Political Mummery and Journalism

I watch Insiders most Sunday mornings and generally speaking find it a easily consumed form of political commentary. This week left me shaking my head a little for they decided that they'd do a summary of the months polls. Now this activity I don't actually have a problem with but the way it was introduced and spoken about, that I did have a problem with. Before I launch into it, my issue was with the flipancy with which it was done. If it was worth investigating and discussing then do it properly. The end result really told me nothing. Sorry if you are offshore, I'm not sure if ABC allows non-local ISP to view video.

The Poll of Polls segment was introduced with the statement; 'While single polls are not meaningless, they're not particularly meaningful either.' This grabbed my attention. It sounded somewhat like that old comment we'd hear at university ... 'Shit in ... Shit out'. How meaningful are the trends to come from polls that are apparently not particularly meaninful in the first place? Yes, trends will be identifiable, but if we are uncertain of the initial data, then can we rely on the conclusions. The invited guest makes the comment that he'd 'averaged and weighted all the polls that are available'. On what basis was this done? The number of responses in the initial data set? the locations of the survey? The distribution of votes or the perceived reliability of the polls? If you want your viewers to pay attention to your information and political commentary then at least sound as though you take it seriously yourself.

Part of my dislike of political polls being used as an accurate reflection of the population's attitudes stem from my experiences of being polled. The polling was via phone call as dinner was beign prepared and one question stuck in mind. It was the most obvious of them all ... on a two party preferred, who did I prefer. Being annoyed that I'd fallen into the trap of a poll (something I avoid on principle) by just not really registering until it was too late, I replied ... neither. This was an unsatisfactory answer, 'No' I was told 'You must pick either Labor or Liberal' so I dug my toes in, as you do, and insisted that I preferred neither. 'But I can't put that down' the lass on the phone protested. 'Well, you have a problem. For starters, we don't have two parties as the main parties in the House of Reps, we have one (Labor) and a coalitian (Liberals and Nationals) who are two distinct parties working together with a smattering of other groups such as the Greens and Family First and then there are the independents. So for starters, your question is based on an eronious assumption. Secondly, you are offering a forced option of two when there may well be a third political party, of which we have many, that I'd love to see rise to power and gain ascendancy in the House of Reps so, my answer remains ... neither'. There was silence on the phone ... 'so I'll put you down for the Liberals'. ... How accurate are the polls when the people surveyed may well not have their opinions accurately recorded, or who, like me are annoyed enough to deliberately misslead the questioner.

It is interesting to note that Roy Morgan Research who have been polling the Australian electorate claim to have had the most accurate polls from the 2007 federal election while Galaxy Research state that 'our polls have been recognised as the most accurate in both federal and state elections'. Pardon my cyinicism, is it even possible for them both to be the most accurate? Meanwhile back in 1975, John Ray concluded that 'polls are in general no better than their market demands ... [meaning] ... a generally fairly low standard except so far as the sampling itself is concerned.' Data Analysis Australia came back to this point some 30 years later stating that 'confidence in polls is a belief that the results will be close to the actual election results' concluding from this that as polls were not going to go away, it is important that they are the best quality possible 'Ideally the users of polls would be able to recognise the differences between the good ones and the poor ones. This means asking for details on how they are conducted'.

The poll of the polls, I had concluded on Sunday, failed this test by not making it clear how this poll was constructed and what assumptions underpinned the weighting of polls.