---

Statistics and Registration Service Bill

Perhaps I may intervene briefly. I think that the reasonable belief clause is important, and I shall give a short account of something that I once did. I had a sample of 39,000 observations and, by cross-classifying in various ways—race, sex and so on—I found that in some cells there was only one observation. A clever person could more or less have identified a certain person, although I would not have known it when I analysed the data. I discovered it only by accident using some statistical tools on the outline observations. You have to allow some defence for a researcher who reasonably believes that what he has done is as statistically competent as possible and accept that someone may be able to get more meaning out of the sample. That may not be satisfactory to the noble Baroness but I ask her to accept that, on that occasion, I would have been prosecuted by whoever that person was. He had had a certain education and was of a certain ethnic background and happened to be the only one in the cell. His salary also happened to be about 20 times the average, so one could easily have found out who he was. It was reasonably possible that I would not have known that when I read the data. When I found out, I stopped my analysis. However, it is possible that statisticians will not suspect that other people may be able to extract more meaning from statistics, so I think it is possible to have a reasonable belief defence for a statistician.