We are fortunate in the quality of the political polls carried out here by several professional organisations. But all polls are subject to a number of qualifications.
First of all - and most obviously - polls provide no more than snapshots of political opinion at a moment in time - and during an election or referendum campaign, opinions can change quite rapidly. Thus it is clear that in the first divorce referendum in 1986 no less than one-quarter of voters reversed their initial pro-divorce stance within three short weeks in response to the campaign by opponents of divorce concentrated on the issue of property rights of spouses.
Second, when asked how they voted at the last election as well as how they intend to vote in the impending contest, some of those polled who in fact did not vote in the last election nevertheless claim to have done so - and these respondents usually also claim to have voted for the winning party on that occasion!
I note that in the most recent Millward Brown IMS Poll published two weeks ago this phenomenon is evident. Thus, whereas in 2002 the official turn-out was only 63 per cent, in this poll no less than 76 per cent of those interviewed claimed to have voted in that election. Not all of this difference can be accounted for by failure to delete from the electoral register the names of people who had moved house or who had died.
Moreover, almost 33 per cent of all those polled claimed to have voted for Fianna Fáil in 2002 whereas in fact less than 26 per cent of those on the register gave a first preference vote to that party. By contrast the poll figure for the other parties was 35 per cent - very close indeed to the actual figure of 36 per cent of voters who gave candidates of other parties their first preference votes. (Note that these figures are all proportions of the total electorate and of the total sample polled - not of those who actually voted or who in the poll said that they voted).
In their recent poll the Millward Brown/IMS organisation sought to deal with this distortion by adjusting downwards by four percentage points the stated voting intentions of those who said they would give their first preference vote to Fianna Fáil in the forthcoming election.
One further problem is that poll data does not allow for the fact that supporters of some parties may be more inclined than supporters of others to actually turn out and vote on election day. Polling in past elections suggests that Fine Gael supporters are more likely than others to cast their votes, with the result that polls tend to underestimate that party's actual support on polling day - a factor which the polling companies also try to allow for in the adjustments they make to their crude data before they publish the poll results.
A final point that needs to be made is that national polls using a sample of 1,000 have an accepted error margin of three percentage points, with, moreover, one poll in every 20 liable to an even larger margin of error. Of course a three-point margin of error has a much greater impact upon the single-figure percentages for smaller parties than it does on figures for the larger parties.
As well as the national polls, nowadays there are also constituency polls, some of them carried out by party supporters employing samples as small as 400, but others carried out by polling companies on behalf of local newspapers, employing samples of 500. With samples of this size the margin of error is of course larger: four percentage points rather than three. And because the votes may be spread over a dozen or more candidates, the margin of error in the case of any individual candidate can be very large indeed.
However, these constituency polls have one significant advantage over national polls: those interviewed are given a mock ballot paper with all the likely candidates listed, and they are invited to fill in their voting preferences. This procedure is much more realistic than simply asking people to what party's candidate they will give their first preference vote. And of course these mock ballot papers with all their preference votes can then be subjected to successive counts until a final notional result emerges.
Although the data from each of these individual constituency polls is subject to a large margin of error, nevertheless if one adds up the figures available from eight constituencies outside Dublin that have not changed their boundaries significantly since 2002, it is possible to get a fairly good sample of 4,000 people, covering one-quarter of constituencies outside Dublin.
These polls were compiled independently of the national polls between late February and mid-April, and what they suggest is that Fianna Fáil outside Dublin had by then lost about one-seventh of its 2002 election support, bringing its vote outside Dublin down from 43 per cent to 37 per cent.
By contrast Fine Gael seems to have increased its vote in these non-Dublin constituencies by about one-quarter, to about 30 per cent, and the Sinn Féin and Green Party votes had risen respectively by 40 and 60 per cent, while both PD and Independent support outside Dublin appears to have fallen by about 40 per cent. These figures suggest very significant shifts in party support since 2002.
Although no similar constituency data has been published with respect to Dublin, (and in the national polls the regional samples for the capital are so small that their figures are simply not reliable enough to use for this purpose), nevertheless if one assumes that party support in Dublin has moved broadly in harmony with that elsewhere in the country, the picture derived from these constituency data suggest that Fianna Fáil's national support in March and early April may have been about 35 per cent, with Fine Gael/Labour together about 40 per cent, Sinn Féin at 9 per cent, the Greens at 6 per cent, the PDs at just over 2 per cent and Others at about 8 per cent.
These figures, although derived from a quite different polling source, are in fact astonishingly close to those from the most recent TNS/mrbi national poll.
Although the campaign to date does not seem to have given cause for much change in public attitudes since these two sets of polls were taken, nevertheless, during the election campaign of the next three weeks, it is of course possible that party support might shift away from this pattern.
All is still to play for!