The Boulder School District has two years experience using clickers in over 30 8th grade science and math classes, each with
about 25 students. Their results are very positive and their assessment quite thorough.
The recently completed thesis by Angel Hoekstra, Univ. of Colorado, shows that in classes outside the sciences (e.g. social scieinces) the largest impact of clickers in on full-class discussions. This should be true in classes even of 20 (you didn't say if these are science classes). In social sciences, it is a question of whether students speak out. Knowing what others think, anonymously, causes much more willingness to join in the discussions.
I think it is worth trying in groups of 20, yes.
In my science seminar class of 10 students they didn't speak out as much as they should have, because they
overestimated what they knew. If one student did something on the white board they often thought,
"yes, that's what I would have answered." But on an exam they wern't as good as they thought.
For that class I adpoted "clicker methods" without the clickers. That is, I often asked questions,
had EVERYONE write the answer on paper, exchange the papers and critique each other, THEN
have class discussion. Exam scores went up 15%.
Doug Duncan
Univ. of Colorado