Buy

Books
Click images for more details

Support

 

Twitter
Recent comments
Recent posts
Currently discussing
Links

A few sites I've stumbled across recently....

Powered by Squarespace

Discussion > Will Steffen deals with "skewed" poll data

Will Steffen, one of Australia's climate guys, from his bio:

Prior to taking up the ANU Climate Change Institute Directorship in 2008, Steffen was the inaugural director of the ANU Fenner School of Environment and Society. From 2004 to 2011 he served as science adviser to the Australian Government Department of Climate Change. He is currently a Climate Commissioner with the Australian Government Climate Commission; Chair of the Antarctic Science Advisory Committee, Co-Director of the Canberra Urban and Regional Futures (CURF) initiative and Member of the ACT Climate Change Council.

He's co-authored a recent article pertaining to - not climate - but an urban planning matter. A recent poll showed a majority of residents didn't support a new public transport proposal. This was the wrong result, hence a reanalysis was in order.

Light rail poll: What do the numbers really say? (June 25, 2015) "We reanalysed the numbers in the light rail poll and got a very different picture, write Will Steffen and Barbara Norman."

So what did reanalysis involve? Let's break it down.

Buried in the detail of that outcome is the most striking number in the poll – only 15.8 per cent of intended Liberal voters support light rail, while for all of the other groups of intended voters (Labor, Greens, Others and Undecided) support for light rail varied between 42.0 and 63.5 per cent.

They identify that party lines are highly correlated with responses - conservatives (32 %) don't want it, progressives/other want it. Urban planning, go figure.

The very low 15.9 per cent of intending Liberal voters who support light rail are indicative of an issue that has become excessively polarising along partisan political lines. Such a strong skew also has statistical implications for the poll itself, and can easily generate a misleading impression of what the poll numbers are actually showing.

Political affiliation of sub-groups can effect the result, or "skew" it in Will-speak, which can be "misleading" because, eh, because that sub-group - I mean the skewed one, not the others which aren't skewed - the skewed sub-group constituting 32 % of polled residents generates a misleading impression because... hey look, is that Mel Gibson riding a unicycle?! *runs*

In our reanalysis, we used all the percentages reported in the Canberra Times article in terms of level of support for light rail according to intended voting patterns. We then removed the intended Liberal voters from the analysis, giving a total of 980 remaining respondents to the poll

We removed every single response of those who support the current elected government, because they're skewed!

The results of the reanalysis are given below.

Support light rail 51.9 per cent
Oppose light rail 33.2 per cent
Undecided 14.9 per cent
This gives a very different picture. Now a majority support light rail. In fact, for the more-than-two-thirds of Canberrans who are are not intending to vote for the Liberals, there is very strong support for light rail, a nearly 20 per cent lead over those who oppose

After removing those we disagree with we get the result we want to see.

Statistics!

Jun 28, 2015 at 6:42 AM | Unregistered CommenterDavidA

Wow.

Jun 29, 2015 at 8:33 AM | Unregistered CommenterTinyCO2

Removing data because it's confounded with a separate issue (party politics) and thus is misleading is quite justifiable.
The error is in only assigning one side of the debate with the separate issue.

If they had concluded "this survey means nothing as it's all too politicised" then it may have been fairer and more accurate.
It may even have helped in creating a more accurate survey.

Jun 29, 2015 at 8:56 AM | Registered CommenterM Courtney

M Courtney, shouldn't we consider the goal of the study?

It seems to me here they just wanted to gauge where public opinion sat, not gauge the objective value of the proposal. I can't see how voting intention comes into it then. And poll rigging would be a separate issue, which wasn't suspected.

Here's a jocular follow up by one of our conservative columnists

Warmist Will Steffen’s rail ruse light on logic

Jun 29, 2015 at 10:47 AM | Unregistered CommenterDavidA

M Courtney, shouldn't we consider the goal of the study?

It seems to me here they just wanted to gauge where public opinion sat, not gauge the objective value of the proposal. I can't see how voting intention comes into it then. And poll rigging would be a separate issue, which wasn't suspected.

Here's a jocular follow up by one of our conservative columnists

Warmist Will Steffen’s rail ruse light on logic

Jun 29, 2015 at 10:50 AM | Unregistered CommenterDavidA

M Courtney
I think that similar logic is used by Brand (Russel), Church and others when protesting about the "Nasty Tories" getting into power. With the electoral system in place the largest group wanted the policies of the party in power, including keeping more of their money to spend how they see fit.

There seems to be, in my view, a difference between those on the right and those on the left. When a result goes against what those on the right wanted they get on with making the best of the situation they find themselves in, those on the left take to the streets in the hope of putting right the grave error committed by the uncaring/ill informed/stupid voters.

Jun 29, 2015 at 11:00 AM | Unregistered CommenterSandyS