The use of grafted polynomials and dummy variables in analyzing time series data

Abstract

Describes an innovative statistical method for analyzing single-S time series data. The S and data in the paper were selected from a larger study which applied operant conditioning to ambulation problems of severely handicapped patients and used positive verbal reinforcement. The method described is a variation of traditional multiple regression analysis described elsewhere as "grafted polynomials" (Draper & Smith, 1967; Fuller, Note 1). An extensive explanation of the use of dummy variables also is provided.