Evaluation Framework
Every other model on the usual comparison lists evaluates a programme. Valerie Anderson's 2007 model, commissioned by the CIPD, evaluates something else: whether an organization's learning provision as a whole is aligned with its strategy, and which kind of evidence its decision-makers will actually accept. If your clients are UK HR teams, this is the model their professional body taught them.
Before measuring any course, ask whether the learning portfolio points at what the organization is trying to do. A beautifully evaluated programme that serves no strategic priority is a well-measured waste. Anderson's research found this alignment question mattered more to senior managers than any efficiency metric.
The model groups evaluation evidence into four kinds of measure, and expects organizations to use a mix:
The stage that makes the model unusual. Instead of prescribing one currency of proof, Anderson says: find out which of the four kinds of evidence your decision-makers value, and evaluate in that currency. A finance-led organization wants ROI. A growth-stage organization may care most about capacity. Reporting the wrong kind of value, however rigorously, persuades nobody.
Anderson's third stage is, quietly, a piece of consulting advice dressed as an evaluation model. Before you design a single survey for a client engagement, find out what their organization counts as proof. The model gives that discovery conversation a CIPD-endorsed name, which matters when the person across the table holds a CIPD qualification, and in UK HR departments they usually do.
It also explains a pattern every consultancy has met: the immaculate evaluation report that lands with a shrug. Usually the data answered a question the client wasn't asking. The fix isn't more rigour. It's stage 3: ask which currency they deal in, then collect for it. The programme-level instruments barely change; Phillips' estimate questions feed the ROI quadrant, expectation questions feed ROE, and the same core survey feeds both.
Anderson's model doesn't replace programme-level evaluation; it decides what kind to run. Once stage 3 picks the currency, the collection looks familiar: reaction and relevance questions after delivery, retrospective before-and-after scales for capability change, delayed application questions, and expectation questions agreed with the sponsor up front. ImpactCheck is the instrument for that layer, whichever quadrant the answers are destined for.
One survey layer, four kinds of evidence. Branded micro-surveys that feed ROI, ROE, or benchmark reporting equally well.
Start your free trial30-day free trial. No credit card required.