It got better. How do you know your work did it? Someone takes part in your programme, and afterwards their situation improves. That’s worth understanding. It doesn’t automatically mean your programme caused it. Credit the wrong thing, and future funding can back the wrong work and miss what people actually found useful. Maybe the person had other help too. Maybe the local job market picked up. Maybe they’d have improved anyway. The improvement you saw isn’t your impact. Your impact is the gap between what happened and what would have happened without you. Nobody sees that second version directly. A fair comparison is how you estimate it. Decide what a fair comparison would look like, ideally before the results come in. It doesn’t have to be a perfect experiment; other designs can give useful evidence if you’re clear about their limits. Check where the groups differ, how the result was measured and what else changed. And if your study can only describe the improvement, say that, r...