Abstract: The experiment took forth graders as subjects, adopted completely random-designof equal groups. The experimental group adopted SAT at the same bine, thecontrolled group followed conventional teaching pattern to do relevant exercises. Theexperiment results were as follows: (1) trained students obviously improved theirability to solve problems, of all the students the middle--level students benifitted most;(2) the thening effect correlated with the intelligence level of the pupils; (3) pupilswelcomed the training and made positive responses.

Key words: applied problems, structure--analysis, ability to solve problems