Polity polls and pondering

Rob Salmond, Labour Party consultant now an author at Public Address, has posted on recent polls – Poll Soup. He complains about poor analysis of recent polls by media, so I’m complaining about his poll analysis.

At the outset, let me say this is not a post that says the polls are wrong, nor that the left are where they want to be.

But the analysis of the polls this week has been poor.

Primarily, there’s been the claim that National’s high 40s ratings show the TPP protests and/or Labour’s tertiary policy launch have had no impact. That claim is wrong, both because an overall poll rating doesn’t say anything in particular about single events, and more importantly because the government has actually lost almost 2% support over the summer break. Here’s the evidence:

There were four public polls in November / December  2015 – two from Roy Morgan, and one each from TV3 and the New Zealand Herald. Across those four polls, National’s average was 49%.

There have been three polls so far this year – two from Roy Morgan and one from TVNZ. Across those three polls, National’s average was 47.5%, 1.5% below its average from November-December.

Over the same period, the government as a whole (National + Maori Party + ACT + UF) is down an average of 1.8%.

So the claim of “no movement” is a stretch.

TNZV made that claim by comparing their poll in February with one in October, some four months ago. There’s been a lot of events over those four months, not just the TPP and Labour’s policy launch, including a bunch of more recent polls to compare against.

Comparing 3 polls at the quietest time of the political year with 4 polls at the second quietest time of the political year is a fairly narrow analysis, especially when one of the polls in December could easily be a high outlier.

Looking at the trends of a single pollster, Roy Morgan…

…it shows National rising slightly. Salmond notes:

For completeness, the latest Roy Morgan poll does in fact show National up slightly from January to now. But I’d hardly be the first to note the jumpiness of that particular poll from one polling window to the next. I always prefer more evidence than that.

More evidence, the last five Roy Morgan results for National: 49, 49, 47, 48, 48.5

Not much jumpiness there, especially taking into account a margin of error of about 3.2%.

I deliberately left out the previous three Roy Morgan results: 43, 50.5, 44.5

That’s quite ‘jumpy’ – if you choose your range you can just about support whatever contention you like.

The ‘jumpiest’ result in the ranges Salmond used happens to be Herald-Digipoll in December of 51.3%, which in a four result sample (the others were 46.7, 49, 47) can make quite a difference.

Here’s another way of cherry picking poll results:

  • Last 5 results: 51.3, 47, 47, 47, 48.5 = 48.16 average
  • Previous 5 results: 50, 47, 49, 46.7, 49 = 48.34 average

Statistically there’s nothing in that comparison.

Then Salmond looks at future prospects of left (Labour+Greens) versus right (National).

National, of course, remains in the box seat. Along with its solid hangers-on (ACT, UF), it sits in the high 40s. Solid supporters of an alternative government (Labour + Greens) sit a little above 40%, and the swinging centre is climbing towards 10%.

If the election were held today, I’ve little doubt National would be returned even though – as noted on One News – Labour + Greens + NZF would be very nearly able to form a government if they wished. On today’s numbers, I think Winston would choose Key if offered the choice.

To be seriously in the game, the putative left coalition needs to at least tie National at election time. Obviously, having the combined left beat National is better again, and the higher the margin the better. But at a minimum, a tie’s required.

That means the left needs to shift around 4% of the population from supporting the status quo to supporting change.

National -4 and Labour/Green +4 is a shift of 8%. It’s certainly doable.

It’s interesting that Salmond is suggesting that “to be seriously in the game” Labour+Greens needs to at least tie with National, otherwise he thinks NZ First would go with National (that’s uncertain).

So their target is about 45%, which could be Labour 33 + Greens 12, or Labour 35 + Greens 10.

But that would still need NZ First to make up the numbers.

Would Winston think that National 45 versus Labour 35, or National 45 versus Labour Green 45? It’s hard to see Labour being level with National on their own.

NZ First are polling relatively high for them between elections, in the range 5.5-10 since the 2014 election. They normally better their polling in elections.

What if NZ First get 10% and Greens 8%?

There are many interesting possibilities. And don’t forget the final numbers, which could be ACT, Maori Party and Dunne if he doesn’t retire. They could still make the difference, as they have for the past two elections.

There’s some interesting discussion in Poll Soup, especially on the ‘missing million’ that some on the left still think is their holy grail.

