ABSTRACT
This chapter describes risky tests in cognitive developmental research and explains how some of Bob's most influential work exemplifies this idea. It describes some attempts, influenced by Bob, to develop riskier tests of theories about the long-run effects of early mathematics intervention, and an argument for riskier tests in correlational research on cognitive development. Experimental research in cognitive development is unique in its application of risky tests to measurement. In the case of children's mathematics achievement, substantially more variance is observed at the trait level than the state level. Three design features that would facilitate more rigorous coherent pattern matching in correlational research on cognitive development are multiple testing occasions, non-equivalent pretests and posttests, and larger samples. Familiarity with effect sizes also enables coherent pattern matching: It facilitates the theoretically important comparisons of estimated effects across time and domains, and between experimentally derived benchmarks and correlational estimates.
