Average V Median wage Ireland?

Are you ever going to give up? How could an earnings series include retired and unemployed? By the way, Eurostat do not carry out surveys. Their figures come from a survey carried out by CSO.

From a calculation point of view, the problem is that Revenue knows how much each person is earning, but it doesn't know whether each person is a full-time worker or not. To calculate the median earnings for full-time workers you'd have to (1) take the Revenue dataset for all earners; (2) drop the data points relating to earners who are not full-time workers; and (3) calculate the median from the remaining dataset, which would represent full-time workers only. Neither Revenue nor CSO have the information they would need to complete step (2); simply knowing the total number of full-time workers is not enough to do this.

In reality, that's not in fact how median earnings are calculated in the real world. Where it's done, I think it's mostly done by stratified sampling of household surveys - as in, you interview a number (i.e. several thousand) of households in detail, asking them questions about all the things you're interested in, including in this context (a) whether the adult members of the household work full-time; and (b ) how much they earn. If you're being thorough, you check this by looking at their payslips, etc, rather than relying on their own estimation of whether they work full-time and how much they earn. You then statistically weight the results to reflect the demographics of the entire country. And from that you can calculate a median.

It's not perfect, but it's close enough to give you a meaningful and useful figure. But it's a lot of work and, therefore, expense.

The CSO does household surveys three or four times a year. Each survey would cover, say, 13,000 or 14,000 housholds - maybe 40,000 people? There'll be a standard set of question that are asked every time - age, sex, education levels, that kind of thing. This enables them to know how much the sample of households they are studying matches or differs from the demographics of the community at large, which means they can weight the responses to extrapolate what they learn from the survey to the whole country.

They they'll ask the questions that they are actually studying. This will differ each time. It's usually co-ordinated with the questions being asked in similar household surveys by other national statistical agencies in the EU, so that EU as well as national data can be obtained, and so that Irish data can be meaningfully compared with similar data from other EU countries. Each time there's a particular focus - it might be on eliciting household data about health and healthcare, of financial consumption, or lifestyle, or . . . well, anything.

It seems to me that what you need is to persuade the CSO (and other EU statistical agencies) that there should be a focus on employment and earnings, so that they will ask questions from which median full-time earnings could be calculated. But you'll need to persuade them that there's a public policy need for this - that the data will be useful for government planning, policymaking and policy delivery.

Per Benedict: Any total number must calculated on individual numbers. You cannot say the total earnings of the ft workers in Anywhere Road is 100k if you don't know what each is earning. Otherwise, how could you be confident that your total is correct.

I don't know about the current situation but, in the past, CSO surveys asked employers to report aggregate total earnings, not individual earnings so above point was wrong then.

What is this "paperwork linked to each of the FT employees" that the CSO have, or have access do? It can't be Revenue records because, even if the CSO had free run of the Revenue system - which they don't - Revenue records don't distinguish between FT workers and other workers. So it must be something else. What is it?

Because the total is estimated from a household sample survey - not a fully documented list of all of them.

Your first statement here is logically flawed - it's obviously possible to know the number of things in a set without knowing the identity of each thing in the set. And we've explained to you several times statistical methods for calculating the size of a set without identifying all the members of the set - real-life statistical methods routinely used by real-life statistical agencies like the CSO.

And, without your first statement, the rest of your argument collapses.

It's time to give up, Benedict. What you should be pressing for is for the CSO to gather the data from which median full-time earnings could be extrapolated; not trying to persuade us that they have already gathered the data but for unstated reasons fail or refuse to make the calculation. You're in conspiracy theory territory there.

Even though the 47k is not the median for FT workers, there will still be plenty of FT workers earning close to 47k.