Salmond explains his Chinese name data analysis

Rob Salmond has posted further explanation for his data analysis of the Auckland property sales on his own Polity blog. His site has been mostly unloadable so I’ll post it entirely here as an alternate source.

How Labour estimated ethnicity from surnames

In response to requests via Twitter, this post walks readers through the general method Labour used to predict the ethnicity of Auckland house buyers from their surnames. This analysis was featured in the New Zealand Herald’s lead story yesterday.

Note that there are two points in this explanation where I will refuse to go into further detail, in order to protect Labour IP. The rest of this explanation has been made publicly in various venues already, so this post does not give away any new secrets.

Part 1: 2014 demographic study

Pre-election, Labour estimated the ethnicity of every person on the electoral roll, via standard Bayesian updating. There are 3.2 million people on the roll. This was one of many demographic estimates we did for everyone in the country. Most serious political parties now engage in this kind of demographic profiling.

To estimate ethnicity, we used public NZ census data on the ethnic distribution of neighbourhoods, and also used data we developed privately about the ethnic distribution of last, middle, and first names in New Zealand. We followed some advice – especially about estimating Asian ethnicities – from prominent US academic studies. I won’t be describing that process further, as that is sensitive IP for Labour.

Using these data, our base method was to estimate people’s ethnicity in a three-step Bayesian analysis:

  • Step 1: Prior: Neighbourhood ethnic distribution. New information. Lastname distribution. Posterior: Neighbourhood / lastname ethnic distribution.
  • Step 2: Prior: Neighbourhood / lastname ethnic distribution. New information. Firstname distribution. Posterior: Neighbourhood / lastname / firstname ethnic distribution.
  • Step 3: Prior: Neighbourhood / lastname / firstname ethnic distribution. New information. Middlename distribution. Posterior: Neighbourhood / lastname / firstname / middlename ethnic distribution.

This process provides a distribution of the likely ethnicities of each person in New Zealand, given their address and their full name.

The distribution covered the probability that a person was each of the following ethnicities, drawn from the level 1 and level 2 ethnic classifications from the New Zealand census: European, Maori, Pacific (not further defined), Pacific (Samoan), Pacific (Tongan), Asian (not further defined), Asian (Chinese), Asian (Japanese), Asian (Korean), Asian (South Asian), Asian (Middle East), other.

For the person-level point estimates, we used the largest single probability. That probability was typically above 0.9.

We refined these estimates further with three tweaks to account for moderate issues we encountered estimating certain ethnicities. I won’t be describing those tweaks further, because IP.

We then tested our predictions against a more-or-less-random sample of around 3,500 known New Zealanders for whom we had ethnicity data. Our best predictions, which we have used since, were 94.8% accurate.

This is an important point. Having developed our method for estimating ethnicity, we then tested it for accuracy against real world data. Only once we were satisfied it was accurate were we willing to rely on it in our work.

Part 2: Applying the predictions to housing data

To apply our general predictions, derived in part 1 above, to the Auckland housing data, we followed a two-step process.

First, we collapsed the 1.4 million Auckland-based ethnic estimates we had by surname only, as that is the only data we had in the real estate data. This allowed us to also partly leverage the earlier electoral roll-based information we gleaned from first names, middle names, and locations as part of our surname-based estimates.

Most of the surnames pointed strongly (pr>0.9) to one and only one ethnicity, although there were some examples with more mixed predictions. It created estimates such as the following (these are the real values):

Name pr(European) pr(Maori) pr(Chinese) pr(other)
JONES 0.938 0.054 0.001 0.007
HOTERE 0.048 0.887 0.000 0.065
LEE 0.481 0.027 0.400 0.092
LI 0.028 0.001 0.957 0.014

Having done that for each individual purchaser, we then summed the probabilities across all 3,922 sales in the dataset. This provided an aggregate estimate, based on the distributions of likely ethnicities in each individual sale, for the overall ethnic distribution of house buyers in Auckland.

In doing this aggregation, we tested various ways of accounting for the fact that some sales had one surname attached, while others had two or even three, accounting for multiple people with diffrerent surnames purchasing a property together. No matter how we cut those observations, the overall pattern remained within 2% of the numbers that appeared in the New Zealand Herald.

