Reducing inequalities in word-problem solving. Working on comprehension before processing numbers.

We examine whether performance of Word Problem Solving (WPS) improves by focusing on problem comprehension rather than on numerical treatment. For validation, we compare performance of two groups (experimental and control) in 22 second-grade classes, with a total of 389 students from priority (REP) and (Non REP) non-priority education areas. For 12 weeks, both groups had been trained to solve the same word problems. Control group (N = 190) were taught to problem solving using conventional teaching. Experimental group (N = 199) solved first the same problems without any numerical values. We evaluated performance on WPS before (pretest) and after (immediate and two deferred posttests) training. We concluded that initial work on comprehension, significantly and sustainably improves performance, and reduces inter-individual differences in WPS, especially in R, compared to classical approach.