Another post having another go at another rubbish poll

Another day another poll. Yesterday Te Karere released their Waiariki poll. The poll put Te Ururoa Flavell on 56%, Annette Sykes on 22% and Louis Te Kani on 22%. If you ask me, it’s certainly true that Te Ururoa is leading, but I assure you he doesn’t enjoy the support of 56% of the electorate. That’s an inflated figure.

Polls polling the Maori electorates usually, if not always, miss the mark. Five days out from polling day in the Tai Tokerau byelection the Native Affairs Baseline Poll put the gap between Hone and Kelvin at one percentage point. Hone went on to win by over 9%. In 2005 and 2008 the Marae Digi-polls were predicting a clean sweep for the Maori Party in the Maori seats. In reality the party won four seats in 2005 and five in 2008 – not all seven (that is wrong and I take responsibility for making a poor call). There is significant variance between what some polls hold and what others do when it comes to the Maori electorates. For example, the Marae Digi-poll on Te Tai Tokerau held that Kelvin Davis was ahead of Hone Harawira. A week or so later the Te Karere poll indicated that Hone enjoyed a comfortable lead over Kelvin.

As you can imagine, I have a few criticisms to make. Firstly, the poll was conducted via landline. Poor people, read many Maori, prefer cellphones over landlines. With cellphones you pay for what you use whereas landlines come with a fixed cost no matter how much or how little you use it. Data from the 2006 Census indicates that only 76% of Maori households in Waiariki have access to a landline and 73% of Maori households have access to a cellphone. That figure would have risen significantly with the increased proliferation of cellphones over the past five years. Many Maori operate two or three cellphones on different networks. The problem with surveying landline only households is that the subset of the population without landlines differs from those with landlines (e.g. the very poor and the very rich tend to operate cellphones only – those in the middle tend to be landline and cellphone users).

Furthermore, 14.6% of workers in the primary sector do not have access to landlines. By my reckoning that figure will rise for Maori primary sector workers. In the Eastern Bay, where Annette is strongest, many Maori are employed in forestry, farming or the pulp and paper industry (primary industries). Also 19.6% of 18 to 24 year olds do not have access to a landline phone. As you can imagine, this group will include a disproportionate number of Maori voters. Annette’s support base is likely to consist of mainly young Maori too. For low income homes 12.5% do not own a landline and, again, that figure will rise when considering poor Maori households only. Again, Annette’s support base is primarily drawn from the poor.

With this in mind, a landline only poll is going to bypass a significant bloc of the Maori population. Maori, especially low-income Maori, are transient too. Often moving from rented home to rented home. As such, these Maori have never had access to a landline telephone and, if they do, they are extremely difficult to reach because they tend to avoid listing their numbers in the White Pages.

The time calls are made will also affect who is reached. I figure a disproportionate number of Maori in Waiariki are shift workers and these workers will be missed because, from my experience and the pollsters call my family every time, the poll is conducted late afternoon early evening when shift workers are beginning their shift/still sleeping. Again, a large bloc is missed. This brings me to another point; pollsters tend to ring the same Maori over and over again. Our household is contacted for every poll they run I think. You’re not going to get reliable results when you poll the same people over and over again.

The poll also worked from a low base, only 400 voters from the Maori role were surveyed, and the margin of error was quite high at 4.9%. Another significant flaw is that the poll was conducted between October 15 and November 8. Polls are a snapshot of a particular moment in time. The snapshot from this poll captures a time where Annette was still in first gear – slowly building her name recognition and momentum.

Polls polling the Maori electorates may also suffer from non-response bias. Some Maori may be unwilling to respond to pollsters so the poll may not be representative of all Maori. The characteristics of the Maori who do agree to be polled may be different those who refuse. For example, those who agree may be more conservative or liberal.

It would also be interesting to know the wording of the questions. For example, were the names of the candidates given (unlike in the Marae Digi-poll). Was Te Ururoa described as “the incumbent” or did the pollsters refer to his “majority” or “record” etc… that sort of wording will influence the direction in which respondents swing.

I’m not sure what, if any, weighting techniques the polling company uses. The deficiencies identified above are significant in my opinion and have skewed the result.

I’m not arguing that Te Ururoa is not ahead at the moment. He certainly is, but he certainly does not enjoy such a handsome lead. For example I was told that yesterday Annette smashed Te Ururoa at a debate at Te Wananga o Aotearoa in Rotorua. Many of the Wananga people who are Te Ururoa supporters (Te Ururoa used to work with the Wananga) were switching their vote to Annette. This is the story I’m hearing consistently. Having said that, I’m also hearing a lot of praise for what Te Ururoa has done for Waiariki (minus the East Coast of course – boy is Te Ururoa not liked up there).

Tim Selwyn and I are lonely voices calling it for Annette. But then again we were lonely voices when we were calling it for Hone too. Despite this, I’m standing by my feeling that Annette is going to win it. Call me a fool.