Psychologist logo
Work and occupational

‘Open the black box, see what’s in it’

Andrew Clement meets Professor Karina Nielsen, from Sheffield University Management School, at the British Psychological Society’s Division of Occupational Psychology Annual Conference

07 August 2018

What are the key things organisations will have to do over the next five to ten years, in terms of wellbeing?
I think there are two questions – what organisations need to do and what researchers need to do to support organisations. We’ve just had a keynote and a symposium on evidence-based work, yet there’s still a big divide between research and practice. Organisations that want to improve employee wellbeing don’t necessarily have the tools, they don’t know how to run these processes.

It’s all very well saying you’ve got a theory that can improve work–life balance, but so often these kinds of organisational interventions don’t work. My argument is that we need to be looking at the processes by which we implement our interventions, and we need to have the right tools to take organisations and their members through the interventions. You can’t just expect them to be able to develop action plans; for example, they need the tools to develop the content of the action plan, to address the most relevant questions or problems they have within the organisation.  

What we often see is that organisations do a lot of initiatives, and they don’t actually evaluate whether those work or not. There’s no organisational learning going on. They don’t actually know what they can do better next time, and of course that creates ‘change fatigue’ in employees. I’m not saying that organisations can’t develop those tools themselves, but that’s where we can become better. Rather than having the typical intervention where we have a pre- and a post-measurement, and we know ‘it didn’t work’ or ‘it did work’ – do we actually know how we can make it work, and what organisations can do themselves to make things work?

What are the best steps for moving us forward?
It just so happens that we have a book out! I have asked researchers who do applied research, who are developing tools, to describe their experiences, focusing very much on the methods and the tools to assess what are the risks. The tool can be a questionnaire, or it might be a dialogue method, a tool to evaluate whether we achieved what we wanted with this intervention. We need to get those tools out there so people can pick and choose what works in their organisation, because they won’t work for everybody; it might be very different if you work in the postal service compared to doing interventions in universities.

It sounds like this is the beginning of that journey.Is there anything that is looking quite promising?
Well… that’s a difficult question. I think what is looking promising is that when I started doing this type of research, there were hardly any of us, and now there’s actually quite a group of us. People are starting to realise that interventions are about changing behaviours, and if people don’t understand why they have to change behaviours, and they don’t have ownership over what is going to happen, then you’re not going to see any changes in behaviour. I think that shift in understanding – that there is a process behind the intervention – I think that’s the most promising thing. 

Do you think the rationale is a key element of interventions?
Yes, it’s a paradigm shift that is happening at the moment. I’ve personally started advocating the use of realist evaluation… are you familiar with that term? Realist evaluation looks at what works for whom in which circumstances, so they explore what is called ‘CMO configurations’. You’ve basically got a Context, the ‘C’. Then you’ve got some ‘Mechanisms’ of the intervention, for example it can be the support of the line manager throughout the process, or it can be employee participation, that might be typical mechanisms that make the intervention work. And then you’ve got the ‘Outcomes’ at the end – does it lead to a change in wellbeing, if that is what your outcome is?  

So we need to understand what kind of contexts an intervention would work in, and what mechanisms make it work. What are the contextual factors that trigger those mechanisms, the preconditions that need to be in place before an intervention can work? To give a popular example, there’s no use in trying to implement a participatory intervention in an organisation where managers and employees do not talk to each other, don’t trust each other… they are never going to sit down in a room and develop good action plans, if employees feel that their suggestions are going to be ignored by management, or if they feel they’re going to be bullied if they do make suggestions that management don’t like.

There needs to be some homework done before intervention?
Exactly. Because organisational interventions are about changing the way work is organised, managed, and designed, before we do them we typically run individual level interventions beforehand. We do workshops and training with line managers, with union representatives, so they know what the process is about, and also try to create some kind of ownership of the intervention. We might say ‘OK, so there’s this intervention, how can you as a line manager benefit from it? How does this fit in with your normal leadership role?’ We’re trying to get them on board up front, and to create a good basis for the organisational intervention.

What kind of research would you like to see happening?
Two types. The realist evaluation approach is crucial. I think we need to analyse the process in much more detail to understand why the intervention did work, why it didn’t work, and how can we make it work next time, trying to dissect the working mechanism.That’s where I think we can start transferring one intervention from another context, saying, ‘You have to have this mechanism in place, this is what’s going to make your intervention work.’ And then also on the tools side, testing out different kinds of tools to see whether, if they work in one setting, they can work in another setting. You translate principles to different tools and methods and different organisations.

Talking about why something did or did not work, do you think that we need to be writing different kinds of articles?
Yes. We’ve started writing the kind of realist evaluation papers. It’s taking that step back and thinking about what we’re measuring. And that process can be measured both qualitatively and quantitatively, so when you have your follow-up questionnaire six months in, you can measure line manager support or participatory support. That is where we have a lot of research to do. What is happening now – and I’m guilty of it myself – is that if you measure the mechanism, you do that at the end. Really we should be asking it every two months, or every month. That way we can feedback to the organisation, saying, ‘Actually your process isn’t on track, you need to do something now.’ That’s going to minimise the risk of a failed intervention. With ongoing measurements we can go back and say, ‘Now you need to make sure people feel involved’, or ‘Your line managers are not supporting the intervention; they’re too busy doing other things.’

This might be a departure from a randomised controlled trial?
Yes, and I have written all about why I don’t think the randomised control trial works. The reality is it is very difficult to find a good control group. For example, if you do an intervention in an HR department, you can’t find a control group within that organisation, because there aren’t two HR departments. And of course individual-level randomisation doesn’t work in organisations. If you and I work in the same department, and my work environment is changed – for example, the procedures for how we work together – that’s going to affect you too. I’m not saying you can’t use the randomised control trial, but it’s not enough, you need to consider what else is needed.

I was also thinking that adjusting mid-course would be unusual for an RCT.
Very much so. We’ve taken the RCT from the medical world, where you give people a pill, and that’s the active ingredient, but the problem with participatory organisational-level interventions is you don’t define the content of the intervention up front. You go through a process where you say, ‘OK, now we are screening, now we are identifying what the problems are, and then we’re developing action plans.’ You don’t actually know halfway through what the content of the intervention is, so the randomised control trial doesn’t really work in that traditional sense. You can have the design of the cluster randomised trial in participatory interventions as well, but there’s a lot happening that you need to measure. Open the black box, see what’s in it. 

Is there one last thought for those who work in wellbeing?
We need to find interventions that address the causes of the problem, not laying the responsibility upon individuals, saying, ‘You can do stress management, you just need to breathe properly’… ‘You have to be…’ – even worse – ‘…resilient.’ That puts the onus on the employee, and that is completely wrong. I’m not saying you can’t have individual-level interventions, but they don’t address the problems. The workplace is so much more complex than individual-level interventions… a lot of people have to change their behaviours when what you’re doing changes the way work is organised and designed.