Today, we see a Herald ‘editor’ demonstrating exactly how useless they are at interpreting polls. Their editorial says
The result of our poll today could suggest voters will forgive any foolish behaviour from this Prime Minister.
Huh? Ponygate only started late last week. When was the poll taken? Audrey Young says in her linked article
Half of the poll was conducted before the news broke that he had repeatedly pulled the ponytail of a waitress at a Parnell cafe despite her annoyance, and half of the poll was conducted after it.
And way way down the end of article
The poll of 750 eligible voters was taken between April 17 to 26.
A light dawns. We have people who either don’t understand polling or dates or someone wilfully covering their knowledge… I’m going to show you a public graph of the page views from The Daily Blog who broke the story covering those dates.
The first post at The Daily Blog was on Wednesday, after the poll had already been running for 3 week days and 2 weekend days.
So on Wednesday, the news was breaking after the story was in TDB in the morning. It was picked up by media through the day, but only really to people actively political and embedded in twitter. It wasn’t until the afternoon after journalists got confirmation from John Key about its essential accuracy that the story went wild.
You can see this in the figures from TDB. I observed that most of their page views for the 22nd were after the evening news on the 22nd. They nearly doubled their page views on the following day when people were reading it in their newspapers. Virtually all of the views were likely to be on the two posts that featured on those two days. The numbers of the otherwise useless “unique visits” almost equalled the number of page views. This indicates that almost all of the page views came direct from links from other sources like social media directly to a post, rather than via the front page of the site (as woudl happen if directed from print of visual media).
The boosted page views was almost entirely from social media if it was anything like our smaller bumped linkages for those three days.
Realistically most people in a poll would have only become aware of the story on the Thursday when they were still reacting to it and making up their minds. They immediately ran into a long weekend with many people leaving work and home on the Friday.
But even without the long weekend, the vast majority of the calls making up the poll would have been in the first half of the polling cycle before the story broke. Only a minority of calls in the sample would have been made in the final days of the poll.
The reason for this is because of the way that a statistical sample used in a poll (especially such a small sample) is either balanced against location (random sampling) or against the population demographics like age, gender, etc (quota) or on some combination between the two. This is to ensure that sample statistically is representative in some way to the population that is being sampled.
Gathering accurate data is only viable while using a sample that reflects a population. In a scientific survey every member of a population must have an equal opportunity to be questioned.
Companies using bought databases are neglecting this aspect, where as DigiPoll produces random numbers to be used in a sample.
With the emergence of new telephone providers it has become increasingly difficult to produce accurate samples. DigiPoll overcomes this problem by having continuously updated “telephone maps” of populations. Accurate maps are also vital for representing ethnic minorities and other hard to reach groups.
and
Weighting survey responses is designed to remove bias from a survey sample and make the results better project the target population. DigiPoll has formulated a weighting methodology that is applied to the raw data in order to reflect the demographic composition of the sample, which has been proven to be successful time and time again.
Almost all of the tail end of a poll is finding the final instances of the types of people who match the missing profile required by the sampling technique. This is the reason why most polls reflect public opinions towards the start of the polling period.
So when the Audrey Young writes
The ponytail-pulling antics of Prime Minister John Key appear to have had no immediate impact on his party or personal popularity, the latest Herald-DigiPoll survey suggests.
Half of the poll was conducted before the news broke that he had repeatedly pulled the ponytail of a waitress at a Parnell cafe despite her annoyance, and half of the poll was conducted after it.
What she is mostly suggesting to anyone with a brain is that she should get an education in basic statistics and polling sampling techniques
When that the anonymous editorial author states
The fact that his support from polled voters is undiminished by the disclosure of his penchant for pulling a young woman’s ponytail may have something to do with his response.
The only fact that they are demonstrating is that they are a politically ignorant jackass. Mind you they prove that by then stating
In the meantime, the unusually sustained popularity of the PM is causing intense frustration among his opponents, not so much in Parliament but outside it, on websites and in some academic circles where resentment has become extreme. These people are doing their cause no favours with their seething hatred of a political figure who everyone else knows to be an economic moderate and social liberal. They are not helping Labour’s recovery, still below 30 per cent in this poll, and the Greens remain around 10 per cent.
What a complete and utter dimwit. After running this very open, public, and largely uncensored webite (at least compared to the Herald) for so many years, I can testify that all social media is are people directly expressing their own feelings and reactions to political events. With a relatively few exceptions like David Farrar or Cameron Slater, they aren’t doing for a particular political party, they are doing it because they like expressing themselves clearly and accurately.
In mature social media channels authors or all kinds get checked in what they say by the many other people who also write around an write around an increasingly sophisticated social media communicating without the filter of various broadcast media like newspapers.
What the Herald is demonstrating is that the quality of the newspaper has descended to farcical levels of inaccurate and cloistered stupidity. But this isn’t exactly surprising after the events of last week.
At the Herald we saw the inexcusable lack of journalistic ethics displayed by editor Shayne Currie in trying to excuse the ghastly behaviour of the gossip ‘journalist’ Rachel Glucina.
How this Herald reporter got in contact with the owners of Hip Group and obtained a interview with Amanda Bailey under false pretences isn’t quite clear. Mainly because The Herald in all of their carefully worded ‘explanations’ seems to have resorted to lying by omission. They completely excluded the level of permission sought and gained from Amanda Bailey and if she even sighted the comments attributed to her.
But it is pretty clear that the interview was gained with deliberate deception and lying to Amanda Bailey by Rachel Glucina and probably her ex-employers in violation of any decent level of journalism or management ethics.
With their actions supporting the disgusting Rachel Glucina, the Herald editorial staff has actively condoned one of the most heinous violations of journalist ethics I have seen in NZ. Now they seem to be adding stupid inaccuracy to it as well.
If anyone wants to sue me over my words here, then I’ll be happy to publish any discoveries that I make during the court process.
Powered by WPtouch Mobile Suite for WordPress