Behaviour support is based on data that the practitioner doesn't gather. The functional hypotheses, the changes to strategy, the limiting practice reporting, the conclusions in the review: it's all recorded by the support staff, at the edges of stressing work shifts, in the system that was once installed by someone. Most practitioners, given the resulting dataset, see the same three artefacts: the blank stretches where busy levels recorded nothing; the vague entries where behaviour and was upset was used to signify anything observable; and the suspicious tidiness of incident counts that decrease the longer the form is, indicating form-fatigue and not behaviour. The first thing you are tempted to do is redesign the form, train harder, it is the wrong diagnosis. Data quality is not a workers' diligence issue. It's a design issue and the design limitation is the shift itself.

Design for the ninety-second window

The honest constraint: A support worker's recording window, after an incident, during a shift is approximately 90 seconds of divided attention and any system which requires more will be lost to arithmetic, not attitude. It is the basis of designing everything. Tick and circle: pre-listed antecedent categories from your functional assessment, behaviour definitions as observable checkboxes, consequence options circled (not composed), with one short optional line for anything not covered by the categories. The practitioner's job is to select the three that this plan's decisions really need, and what matters is not everything, but what matters is counted. And place recording where it can be utilised in the house or arrangement: somewhere on the fridge with a clip on the board, the provider's app, a shared note, it doesn't matter as long as it's available within the window.

Define behaviour so a stranger could score it

Vague-entry is a definition problem disguised as a diligence problem. Aggression is four thresholds and hit, kicked or threw an object at a person is one. Each measured behaviour should be defined as far as it's possible to make it concrete so that a new casual reading the summary in the briefing gate outlined in Part 1 of this series will score exactly the same incident as the senior worker and that's the test; if the two workers on one shift would score differently, the definition is not complete. The same discipline has to be applied to the antecedent categories: for example, a worker need not learn the word transition, and can get the functional information your analysis requires, in tick form, from the workers.

Close the loop or lose the collectors

Effort is not the reason for data decay, it's because it's pointless: workers cease to record what they see simply going into a folder. Bring the data back, the counter is too cheap and it's wonky. A one-page graph at the monthly review, look what your records show, incidents changing from morning to transitions, the new strategy tracking downward, teams that have seen their data move a plan record differently forever after. Also, your quality control comes with you in the loop, and gaps and drift naturally emerge when you examine the data together, on a monthly basis, instead of digging them up at review.

How this plays out in practice

Imagine that a woman's plan review is coming up in supported accommodation, and six months of her records are missing, and the other six months are all read “escalated, settled with support”. Three measures selected, incidents of behaviour of a clearly defined nature, prn free days, participation in the two activities valued in the plan, the tick sheet created from the functional assessment categories, and 90 second compliance testing with the senior worker, the graph promised and delivered month after month. Two quarters later, the review is in the real signal, incidents are piling up on staff-changing days, the skill-building plan is clearly effective and one strategy is squarely dropped on the floor. The workers were not made more hard-working. The system went on to become collectable.

Next steps

Then take the worst set of data you have now and use the 90 second test: can a tired worker, working in the window that a shift allows, accurately complete one entry? Redesign until answer is yes, if so, then schedule for first graph.

Find Support Workers

Use your Funding

Supported Independent Living

NDIS Provider Workforce Solutions

NDIS Support Coordination Platform

Aged Care Providers & Coordinators Staffing Solutions

Business Solution

Support Services Available

Conditions We Support

Cultures We Support

Getting Usable Data From Support Workers

Table of Contents: