Wednesday, June 06, 2007

Journalistic laziness

I was woken up this morning by the radio blaring that a poll on the world's largest Maori website had found that Maori were overwhelmingly in favour of outlawing gangs and throwing children in jail. The story is ultimately sourced from a press release from the Sensible Sentencing Trust, and has since shown up on TV3 as well, but you'd think that our journalists would at stop and think a little before checking the broadcast - or at least look at the polls in question. Unfortunately, they didn't. So, here's the bits they should have noticed.

Firstly, as I keep having to remind people, online polls are not statistically reliable. The sample is self-selecting, rather than randomly selected, and thus is unlikely to represent a true cross section of the community's views. They should therefore be treated with a grain of salt, rather than as serious news.

Secondly, looking at the poll results in question, we find that a grand total of 68 people expressed an opinion on whether gangs should be outlawed, and 43 on whether children accused of serious crimes should be tried and punished as adults. Both polls were held open for a month on what the Sensible Sentencing Trust calls "the largest Maori membership website", so its hardly a ringing endorsement.

Thirdly, even if by some miracle the self-selecting samples were random, rather than consisting mostly of angry rednecks wanting to vent their spleen, the margin of error on the two polls would be 12% and 15% respectively. Nobody, not even a social scientist (who are actually quite strict about such things) would consider that to be a credible or useful result.

In short, this story is a perfect example of lazy journalism, of reporters copy-pasting a press-release without bothering to check the facts for themselves. Meanwhile, if the "hang 'em high" brigade are down to relying on a handful of people in a self-selecting poll on an obscure website for support, they really are clutching at straws.


"In short, this story is a perfect example of lazy journalism, of reporters copy-pasting a press-release without bothering to check the facts for themselves. Meanwhile, if the "hang 'em high" brigade "

It is also very similar to some of the coverage of the mercury energy decision.......oh hold on, werent you one of those people saying they should be charged without checking the facts as you were part of the hang em high brigade?

Posted by Anonymous : 6/06/2007 01:52:00 PM

If the poll was self selecting, surely you could take that as 68 people out of the entire number of users of that website wanted things changed? Seems they are pretty happy with the status quo to me.

Posted by Anonymous : 6/06/2007 03:12:00 PM

The Sensible Sentencing Trust are part of the problem. They only seem to have one solution to crime and anti-social behaviour - throwing people in jail for as long as possible.

I bet McVicar wouldn't want to stump up with the extra tax required to support his 'vision' either.

Posted by Anonymous : 6/06/2007 04:32:00 PM

Excellent work, I/S. I wish it were true that all social scientists would condemn such unscientific polls but I know of one in particular, who flaunts her PhD, who thinks internet surveys are "good enough" so long as they attract a large number of people. Perhaps the lure of being published outweighs the fear of peer criticism.

As for Sensible Sentencing, they seem to be everywhere these days. Garth McVicar was a keynote speaker at a recent Rural Women's conference in Rotorua, on a platform full of women with far better credentials for addressing that audience. I wonder who they are fronting for, is it the same rich white christians who are behind For the Sake of Our Children and Family First? I know they've held joint meetings and they seem to have a common anti-Labour agenda. I guess securing the conservative female vote is part of that. Unfortunately for McVicar, Jeanette Crean, a well-known Rotorua businesswomen, undermined his "lock 'em up" message with a moving account of some of the marginal people she had met on her road to success and how she'd helped them to turn their lives around.

Posted by Anonymous : 6/06/2007 10:49:00 PM

Since we're being "accurate" here, you want the self selected sample to be representative or unbiased. Random is the technique you use when you don't have any designed way to get representative. I like this acronym for self selected polls like this: SLOP (Selection Limited Opinion Polls). It appears occasionally in the literature.

Posted by Anonymous : 6/07/2007 10:49:00 AM

The silence of the SST with regards to the recent prison van bashings, as well as the rise of corporate crime is deafening.

The SST are in favour of tougher sentencing only when the offenders are brown, poor or both, and let us not forget the fact that the SST also favour privately run prisons, with the prisoners working to earn their "food and board". Does this mean that the SST desire a prison based cheap labour system? An army of prisoners to undercut free workers? It would be insteresting to see who is bankrolling the trust? Wackenhut Security? The People's Liberation Army?

And they are just as bad with regards to "blaming", ie not enough hidings, too much welfare, focus on rights, etc etc...

I suppose their hearts in the right place - who wouldnt want to walk the streets safely?

But, the SST does not speak for me. I belive that there needs to be a new approach to crime other than chucking people in prison for stealing a piece of bread, or allowing parents to bash their children.

Brendon "Millsy" Mills
Left-wing middle class deviant.

Posted by Anonymous : 6/07/2007 11:37:00 AM

Prior to the last election ran polls on which party people would vote for.

Typically National polled at 60%+ in polls attracting thousands of voters.

What happens a couple of days later?

National gets 30 odd percent of the ACTUAL vote.

The margin of error in online polls is virtually impossible to calculate as it depends on what people are voting on.

Unless people also note down their income, level of education, ethnicity etc and a random and even sample of those peoples votes is counted then online polls are straight up f*&cked.

Posted by yamis : 6/07/2007 05:00:00 PM