Thanks, Alice! It's quite unique to have a statistician in the central administration of the university.; potentially a great opportunity to promote the principles of reproducibility in research. The very upbeat article that John Maindonald mentioned (and Adrian Barnett shared) in the thread ("Statisticians may be the key to solving the reproducibility crisis") may point to some ideas for how to incentivise researchers and lab leaders to adopt good statistical and reproducible practices.

Thank you, Teresa, for that information.  My parting comments were intended as general comments on the experimental science scene, with no particular reference to ANU.   I had thought to say, and should have said, that the SSN structure, if managed well and given appropriate support, will open up possibilities that were not previously available.

I I found it a priviledge, as well as an intellectual challenge, to be involved in the ANU computational genomics initiatives in the early 2000s.  I guess that the 2014 1-week course in which you Teresa as well as myself were involved, was my swan song for  work with the Genomics Discovery Unit --- by that time there were PhDs and postdocs on the scene who made most of that I could contribute redundant.  BDSI does appear an intellectually invigorating place in which to work, just as the Computational Genomics Unit was for a time after it opened up in 2000 or thereabouts.

BDSI is however working in areas that are moving very fast, where the statistical and computational demands are intense, and where (even before Covid-19) demands placed on those areas of work have been increasingly in the public eye.  The demands for statistical support in other areas of the University, while often of a more traditional kind, also require attention.  How much has changed since the Stark and Saltelli paper appeared. ("Cargo‐cult statistics and scientific crisis." Significance 15, no. 4 (2018): 40-43), in the research scene generally, as well as within specific university setting?

Alice Richardson wrote:

Greetings Australian and New Zealand statisticians: I’m writing to let you know the outcome of the proposal to reorganise statistical support for research students and staff at the Australian National University.

Many of you will be familiar with the Statistical Consulting Unit (SCU) which was established in 1982. It flagship activity of one-on-one consultations has resulted in methodological advances as well as contributions to knowledge in disciplines right across the ANU. I’d like to thank all the Directors and consultants in the SCU who have contributed to its success over the last 39 years.

You’ll recall that in March 2021 I let you know that the ANU was planning to shut down all centrally-funded statistical support for research students. I’m grateful for the support the SCU received from ANZSTAT, Statistical Society members and the SSA Executive in our partially-successful attempts to defeat this.

From 20 August 2021 the SCU will close and be replaced by the Statistical Support Network (SSN). I’m the Lead of the SSN. My initial goals including building a network of statisticians across campus who can support students in their own disciplines, as well as organising drop-in sessions, workshops, and developing a web portal of statistical tutorial material. There’ll be a period of transition over the next 5 months for the current clients of the SCU, and updating the SCU website is also a priority.

I look forward to continued support from the ANZSTAT community as this new venture gets under way – thanks in advance, Alice.

Congratulations on what you have achieved.

Evidence on the very serious problems that exist with the science funding and publication processes continues to emerge, with statistical analysis issues a  large part of it.  Why does it seem that so little is being done in most of the areas affected (psychology is one area where there has been change) to deal with it?  Why is there not more concern within the academic community?

The areas where publication processes appear to be functioning well are those where what is done relies on contributions from scientists who share data and expertise, all contributing their own skills, so that the refereeing that matters occurs before papers are sent for publication --- such areas as climate science, geophysics, earthquake science, the study of viruses and vaccines, modelling of epidemics, and so on.

In areas where what is presented is the work of one scientist, or of a small tight-knit group, the refereeing that matters is what happens, if at all, after the paper is published. Examples are the May 2020 Lancet and New England Journal of Medicine studies, claiming to be based on observational data, arguing that use of the drug hydroxychloroquine as a treatment for Covid-19 was increasing patient deaths. Issues (with the analysis as well as with the credibility of the data) with these papers were quickly identified because they made claims that bore on an issue of major concern, and attracted attention from readers who carefully scrutinized their detailed statements. They were quickly retracted. How much that has no sound basis does not attract such attention, and is never challenged? 

There are apt comments in

Stark, Philip B., and Andrea Saltelli. "Cargo‐cult statistics and scientific crisis." Significance 15, no. 4 (2018): 40-43.

The mechanical, ritualistic application of statistics is contributing to a crisis in science. Education, software and peer review have encouraged poor practice–and it is time for statisticians to fight back.

Not just statisticians, I'd suggest, but scientists who care about public regard for science.

A recent book that highlights many of the issues is:

Ritchie, Stuart. 2020. Science Fictions: Exposing Fraud, Bias, Negligence and Hype in Science. Random House.