A picture that looks complete
In a hypothetical workshop, maintenance requests take a long time to complete. The team introduces a handover card specifying the request, inspection result and next responsible role. Requests appear faster in the following period. The presentation seems straightforward: a new procedure followed by improvement. Yet the incoming workload has also fallen, a more experienced worker has joined, and a season of particularly complicated repairs has ended.
This account establishes that change and improvement occurred in sequence. It does not establish how much improvement the procedure caused, if any. A reading through boundaries and connections might suggest that the card clarified a transfer point. That reading cannot by itself rule out workload, experience or repair complexity. The causal question begins here rather than being settled.
The comparison we cannot directly observe
The precise question is how completion times for the same requests in the same period would differ with and without the card, under a specified definition of using it. We cannot observe both complete paths for one request at the same moment. A comparison design must therefore justify using other information to estimate the path we did not observe.
A simple comparison between an earlier and later period substitutes the past for that missing path. This can describe change, but it relies on assumptions about continuity of conditions. If request types, working hours or time recording changed, the difference between periods would not represent the card's effect alone. Giving the indicator a precise name does not remove this problem.
Three possible routes to the outcome
The diagram draws three arrows toward completion time: one from the handover card, one from workload and one from worker experience. These are possible explanatory pathways to examine, not causal findings established in this workshop. There is no arrow from the card to reduced workload because the scenario does not say that either caused the other.
The drawing helps resist reducing the scene to one convenient relationship. We then ask about omitted pathways. Did repair difficulty change? Did time recording start later? Were slow requests left out of the report? Selection and measurement can alter what the observed difference means even before we consider what caused it.
A comparison closer to the question
Where appropriate for the work and feasible without harming service, we might plan to assign handover methods to comparable work units using an explicit random mechanism. The assignment unit needs care: a worker who learns the card on one request might change practice on another. We define the outcome, monitor adherence and choose follow-up sufficient for the question and anticipated uncertainty.
Where that arrangement is unavailable, an observational comparison can still be more careful than two separate snapshots. We can record request types, workload and experience consistently, examine time trends and seek suitable comparison work. Conclusions remain conditional on assumptions and limitations. A larger table or more attractive chart cannot supply strength absent from the design.
We may gain a description without settling causation
Suppose recorded time falls while handover errors remain unchanged. That observation invites reconsideration: flow may have improved for other reasons, or the card may affect a different part of the process. We therefore specify a primary outcome and meaningful supporting outcomes beforehand, rather than selecting whichever indicator moved favorably after observation.
The record might also reveal that workload fell only for the group using the card. Hiding this would simplify the story at the expense of the question. An honest conclusion may describe improvement accompanying several changes while leaving causation unresolved. That is a useful foundation for further work, stronger than certainty undone by one question about other conditions.
A sentence the evidence can support
Before writing that clarification caused improvement, we check what the design actually permits: who entered the comparison, how they were selected, what changed alongside the intervention and how uncertain the estimate remains. With only a before-and-after observation, we can report improvement after the procedure during the observed period and identify other known changes.
The framework's language can suggest a possible intervention point. Research then examines that intervention using the tools of causal inquiry. Each contributes a different part of the work. The first attempt may chiefly reveal which data and comparison are needed next, so that 'because of it' eventually becomes a defensible conclusion rather than a convenient abbreviation for 'after it.'