It is that overall distribution, not data cherry-picked from any particular sale, that we then compared with various other aggregate datasets about the ethnic distribution of Auckland residents, or various subsets of Auckland residents. Many of those comparisons are detailed in the Herald article and in my Public Address blog post yesterday.

Bridges too far?

Did Simon Bridges got to far in seeking cost details on Northland bridges?

Mr Bridges’ office asked the NZ Transport Agency for information on the bridges and estimated costs of upgrading them prior to the byelection announcement that National would upgrade 10 one-way bridges.

Andrew Little thinks he did.

Labour’s leader Andrew Little said that was a clear breach of the rules for ministers’ use of public officials and Mr Bridges should be sacked.

John Key thinks he didn’t.

Mr Key said he did not believe it was a breach.

“My understanding is it’s quite okay to ask for information. You’re quite free to do that. The issue is whether you’ve got policy advice and Mr Bridges didn’t do that.”

The Cabinet Manual seems unclear.

The Cabinet Manual states that “any requests [ministers] make for advice or information from their officials is for the purposes of their portfolio responsibilities and not for party political purposes”.

Bridges would be responsible for fulfilling the bridges bribe so should be basing decisions on advice and information. Many policy decisions can be both part of Governance and for party political purposes – trying to get re-elected.

Anthony Robins at The Standard thinks it’s clear in Burn the Cabinet Manual:

Key won’t take any action over Simon Bridges’ clear breach (excellent work by Rob Salmond at Polity) of the Cabinet Manual. So, might as well burn the thing, at least for the remainder of this government’s term. Key has no intention of being held to account, or holding his ministers to account, by or for anything at all.

Did Helen Clark and Michael Cullen get advice and information before making their famous election rescuing Student Loan bribe? Was any Minister sacked as a result of that? I’m sure there are numerous examples of advice or information from officials being used for election (party political) purposes.

David Farrar at Kiwiblog calls it A beltway beltway issue.

I don’t believe that anything Simon Bridges did, is a breach of the Cabinet Manual. But regardless this is what you call a classic beltway issue. The number of people who get excited over this is miniscule. Mrs Jones in New Plymouth and Mr Smith in Hamilton want jobs, incomes, decent schools, good healthcare etc.

The sort of people who think this is great politics are the same sort who orgasm over who won question time in the House. I know, because I used to be one of them.

Ecch. But he may have a point, no matter how awfully he has put it.

In comments yesterday on Your NZ Alan Wilkinson commented:

This is b.s. If a Government makes a promise before a by-election it has to implement it and therefore it has to cost it responsibly and accurately.

Totally different to before a general election when it may not be reelected. No matter what the Cabinet manual says the Minister was making a promise in his ministerial capacity which he would have to implement and therefore fund.

Just to add the obvious corollary to this, in a by-election if the Cabinet Manual rule were to be applied it would mean the Government’s opponents in the by-election would be free to promise anything they wished and the Government’s candidate would be unable to promise anything new. Farcical nonsense. It shows exactly how incompetent or biased MSM journalism is that this is not pointed out and the opposition’s arguments rubbished.

There might turn out to be some sort of technical breach of the Cabinet Manual but Alan’s comments make sense to me.

Flipper at Kiwiblog:

The closest that anyone has comes to the true worth of “The Cabinet Manual”: is Helen Clark. She amended “it” to suit each circumstance…and to her benefit.

The reality is that the manual is just a collection of “thou shalt nots” (well if it suits the PM), and “:thou shalls”. It has no stranding in law because it is not backed (compiled pursuant to) by a statute. Many matters upon which it offers guidance may well (probably are) covered by Statute. At best the manual is a collection of Executive fiats.

Back to the instigator of the beat-up to far, Rob Salmond at Polity, who responded to Farrar’s post in The “beltway” response:

By “acted in a political way,” of course, he means “breached the rules of his office.” Also, good luck passing off the actual job of Her Majesty’s Loyal Opposition, to hold the government to account for its actions, as “crying wah wah.”

I agree about Mrs Jones and Mr Smith, though. This is not an election defining issue. I’m guessing Labour’s 2017 election campaign won’t have much to do with this issue, in the same way National’s 2008 campaign didn’t say too much about Taito Philip Field.

The thing about so-called “beltway” issues is that they aren;t much good at election time in their own right, but if a number of similar issues emerge around a government then it forms a more general impression which does matter in elections. That was how National used Field. In National’s case, that general impression might be “arrogant” or “liars” or “duplicitous” or “corrupt.” They’re certainly handing out plenty of material…

So Salmond doesn’t seem to think think this is much of a big deal but is trying to chip away at National’s credibility.

Rob would help his own credibility on this if he didn’t try and compare what Simon Bridges did with what Taito Philip Field – Field was charged with “15 counts of bribery and 25 of attempting to pervert the course of justice”.

Field was jailed for six years on corruption charges, with the sentencing judge saying his offending threatened the foundation of democracy and justice.

Likening this to Bridges going too far seeking Ministerial information and advice looks like a beat-up too far.

National’s by-election campaigning costs

National have offered voters of Northland many millions of taxpayer funded pork to try and rescue the electorate from a Peters and media onslaught.

They have also piled probably unprecedented resources into the campaign itself, with a swarm of Ministers including the Prime Minister using a significant state funded travel advantage.

Rob Salmond at Polity claims:

National has poured massive, massive resources into Northland, most of which won’t show up in the financial returns. Polling, focus-grouping, canvassing, MP visits, Ministerial cars, taxpayer bribes, flying squads to drive people to the polls. All of it is off the by-election books. I have heard rumours that National’s total outlay is close to $250,000, not to mention the bill the taxpayer will carry.

They have a huge financial and organisational advantage. Comments add to this. George D:

If National’s outlay is in the order of $250,000 then there are surely things that should come to the attention of the Electoral Commission?

Deborah Russell:

Lots of things aren’t caught by the rules. Venue hire, MPs’ time, ministers’ time, petrol, polling, focus groups, paying organisers’ wages, wood for billboards, wages for people to put them up. I can see how the costs could mount up very quickly. I might find it a little hard to get to $250,000, but even so, I can see how lots and lots and lots of money could be spent.

So what are the spending rules:


6.2 Expenditure limit

The regulated period for the 2014 General Election will start on Friday 20 June 2014, and will end with the close of the day before election day (Friday 19 September 2014).

An electorate candidate’s election expenses during the regulated period must not exceed $25,700 (including GST).  It is a serious offence to spend more than this.

If you are representing a registered party, you should stay in touch with your party secretary on advertising.  This is because there can be boundary problems between advertising by candidates and advertising by the party, with consequential effects on the expenditure limits and expenditure returns of the candidate and the party.

The election expense regime does not apply to people who are list candidates only.  Any spending by those candidates promoting the party is an election expense of the party and must be authorised by the party secretary.


6.2 Expenditure limit

An electorate candidate’s election expenses during the regulated period must not exceed $25,700 (including GST).  It is a serious offence to spend more than this.

This is for a general election but I presume it applies to a by-election as well.

6.3 Election expenses

A candidate’s election expenses are the costs of advertising in any medium that:

  • may reasonably be regarded as either encouraging voters to vote for the candidate, or discouraging voters from voting for another candidate, or both (whether or not the name of the candidate(s) are mentioned),
  • is published, or continues to be published, during the regulated period (from 20 June to 19 September 2014), and
  • is promoted by the candidate or any person (including a registered promoter) authorised by the candidate.

[See section 205 of the Electoral Act].

Candidate election expenses include:

  • the cost incurred in the preparation, design, composition, printing, postage and publication of the advertisement,
  • the reasonable market value of any materials used for the advertisement, including materials provided to the candidate for free or below reasonable market value,
  • the apportioned costs for advertisements that promote two or more candidates, or a party and a candidate  (see paragraph 6.5 below for further information on apportionment).

[See section 3E of the Electoral Act].

A candidate’s deposit or the costs of food, hall hire, surveys or opinion polls, free labour, or replacing materials destroyed through no fault of the candidate are not election expenses.  The cost of any framework that supports a hoarding (other than a commercial framework) is no longer an advertising expense.

So National have a huge financial advantage in general.

But it’s not all going their way. Winston Peters has had a huge and free amount of support from the media, getting effectively promotional that money can’t buy but can be priceless in a campaign.

Such is the corruption of our electoral system. Money and media rules.

Polity picks Osborne in Northland

Rob Salmond, a Labour pollster, has picked Mark Osborne to win the Northland by-election based on National having a well organised machine in action versus Winston Peters with little established electorate organisation and Labour giving up trying.

