Great question and one I have had cause to reflect about.
I believe that there are three circumstances where specifying an actor can be valuable:
1. Where you are looking at mechanisms across multiple levels - this was the case for our research capacity realist synthesis where activities could take place at national, regional, network, organisational, managerial and individual levels:
https://health-policy-systems.biomedcentral.com/articles/10.1186/s12961-018-0363-4 and for a current nutrition in Africa project where activities operate at a national policy level and at a local frontline level.
2. Where you are exploring the programme theories of multiple stakeholders (what I call "relativist synthesis"!) - if a manager has a different view of how an intervention works from a clinician and a patient then attributing to an actor can be valuable and help to remove ambiguity(as can attributing a perspective as well!).
3. Where the purpose of the realist project is to generate recommendations or an action list - in order to apportion clear responsibility that the action gets done. Specifying as "this must happen" loses this sense of ownership.
In terms of authority for this approach from the literature I have to confess that the 2nd edition of our Systematic Approaches to a Successful Literature Review (Sage, 2016) states this! Pages 242-243. This was based on our experience that was underway at the time we wrote the book -on the research capacity realist review above and for a realist review of involvement ins service development for persons with dementia. In the latter case certain success factors were triggered by group facilitators and others by persons with dementia or their carers either individually or collectively (making it important to distinguish between individual actors specifically and collective actors!)
Hope this helps
Andrew
