Doug PollardThe Murky Waters of Online Polls
Polls in general are a strange, and a ridiculous way of gaining information. The major ones, like those driven by media conglomerates (like Fairfax and News Limited) are always of the Yes/No format, which in itself is an extremely problematic representation of information. Now with virtually every house hold surfing around the net, polls are everywhere with every man and his dog having their say with a mouse click.
But do these polls really represent what we think as a society?
If 63000 people were to vote on an online poll in one day, would you consider it an accurate representation of the Australian population as a whole? This is approximately the number of people voting on a Ninemsn poll, responding to the question "Would you drink recycled water?" The numbers were evenly split with the Yes vote having 31815 (50%) and the No 31966 (50%).
These numbers would tend to imply that as a nation we are evenly split, half of us would not drink recycled water and the other half would. However, these numbers have been deliberately engineered by the phrasing of the question. Had the question more accurately represented the situation that it is asking about (i.e. the water shortage in Tawoomba), there would have been completely different statistics. For example, had the question asked "Would you drink recycled water if it was expected that fresh water would run out in 6 months?" a greater proportion would have voted for yes. Alternatively if the question were "Would you drink recycled water considering the possible health hazards?" it is also obvious what the results would be.
This is clearly the nature of all polls, all of the yes/no variety of questions are engineered (whether the voter is conscious of it or not) to get certain answers. Here are some examples (again, specifically of the Ninemsn variety):
Tuesday, 02 May 2006: Has there been a cover-up surrounding Private Kovco's death?
Yes: 54258 (82%)
No: 12136 (18%)
I find it difficult to believe that some 66000 people have been privy to the intimate details of the situation. How could it be asked of someone completely uninformed as to whether there has been a cover-up or not? The question in itself demands that there has been a cover-up, if there was no such instance there would be no reason to speculate as to its nature. It would be a much more appropriate question to ask "Do you think there have been serious flaws in the investigation surrounding Private Kovco's death?" The voter at least has some information as to the nature of the investigation and its circumstances, in particular due to media reporting.
And another:
Tuesday, 21 March 2006: Do you think police should be patrolling high schools?
Yes: 31110 (57%)
No: 23595 (43%)
Why should police patrol high schools? How can you possibly justify wasting an absolutely massive amount of resources for one of the smallest areas of crime? And yet 57% of voters said yes, the essential reason being that they voted yes because the question was asked, implying that it may be necessary for police to patrol schools. The yes/no format does not allow a voter to express any particular view. The question also relates to a system of managing a school environment for the safety of the majority of students. To get an accurate representation of parents’ feeling on the safety of schools, a question like “do you think schools present a physical risk to the majority of students?” would be a more appropriate and give (slightly) more of an informed insight.
And finally, this poll demonstrates the amount of bias present within this system of representing information:
Friday, 24 February 2006: Should migrants who don't observe Aussie values be kicked out?
Yes: 91815 (86%)
No: 14874 (14%)
What is an "Aussie Value"? I think multiculturalism is one of them. Religious tolerance maybe? Respect for self determination (that is, a person determining and acting on their own desires)? A "Fair Go"? This question and its overwhelming support in the yes category seem to sneer in the face of all of these ‘aussie’ values. This brings up a question for later debate, does the faceless medium of the internet allow us to express the views that we really hold?
In summary, polls in general are a horrible way of representing information, and do not accurately represent a population. They are all engineered with inherent bias, particularly media driven polls and ones that appear on this special medium we call the internet.
But do these polls really represent what we think as a society?
If 63000 people were to vote on an online poll in one day, would you consider it an accurate representation of the Australian population as a whole? This is approximately the number of people voting on a Ninemsn poll, responding to the question "Would you drink recycled water?" The numbers were evenly split with the Yes vote having 31815 (50%) and the No 31966 (50%).
These numbers would tend to imply that as a nation we are evenly split, half of us would not drink recycled water and the other half would. However, these numbers have been deliberately engineered by the phrasing of the question. Had the question more accurately represented the situation that it is asking about (i.e. the water shortage in Tawoomba), there would have been completely different statistics. For example, had the question asked "Would you drink recycled water if it was expected that fresh water would run out in 6 months?" a greater proportion would have voted for yes. Alternatively if the question were "Would you drink recycled water considering the possible health hazards?" it is also obvious what the results would be.
This is clearly the nature of all polls, all of the yes/no variety of questions are engineered (whether the voter is conscious of it or not) to get certain answers. Here are some examples (again, specifically of the Ninemsn variety):
Tuesday, 02 May 2006: Has there been a cover-up surrounding Private Kovco's death?
Yes: 54258 (82%)
No: 12136 (18%)
I find it difficult to believe that some 66000 people have been privy to the intimate details of the situation. How could it be asked of someone completely uninformed as to whether there has been a cover-up or not? The question in itself demands that there has been a cover-up, if there was no such instance there would be no reason to speculate as to its nature. It would be a much more appropriate question to ask "Do you think there have been serious flaws in the investigation surrounding Private Kovco's death?" The voter at least has some information as to the nature of the investigation and its circumstances, in particular due to media reporting.
And another:
Tuesday, 21 March 2006: Do you think police should be patrolling high schools?
Yes: 31110 (57%)
No: 23595 (43%)
Why should police patrol high schools? How can you possibly justify wasting an absolutely massive amount of resources for one of the smallest areas of crime? And yet 57% of voters said yes, the essential reason being that they voted yes because the question was asked, implying that it may be necessary for police to patrol schools. The yes/no format does not allow a voter to express any particular view. The question also relates to a system of managing a school environment for the safety of the majority of students. To get an accurate representation of parents’ feeling on the safety of schools, a question like “do you think schools present a physical risk to the majority of students?” would be a more appropriate and give (slightly) more of an informed insight.
And finally, this poll demonstrates the amount of bias present within this system of representing information:
Friday, 24 February 2006: Should migrants who don't observe Aussie values be kicked out?
Yes: 91815 (86%)
No: 14874 (14%)
What is an "Aussie Value"? I think multiculturalism is one of them. Religious tolerance maybe? Respect for self determination (that is, a person determining and acting on their own desires)? A "Fair Go"? This question and its overwhelming support in the yes category seem to sneer in the face of all of these ‘aussie’ values. This brings up a question for later debate, does the faceless medium of the internet allow us to express the views that we really hold?
In summary, polls in general are a horrible way of representing information, and do not accurately represent a population. They are all engineered with inherent bias, particularly media driven polls and ones that appear on this special medium we call the internet.
Kylie