SIZE matters. There is all the political and policy difference in the world if it is 44pc or 10pc of people with young children who would be better off not working.
But it is not everything. The unprecedented row over the ESRI working paper on the costs of working raises important issues, irrespective of the number. They include costs in the economy, social welfare policy and how public debate is conducted.
On a day when the troika's latest progress report surfaced, it even has a bearing on the fears that, even though we are two-thirds of the way there, Ireland may not succeed in its programme to fix the public finances.
As to the numbers row, the situation became, if anything, more confused last night. The ESRI issued a statement saying that a revision by an un-named expert "referee" had come up with the 10pc estimate. Prof Tol said he knew nothing about it and no-one had given him any reason why his research was flawed.
Even if it is 10pc, the devil would be in the detail. How much extra disposable income do the 34pc in between get from working? Finding out is not simple.
Dr Tol had to take one set of data on incomes, and another on benefits, and try to match similar people from each to figure out how they would be affected by working or not working.
The data collected is in line with EU and UN statistical methods. One cannot complain about that but it does mean that conclusions, even in a final paper, would be best estimates, rather than precise numbers.
Reducing them by three-quarters, to 10pc, could only mean that something was wrong with the sums.
Statisticians regularly argue about "methodology", never mind the conclusions. The big question is why the arguments could not take place in an open, reasoned manner.
Part of the reason may be the character of Richard Tol himself. He made no secret of his frustration at the lack of public research or evidence behind most Irish government decisions. He is not alone in that -- indeed it is hard to disagree -- but his Dutch bluntness did not help.
He was also blunt in his views on the ESRI, from whom he parted on bad terms, accusing it of a lack of transparency and independence. The institute is not a "government think-tank," but the Government is by far its biggest customer for research.
Yet the problem goes deeper than paying pipers and calling the tune. Irish governments, of all persuasions, do not take kindly to public pronouncements, whether right or wrong, which cause them political difficulty. Who can forget Brian Cowen's four-letter description of the consumer council, overheard on a live microphone, when its defence of, er, consumers, trod on political toes?
The important issues raised by the paper risk being buried in a welter of accusations, denials and recriminations. Fortunately, there was a reasoned, reasonable response from a trade union economic researcher, Michael Taft, which attracted quite a lot of attention.
He rightly criticised knee-jerk reactions that this welfare/work conundrum means that social welfare payments are too high. The other two bits of the equation are wage rates and costs, especially, but not exclusively, for childcare. Which should take the strain?
BUT that makes things even more uncomfortable. Irish wages are high by EU standards, not low. Even after the austerity Budgets, net taxes (allowing for universal payments like child benefit) are among the smallest. So disposable income remains among the highest in the EU, and it is going to have to fall to fix the budget deficits.
But Irish costs are also among the highest in the EU. The two things are connected and reflect the impact of all those billions of borrowed money during the bubble. Prices have been falling but not as fast as consumers' purchasing power is shrinking.
Matching this reduction in workers' living standards with equivalent falls in social welfare payments is a political nightmare. One lesson from the ESRI farrago is that we should stop being so political about it, and try to base what will be unpleasant actions on clear factual analysis.
Dr Tol may or may not have got his sums right, but he was correct to address the question. If the ESRI thinks he was wrong, it now has a duty to supply the answers.