Rules in the balance - Classes, strategies, or rules for the Balance Scale Task?

The Balance Scale Task was recognized in the early 1980s as a way of eliciting different rule-governed response patterns for proportionality reasoning (Siegler, 1981). However, soon other rules than the four defined by Siegler were suggested and doubt was expressed whether task behavior was rule-governed at all. Doubt could emerge since the scoring laid down by Siegler lacked a firm empirical basis. To redress this lack exploratory latent class analysis (LCA) was used with data on 484 children. The number of classes of answer-pattems found was six. Three classes reflected Rules 1, 2, and to a lesser degree 4. One class reflected a rule not considered by Siegler. The last two classes, however, were not rule-like at all. No simple straightforward strategy seemed to be involved. Nevertheless, age trends suggested that at least one of these classes reflected a variety of a complex strategy that replaces Siegler's Rule 3. The paradoxical results of this and previous research can be solved by distinguishing between "classes" as empirically based probabilistically defined collections of responses (bottom up), and "strategies" as involved in responses that are consistent in view of being interpretable as rational. (C) 2001 Elsevier Science Inc. All rights reserved.