Feeds:
Posts
Comments

Archive for November, 2010

The 2010 Mana by-election will be held today. Unfortunately, we haven’t seen any polling in the electorate, and national-level polling over the past few weeks has been sparse, so it’s not very obvious what to expect.

In spite of the problems mentioned above, I have attempted to predict the outcome of the election and the share of the votes won by each of the major or minor NZ political party’s candidates using a Monte Carlo simulation of 50,000 NZ general elections held today based on recent political polling results. The results for the Mana electorate are determined using the Uniform National Swing model of electorates. The results of the simulation are shown in the table below, and the results of the 2008 General Election for the Mana electorate are included for reference.

Table showing simulated results of 2010 Mana by-election.

Table showing simulated results of 2010 Mana by-election, as calculated on 20 November, 2010.

For reference, similar predictions and results for the 2009 Mt Albert by-election are shown in the table below. The results of the 2009 by-election are of course already known, so you’ll just have to take my word for it that they were not used to cheat by including them as inputs in the simulation.

Table showing simulated results of 2009 Mt Albert by-election.

Table showing simulated results of 2009 Mt Albert by-election, as calculated on 13 June, 2009.

As you can see, the predictions aren’t particularly brilliant. The simulation significantly underestimated the results for the Green Party, and overestimated the results for National. The results for Labour aren’t significantly out, but given that the flow of support was most probably National to Labour and Labour to the Greens, rather than National to the Greens directly, then this was probably just a lucky coincidence.

There are several likely possible causes for discrepancy:

  1. The UNS model for the electorate results is not reasonable.  Empirically, the model has performed well in median seats, but is not very realistic when used in party strongholds such as Mt Albert and Mana, which have historically been safe Labour seats.
  2. There is usually a swing against the government in by-elections relative to that which would be expected from national-level polling, something which the simulation doesn’t take into account.
  3. By-elections tend to attract a different selection of candidates than a general election typically would.
  4. By-elections tend to attract a different selection of voters than a general election typically would, with voter turnout affecting the result in unpredictable ways.

Given the above caveats, the predictions for Mana should be interpreted loosely.  I’m not going to take responsibility for them if they are significantly out, and conversely I will not be taking any credit if they prove accurate.

As for today’s Mana by-election, the only other predictions of the result of the election that I am aware of are from New Zealand futures market iPredict, who as of 15:40 (see screen captures below) are giving the Labour candidate a 96.99% chance of winning and the National a 2.36% chance of winning. iPredict also predict that the Labour candidate will win 46.67% of the vote, the National candidate will win 36.97% of the vote, and all others combined will win 16.80% of the vote. I’m not aware of anybody else making quantitative prediction on the results, but if you know of anybody, or if you would just like to take a punt, then feel free to leave a comment below.

iPredict Mana by-election winner predictions

Screen capture of iPredict's Mana by-election winner predictions, as of 15:40 on 20 November, 2010.

Screen capture of iPredict's Mana by-election vote share predictions, as of 15:40 on 20 November, 2010.

Screen capture of iPredict's Mana by-election vote share predictions, as of 15:40 on 20 November, 2010.

Advertisements

Read Full Post »

It’s been two years to the day since the 2008 New Zealand General Election, so I thought it would be a good time to look back and see what has happened with the polling.

The graphs below summarise the polling averages for the party vote for each major or minor NZ political party over the last two years. The horizontal axes represent the date and start on 8 November 2008, the day of the 2008 NZ General Election, and finishing on the present day. The solid lines with grey error bands show the moving averages of the party vote for each party, and circles show individual polls with the vertical lines representing the total errors.

Party vote support for the National party over the past two years

Party vote support for the National party over the past two years as determined by moving average of political polls.

Firstly: National. After getting 45% of the vote in the 2008 Election (slightly lower than the polls predicted) they had a fairly rapid gain in popularity to about 54% +/- 1% by early 2009, where their polling stabalised for about the next year or so. In early 2010 the polling average underwent a significant decline to about the 51% +/- 1% level, where it again stabalised. There have been several polls out this year putting National at 48%-49% support, and causing the polling average to momentarily drop below 50%, although in hindsight these appear to be fairly typical statistical fluctuations.

Party vote support for the Labour party over the past two years

Party vote support for the Labour party over the past two years as determined by moving average of political polls.

