Problem length as a structural variable in verbal arithmetic problems

Of five structural variables that in previous studies accounted for a significant amount of the observed variance in the error rate in verbal arthmetic problems, the length variable (number of words in the problem statement) apparently was more important in the upper grades than in the lower grades. In this study three forms of a verbal problem set in which the number of words in the problem statements were systematically varied were administered to classes of students in Grades 4–9. Using regression analysis, the investigator found that although the length variable did not measure a significant amount of variance between two of the forms, with the third it did.This study was begun while the author was a staff member of the Institute for Mathematical Studies in the Social Sciences at Stanford University. The research was partially supported by National Science Foundation grant G18709.