STEM Central

A Community of Practice for NSF STEM Projects

logic model wisdom from the AEA list serve.

From:    Bill Fear <williamjamespsych@GMAIL.COM>
Subject: Logic Models in Action

Logic models that use Input - process/action - output are really all you need.  Everything else is argument over definition that was recognised as an argument over definition many years ago.

You can develop the logic model either way.  Start with the goals.  Reduce these to meaningful outputs.  Work out what actions you need to take to achieve the outputs.  Determine the required inputs.

Advantage:  creative.  Unbounded.

Disadvantage:  very difficult to do.  Time consuming.  Frustrating when you realise that the inputs are actually going to constrain the actions and therefore reduce the probability of achieving the desired outputs.

ALternative: Start with a realistic assessment of available inputs.  Work out plausible and desirable actions in relation to the inputs.  Track through to see what the probable output is.

Advantage:  you know where you are at all times.  Surprisingly, creativity does not have to be constrained as much as one might think.  Fidelity is, in principle, clear as actions are constrained in the first place.

disadvantage:  discouraging to realise the outputs are remarkably hard to achieve and the level of constraint required to keep fidelity.  limits creativity.  Can lead to wild optimism bias despite the constraints being known in advance.

I think there is a virtue to having some very clear questions before rushing into a logic model:  what are we trying to achieve and why?  When, where and how has this been done before (everything has been done before) and did it 'succeed' / 'fail' - why?  how?  What is different this time round (if anything).  What has been learned?  This is similar to prospective evaluation and a logic model will start to fall out of this by default as you realise - we will need to...if we do X then Y will/will not happen...to achieve C we will need to do B and to do B we will need to have A...and so on .

FInally, you can do both a backward and forward mapping logic model separately and compare them.

There is quite a good set of tools here

http://www.uwex.edu/ces/pdande/evaluation/evallogicmodel.html

and some more historic ones here

http://www.sagepub.com/upm-data/23938_Chapter_3___Creating_Program_Logic_Models.pdf

http://www.innonet.org/client_docs/File/logic_model_workbook.pdf

And if you are really serious about it then you will need to consult the Public Administration journals as they have been publishing on this for about 10 years now and they inform the high level government and NGO strategists.

e.g. Public Administration and JOurnal of Public Administration Research and Theory