The Analysis of a Factorial Experiment (with Confounding) on an Electronic Calculator

Abstract

Summary (a) The analysis of a 24 factorial experiment confounded into blocks of two carried out on an electronic computer (the EDSAC) is described. Treatment effects were calculated in the factorial analysis and summarized. The factorial analysis, which would have taken 10 or 12 weeks with a desk machine, was carried out in about 13 days. The summary represented a greater saving of time, being carried out in 3 days. (b) Three-quarters of the time needed for the computation of the factorial analysis was used in punching the data tapes; calculators capable of reading from punched cards are available and may be more suited to statistical calculations of this type. No data tape had to be punched for the summary analysis since the input for this was the output of the factorial analysis. (c) The total machine time needed for the factorial analysis was about four hours. (d) About 450 orders were used in each part of the analysis leaving storage space for about 190 numbers of 10 decimal digits.