In the 1980s there was a growing realisation, particularly among medical statisticians, that p-values simply weren't the most helpful way to express the uncertainty attaching to research findings due to limited sample size. Doug was one of the leading voices presenting persuasive arguments why confidence intervals were much more informative.
At that time, a physician colleague sought my advice on presenting the results of a very small comparative study with a binary outcome, which showed a very large difference between two independent groups. The proportions of positive individuals in the two groups were something like 11 out of 12 and 2 out of 11, and he wanted to calculate a confidence interval for the difference between these proportions. At that time I was unaware of Miettinen & Nurminen’s seminal 1985 paper, and I conceded it would be difficult to calculate a good interval and I had no idea how to do so. The textbook asymptotic methods simply don't work well when sample sizes are small or cell frequencies are at or close to zero. The issue stuck in my mind, and I started to examine how intervals for proportions and their differences were treated, first in widely used textbooks, then in research articles. I came to realise that for a difference between proportions based on paired samples, none of the widely recommended methods were sound.
All of this led to extensive research to establish what methods we should advocate to calculate confidence intervals for proportions and related quantities. This became my main research interest and culminated in the publication of my book Confidence Intervals for Proportions and Related Measures of Effect Size, which is available at http://www.crcpress.com/product/isbn/9781439812785 See http://www.facebook.com/confidenceintervals
I also developed corresponding computational resources for calculating these confidence intervals. These are designed to be highly user-friendly MS Excel spreadsheets and are available at http://profrobertnewcomberesources.yolasite.com/
All of this - and to a considerable degree, the personal chair Cardiff University awarded me in 2005 - is thanks to inspiration from Doug, probably more than anyone else.
Robert Newcombe
Emeritus Professor of Biostatistics
Cardiff University
-----Original Message-----
From: A UK-based worldwide e-mail broadcast system mailing list [mailto:[log in to unmask]] On Behalf Of Martin Bland
Sent: 03 June 2018 16:18
To: [log in to unmask]
Subject: Doug Altman
Dear friends,
I am very sad to tell you that my dear friend and collaborator Doug Altman passed away today. We were working together to the end and exchanged emails only on Friday.
Doug and I have been friends since 1972 and have published together 100 times, with more in draft. They will be finished. Doug is one of the most influential medical statisticians of the past 40 years and did so much to improve the methodological quality of medical research, thus benefiting thousands of lives.
Doug had been treated for bowel cancer. Recently, he told me that there were metastases, but I thought we still had time. I am bereft.
Martin
--
***************************************************
J. Martin Bland
Prof. of Health Statistics Emeritus
Dept. of Health Sciences
Seebohm Rowntree Building
University of York
Heslington
York YO10 5DD
Email: [log in to unmask]
Phone: 01904 321334 Fax: 01904 321382
Web site: http://martinbland.co.uk/
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|