Thursday, 10 December 2020

Reflections on Interdisciplinary Research (2013-2020)

I've always found interdisciplinary research rewarding. It also fits with my view that, at least from an applied perspective, no single discipline has all the answers. Probably also says something about my attention span. 

Anyway, I'm going to share three things that early-career interdisciplinary researchers in the UK (or who are thinking about working in the UK) might find useful. Somewhat timely as we come to the end of another REF cycle. For those outside the UK, the REF is used to quantify the quality of research outputs across institutions every 7 years or so. It determines how much funding is allocated to each institution by the UK Government. 

The below is therefore based on my own and colleagues experiences while conducting interdisciplinary research at research intensive institutions across the UK.

1. Take time to appreciate disciplinary hierarchies because it may impact where you can work in the future.

Publication expectations placed on researchers in terms of the REF appear to be somewhat incompatible with an interdisciplinary research agenda. Papers make up the bulk a REF submission and outputs are submitted to traditional subject panels. Psychology, psychiatry and neuroscience form a single panel, for example. 

Unfortunately, every discipline has its own hierarchy of what might be considered excellent research. Sometimes these hierarchy's are based on a list of journals in someones head that you can guess (e.g, psychology). At the other end of the spectrum, medicine can set high expectations with highly selective journals dominating the landscape. Management schools rely on the ABS list. This ranks dozens of journals from 1 to 4 with a special category for a few select 4* journals...so..em..1 to 5 then. 

It is unrealistic to suggest that any of these hierarchies are perfect. For example, I believe the intention of the ABS list was to allow outputs to be compared in an area that is naturally interdisciplinary. Makes sense, but the current list is pretty limited and changes infrequently. Most psychologists would be amazed to see what psychology journals are both present or absent let alone how they are ranked. There are no major medical or general science journals (Science, Nature, Lancet and JAMA publications are all missing) and no open access only publications (e.g., PLOS). In fact, open science practices embraced by other fields including data sharing and pre registration are not really a thing (yet). 

With or without such frames of reference, to appreciate if a paper is any good, you'll have to read it or get someone who knows the area to read it and come to a conclusion. 

Believe it or not, this is actually what the REF aims to do. In practice, these hierarchies including related metrics like impact factors, are not meant to impact REF outcomes. When the results are announced, an individual never finds out how high or low their work is rated and institutions are only provided with summary data. Papers can of course be screened by other academics before submission and many institutions pay outside sources to help prepare submissions.  

Despite all this, getting papers in the ‘right’ places remains core to hiring, probation and promotion decisions in the UK. Having worked in several places, the process rarely deviates. 

Researchers who wish to pursue interdisciplinary research have little choice but to publish across multiple fields in order to meet both personal research goals and those pre-determined by a primary discipline to ensure long-term success. This places early-career academics under significant pressure as they try to establish themselves as future leaders. 

Attempts to mould findings or methods to a specific journal can come at the expense of conducting the best science possible. In turn, this can reduce the potential impact of work that might be better placed elsewhere, which is another core part of the REF incidentally. However, the level and type impact is also an area that varies tremendously between individual disciplines and again interdisciplinary impacts challenges these norms. 

Yet interdisciplinary research is what UKRI are keen to fund. This is another core mission of all research intensive universities. 

For example, if I was based in computer science and working with colleagues in medicine, there is no way they or I would be happy to go for a computer science outlet. We could force it, but the process would be disastrous and not appropriate for the work. It will also limit the real-world impact due to differences in readership. Regardless, I need to cover my bases to ensure I defiantly pick up a more suitable discipline specific paper elsewhere. 

So you do what is right by your colleagues and your science, but accept it will increase your workload.

I do not personally resent this work at all, but occasionally wonder if someone had told me all this 5 years ago - would I have planned things a little differently? Probably not, but I didn't really anticipate all the above until fairly recently! I just did stuff I genuinely wanted to do! 

All this gets easier as you build up research networks, which interdisciplinary work requires by default. This provides a host of other benefits including opening doors to new funding streams. Further, if the aim is to move over to another discipline in the future, then none of this should be viewed as a problem.  

But go in with your eyes open. 

2. The process of conducting and publishing research can be radically different between disciplines even for areas that appear analogous.

So if the above hasn't put you off, the language used to describe the same thing in different disciplines remains challenging. I still struggle a bit with some computer scientists describing the running of different statistical models as experiments!

In terms of publishing specifically, I like to imagine how different disciplines might behave if they were all at the bar and attempting to order drinks:

Computer science places more value on conference proceedings, rapid turnarounds and getting stuff finished. They want to get drunk quickly and move onto the next pub. 

Medical journals still have incredible copyeditors and illustrators. Medicine arrives well presented and popular. They leave the pub looking immaculate despite drinking 12 pints of high impact lager.  

Psychology will start a fight at the bar as it is unable to agree on how drinks should be measured. After another crisis of confidence, they settle for tap-water, which tases different from last time. 

Management will think carefully, attempt to calm the situation and spend a long time at the bar studying their own menu (called the ABS list). They will then order what they believe to be 4* drink. No other discipline has ever seen the menu or heard of the drink in question.  

On a serious note, establishing a common problem can help break down barriers when working as part of an interdisciplinary team. Even when the above becomes confusing, getting back the to task in hand encourages people to articulate how their perspective or method can help. But this doesn't mean it shouldn't be challenged. This is one of the reasons interdisciplinary research takes longer. Getting through disagreements or misunderstandings can take a long time, but the end result is always stronger. 

Again, probably should have realised this 5 years ago. 

3. Have faith in good mentors and others who are helping to shape what excellent science looks like. The landscape is changing.

REF panels now have interdisciplinary members, although I'm less clear on how this will work in practice. While this appears to have had little effect on the selection of papers submitted by departments, it is an encouraging sign. 

Perhaps more importantly, PhD students are often part of doctoral training centres (like this one) and are already being trained as part of an interdisciplinary enterprise. Masters programmes are heading in a similar direction.

At the same time, some senior academics are encouraging a variety of several sensible changes or launching initiatives that will improve science more generally (e.g., Psychological Science Accelerator). All this means increased transparency and larger numbers of authors from different disciplines working together. Even if existing reward structures haven't quite yet caught up with reality, the direction of travel is probably irreversible. I suspect current PhD students and post-docs will drive this change further. 

Conversely, university's might need to acknowledge that other senior academics have come through a very different system that was driven by the loan genius model of academia. This tends to prioritise individual success over the advancement of science. For good or bad, what the majority think makes a good scientist is changing and I can see some tearing in the space time continuum when listening to advice from some professors nearing retirement on what ECRs should be prioritising. There is no point pretending that what worked for the last generation of professors will be the path for the next. 

Despite pressures placed on institutions, I am a firm believer in allowing researchers to do things they genuinely care about and trusting them to do excellent science. Ensuring this happens makes the things we get grumpy about as academics easier to accept. 

Case in point, while I am throwing out some words of caution that go against what I would like in an ideal world, my intention is not to discourage others from embarking on an interdisciplinary career. 

Quite the opposite.