Mr. Pádraig Dalton:

There were questions on the census in particular. The first point to note that the census projects for 1901 and 1911 were undertaken under legislation where there was no provision laid in respect of confidentiality of the data recorded. In 1926 the Statistics Act 1926 applied, and the legal framework at the time was that there would never be provision given to the census records at the time. That is the distinction between the 1911 and 1926 records, and it is why work on the 1911 project was able to commence well in advance of the 100-year rule; it did not apply to the 1911 census.

The Deputy mentioned the release of data in the US. Data is released after 72 years there, and in Canada the period is 92 years. In Australia the period is 99 years and the UK and New Zealand have 100 years as the period. There is a range of countries in Europe where census records are destroyed after a prescribed number of years, including Sweden, Luxembourg, Belgium, Portugal, Germany, Switzerland, Poland, France and Russia. I could go on. There are varying practices in that range of countries.

We can only operate within the law, and as it stands that is the Statistics Act 1993, which states that census records can only be put into the public domain after 100 years.

Until such time as the law is changed that is the way things are. Of course the CSO has no role in terms of legislation. Therefore, we can only advise if we are asked for our opinion.

The Deputy is correct that we have a position on the 1926 records. Our position is that we need to hold to the legislation that exists. It is important to remember that we are talking about a retrospective change in the legislation which could have unintended consequences, not just for the census of population, but right across the CSO's range of statistical activities. People do not look at our activities in a compartmentalised manner. Therefore, if we make a mistake in one area it will have knock-on consequences that will have an impact right across all of our statistical activities. For example, if we released 1926 data in 2016 it would immediately lead to the disclosure about a significant number of people who are still alive. Some of them may still live in Ireland and some elsewhere. If one adopted the 72-year rule like in the US then one would release information of up to 310,000 people. That is a significant issue. The most important thing to remember is that at the time the 1926 census data was collected, a legal guarantee was given to respondents that the data would never be put into the public domain. The introduction of the Statistics Act 1993 brought a change and a compromise that records could be put into the public domain 100 years after the date of the census. That suggests that the 1926 data could go into the public domain in 2026. However, we are concerned about the signal that a second retrospective change to the legislation could have on the broader statistical system, in particular to our survey respondents, business respondents, and personal respondents through our household surveys. It would negatively impact on our response rates and the quality of the official statistics.

We need to be conscious of the environment in which we operate. We compile a lot of sensitive information. The quality of our key outputs such as GDP and GNP, debt and deficit and unemployment and unemployment - and I mentioned many of these areas earlier - are monitored very closely, not just by national users but internationally. The environment in which we work has changed dramatically over the years and a high premium is now placed on the quality of official statistics. Quality is related to response rates. Our ability to maintain response rates is directly related to the legal guarantee that we provide to respondents of confidentiality regarding the data that they provide to us. Without that legal guarantee we would not have an effective and fully functioning statistical system in Ireland. A legislative change would have implications, not just for the census area, but right across the range of statistical data that we provide. Therefore, we need to be very careful about undertaking any initiative that could undermine the quality or integrity of the system. To change the law retrospectively once could be classified as an event but to change it twice could be seen as a pattern or track record. That is the major concern about the second proposal to retrospectively change the law. That is why we have adopted the position to hold to the 100-year rule and want to provide access to the 1926 census records in 2026.