Cognitive Development, 7, 259-268 (1992) The Role of Situational Context in Solving Word Problems Elsbeth Stern Max-Planck-Institute for Psychological Research, Miinchen, F.R.C. Anne Lehrndorfer Centrum fiir Information+ und Sprachverarbeitung, Miinchen, F.R.G. Word problems depicting the comparison of quantities have been shown to be difficult for elementary school children in several studies. One reason for this may be that young children lack situational understanding because they are not familiar with the quantitative comparison of sets. To test this assumption, we presented 45 first graders with “compare” problems that were embedded in a familiar situa- tional context. The results showed that children who received compare problems following stories that induced situational understanding of qualitative com- parisons performed better than children who received compare problems follow- ing stories that had nothing to do with the comparison of sets. The data suggest that most first graders have access to the mathematical problem-solving’knowl- edge necessary to understand and solve compare problems only when these problem are embedded in a familiar situational context. In several studies, it has been shown that simple arithmetic word problems that demand addition or subtraction of two numbers are not necessarily equal in difficulty (Briars & Larkin, 1984; Carpenter & Moser, 1984; Cummins, Kintsch, Reusser, & Weimer, 1988; Kintsch & Greeno, 1985; Riley & Greeno, 1988; Riley, Greeno, & Heller, 1983; Stem, in press). Problems dealing with the comparison of sets (e.g., John has 5 marbles. Peter has 2 marbles. How many marbles does Peter have less than John?) are more difficult than problems about the exchange (e.g., Peter had 3 marbles. Then John gave him 2 other marbles. How many marbles does John have now?) or the combinafion of sets (e.g., John has 5 marbles. Peter has 2 marbles. How many marbles do John and Peter have altogether?). Riley and Green0 (1988) reported that 67% of the first graders were able to solve exchange problems, 60% performed correctly in combine prob- lems, but only 19% solved compare problems. These results indicate that We would like to thank Merry Bullock, Norman Freeman, Zemira Mevarech, Tine Whitley, and two anonymous reviewers for their helpful comments on former versions of this article. Correspondence and requests for reprints should be sent to Elsbeth Stem, Max-Planck-Institute for Psychological Research, Leopoldstrafie 24, 8000 Miinchen 40, F.R.G. Manuscript received December 18, 1990; revision accepted October 23, 1991 259