A context analysis of contingent teacher attention.

Manabe (1990) watched two classrooms for long stretches. The kids had intellectual disabilities or were typical learners.

The team counted every time a teacher praised or looked at a child who was working. They wrote the numbers two ways: raw counts and the share of all attention that was earned.

The share version won by a mile. When 80 % of teacher looks and words went to on-task kids, seat-work stayed high.

Raw counts told almost nothing. A teacher could give fifty praises, but if most went to goofing-off kids, work still fell apart.

Taber et al. (2017) extends the idea. They showed staff a five-minute clip of ideal praise ratios. After one viewing, new teachers hit the same high shares Manabe (1990) says matter.

Iwata et al. (1990) ran the same year and also checked classroom codes. They proved 15-second momentary time sampling is close enough for most behaviors, so you can copy K’s method without all-day filming.

Baer et al. (1984) adds the long view. They argue you must later fade social praise so kids work for the math or reading itself, not just for teacher smiles.

Stop tracking “five praises per minute.” Track the percentage that lands on the behavior you want. Aim for four out of five teacher looks and words to follow on-task acts. You can hit that target with a simple tally sheet or the 15-second method Iwata et al. (1990) validated. Do it for one week and watch engagement climb.