Interobserver Agreement among a Staff Member and Visitors at a Wolf Sanctuary

Researchers asked everyday zoo visitors to watch wolves and tap a free phone app when they saw a target behavior.

Half the guests tracked one behavior, such as pacing. The other half tracked two behaviors at once, like pacing and howling.

A trained staff member watched the same wolves and scored the behaviors for comparison.

Visitors who watched one behavior agreed with the expert at least 80% of the time.

When guests had to watch two behaviors, their agreement stayed high, but a stricter kappa score dipped below the good range for some.

The app worked best when the rules were short and the behavior was easy to see.

Marano et al. (2020) also showed that people with little training can judge behavior well, as long as they first watch and rate short clips.

Yarzebski et al. (2024) proved that a quick video model can teach caregivers near-perfect technique, matching the low-cost spirit of the wolf study.

Rice et al. (2009) quadrupled staff greeting accuracy with a five-minute talk and praise, echoing how simple instructions gave visitors 80% accuracy here.

García‐Leal et al. (2019) warns that extra effort can warp data; likewise, asking guests to score two behaviors at once lowered their kappa scores.

You can hand a tablet to a parent, volunteer, or new tech and collect solid data on a single, clear behavior. Keep the definitions short, add photos, and check the first few sessions. If you need two behaviors, train a bit more or split the job between two people.