2 min read


Norman Breslow died early this month.  If you’ve had any involvement with medical statistics you have used his work. There isn’t really any need to expound on his contributions.  I have a few Norm memories.

In my first quarter at the University of Washington, I took BIOST 570 (generalised linear models) from Norm. One day, about halfway through the quarter, he appeared with a copy of ‘Science’ and asked me why I hadn’t been a co-author on a paper from the Sydney Blood Bank. This was research about long-term HIV non-progressors, and I’d done a bit of consulting that led to an acknowledgement buried in the references. 

Later, when I was looking for a PhD supervisor, I had some ideas of a topic related to correlated data with correlation structures more complicated than just clustering. I went to Norm, based on his paper with David Clayton on generalised linear mixed models. He said he wasn’t up to date with that literature and suggested Patrick Heagerty instead. Patrick claimed to be too new and inexperienced, and sent me back to Norm.  Some people might have been put off by this.  But, as Norm was always at pains to point out when I later told this story, he was right. Patrick was an ideal supervisor, and Norm (who did end up on my committee) was working on different topics by then. 

The scary thing about talking with Norm in the audience was his interested and slightly puzzled expression. It looked the same whether he was learning something new or surprised that you’d say something that silly. 

And finally, one of the most endearing things about Norm was his pride in the success of his students: for example, two winners of the COPSS Presidents’ Award, four winners of the Spiegelman Award, and other honours. One of the last times I saw him, he showed me his copy of the COPSS book “Past, Present and Future of Statistics” and pointed out his students among the fifty people on the cover.