The importance of questions

Almost nothing is more important in polling than setting the questions.  And two recent reports of polls reminds me of this:

The first on cellphone use is reported as:

A new Research New Zealand (RNZ) poll showed 86 per cent of New Zealanders believed that phoning/texting and driving should be illegal.

Now they don’t state the actual question asked, but my first reaction is that by including texting and phoning together could lead to very different results to asking separately. Also talking about “phoning” rather than “using a phone” could get different results as one conjures up dialling out on a phone, rather than merely answering a call. I think texting someone while driving is pretty reckless, but answering a phone call is far less of a distraction.

A second is on Easter Trading:

Research New Zealand surveyed 500 people to find out whether they thought shops should be open on Good Friday and Easter Sunday.

Director Emanuel Kalafatelis says only 32 per cent of those polled want the law to change.

Now if you asked “Should shops have the option of opening” as opposed to “Should shops be open”, you may get different results. For example I don’t think shops should open on ANZAC Day morning, but I don’t think the law should prohibit them.

Now I am not saying in either of these cases, that a differently worded question would change the majority support into majority opposition. But that the questions, as reported, may make people more likely to respond a certain way.

This is not a criticism of Research New Zealand, a very good company.  Apart from anything else I have not seen the actual questions asked – only a media summary. The reason for the post is to highlight how important the wording of questions are, and how media reports which do not include the actual question used need to be treated with some caution.