians do and say] this week? That’s a HUH? from me.

Julie Anne Genter – YES

What a week for Julie Anne Genter.  Genter’s Equal Pay Amendment Bill passed with unanimous support on Thursday night. The bill facilitates industry disputes to allow “female-dominated” industries (think nursing, education) to bargain for higher pay across all workers. Genter herself described it as “one of the biggest gains for gender equity in the workplace since the Equal Pay Act 1972”. And if that’s not enough, last week saw Genter’s (and Phil Twyford’s, actually more Phil Twyford’s but he’s been otherwise underwhelming) national policy statement on urban development published, which will prevent councils from restricting building heights or enforcing minimum parking requirements on urban developments. In other words, there are now fewer barriers in the way of higher density housing. This is good.

Simon Bridges – YES

Yes this is old news but last week saw enough failure to remind us all that Simon Bridges was in charge of the National Party honestly not that long ago. Remember when everyone was like wow Simon Bridges sux? Now National is polling at 25% and everyone’s like wow Simon Bridges was maybe great? He doesn’t have to worry about polls any more so he’ll head into this campaign doing piss-all and having the time of his life. The one good thing about being turned on by your own so-called friends is that going forward, no matter who stuffs up, whether it’s your opponents in government or your opponents in your own party, you get to feel smug.

Andrew Falloon – YES

Before this month, Andrew Falloon was an unknown, and in the world of politics, there’s nothing worse than being unknown. Now everyone knows Falloon’s name. He’s the man of the hour! You honestly couldn’t buy this sort of exposure. That has to be good. No such thing as bad publicity, right?

Andrew Falloon – OH GOD WAIT NO

News just in: there is such a thing as bad publicity and it’s called *checks notes* being found to have sent several young women, at least one of whom was teenaged, unsolicited sexually graphic material. Falloon was in a safe National seat. The bar to clear in order to keep being paid $163,961 a year wasn’t on the ground but it was bloody close to it. Unfortunately, Falloon had a shovel and not enough real work to stop him from using it.

Note: It has only been one single week since Andrew Falloon sent that fateful press release. I hate this year. 

Iain Lees-Galloway – NO

When I was at school, we weirdly had to sit in on our own parent-teacher interviews so my siblings and I miserably accompanied our parents to school one night, knowing that instead of simply getting told off at home, we’d have to listen to the teacher decide our fates in front of our eyes and then get told off at home.

Mine was first. It was OK but not great, and as we walked out I could sense that it would be a long night. My sister’s was next and hers was also OK but not great. Except hers was last, which meant my mum forgot all about me and instead spent all night “just having a talk” with my sister.

My sister was Iain Lees-Galloway resigning because of an inappropriate affair with a staffer 24 hours after the Andrew Falloon news broke. Can anybody say … lads lads lads?

Winston Peters – NO

Winston Peters dismissed an opposition MP in the house on Thursday by telling her to “scream out as long as you like, lady.” When he was told by the speaker that “lady” wasn’t appropriate, Peters demurred, claiming that were he to use the word “woman” he would consider it “thoroughly a put down”. Putting aside the fact that Peters could’ve used literally any term at the end of that sentence and it would’ve remained incredibly patronising, it’s very funny to me as a LADY to think old Winston Peters is so progressive that even woman is a derogatory term.

Helen White – HUH?

Last week, RNZ asked pedestrians on Ponsonby Road (squarely within the Auckland Central boundaries) to identify both Helen White and Chlöe Swarbrick, the Labour and Green candidates respectively in that electorate.

No one recognised Helen White, despite her contesting the same seat in 2017.

Sorry to this woman.

With less than two months until the election, something must be done if White wants to consolidate her advantage. I dunno, befriend some local comedians or something, that usually does the trick.

What doesn’t do the trick is telling everyone that your opponent, who has a reputation for being a hard-working first-term MP, is just “a celebrity”. (Un)luckily for White, it’s election time which means everyone will forget about her and her faux pas in approximately 36 hours.

Until next time.

Keep going!
Gerry Brownlee. (Photo: Hagen Hopkins/Getty Images)
Gerry Brownlee. (Photo: Hagen Hopkins/Getty Images)

PoliticsJuly 27, 2020

Brownlee says National is the victim of a ‘rogue’ poll. Can that really be true?

