Online polls are evidence only of the existence of online polls, oh, and that any conclusions drawn from them are absurd.
A recent online poll conducted by the New Zealand Herald showed that public support for gay marriage , counter to earlier polls which showed huge public support for gay marriage, had dropped to 48%.
Okay, an online poll was conducted. That is all we can reliably say about this poll. However, the Herald began using it in stories to demonstrate that “support had dropped” as the parliamentary debate about the proposed gay marriage bill went on.
Can we really say this? Can the results of an online poll be used to draw such conclusions? Or indeed, any conclusions at all?
Methodology? What methodology?
The reliability of any survey or poll comes down to the methodology employed in collecting the responses. Is it a representative sample of the population? Can it reliably ensure only one response per person? What is the sample size? Is it statistically significant?
Online polls typically violate all of these measures and controls. Other than that, it is totally possible to ‘game’ an online poll by rallying supporters to vote one way. My conservative uncle even tried to get me to vote like this on the Herald digipoll in question.
With regards to said Herald digipoll on gay marriage, the sample is self-selected — in other words people voluntarily choose to vote on it. Who would choose to vote on such a poll? Not only those who are emotionally engaged (which touches both sides of the argument) but specifically, those who feel incensed by gay marriage, predominantly those of a conservative persuasion.
This is the same phenomenon that research shows has been employed on social media: negative views on an issue outnumber positive views.
Probably the best illustration of this is review websites. For example, 10,000 people go to a hotel — 9900 have an enjoyable experience while 100 do not. Which reviews are most likely to end up on the hotel review website? Negative views on an issue motivate us to speak out and “right the wrong” so to speak.
Absurd methodology, absurd conclusions
The Herald did repeatedly refer to the result of the Digipoll as a change in opinion — something that cannot be justifiably determined from a poll that was clearly being gamed.
Worse still, Bob McCroskie, director of Family First NZ took these results as confirmation that his campaign of intolerance has swayed public opinion in his direction and that now it’s just politicians who are failing to realise this.
It is ridiculous to suggest an online poll has any correlation with real public opinion. Emotionally heated issues such as gay marriage are always going to attract those who are emotionally invested in the debate and for the most part it is the outraged — the conservative minded who feel gay marriage will signal the end of civilisation.
The unreliability of online polls, indeed any self-selected poll has been noted before. However, media outlets without actual data are using these polls as if they have some validity. This in turns feeds into people’s biases which in turn doesn’t help fuel a productive public debate.