Note that this was posted before yesterdays 3 News poll:

  • Winston Peters (NZ First) 54%
  • Mark Osborne (National) 34%
  • Willow-Jean Prime (Labour) 10%
  • Other 2%

But Salmond’s point still stands. There’s a difference between sticking one up National when someone rings and asks for your off-the cuff opinion and getting out and voting.

In Northland, the National supporters are organised by the National Party nationwide machine. Winston Peters’ supporters, by contrast, aren’t that well organised. That’s why they’ll likely lose.

And he details the reasons.

But this this by-election the turnout is enormously higher than in the most recent general election. It is *up* around 70%, compared to the general election just six months ago. Normally, it would be down 50%.

… where is it coming from?

  • Labour’s machine? Categorically nope.
  • New Zealand First’s machine? Nope. They don’t have much of a turnout machine.
  • Sudden discovery of advance voting by Northlanders over the past six months? A stretch.
  • Northlanders care much more about the by-election issues (bridges, arts centre accounting, ferry ride discounts) than the general election issues? Another stretch.
  • National’s machine? Yes. That is the cause.

So, my prediction remains a solid National win, not borne of popularity, but borne of organisation.

This is supported by a comment by a Labour campaigner:

Speaking to Willow Jean earlier today she says the Nat’s have a huge on the ground team, where as Peters has very few.

I don’t know if Rob’s prediction still stands but the result could be much closer than the poll suggests due to it being much less effort answering a question on the phone than going out and voting.

Polity: Northland: Countdown-to-letdown

Salmond predicts easy National win in Northland

Rob Salmond is a strongly pro-Labour pollster. At his blog Polity he predicts what looks obvious to those with a grip on reality.

Prediction: Easy National win

With Winston Peters’ confirmation that he’s running in Northland, and Labour’s confirmed candidacy, I predict National will win the seat comfortably. The seat is a National stronghold, and a split opposition vote only helps them further in an FPP-style environment. So whoever wins the National selection, wins.

It would be remarkable if there’s an upset, and National have too much to lose to allow that to happen easily.

Since Peters has announced he will stand iPredict has swung against National but they still have a large majority. Mostly over that past weeks National have polled in the 90s but yesterday swung in the 70s, and they are currently at 82%.

Other (not National or Labour) party to win Northland by-election in 2015 has risen to 17% after peaking yesterday in the low 20s.

And more from Salmond a couple of weeks ago in Poll wordings in Northland that shows why caution is needed when peop;le claim ‘private poll’ support for their cause.

For the most dramatic poll result – “Winston could win!” – you would ask:

There is a by-election in Northland on 28 March. Which of the candidates are you most likely to vote for:

  • Willow-Jean Prime
  • Grant McCallum [or whoever it ends up being]
  • Winston Peters
  • Another candidate?

Winston gets a huge boost in this poll wording because he’s the only one with really wide name recognition across the seat, and also gets a smaller bonus for being the last named candidate, meaning his name is freshest in the memory when the person is required to give their answer. This question will hugely overstate Peters’ chances, at the expense of both National and Labour candidates.

For a more accurate poll result, you might ask this instead:

There is a by-election in Northland on 28 March. Which of the candidates are you most likely to vote for [in randomised order]:

  • The Labour party candidate, Willow-Jean Prime
  • The National party candidate, Grant McCallum [or whoever]
  • The New Zealand First party candidate, Winston Peters
  • Another party’s candidate?

This question provides people the information most of them will actually rely on when making their choice – party affiliation. That information, I understand, is printed on the ballots themselves, and it is the heuristic most people use when choosing local election candidates. (Also, the order is randomised from person to person, meaning the fresh-in-the-memory effect goes away across the whole sample.)

This is a much better question, and media outlets would be wise to insist on it, or some other variant that cues party affiliation. They’ll get less egg on their face on election night that way.

Third strike and out for TU, third Cunliffe fiasco ‘unhelpful’

Rob Salmond posted Three ramshackle PR fiascos and you’re out, having a dig at the Taxpayers’ Union:

Three strikes – they’re out.

The same could be applied to Cunliffe. Not Salmond though:

The sight now of of a leading Labour politician, with the help of a blogger at The Standard, setting up a trust to shield substantial leadership donors from public view, is most unhelpful.

