Monday, November 12, 2001

NRC shelled out real dough for bogus statistics

Professor Dr Michel David was awarded Grant NRC7035 to advance geostatistics. In those days NRC stood for National Research Council. So David plugged away and his work was printed in 1977. But why did David come up with such a curious caution? He cautioned that professional statisticians would find unqualified statements. How about that? He touched upon it on page VII of what he had come to call Geostatistical Ore Reserve Estimation. He didn’t point out any unqualified statement. But I found out he was right after I had bought my own copy of his book.

David was as smitten with geostatistics in the 1970s as young Matheron was with applied statistics in the 1950s. Alas, Matheron’s pursuit of applied statistics was not to last. On the contrary, Professor Dr Georges Matheron in the 1970s praised geostatistics as a new science. He did so because he had failed to grasp that all functions do have variances. It is a fact that each distance-weighted average has its own variance in applied statistics. All the same, Matheron got stuck with the variance-deprived distance-weighted average. So much so that he got into calling it a kriged estimate. He did it to honor D Krige who put distance-weighted averages to work at Witwatersrand gold mines in South Africa. The problem is that the variance of the kriged estimate vanished on Matheron’s watch.

A strong case can be made that eulogies be written long before one’s time on this planet comes to an end. One might ask a lawyer to assist in assuring the veracity of one’s credentials. A case in point is Professor Dr George Matheron’s 2000 eulogy. Dr F P Agterberg was his eulogist. He remembered him as the founder of spatial statistics. As luck would have it, Matheron never tested for spatial dependence by applying Fisher’s F-test to the variance of a set of measured values and the first variance term of the ordered set. Given that Matheron’s magnum opus is posted on its own website, anyone could spend a long time to study his life and times.

Professor Dr Michel David passed away on May 10, 2000. Professor Dr Roussos Dimitrakopoulos and Michel Dagbert put together his obituary. The International Association for Mathematical Geology had awarded him in 1988 the W C Krumbein medal. David became a Fellow of the Royal Society of Canada in the same year. CIMMP recognized his worldwide achievements in 1989 with the award of the Selwyn G Blaylock medal.

My son and I knew precisely what had gone wrong with geostatistics when I was face-to-face with David on Saturday, March 23, 1991. It was at a seminar on Sampling and Ore Reserves at the Royal York Hotel, Toronto, Ontario. CIM Bulletin in 1990 had rejected Precision Estimates for Ore Reserves. We had shown how to test for spatial dependence and how to derive unbiased confidence limits for gold content and grade. David saw fit to nitpick that twenty years of geostatistical literature is missing in our work. He did not ask me a single question. Take a look at what is wrong in Matheron’s new science of geostatistics.

Marechal & Serra, Random Kriging, 1970, Figure 10
David, Geostatistical Ore Reserve Estimation, 1977, Figure 203

Professor Dr Michel David claimed: "Writing all the necessary covariances for that system of equations might be a good test to find out whether one really understands geostatistics"! David didn’t know that deriving the variance of each equation and counting degrees of freedom for his system of equations is a good test to find out whether one really grasps applied statistics.

The National Research Council is the Government of Canada’s institute for research and development. As such it has been active since 1916. NRC’s task is to stand on guard for ethics and integrity. It did not know in the 1970s that Matheron’s new science of geostatistics is an invalid variant of applied statistics. Neither did Dr Roger A Blais, a Professor in Economic Geology and a Fellow of the Royal Society of Canada. He made it possible for David to write so much about so little. David confessed to be indebted to Matheron. David’s writing added up to a batch of bogus statistics. NRC7035 was in place not only for his 1975 Geostatistical Ore Reserve Estimation but also for his 1988 Handbook of Applied Advanced Geostatistical Ore Reserve Estimation.

Dr Isobel Clark in her 1979 Practical Geostatistics derived the variance of the distance-weighted average AKA kriged estimate. Professor Dr Michel David wrote about the “famous Central Limit Theorem” but did not know how to apply it. The Royal Society of Canada no longer list David under Obituaries of Deceased Fellows. Sic Transit Gloria Mundi!