Test the mechanism before naming a cause
5 min read
Lesson objectives
- Frame an open advisory brief, map affected parties, test competing mechanisms, and prioritize decision-relevant evidence.
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.
A stakeholder map tells you who may be affected, but it does not establish how an effect arose. Bus trips might fall after a fuel price increase. It is tempting to name the price as the cause, yet the operator may have cut routes during the same month or a holiday period may have begun. Advisory analysis needs a mechanism: a sequence of steps that would explain a result if its conditions hold. Start with the authority's action, move to an intermediary price, then to observable behavior, and finally to an outcome the decision maker values. Beside each arrow, state what could block it. A higher fuel cost will not raise every fare if contracts fix tariffs or the authority absorbs the difference; a fare rise will not necessarily reduce trips where riders have no alternative.
For subsidy reform, draw the first pathway: subsidy reduction, higher operator fuel costs, higher fares if pass-through is allowed, fewer trips for some households, and impaired access to work. These links have different evidential strength. The cost link needs purchase records; the fare link needs contracts and actual tariffs; the trip link needs use data by group; the access link needs comparison of opportunities, not only complaints. Draw a rival pathway beside it: an operational timetable change reduces use even without a fare change. Both mechanisms may operate. The task is not to force a single winner, but to write distinct predictions for a specified time period before you inspect the results, so an explanation cannot be endlessly rewritten after the fact.
Take training figures: average fares rose 10%, observed trips fell 15% over three weeks, and the operator cut scheduled service 20% in the same period. Do not subtract 15 from 10 and call it a net effect; those percentages use different denominators. Do not infer demand elasticity from two points because supply changed and passenger composition may have shifted. Request a prechange series, dates for each intervention, and routes whose service remained stable if available. Route data may show that use fell mainly where service was cut. They might instead show a marked fall after a fare rise on routes with unchanged service, strengthening the price account without turning it into a guaranteed experiment. Explain how sample coverage and concurrent changes limit the conclusion.
A causal diagram helps select measures and avoid controlling for the wrong thing. If the service cut was itself caused by more expensive fuel, removing it from an estimate could erase a genuine part of policy impact. If it was an independent decision made before reform, it is a competing explanation. Event order makes the difference. Ask for the operator's decision record and the announcement and implementation dates; separate a direct fare channel from an indirect service channel. Do not equate a trip count with household welfare. Families might switch to a more costly mode, leaving total trips stable while increasing their burden. Add transport expenditure and access to work as measures, reducing reliance on an easy but incomplete indicator.
Present three levels of judgment rather than a confident single figure: an observation of what occurred, a conditional leading explanation, and a causal effect that the evidence does not yet identify. Tell the official: 'Observed trips fell after fares changed, but the simultaneous service change prevents assigning the entire fall to price.' Specify a learning test within the advisory role: weekly comparisons of routes with similar fare policies and different service patterns, or a carefully sampled set of rider accounts, designed by the authorized institution with privacy safeguards. If most losses preceded the fare rise, revise compensation advice based on price alone. If both fare and service channels matter, protect access rather than calculate transfers solely from fares. A rival explanation improves the decision; it does not excuse a decision made blind.
Lesson summary
A stakeholder map tells you who may be affected, but it does not establish how an effect arose.