A week or so before last Friday's referendum on the status of local government RTE broadcast the results of a vox pop from Trim, home base of the minister responsible for the referendum. Only four out of 40 people asked knew there was a referendum.

Last Friday, over 100,000 of those casting a European or local election vote returned empty ballot or spoiled ballot papers, many apparently complaining they did not have sufficient information to make a judgment on the matter. This compares to 33,000 spoiled votes on the Amsterdam Treaty and 10,000 on the Belfast Agreement.

Who is to blame for this unhappy situation? Until quite recently, the Government played a leading role in referendum campaigns. Following the Supreme Court's judgment in the McKenna case, which was that the Government had no right to spend our money to promote one side of the argument, the job was eventually turned over to a Referendum Commission.

Its task is to explain the matter of the referendum as simply and effectively as possible, ensuring that the arguments of those for and against any amendment are put forward in a way that is fair to all concerned. In other words, it should provide us with sufficient knowledge to make an informed decision at the ballot box. To date, it has been active in the referendums on the Amsterdam Treaty and the Belfast Agreement as well as the vote last week.

Many voters complained they knew too little about the Amsterdam Treaty to make up their minds about it, and this was obviously repeated this time around. So where has the commission gone wrong?

One problem with the Amsterdam Treaty was its complexity. Treaties are legal documents, not written primarily for public consumption, and arguably it was always going to be hard to get the main points of the treaty over to the public. However, this last referendum was pretty simple and straightforward, giving a constitutional status to local government and ensuring regular elections. Why did this cause a problem? One reason was that the decision to hold this referendum was very much a last minute affair.

The commission complained in its report of the 1998 referendums that it was not given enough time to do its job properly, and this seems to have been ignored. A second factor is that the commission appeared to forget what it had learned on the last occasion, which is that newspaper advertising was very ineffective.

Given this, and the fact that the central message of the massive international research done on political campaigns is that television provides most people's primary source of political information, it seems strange the commission put so much faith in a newspaper campaign, and a pretty yawn-inducing one at that.

It could have done much more, as it did in 1998: leaflets and booklets to the home and in libraries and so on, and TV spots, for instance. These would certainly raise the profile of the referendum, but there are serious questions as to whether the problem is simply a quantitative one.

My feeling is that the current strategy makes false assumptions about the way in which people absorb information, and the sort of information they need. The commission adopts the position that people need some details about the proposal and need to hear the arguments for and against. People can then reflect and make up their minds.

Much of what we know about how political campaigns work, and how voters actually do make up their minds, does not fit very well with this viewpoint. More knowledge in itself does not necessarily help. In fact, at times it may be confusing.

Voters generally act on very little information and as several recent studies have argued, can still act pretty sensibly.

Research on how far people can make sensible decisions in such circumstances suggests that people use "cues" for sorting out candidates, and sorting out propositions. They think about who they should trust, for instance, and can ask themselves: why should this person be lying to me? The disembodied overly rational approach of the commission makes this impossible.

People also need to engage with information in order to learn anything.

Learning must be active. Providing booklets or newspaper ads may put information into the public domain, but does nothing to convince people they should take any notice of it. The commission admitted that very few people read their leaflets in 1998.

Various ways to run referendum campaigns are detailed in a recent book edited by Michael Gallagher and Pier Uleri entitled The Referendum Experience in Europe (1996). The Government has chosen something akin to the Australian model, although somewhat more de-politicised.

A different model is the Swiss, where no money comes from the public purse but everything is left to interest groups, who must rely on their own resources. We have moved away from that, both in referendums and elections, largely on grounds on equity. A third way is the Scandinavian one, where public money is used but responsibility for campaigning is largely subcontracted out to civic bodies - interest groups, promotional groups and political parties.

In such a context voters know who is making the arguments, and the various groups can direct their campaigns to the aspects of the decision most salient for the people they target. Such a campaign has more life, and is more likely to attract attention, and provides voters with better information. It also encourages political participation and the institutions of civil society in a direct way.

Of course all this can happen anyway. There is nothing to stop organisations campaigning, and a good case can be made that the government parties in particular fell down on the job completely this time. However, if the referendum device is to be used, and public money is to be spent to make it more effective, it should be channelled through civic organisations, not used to bypass them.

Dr Michael Marsh is Head of the Department of Political Science at TCD