Before blaming the person, an investigation should ask what about the system made the slip likely

What it is

Systemic causes taking priority over individual attribution is an analytic principle that applies well beyond formal accident review: in a usability test, in a log of everyday errors, any time a user is observed doing something wrong, the order of analysis should first exhaust what can be explained at the system level — is the information presented clearly, are the right constraints in place, does the flow match the user's mental model — before turning to whatever's left that the system level can't explain. It should not run the other way, where an individual gets labeled "careless" or "not skilled enough" first, with the system-level possibilities treated as already ruled out.

Why it happens

The order matters because settling on an individual attribution first psychologically shuts down the motivation to keep asking about system conditions. An explanation like "this person wasn't careful enough" is almost always available, because tracing any error back far enough will turn up some moment when the operator could have been more careful — and that easy availability hides how shallow the explanation is. It doesn't say why the system allowed that lapse to turn into a consequence, and it doesn't answer whether someone else with the same background, in the same conditions, would have produced the same result. Analyzing system conditions first, by contrast, gives a falsifiable test: if errors still concentrate on a small number of people after the interface, process, and information presentation have all been improved, that's when discussing individual factors has grounds; if the error rate drops broadly as system conditions improve, that shows the original weight on individual factors was misplaced from the start.

Studying it

Checking whether this ordering is actually followed in usability testing or error-log analysis means looking at the structure of the analysis record itself: does it list system variables — interface elements, task flow, terminology — first, or does it lead with individual attributions like "the user wasn't paying attention" or "the user is a novice, so errors are expected." Methodological caveat: individual attribution often disguises itself as system analysis in the form of "user persona differences" — chalking errors up to a trait of low-skill users without asking whether the interface adequately supports that population. That's still shifting a system problem onto a user-group label, and it needs to be specifically watched for when reviewing analysis records.

Where it stops holding

Systemic-cause priority is an ordering principle for analysis, not a conclusion that the system is always the sole cause: if a person knowingly and capably chose to deviate from a procedure, that behavior needs to be handled at the individual level on its own terms, and systemic causes shouldn't be used to sidestep it. The principle also doesn't apply once system conditions have been repeatedly verified as clear enough and constraints adequate enough — if errors still consistently concentrate on a specific individual at that point, continuing to focus on the system level is itself a way of avoiding an individual-level issue that already has evidence behind it, such as whether training was actually adequate or whether the role's competency requirements match where this person currently stands.

Applying it

Build this ordering into review and usability-issue templates: put the system-variables column ahead of the individual-attribution column in the analysis table, and leave the individual-attribution column blank unless the system-variables column has been filled in seriously and still can't fully account for the observed error. Ban closing lines like "the user wasn't careful" or "user operator error" from standing alone as a conclusion in analysis records when they point to no specific system condition. Verification: sample recent usability test reports or error records and tally what proportion of conclusions land on system variables versus individual attribution. If individual attribution dominates and there's no corresponding system-variable analysis on record, this ordering principle isn't actually being enforced.

Related

• Same group: A10.09.2 treating "human error" as a conclusion masks systemic causes · A10.09.3 blame-free error reporting · A10.09.5 skilled and novice users have different error profiles, so safeguards can't be one-size-fits-all
• Nearby: A10.07 Swiss cheese model · Y7.01 systemic causes
• Search terms: systemic cause · individual attribution · root cause priority