No matter how you view it, Bush is on the nose

Finally, there are the ways in which changes in response options can inform our understanding of what it is we are trying to measure. Asked how important the US alliance under ANZUS was for protecting Australia's security, 79 per cent said it was very important (37 per cent) or fairly important (42 per cent); few said it was not important (11 per cent) or not at all important (5 per cent); while 6 per cent didn't know or refused to say.

Contrast this with a poll, conducted a few weeks earlier for the Lowy Institute, in which no more than 63 per cent said our alliance relationship with the US was very important (36 per cent) or fairly important (27 per cent). In that poll, 27 per cent said the alliance was somewhat important, 9 per cent said it was not at all important and 1 per cent didn't know or refused to say.

An explanation for the contrast is not far to seek. Unable to choose Lowy's response of somewhat important, most of those in the USSC study who might have done so chose the more positive option instead, hence the higher proportion (41:27) choosing fairly important. A smaller proportion opted for not important or found the choice too difficult (6:1). In short, the USSC poll reported a different result, one more favourable to the alliance, because it offered respondents a starker choice.

As with differences, or the absence of them, generated by questions that are differently worded or by the order in which questions are asked, the consequences of varying response options can be significant, not only statistically but politically, too.