After getting 34% of the vote in the 2008 Election (about what the polling was predicting) Labour dropped to below 30% support in early 2009. Support has since fluctuated as it made an undulating climb back to about 33% +/- 1%. To some extent the rises and falls in support for Labour over the last two years have been anti-correlated with support for the Green party, indicating that combined support for the left-wing parties may have been slightly more consistent than it appears in the above graph.

Party vote support for the Green party over the past two years

Party vote support for the Green party over the past two years as determined by moving average of political polls.

The Green Party won 6.7% of the vote at the last election, significantly below the polls which put them on about 8% +/- 0.5%. Other than a few small dips in late 2009 and early 2010 their polling has been fairly consistent in the high 7% range.

Party vote support for the ACT party over the past two years

Party vote support for the ACT party over the past two years as determined by moving average of political polls.

The ACT Party won 3.6% of the votes at the last election, significantly higher than their polling at the time which was in the high 2%-range. By early 2009 their polling had stabalised at just under 2%, where it stayed until mid-2010 when it suffered a significant drop to about 1% +/- 0.5%. I’m reluctant to attribute movements in polling data to political events, but the drop in support did happen suspiciously close in timing to the rolling of deputy leader Heather Roy and the David Garrett passport scandal.

Party vote support for the Maori party over the past two years

Party vote support for the Maori party over the past two years as determined by moving average of political polls.

The Maori party polling has been perhaps the most consistent of any of the eight major or minor political parties. After winning 2.4% of the votes in the 2008 Election they have consistently polled in the mid-2% range.

Party vote support for the Progressive party over the past two years

Party vote support for the Progressive party over the past two years as determined by moving average of political polls.

It’s difficult to draw any meaningful conclusions from the polling numbers for the Progressive party other than to say that they have most certainly not broken through the 0.7% support level in the past two years. The polling results are normally at the level of the margin of error introduced by rounding, so it’s pretty much impossible to make out any statistically-meaningful trends in the data.

Party vote support for the United Future party over the past two years

Party vote support for the United Future party over the past two years as determined by moving average of political polls.

The same comment above for the Progressive party largely applies to the United Future party as well, albeit with a slightly wider error band on the moving averages. The United Future party have most likely not broken through the 0.9% support level in the past two years.

Party vote support for the New Zealand First party over the past two years

Party vote support for the New Zealand First party over the past two years as determined by moving average of political polls.

And last but not least, the New Zealand First party. After winning 4.1% of the vote at the 2008 Election, slightly higher than the result predicted by the polling, they suffered a fairly obvious drop in support to the low-1% range. Since mid-2009 though they have started on a gradual comeback, and are now polling in the mid-2% range; still lower than they were at the last election, though.

In hindsight, there weren’t many occasions in the past 24 months where National would have been predicted to lose an election. But with 12 months or so until the most likely dates for the next election obviously anything can still happen. If Labour can win about 3% points of support from National in that time then the Maori party will probably hold the balance of power in Parliament, which could make for a very unpredictable 2011 Election result.

Read Full Post »

Please accept my apologies for the slow updates. Since the last update we have seen the release of four political polls: three from Roy Morgan Research (1, 2, 3) and one TV3-Reid Research poll. The addition of the four new polls to the calculations does not alter the moving polling averages by much. The averages now have National on 51.8% +/- 1.8% and Labour on 33.7% +/- 1.7%.

As usual, the two graphs below summarise the polling averages for the party vote after the new poll. The horizontal axes represent the date, starting 60 days before the 2008 NZ General Election, and finishing on the present day. The solid lines with grey error bands show the moving averages of the party vote for each party, and circles show individual polls with the vertical lines representing the total errors.

Party vote support for the eight major and minor NZ political parties

Party vote support for the eight major and minor NZ political parties as determined by moving averages of political polls. Colours correspond to National (blue), Labour (red), Green Party (green), New Zealand First (black), Maori Party (pink), ACT (yellow), United Future (purple), and Progressive (light blue) respectively.

Party vote support for the six minor NZ political parties

Party vote support for the six minor NZ political parties as determined by moving averages of political polls. Colours correspond to Green Party (green), New Zealand First (black), Maori Party (pink), ACT (yellow), United Future (purple), and Progressive (light blue) respectively.

The latest scenario analysis graph is also shown below:

Scenario analysis for 6 November 2010. The bar graph shows the probabilities for different possible outcomes for a NZ General Election if held on that date. The National Party are estimated to have a roughly 84% probability of winning an outright majority of the seats in Parliament, and 99% chance of being able to form a governement without Maori Party support.

As always, please check the Graphs page for further simulation results.

Read Full Post »