Hello everyone,
Following Gill’s ‘back to basics’ approach to mechanisms, I find it helpful to consider the critical realist roots of mechanisms. I apologise if all of this
is familiar but is has helped me to get to grips with the key ideas – and I do eventually get to Sue’s question!
Some of the key concepts of critical realism that relate to realist evaluation are the recognition that intentional human behaviour is caused by reasons; and
a transformational model of social activity.
Bhaskar’s transformational model of social activity suggests that humans, like other things, possess liabilities and powers which arise from the physical, biological,
spiritual, psychological and socio-cultural structures that constitute human beings in society. These mechanisms interact with those of other humans and other objects – that is with social structures - to cause events. An important mechanism in this scheme
is human agency: the power to plan a project to achieve one’s goals and to produce speech and action directed towards achieving those goals in the world. Here, agency exists in the realm of the real as an (unseen) mechanism, and speech and action exist as
part of the observable realm of the actual. However, human agency exists together with other mechanisms and structures which may enable or constrain the agent in the achievement of their goals. In turn, human beings have the capacity to reflect on planned
courses of action and on structural mechanisms, and to change their actions to account for enablers or constraints, or to modify the structures themselves. Bhaskar expresses these ideas in his description of a transformational model of social activity:
‘People do not create society. For it always pre-exists them and is a necessary condition for their activity. Rather society must be regarded as an ensemble
of structures, practices and conventions which individuals reproduce and transform, but which would not exist unless they did so. Society does not exist independently of human activity (the error of reification). But it is not the product of it (the error
of voluntarism)’ (Bhaskar, 1998, p.36)
This places critical realism squarely in the body of work that sees the social world as a product of structure and agency. Social structures are the relatively
enduring ideas and relationships that already exist in the world before we come to it. Things such as language, marriage, health care, capitalism, nation states, money, property - and all the roles, relationships and ideas that go with them. These are structures
that we find are already there when we are born, and which shape us as we grow in our world. So we can see that social structures have powers which can act as mechanisms to produce events. However, human beings are not passively shaped by the social structures
in which they live because, as Bhaskar notes, these social structures cannot exist apart from human activity. Furthermore, human activity is not a product simply of automatic responses to social structures. Human beings possess agency. Archer observes that,
‘Agents possess properties and powers distinct from those pertaining to social forms. Among them feature all those predicates, such as thinking, deliberating, believing, intending, loving and so forth, which are applicable to people, but never to social structures
or cultural systems.’ (Archer 2003, P. 2)
These ideas are evident in Pawson and Tilley’s description of realist evaluation. For example, they argue that, ‘Social (and physical) reality is stratified.
That which we can observe, including the most manifest and routine regularities, is produced by the operation of underlying generative forces which may not be immediately observable.’ (Pawson & Tilley 1997, P. 216) These ‘underlying generative forces’ are
elsewhere identified as mechanisms. Mechanisms are characterised as a weaving together of resources and reasoning which enables program participants to change their behaviour, and so fulfil the intentions of the program designers. However, in congruence with
Bhaskar’s transformational model of social activity, Pawson and Tilley also argue that programs are always introduced into an existing set of social relationships, which they call ‘context.’ Context is seen not simply as ‘the spatial or geographical or institutional
location into which programs are embedded’, but as ‘the prior set of social rules, norms, values and interrelationships gathered in these places which sets limits on the efficacy of program mechanisms’ (Pawson and Tilley, 1997 p. 70). In other words context
is equivalent to Bhaskar’s ensemble of structures, practices and conventions. The signal contribution of Pawson and Tilley is to take these core ideas and show how they can be used to explain how programs work in terms of the interaction of context, mechanism
and outcome. This can be seen in their ‘logic of realist explanation.’:
‘The basic task of social inquiry is to explain interesting, puzzling, socially significant regularities (R). Explanation takes the form of positing some underlying
mechanism (M) which generates the regularity and thus consists of propositions about how the interplay between structure and agency has constituted the regularity. Within realist investigation there is also investigation of how the workings of such mechanisms
are contingent and conditional, and thus only fired in particular local, historical or institutional contexts.’ (Pawson and Tilley, 1997 p. 71).
It is clear that realist evaluation explains observable patterns of human behaviour as products of the mechanisms exercised through the interaction of human
agency and social structures, in the context of a pre-existing but contingent social context.
Turning to Sue’s question about ‘how the overarching selected mechanisms for systems are related to the way we look at individual level evaluations and individual
level reactions such as fear, empowerment or motivation etc.’ – I think these belong together as structure and agency. When we say a programme is successfully implemented, what we mean is that people behave in a way that is consistent with that programme.
If we take agency seriously, then they are doing that because they are, in Archer’s words, ‘thinking, deliberating, believing, intending, loving and so forth.’ I would argue that these unseen thoughts and feelings are the mechanisms at work. So, individual
level evaluations are vital to understanding how an intervention works. But so are the ‘ensemble of structures, practices and conventions which individuals reproduce and transform’ and which are present in every organisation or ‘system.’ In my experience,
there are a relatively small number of mechanisms at work in relation to agency (because there is a commonality in human nature) but a myriad social structures in a given context. So, in the four African countries I would expect a degree of commonality in
the agency related mechanisms and greater variety in the structures that make up the context and which hinder or help the intervention.
