Improvisation demands and time of opinion measurement were varied to test the prediction that verbal intelligence would facilitate opinion change through active participation (essay writing), compared with passively reading messages, only when a high degree of improvisation is required.
Each of 79 subjects participated in one of the following conditions: total improvisation, partial improvisation (where the subjects were given three basic points to develop in their arguments), and reading persuasive communications. Opinion was measured both immediately and two weeks after
the experimental treatments. The prediction was confirmed by the fact that a significant interaction effect was obtained at each time of measurement between verbal ability and those methods of opinion change involving total improvisation and reading standard messages, with intelligence facilitating
opinion change in the former and inhibiting it in the latter condition. Under conditions of partial improvisation, the effect of verbal ability not only washed out, but actually reversed in direction, although this trend was not significant.