The earliest studies of election campaigns, using repeated surveys of the same people, confounded the researchers when their results suggested that campaigns had little effect on the outcomes, because the great majority of people had already decided how they would vote.
More recent wisdom is that campaigns do matter.
They do help people to decide, but that is not to say that those decisions are unexpected.
Campaigns are one of the factors that help people to make a choice that would have been predicted before the election, even though those individuals might say they had not made up their mind.
This is because people filter information in the light of their predispositions, and interpret it according to their personal biases.
This is not to say that we do not sometimes see changes in the distribution of party support through campaigns (recent UK elections are examples), both because some who claim to be undecided make up their minds and because some who are decided then change their minds.
Voters also become more knowledgeable, with those who know least at the start learning most, but research also suggests that they also come to think a little more positively about the democratic process and their role in it.
One set of available evidence from Irish general elections suggests that the campaign is when most people decide who to vote for.
Exit polls carried out for RTÉ since 1997 have asked voters emerging from the polling station at what point did they make up their mind which party to support.
[The question is: Can you remember exactly when you finally made up your mind about which party or independent candidate you would give your first preference vote to in this election?]
Overall, taking these elections since 1997 together, 43% said they decided sometime before the election was called, with 57% deciding during the campaign.
A quarter decided in the first couple of weeks after the election was called, with 18% deciding in the last week and 14% making their minds up in the last couple of days.
Clearly, on this evidence, with two weeks to go there is a lot to play for.
Those who make up their mind early tend to stick with the party they supported last time, although 20% of the early deciders did opt for change.
In contrast, only 40% of very late deciders opted to stay with the same party with the majority opting for change.
Those deciding in the last week look very similar, while those deciding in the first week or so of the campaign tend to stay with the same party.
The timing of decisions has not changed all that much over the last quarter of a century.
(Were evidence to be available, it might well be that decisions were made earlier 50 years ago, when people identified more clearly with a party, but we simply do not know.)
Since 1997, there is some sign that decisions are being made later.
In 2016, 64%, almost two thirds, decided either relatively early in the campaign or before it started, compared with 73% in 1997.
However, 2011 saw only 59% deciding early, so there is fluctuation in the downward trend.
Early deciders in every election were more likely to stay with their party and late deciders more likely to change, with those who decided before the campaign significantly most likely to stick with the same party.
Which parties do the campaign deciders tend to favour?
The short answer is that they tend to favour FF and FG, just as do pre-campaign deciders, but significantly they do not do so to the same degree.
Over the last five elections FF did much better among pre-campaign deciders than those who decided during the campaign, and the same is true, though to a lesser extent of FG and SF.
In contrast, those deciding during the campaign were more likely than early deciders to support other parties, including Labour, and Independents.
If we look at the most recent elections, we see the same pattern, but now FG does much better among early deciders with FF just marginally better.
Even so, for almost all parties, most of their support comes from those who decided after the campaign has started.
One exception is SF, at least in 2016 who got just 47% of its vote from campaign deciders, as did FF before 2011.
This analysis suggests there is a very substantial proportion of the electorate who are still up for grabs, although another set of data suggests we should not expect to see huge changes.
In 2016 several companies ran more than one poll and in some cases, several polls through the campaign.
Remarkably, there are few signs of significant changes in support for parties.
All companies showed increases in the FF vote of 1-2%, and a similar fall in support for SF.
There was no such pattern for FG, and in general some companies showed an increase and some a decrease for other parties.
All changes were within stated margins of error, so there is no hard evidence that anything happened in terms of party support levels.
Much the same is true of 2011, although there was more evidence of a decline in Labour support over that campaign.
Perhaps we should not expect to see a lot of change in next week's polls, but for those who are spending their time campaigning solidly there is at least some reason to think that minds are still being made up and there are potential gains to be made.
And if not, and this simply leads to voters being more positive about politics, that is a worthwhile outcome.