Hope that is helpful.
Yours,
Pete
Dr Peter O'Halloran
Lecturer
School of Nursing and Midwifery
Queen's University Belfast
Medical Biology Centre
97 Lisburn Road
Belfast BT9 7BL
Tel 028 9097 2490
Fax 028 9097 2328
email: [log in to unmask]
Web page:
Peter O'Halloran
From: Realist and Meta-narrative Evidence Synthesis: Evolving Standards
[mailto:[log in to unmask]] On Behalf Of Gill Westhorp
Sent: 10 March 2015 01:29
To: [log in to unmask]
Subject: Re: Mechanisms in realist evaluation of systems
Hi Sue (and all)
I think there are multiple avenues in to thinking about this kind of issue. For me the process is always to go ‘back to basics’ and build from
there.
1.
Related to ‘the nature of theory’: realist evaluation is ideally intended to work not just with program theory, but also with formal
or substantive theory.
2.
Starting with program theory: Mapping out the program theory should enable one to identify the levels and nature of the systems involved,
and ‘what it is that the program intends to affect’ for that level/nature of system.
Once one knows those two things, one can work out who the decision-makers are and the nature of the decisions that they
make, and what it is about the program that is expected to affect those decisions. This kind of approach enables the ‘reasoning and resources’ construct of mechanism to be applied at multiple levels of systems. NB that ‘decision-makers’ are not necessarily
individuals – they can be groups, organisations, structures… but they still make decisions and therefore one can still investigate the ‘reasoning’ involved in those decisions and the factors that affect that reasoning, therefore the decisions, therefore the
actions that follow... (etc etc).
3.
Once one knows the same two things (i.e. the level/nature of systems involved and ‘what is intended to be affected’ for that system),
one can also think about / investigate the kinds of formal theories that are relevant to the different levels. Formal theories and the research that has been done into them, or using them, are a great source of ideas and information about both mechanism and
the contexts in which mechanisms do and don’t fire. (For a worked example of this, see the second article in my two part set on ‘layered theory’ -
http://evi.sagepub.com/content/19/4/364.abstract)
For example, one might draw on political science to find theories to understand how overall changes in political systems
– often supported or encouraged by international aid – might contribute to (or detract from) health system changes. (I have in mind here things like democratisation, or international movements to build accountability of governments and service providers to
populations or communities. There’s lots of work in these areas in the international development sector). One might then draw on the organisational management and change literature to find formal theories to help understand change at those levels; or on
economic theories to understand how funding systems affect program operations. Similarly one might draw on community development literature to understand the community engagement and responsiveness level and on health promotion (including health literacies
etc) to find theories to understand the individual level.
4.
Related to ‘the nature of mechanisms’: I’m pretty sure I’ve posted about this before but in brief: Much as I love the construct and
find it extremely useful, I don’t think we have to be constrained to the ‘reasoning and resources’ concept of mechanism. Realism posits that there are mechanisms already operating in all kinds of systems (material, biological, psychological, organisational,
community, political etc etc). One of the ways that social programs ‘work’ is by changing the context in such a way that existing mechanisms fire ‘to different extents’ (some are enabled, some are disabled). If the mechanisms were already there, in
different kinds of systems and at different levels of systems, it follows that they weren’t all necessarily at the individual level and/or they weren’t all mechanisms of the ‘reasoning and resources’ variety. It therefore follows one needs to identify what
the existing mechanisms were and what it was about the program that enabled/ disabled/ strengthened (etc) them. Another way that social programs work is to introduce new mechanisms… for example, new ‘reasoning’ is involved in pretty much any program that
involves a learning component.
5.
Even when thinking about program mechanisms they might be (and are, in the literature) thought of in different ways: as a force, as
an interaction, as powers and liabilities, and as feedback and feed-forward processes. Examples of each of these can be drawn at different levels and/or different types of systems. Forces either push or pull: gravity is an example from physics; peer pressure
is a social example. An interaction involves a transfer of something from one party to another that results in a changed state. Gunpowder + spark (physical); contract to buy a house (institutional level in the sense that it’s embedded in legal and economic
institutions, resulting in the homeowner having different ‘powers and liabilities’ once they’ve bought the house). Powers and liabilities exist whether or not they are currently in use: trees can grow; humans can learn; workers can work; states can make laws.
Feedback and feedforward processes are ones where the next state is a necessary outcome of the previous state (think genetic inheritance, or stock-market crash).
6.
One of the tricks for ‘getting to’ program mechanisms is to remember that “the mechanism is whatever generated the outcome”- therefore
one can’t identify mechanisms without knowing what the outcome (or intended outcome) is. Is post-partum care intended to improve maternal health? There may be biological mechanisms involved (e.g. through hygiene) as well as learning mechanisms (e.g. through
improving maternal health literacy). Is it intended to improve infant wellbeing? Biological mechanisms may relate to nutrition or clean water; developmental mechanisms might relate to secure attachment. Is it intended to improve access to health services?
