Use observation protocol
4 min read
Lesson objectives
- Produce a sampling sheet and comparable rates from the case facts and state 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.
Public-life observation describes events in a defined window; it is not a census of residents. Begin with the decision question: do you need passage volume, places where people stop, activity types, or differences between periods? Each question needs a different counting unit. A passage is an event at a count line, and one person may generate more than one event. Sitting or play recorded in a snapshot is a state at an instant, not a trip or dwell duration. Dwell time requires a suitable entry-and-exit method that respects privacy; dividing seated people by passersby cannot create it. Without a declared unit, even arithmetically correct tables are hard to interpret.
Choose a clear count line and avoid crossing several routes without a rule. Mark it on the site plan, and define travel direction if the decision requires direction. Specify window start and end, treatment of interruptions or crowding that makes counting unreliable, and observer and checker roles. For stationary activity, draw non-overlapping zones and take counts at scheduled moments. Record how many people are sitting, standing, or playing in each zone without identifying them. Another observer should be able to repeat the same categories. Where descriptions overlap, predefine a primary category: for example, a seated person talking counts once as seated, with interaction noted separately, instead of being added twice to the total people.
To compare passage counts from unequal windows, convert them to an hourly rate: passages divided by observation minutes times 60. Standardizing time does not remove differences between market and ordinary days, hot and mild weather, or an event and a routine period. Add context fields for day, observed weather, nearby event, entrance closure, and observation period. When contexts differ, label the comparison descriptive or collect a closer match. Do not drop an inconvenient window after seeing its count to create a preferred trend; predeclare exclusion reasons or report it separately.
Test counting reliability with a short shared observation by two observers. Agree how to treat a child walking with a caregiver, someone who enters and exits, and a group walking along the count line. Record disagreement rather than silently averaging incompatible definitions. Revise the rule and recount if possible; otherwise disclose uncertainty. If crowding is too dense for a reliable manual count, use a shorter window or a suitable method with privacy review instead of reporting spurious precision.
Observation becomes decision evidence when the indicator, denominator, and inference limit are explicit. ‘Use increased 30%’ is unsound if the numerator counts passages but the claim concerns unique users. Increased passage at one gate after its opening may represent route substitution rather than an increase in total visits. Connect the metric to a specific goal and include a stop indicator: did circulation improve while wheelchair access worsened? If an actual access obstruction appears, contain and refer it immediately through the responsible owner; do not wait for statistical significance. Bound conclusions to the observed sample and timing, and plan a verification window before permanent decisions.
Lesson summary
Accept a table with 108 and 72 passages per hour, an event and weather comparability caveat, seating described as snapshots, a plausible matched repeat, and at least one valid inference limit.