Topic 19 o 64· Social value an impact meisurement
Logic model
A logic model is a linear diagram that links inputs, activities, ootputs, ootcomes and impact for a programme, read frae left tae richt as a chain o accoontability: resources gae in, activities tak place, ootputs get produced, ootcomes chynge for the beneficiaries, and impact builds up ower a wider or langer timescale. It is the staundard structure that funders and auditors expect a programme tae be reportable agin, and the forrit-lookin coonterpairt o a backward-mapped theory o chynge.
Why it maitters
HM Treasury's Magenta Book (the Treasury's guide tae evaluation) lays doun the logic model as a requirt pairt o programme evaluation design, and funders sic as the National Lottery Community Fund big their application and reportin templates roond exactly this five-column chain. Its value is that it forces a programme tae state, in ae diagram, whit it will spend, whit it will dae wi it, whit it will produce, and — cruicially — whit sud chynge as a result, at a level o detail that a paragraph o prose tends tae hide. A logic model wi a filled-in inputs and activities column but an tuim or vague ootcomes column can be diagnosed at a glance, which is juist why funders ask for ane.
The maths
The logic model is a structural chain raither nor a formula:
Inputs Activities Outputs Outcomes Impact
(resources (what is done (direct, countable (change for (long-term,
committed) with them) products) beneficiaries) population-level
or systemic change)
Ilka column sud be mair specific nor the last: inputs are whit ye spend, activities are whit ye dae, ootputs are whit gets delivert nae maitter the effect, ootcomes are whit chynges as a result — the distinction covert in fu in ootcomes vs ootputs — and impact is the lastin, aften only-pairtly- attributable, lang-run chynge.
Wrocht example
Local authority (digital debt advice service):
- Inputs: £180,000 annual budget, 4.0 FTE advisers, a case management system.
- Activities: ootreach sessions, one-tae-one debt advice appointments.
- Ootputs: 900 appointments delivert; 750 debt and benefit plans issued.
- Ootcomes: o clients reachin a 6-month follow-up, 60% (450 o 750) report reduced arrears, averagin a £1,200 reduction per client — £540,000 in aggregate arrears reduction.
- Impact: a measurable faa in homelessness applications frae the service's client base ower twa year, only pairtly attributable tae this service alangside ither interventions (see coonterfactual analysis).
Charity (food bank referral partnership):
- Inputs: £45,000, 1.5 FTE coordinator, pairtnership agreements wi 12 referral agencies.
- Activities: referral triage, parcel packin and distribution.
- Ootputs: 5,000 food parcels distributit tae 1,100 households.
- Ootcomes: 68% o surveyed households (748 o 1,100) report improved food security at a 4-week follow-up call.
- Impact: contribution tae reduced local crisis-service demand, evidenced only in aggregate area statistics, nae attributable tae this charity alane.
Link tae saftware ingineerin
The logic model is gey near a literal data model for an ootcomes system: inputs and activities are operational data ye already hae (spend, staffin, session logs); ootputs are easy tae instrument because they are coontit at the pynt o delivery; ootcomes need follow-up data collection (surveys, admin data linkage) that has been deliberately designed and that winna exist unless somebody biggs it; impact usually needs linked, langitudinal, or population-level data ayont ony ae programme's systems. Ingineers biggin reportin tools sud push commissioners tae define ootcome and impact indicators at design time, raither nor defaultin tae an ootputs-only dashboard juist because that is whit the transactional data already supports. See social return on investment for a method that values the ootcomes and impact columns specifically, and benefits realization for trackin whether the impact column wis actually delivert.
Pitfaws
- Stoppin at ootputs. A dashboard that reports appointments delivert or parcels distributit and implies benefit is reportin activity, nae results — see ootcomes vs ootputs.
- Nae stated causal link atween columns. A logic model states the chain but nae why activities sud produce ootputs that sud produce ootcomes; that reasonin belangs in a theory o chynge, and a logic model wi nane ahint it is untestit.
- Treatin it as a yin-aff bid document. Logic models made juist tae satisfy a fundin application and never updatit stop reflectin whit the programme actually does.
- Attribution creep at the impact column. Claimin population-level chynge as caused solely by ae programme, wioot a coonterfactual, owerstates whit the evidence supports.
Sources
- HM Treasury, Magenta Book (2020), Chapter 3. https://www.gov.uk/government/publications/the-magenta-book
- National Lottery Community Fund, logic model guidance. https://www.tnlcommunityfund.org.uk/
- W.K. Kellogg Foundation, "Logic Model Development Guide" (2004). https://www.wkkf.org/resource-directory/resources/2004/01/logic-model-development-guide