# Welcome — the assumption you didn't write down

A 45-minute session on the weakest part of most theories of change: the step between two boxes that everyone in the room believes and nobody has stated.

You will need one theory of change you have worked on. A photo of a whiteboard is fine. A perfect diagram is not required and may get in the way.

Mark each step **done** as you finish, or **stuck** if you want the facilitator.

# What an arrow actually claims

A theory of change is mostly arrows. Each arrow is a claim that one thing produces another, under conditions nobody has written down.

Take a common chain: *train health workers → health workers counsel mothers → mothers change feeding practice → child nutrition improves*.

Three arrows. Each one hides at least one assumption:

- Trained workers will actually counsel — which assumes they have time, and that counselling is rewarded rather than merely permitted.
- Counselling changes practice — which assumes the mother controls what the household cooks and when.
- Changed practice improves nutrition — which assumes food is available to change *to*.

None of those is unreasonable. All three are empirical, and all three can fail while the programme is delivered exactly as designed.

# Find your weakest arrow

Look at your own theory of change and pick the arrow you would least want to defend to a sceptical funder.

Write, in one sentence each:

1. What has to be true for this arrow to hold.
2. Who has to do something they are not currently doing.
3. What would you see, early, if it were failing.

The third one is the important one and it is the one usually missing. An assumption you cannot observe failing is not being monitored, whatever the logframe says.

#[quiz] Assumptions

## Which of these is a testable assumption rather than a restatement of the activity?

- [ ] Health workers will be trained
- [x] Health workers will have unfilled time in which to counsel
- [ ] Counselling will be delivered
- [ ] The training curriculum will be approved

## An assumption with no observable early failure signal is best described as:

- [ ] A risk, to be listed in the risk register
- [x] An untested belief the monitoring system cannot see
- [ ] An external factor, outside the programme's control
- [ ] A design constraint

# The harm pathway

Now add something most theories of change omit entirely: a pathway for the programme making things worse.

Not risk-register hedging. A real mechanism. For the nutrition chain above, one candidate: counselling that names the mother as responsible for a child's nutrition, in a household where she does not control food purchase, shifts blame onto the person with the least power to act.

Write one harm pathway for your own theory. Be specific about who is harmed and by what mechanism.

If you genuinely cannot construct one, that is worth saying out loud — but try for five minutes first. Most people find one.

# What you would measure differently

Look at your weakest arrow and your harm pathway together, then answer:

**Does your current indicator set contain anything that would move if either came true?**

For most programmes the honest answer is no. Indicators track delivery, because delivery is what the team controls. The assumptions and the harms live in the space between the boxes, where nothing is counted.

Pick one indicator you would add. Just one. It should be cheap, observable early, and capable of embarrassing you.

#[quiz] What a theory of change is for

## The main practical value of writing a theory of change is:

- [ ] Satisfying a donor template requirement
- [ ] Communicating the programme to outsiders
- [x] Making the claims explicit enough that they can be checked and disproved
- [ ] Sequencing activities across the workplan

# Close

Take back one sentence: the arrow you would least like to defend, written out with what you would see if it were failing.

Add it to the next monitoring conversation you have. Not the next redesign — the next conversation. Theories of change decay because they are written once and consulted never.

**Going further.** The [Theory of Change Studio](https://www.impactmojo.in/Labs/toc-lab.html) walks through the same process with structured prompts, and the [LogFrame Builder](https://www.impactmojo.in/Labs/logframe-builder-lab.html) turns the result into an indicator set. Both are free and nothing is uploaded.