I am really pleased, however, to see Selwyn Pellett, Perry Keenan, and Tony Gibbs openly declare their generous support for Cunliffe’s campaign. I thank them for their generosity towards social democracy in New Zealand. All are substantial figures in the business community, and their support speaks volumes about Labour’s visionary plans to help new Zealanders add value to our economy.

He calls that post “Untidy and unnecessary” and describes the third PR fiasco of Cunliffe this year as “most unhelpful” – and then turns his post into a Labour promo.

Rob’s bio on his Polity blog…

Rob has wide experience relevant to public affairs. He has been a Parliamentary adviser to two leaders of New Zealand’s Labour party (Helen Clark, David Shearer), and through Polity continues to work with Labour leader David Cunliffe.

Perhaps Rob wants to continue to work with Cunliffe so is waiving his Three Strikes principle.

Cunliffe’s trust and 3 strikes

It looks like mounting pressure on David Cunliffe has forced action on his secret trust – see Cunliffe “happy to be open” about keeping donations secret.

Claire Trevett broke this story on Sunday and continues today:


Cunliffe has named 3 donors who were willing to go public, including Selwyn Pellett, Tony Gibbs. will return donations of 2 others who won’t

Also Cunliffe got donation from Perry Keenan. Cunliffe says didn’t know donors names till recently. Says using trust was error in judgment.

It’s taken Cunliffe two days to admit that. He must be copping some flak from uncomfortable places.

This is yet another in a series of cases of poor judgment by Cunliffe. Taking belated action will repair some of the damage but some of the dents will remain prominent.

Ironically at The Standard, just under Bunji’s post How short are memories? defending Cunliffe is a repost of Rob Salmond with Polity: Three ramshackle PR fiascos and you’re out. Salmond was referring to the Taxpayers’ Union, not Labour but…

Polity warns opposition parties – Dotcom sounds pretty icky

Rob Salmond at Polity points out the obvious dangers of opposition parties getting involved with Kim Dotcom in Kim Dotcom’s 5% gambit (also posted at The Standard) He refers to Dotcom’s tweet yesterday…

If #InternetParty won’t poll 5+% before ballot papers are printed we’ll self destruct & put our weight behind a party adopting our policies.

Like most pundits and journalists Rob thinks “it is almost certain that the Internet Party will not be polling 5% at any point this year”.

If I am right about that, then come ballot-printing day Mr Dotcom will be throwing his weight in with someone else. And by “his weight,” I presume he means large buckets of money. That sets up an silent auction for parties to compete for Dotcom’s money on the basis of policy promises, first and foremost about Dotcom’s own extradition case. That is, if parties decide they want to play.

I think the opposition parties should all take a pass. 

To me, it all sounds pretty icky. One of the reasons the left parties worked hard to try and make election funding fairer in the late 2000s was to limit the influence of individuals seeking to essentially buy government policy for cash. (These measures were, naturally, rejected by the right, citing freedom of speech and freedom of spending and so on.) Breaking it down, this gambit looks exactly like a convluted version of a rich guy offering up cash in exchange for personally favourable policies. Yuck.

We’re now in this odd position where left parties that actively compete in the policy space for Dotcom’s affections will be hypocrites and, by the same token, many of the right wingers who would cry foul about that will be hypocrites, too.

It’s not just hypocrisy that’s at risk. As Rob says, “it all sounds pretty icky”.

But is it too late for opposition parties to take a pass?

Russel Norman has met Dotcom at his mansion at least twice and has made some risky statements about his position on Dotcom’s extradition.

Winston Peters appears to have met Dotcom more than three times – Peters denied the number three but ducked for cover when asked anything else about it.

David Cunliffe appeared to carefully avoid denying any Dotcom meetings on Firstline this morning. He denied meeting with Dotcom on one specific issue only and no mention of any other Labour representatives meeting.

Potentially for all of the Green Party, NZ First and Labour it all sounds pretty icky. 

Associating with Dotcom has been bad  for John Banks, Martyn Bradbury (that could also have been bad for Dotcom), and Alistair Thompson who was forced to resign as Editor from Scoop.

It’s already looking icky for Russel Norman.

How deep are Peters and Cunliffe in Dotcom ickiness?