Maintenance factors in coercive mother-child interactions: the compliance and predictability hypotheses.

The team watched three mother-child pairs at home for many days. All kids had serious behavior problems.

They wrote down every coercive chain: times when yelling, hitting, or whining kept going back and forth.

They tested two ideas. Compliance hypothesis: kids behave worse when moms give in. Predictability hypothesis: kids act out when mom’s reactions are hard to guess.

Only the predictability idea held up. When a mom sometimes gave in and sometimes got tough, the child stayed coercive longer.

How often the mom gave in did not matter by itself. What mattered was whether the child could guess what would happen next.

Koegel et al. (1992) ran a bigger test six years later. They split two traps: the compliance trap and the inconsistency trap. Their path numbers back the same point—unpredictable mom reactions feed the chain.

Soulières et al. (2007) looked at mothers of disabled children. They found moms mix coercive and gentle moves on purpose. That work widens the 1986 view: unpredictability may be chosen, not random.

Holmes (1990) mapped tiny ABC loops in typical mother-baby talk. The grain-size method is the same; the finding is different. In language, clear contingencies help learning. In coercion, unclear contingencies lock in trouble.

Before you teach a parent new skills, test how steady their reactions are. Count five tricky moments and see if the child can guess the outcome. If the answer keeps changing, teach the parent to pick one clear response and stick with it. Stable contingencies first, compliance programs second.