We have seen in recent days how the findings from public opinion polling can directly and malignantly influence public policy.

Thanks to the leaked Liberal Party research conducted by the party’s pollsters Crosby/Textor, we can plainly see that John Howard’s recent series of attacks on the States and Territories (intervention in Indigenous affairs, Mersey Hospital, Queensland local government) is driven by poll data suggesting there might be a vote or two in it for him if he attacks the Labor-governed States and Territories.

The three examples are all rotten public policy.

The onslaught on Indigenous communities has, among other things, necessitated over-riding the Racial Discrimination Act. The hospital takeover has been described by some health experts as a subversion of public health. The proposed merger of local government areas will be presented to the Queensland Parliament, which has constitutional power in this area.

All this adds to the debate over the roll of opinion polls in our democracy. Some bloggers are highly critical: “Stephen” (Crikey 5 July) described polls as “the poll(ution) of ideas and debate”. He quoted Pierre Bourdieu’s argument that “widespread use of polls tends to devalue other forms of political action such as strikes, demonstrations, or the very elections that opinion polls try to mimic”.

This argument sounds a bit romantic. It is as if strikes and demonstrations are in some way more representative of public attitudes than what the polls tell us. Except for very large expressions of public opinion such as the anti-Vietnam War Moratorium demonstrations, it is very unlikely that strikes and marches are representative of wider public opinion at all.

Strikes in particular are usually conducted by people with a direct financial interest in a very specific industrial outcome, the merits of which are a mystery to the general public. This is not to say that they they are improper but that they do not necessarily reflect widespread support.

Forms of political expression have changed. These changes, like so many others, have been driven by technology: the almost universal access to telephones, the invention of computerised polling systems which allow polls to be conducted more rapidly and reliably than ever, and the capacity to make the findings public almost immediately via the internet.

It is quixotic to oppose these developments. The challenge surely is to assess their strengths and weaknesses so as to promote the first and try to minimise the effects of the second.

One way in which polls are said to weaken our political process is that they are a construct of the pollsters and programmers who devise them and the media who publish them. It is argued that polls thus create an artificial reality, and any connection with the “real” reality is merely coincidental.

Might this be true?

Elections provide the only opportunity to test this and the history of the main polls in Australia over the past quarter of a century is that most get within sampling error of the actual election result most of the time.

This suggests that on voting intention at least, polls are generally not operating in a parallel universe.

But what of attitudinal surveys, say on issues such as immigration or climate change?

Some people know a great deal about a particular subject and others know very little. The more a subject is familiar to a person, or the more it affects the person, the more he or she is likely to know about it or to have thought about it.

If a poll asks people about a subject that they have had little chance to think about, the initial responses are likely to reflect respondents’ emotional responses. This is not meaningless. It does represent a real, if not necessarily reasoned, response. If the subject is pursued over time, people’s views about it may mature as they learn more about it or become more familiar with it. Thus the answers may reflect real trends in attitudes.

All this depends on the questions being fairly designed, the sample being properly constructed, and the data being honestly interpreted.

After that, it is a matter of what the politicians and media do with it. If they use it to create headlines to try to make political capital, that is a question not about the effects of polling but about the effects of how poll data are used.

They are distinct questions and need to be considered separately when assessing the impact of polls on our political processes.