Different mechanisms again. One of the intriguing clues in your original email, Sue, was “a range of (different) post-partum care packages”. Instant question for the realist program theory development: what are the ‘differences that make a difference’? Are
they intended to generate different outcomes? Affect different kinds of systems or different levels of systems? Etc etc.
Sounds like a fantastic project – good luck with it!
Cheers
Gill
From: Realist and Meta-narrative Evidence Synthesis: Evolving Standards
[mailto:[log in to unmask]] On Behalf Of
Jagosh, Justin
Sent: Tuesday, 10 March 2015 2:52 AM
To: [log in to unmask]
Subject: Re: Mechanisms in realist evaluation of systems
Sue,
Thanks for starting this needed discussion. Joanne, thanks for the papers. Admittedly I have yet to read either yours or Gill's in detail so will read with interest!
Sue, there are also people at Trinity College Dublin who have started a 'Global Health Realist Group' and are actively engaged in projects of the scale which you describe. You might want to contact them for advice or direction to their publications to date.
See Brynne Gilmore et al:
https://global-health.tcd.ie/research/ghrg/
I think at the heart of the question is how do we define and apply the concept of mechanism across the dimensions of micro, meso, and macro analyses. Joanne or Gill, based on your systems analyses do you have some specific recommendations here? If we start
with the idea of mechanism as meaning the reaction to resources offered, that's a sensible definition within the context of, shall we say, micro-level, 'granulated' or as Ray described 'textured' analyses of social or behavior change programs.
But what do people do if they are commissioned to evaluate large-scale funded program, which consist of sites across many countries contexts and some resultant level of heterogeneity in intervention type for differing contexts? It would seem to me that there
are two options: 1. insist to funders that the realist evaluation must remain at the micro-level in order to capture the mechanisms in their contexts, which can then be transferred out of the particular context to be tested in the other sites, or 2. Do a large-systems
realist evaluation in which the mechanism is at a higher-level of abstraction which may not reflect the definition as set out by Pawson & Tilley in their seminal work.
If one of the many bright thinkers on this discussion board could provide some insight here, that would be wonderful.
thank you,
J
Justin Jagosh, Ph.D
Senior Research Fellow
Centre for Advancement in Realist Evaluation and Synthesis (CARES)
University of Liverpool, United Kingdom
www.liverpool.ac.uk/cares
________________________________________
From: Realist and Meta-narrative Evidence Synthesis: Evolving Standards [[log in to unmask]] on behalf of Joanne Greenhalgh [[log in to unmask]]
Sent: March 9, 2015 6:36
To: [log in to unmask]
Subject: Re: Mechanisms in realist evaluation of systems
Hi Sue
Great to hear you enjoy Justin's course and feel brave enough to post here - welcome!
I have two immediate ideas in response to your question. First - yes - that is incredibly complex and it's not always easy to connect the macro, meso and micro levels together. One solution is to use Gill Westhorp's strategy of combining complexity theory with
a realist philosophy of science to 'layer theories' in order to explain how mechanisms firing at one level can influence those that fire at another level - from the micro up to the macro.
http://evi.sagepub.com/content/18/4/405.short
In a similar vein to this idea - I worked with Ray Pawson last year on a realist review of demand management for planned care. From an analysis of the programme theories underlying the different demand management interventions, we also found that system change
is layered but that different interventions are often directed at one particular 'layer' of the system. However, what happens at one 'layer' of the system can be hindered or supported by what happens in other layers of the system. Successful change required
integrated change at all levels of the system:
http://www.sciencedirect.com/science/article/pii/S0277953614003268
Hope this helps and good luck with you project
Best wishes
Joanne
Dr Joanne Greenhalgh
Associate Professor
School of Sociology and Social Policy
University of Leeds
Leeds LS2 9UT
Twitter: @Greenhalgh_Jo
-----Original Message-----
From: Realist and Meta-narrative Evidence Synthesis: Evolving Standards [mailto:[log in to unmask]] On Behalf Of Suemann
Sent: 09 March 2015 13:06
To: [log in to unmask]
Subject: Mechanisms in realist evaluation of systems
First to say thank you to justin Jagosh for very stimulating course in liverpool last week - and for inspiring me to be brave enough to post here.
I am a relatively new recruit to realism and have been plunged into a complex realist evaluation of a health systems research project designed to design, implement and evaluate a range of (different) post partum care packages in four African countries. I am
drawing on trish greenhalgh's methodology for evaluating the south london modernisation initiative. I have found the approach of looking at supportive/ unsupportive contexts for the mechanisms fits well here but I am unsure how the overarching selected mechanisms
for systems are related to the way we look at individual level evaluations and individual level reactions such as fear, empowerment or motivation etc.
Can anyone advise ( I hope this makes sense!).
Finally to say we are starting a small skype-based journal club as a way of building skills for some of us earlier on in using the methodology. If any of you experts out there would be happy to join us occasionally that would be fantastic.
Thanks for any help
Best wishes
Sue Mann
UCL NIHR knowledge mobilisation fellow
Sent from my iPhone