Gerry Brownlee. (Photo: Hagen Hopkins/Getty Images)
Gerry Brownlee. (Photo: Hagen Hopkins/Getty Images)

Everyone in politics is talking about rogue polls, after a Newshub Reid Research survey put the National Party on just 25%. Statistician and University of Auckland professor Thomas Lumley explains whether we can see this as one of them. 

“Even with the most rigorous methodology, one in 20 polls will always be a rogue and this is clearly one of them.”Gerry Brownlee

The deputy leader of the National Party is correct that focussing on a single poll result is a bad idea, because the uncertainty in individual poll results is quite large. But it’s not helpful to confuse the usual sampling variation in opinion polls with the idea that some polls are just wrong.

In a mathematically perfect world, given a very large number of opinion polls each sampling 1000 people, 19 out of 20 will get within 3% of the true support for a major party, and one out of 20 won’t, with the one out of 20 typically being off by just a little more than 3%. Here’s a picture of 60 computer-simulated polls of a party with 30% true support. Each line is a poll, with the dot being the headline support estimate and the line being the uncertainty interval. The line is grey if the poll was within 3% of the truth, and red if it wasn’t.

Out of 60 polls, two were red – about the one in 20 you would expect. But these aren’t rogue polls. The largest sampling error was 3.5%; bigger than 3%, but only by a little.

Real polls are more complicated than this and they show more variation. Real polls don’t have the same chance of sampling everyone in the population. Some people aren’t home; some are busy; some just don’t like polls. Real polls use weighting or quota sampling to try to correct for the unequal sampling, and they remove the bias pretty well, but there’s a cost in increased uncertainty.

Worldwide experience is that opinion polls differ a bit more from each other than they would in a mathematical utopia. Rather than being within 3% of the truth, it might be a better approximation to say that 19 out of 20 would be within 4-4.5% of the truth. Here’s the same set of computer-simulated polls, rescaled to allow for the additional variability. Again, a red line means the result is off by more than 3%.

Now, out of 60 polls, seven were red. The worst poll was off by nearly 5%. You wouldn’t want to rely too heavily on a single poll with this level of accuracy, but it would be unreasonable to call any of them ‘rogue’.

Sometimes, though, you get a poll that just seems to be wrong – it’s inconsistent with previous polling, and there isn’t any good reason to think the true level of support has changed. The idea of a ‘rogue poll’ is that you’re seeing something different from the slight excess deviation that is expected in one poll out of 20, and that you would be better off ignoring it, rather than incorporating it into your mental accounting of the election prospects. In the 1970s and 1980s there was lot of statistical research into ‘robust statistics’; automated ways of handling outlying (or ‘rogue’) observations that are more wrong than a simple mathematical model would expect.

Here’s a plot of all the published NZ polling results this year, with two trend lines. The blue line is a standard robust smoother (called lowess), which will try to ignore rogue points; the black line is the same smoother with the robustness turned off.

It’s clear that something unusual has happened in the past few polls. The last three can’t be reconciled as estimates of a constant level of true support with the usual accuracy. This is the sort of setting where ‘rogue poll’ might make sense as a description.

Statistically, the 38% in late June looks like the outlier, and the rest of the points fit a smooth curve. That’s not the only possibility. It’s conceivable that support for National really shot up by 10 percentage points and then back down again, though such a dramatic change doesn’t seem likely. It’s also possible that the new poll is the outlier and that support for National is up to nearly 40%, or that both polls are ‘rogue’ and the truth is in the middle. It will be easier to be sure when we have results from another poll, which should happen soon.

There’s a general problem that opinion polls are getting less informative in New Zealand. There are fewer polls being published, and by fewer pollsters. Response rates are probably also going down, as is happening generally with phone surveys around the world. It’s also possible that the whole sampling and bias correction infrastructure is being affected by Covid-related changes in working from home, and that all the polls are less accurate this year.

As always, it’s important to take the opinion polls in context, and not to over-interpret individual ones. The polls are a fairly crude assessment of popular support for National or Labour or NZ First or the Greens, but they’re still going to be more representative than going by what your mates reckon – and that’s a big part of their social value.

But wait there's more!