Wednesday, October 04, 2006


The following appeared in the Melbourne newspaper The Age:

THE Australian Democrats are refusing to publish an online survey about God and government after a campaign by Christian groups "skewed" the results.

Democrats leader Lyn Allison said 40 times the usual number responded to the survey, and overwhelmingly took the position advocated by some Christian leaders. Normally the party would be happy with 1000 responses, but the church and state survey got 40,000.

Senator Allison said it was ironic that a survey on the influence of churches should attract such an intense effort by churches to apply influence.

Christian groups have urged the Democrats to release the results, saying it was dishonest that a party that was founded on a claim "to keep the bastards honest" should keep the results secret because they were not what the Democrats wanted.

[The Age - National]

When this appeared in the Sydney Morning Herald's blog column "Stay In Touch", there was vigorous (even rancorous) debate about whether the Democrats should publish the results, and the significance thereof.

But what intrigued me more was the fact that the Democrats typically expect only about 1000 people to respond to their online surveys.

Online polls are actually fairly common, but as a few comments on the SMH blog pointed out, they hardly qualify as good scientific or statistical method, and no one should make too much out of the results. One comment noted that the Democrats website itself states "online surveys are useful because they are fast, easy and inexpensive but they do not typically gather in-depth, rigorous scientifically valid information" [here] and then asked the obvious - "then why do it in the first place"?

This then got me thinking - obviously the Democrats have done these surveys before, and have a fair idea of the typical response pattern. Also fairly obvious is that they have published the results on previous occasions, when the number of responses has been around the 1000 mark. Given that 1000 responses is about one-hundredth of 1% of the voting population of this country, do they believe that the results of such polls are in any way indicative of the overall views of the voting public? (As distinct from "rigorous scientifically valid information", I might add.)

It seems a fairly safe deduction that online polls will usually measure predominantly, perhaps almost exclusively, the opinions of those who frequent your website. Who else is going to go to your website? SBS Sport frequently polls its viewers on the SBS website, but if you don't watch SBS Sport or frequent the SBS website, you won't even know the poll is there, let alone participate in it. Ergo, the results reflect the opinions of the watchers of SBS Sport, not the opinions of sports enthusiasts in general.

In the same way, the Democrats' online surveys, which are not advertised broadly in the media, will usually measure only the opinions of those people who are aware of the poll, i.e. those who frequent the website or who keep track of Democrats-related news and events. In other words, those who are interested in the Democrats and their policies, which will primarily be supporters of the party.

It's therefore quite interesting to see that they will not release the results of the latest online survey because these results have been "skewed" - the results of the Democrats' online surveys are almost certainly skewed in any case, normally in the direction of those who agree with the Democrats.

In a nutshell, online polls measure the responses of interest groups of one kind or another. In the case of this latest poll, the subject of the poll intersected another, larger, interest group - Christians. It's a shame that the Democrats won't release the results.

And now I find myself being a touch cynical. I find it hard to credit that the Democrats would not be aware that the usual results of their surveys are dominated by Democrats supporters, so why publish results which you know do not in all probability reflect the opinions of the wider public? Sadly, there is an obvious answer.

The upshot of the Democrats withholding the survey results is they have made themselves look either incompetent, churlish or duplicitous, depending on who you talk to. If political parties are going to be taken seriously on the web, they need to avoid silly games like this, and think very carefully before conducting online surveys.

A final observation in light of all this, useful for both online surveys and elections:
Don't vote, it only encourages them. (Author Unknown)

No comments: