This paper presents a procedure for identifying possible interaction effects of factors in distributions of intelligence test scores through the construction of analogs to the development of intelligence. The procedure is based on the Pascal triangle for generating coefficients in the binomial distribution, but incorporates a technique for modifying coefficients. The most adequate analog is one which produces coefficients identical to an empirical distribution while requiring the fewest modifications of elements in the Pascal triangle. The procedure was applied to IQ test data reported by Burt on London school children. In general, the analogs for these data suggest different effects of social class on intellectual development at the upper and lower IQ ranges.