Hi all,

Please see attached realist review methodology reflection article that we just published in Research Synthesis Methods.

thank you,

Justin

Justin Jagosh, Ph.D
Canadian Institutes of Health Research Post-Doctoral Fellow

Centre for Participatory Research at McGill (PRAM)

Department of Family Medicine
McGill University

[log in to unmask]

Office and Mailing Address:

Rm 426 - School of Population and Public Health

Universtity of British Columbia

Vancouver, BC, V6T 1Z3

604-822-3814

From: Realist and Meta-narrative Evidence Synthesis: Evolving Standards [[log in to unmask]] on behalf of Geoff Wong [[log in to unmask]]
Sent: October 26, 2013 3:17 AM
To: [log in to unmask]
Subject: Re: [IMAGE] Can/should I not use this codes+subcodes-->categories-->theme-->assertion-->theme for a RS?

Hi Mickey,

Thanks for your postings as they raise two important practical points. I hope my comments aren't stating the obvious.

1) The point I think Ray and Ana are trying to raise is that somehow CMO analysis needs to have a purpose. Overall the purpose is sense-making of the literature (stating the obvious I know) but the $64,000 question is how this done in a realist review.

CMO analyses is one such process.

However we also need some overall organising structure to all these CMOs - or as they put it the data is 'atomised' or as I call it, you get 'CMO soup'.

To avoid this, any CMO analysis undertaken should be directed by any initial rough programme theory you have. So a rule of thumb is only do the CMO analyses that help you to make sense of your programme theory.

2) Mechanisms are not 'free floating'

Mechanisms are tied to outcomes. So as a rule of thumb try to work backwards from an outcome asking "I wonder what caused this to happen?" (i.e. what might the mechanism(s) be). The next question is then "I wonder when these mechanism(s) are triggered?" (i.e. what are the triggering contexts).

The outcomes you use are the ones from your programme theory.

We have set this out in more detail in our training materials for realist syntheses:

http://www.ramesesproject.org/media/Realist_reviews_training_materials.pdf

for example pages 11/12

Good luck in your review.

Geoff

On 22 October 2013 18:55, Mickey Sanders <[log in to unmask]> wrote:

Greetings. If I allow the left-most codes to be codes indicating CMOcs, such as codes in the form C1.M1.O1 . . . and if I let subcodes be such things as C1a.M1.O1a . . . then do you think I could follow the process depicted in the attached image as part of my synthesis? For instance, might it make sense to group some Ms into categories, for example, and then run analysis of what types of Os and Cs "tag along" in the category? That might allow me to ascertain the existence of patterns, yes? Any thoughts, ideas, cautions, warnings, suggestions, resources, examples? Thank you, in advance, so very much for your thoughts! Please excuse any typos. Mickey