Define the outcome, population, and baseline
6 min read
Lesson objectives
- Apply define the outcome, population, and baseline to a development case, explaining inputs and decision limits.
Please sign in to use all the tools
Once you sign in, the reading tools are yours: larger or smaller text, typeface, line spacing, focus mode, highlighting and bold, and notes saved to your account and gathered in My study.
An initiative may begin with an appealing ambition such as ‘improve young people's job prospects.’ That phrase tells the design team neither whom the intervention will reach nor what change will count as success. Before selecting training or finance, write the decision question: among eligible young people in a defined area, did the share obtaining sustained paid work within six months increase relative to what would have happened without the initiative? The final clause does not assume that every improvement was caused by the initiative; it identifies the outcome, population, period, and alternative. Keep this outcome separate from activity outputs. Registrations or training hours may show service reach, but neither is a job obtained by a learner. If a team is rewarded only for registrations, it may select easy-to-enroll people and neglect those who need more support. An outcome card therefore pairs a reach measure with an outcome measure and a protective measure, and explains the proposed link among them.
Define the population before calculating a rate. ‘Participants’ might mean registered people, those who attended one session, or those who completed the program; each definition has a different denominator. If the objective is population level, the equity denominator may be all eligible people in the service area, rather than completers alone. A completion rate, by contrast, uses enrolled people. Record eligibility dates and criteria before recruitment, along with counts of eligible and excluded people and reasons for exclusion. Do not change the denominator after seeing the result. A compact flow table should contain the numbers eligible, offered the service, started, completed, and reached the outcome at the scheduled measurement point. This sequence helps explain where impact was lost: did the service fail to reach people, did people fail to complete it, or did new skills fail to translate into jobs? It also prevents a success rate among completers from being presented as an effect on everyone targeted.
A baseline is the indicator before exposure to the intervention, measured as consistently as possible with follow-up. A late question about someone's position a year ago is weak when employment records are available; current experience can alter recall. Define the unit of observation: a person rather than a job application, and a household rather than a visit, unless the question itself concerns applications or visits. Give sustained paid work an operational definition, for example paid work lasting at least 90 days within a common follow-up window, without suggesting that this is a compulsory international standard. Specify the data source, verification method, and who could not be followed up. If those with the worst outcomes disappear from measurement, the reported rate may rise misleadingly. Store the date, numerator, denominator, and collection method with every value, because a bare number cannot be audited or compared.
After fixing the denominator, disaggregate outcomes by groups whose access or benefits may differ, while protecting privacy where cell counts are small. An overall rate may rise because better-educated people enrolled more often while those with less education saw no gain. Do not solve this by removing a group from the denominator. Show offer, start, completion, and employment rates by group, then locate the stage at which people leave the pathway. Suppose 80 of 200 eligible women and 220 of 300 eligible men were offered service. Reach is 40% for women and 73.3% for men, although overall reach is 300/500 = 60%. The gap alone does not prove intentional discrimination; eligibility rules, contact channels, or session times may differ. It does warrant examining the offer mechanism before calling the 60% average equitable. Document the offer rule and test whether service hours or invitation channels can change without undermining eligibility. If group definitions or follow-up windows change midproject, retain separate series or recalculate on a common basis instead of joining incomparable points into one trend.
In the final measurement card, never leave ‘improvement’ without a direction and unit. Write, for example, ‘share of eligible people with sustained paid work at six months,’ then define sustained work, the start of the six-month clock, and treatment of lost follow-up. Add the baseline, its source and update date, and a separate cost-per-eligible-person-reached measure. A completer employment rate alone cannot answer whether area-wide unemployment fell: the initiative may be too small to move that rate even if participants benefit. Conversely, the area rate may improve during general economic growth while the training service fails. Keeping each measure tied to its own question prevents a leap from delivery success to economy-wide impact.
Lesson summary
Define the outcome, population, and baseline