Published in Crikey, June 13 2007
reproduced with the permission of Crikey
Saulwick and Denis Muller write:
Christian Kerr does not need to pussy-foot around the question of push polling. There is no need for the media buffs to wonder if the Galaxy questions were push polling Ė at least one of them bloody was.
This is both a disgrace and a warning. A disgrace because this kind of work is unprofessional and is likely to mislead the public, and a warning because as the ultimate poll approaches the temptation for the spinners of all kinds to lean on pollsters to ask dicey questions will increase.
Letís look at this sorry example. The question asked and the results reported, as referred in Crikey on 7 June was:
The report states: "These surveys were conducted by Galaxy Research. The most recent survey was administered on the weekend of 1-3 June. The results are based on the opinions of 1021 voters. The data has been weighted and projected to reflect the Australian population".
You can reflect the demographic profile of the Australian population Ė but you canít make sense out of bad questions.
How is the question bad? To be unbiased each of these phrases should have been offered with its opposite and the respondent asked to choose between them. For example:
And so on. There are other ways it could be done but certainly not the way it was done. We hope there will not be more to come but we fear there may be.
[Read my contribution the next day]