Take 10 with... Beatrix Jones

Dr Beatrix Jones from the Department of Statistics gives us 10 minutes of her time to discuss how analysing data can reveal new information.

1.  Describe your research topic to us in 10 words or less.

Models for high dimensional covariance matrices.

2.  Now explain it in everyday terms!

We have the capability to measure many, many things on each individual (whether it be a person, animal, or wine sample). How do these various quantities relate to each other? Can it be simplified to some understandable relationships?

3.  Describe some of your day-to-day research activities.

I spend a lot of time analysing data. I try a few different things and see what structures are uncovered, or what structures seem to be there that aren’t well captured by current models. It is the latter case that leads to new models and methodology.

4.  What do you enjoy most about your research?

Discovering things is fun— including just discovering things for yourself. I found some nice structure in a data set I was preparing for a homework assignment the other day. This data set has been analysed a million times, so it certainly wasn’t anything new, but it was still fun to see it emerge from my analysis. When it is new data addressing an interesting problem it is even more fun.

5.  Tell us something that has surprised you in the course of your research.

In statistics, usually if you have a large enough sample size everything is clear. A few years ago, I was doing some simulations and regardless of the sample size I couldn’t settle on which of two model possibilities was better. One would be favoured, and then the other. At first I thought it must be a mistake, but eventually we developed some theory that suggested why this was the case.

6.  How have you approached any challenges you’ve faced in your research?

It's a challenge to balance providing a timely answer to people who have collected data, and taking time to develop a better statistical method. Usually my collaborators publish the “fast” answer, and I keep working on the slow answer. A new method is hopefully ready to use “for real” by the next time a dataset like that comes in.

7.  What questions have emerged as a result?

How can we combine different types of data to best effect? For instance, how might we use metabolomics (measurements of small molecules in the blood) to make better use of data from diet surveys? Can we generate robust hypotheses about how diet affects health at the same time we establish associations between diet and health outcomes?

8.  What kind of impact do you hope your research will have?

I want to make the most of modern data generating technologies like mass spectrometry, while making sure the variability and uncertainty are correctly represented so that results aren’t over interpreted.

9.  If you collaborate across the faculty or University, who do you work with and how does it benefit your research?

I collaborate with Associate Professor Silas Villas-Boas in the school of Biological Sciences; he is responsible for the metabolomics lab where a lot of the data I work with is generated. I also recently began collaborating with Associate Professor Clare Wall, a nutrition researcher in Faculty of Medical and Health Sciences, who collects data on what people eat at crucial life stages—pregnancy and childhood.

10.   What one piece of advice would you give your younger, less experienced research self?

When I was younger I wanted to foresee what statistical innovations would be needed before I started on a collaborative project—with the idea that if I wasn’t going to be creating new statistical techniques I shouldn’t be working on that project. I would tell my younger self to relax about that, both because I feel like I make a worthwhile contribution to projects even when I am using “standard” techniques, and because usually an area where statistical innovation is needed eventually emerges.