Review Of Statistical Methods For The Analysis Of Comparative Experiments

The analysis of the results of metallurgical experiments is often complicated by experimental errors and uncertainty, the complexity of the system being studied, and the confusion of correlation with causation. Consequently, comparative statistics are used to determine if a difference exists between the outcomes of tests, an estimate of this difference, and the precision of the estimate. This paper reviews the use of significance tests and confidence intervals for the comparison of results from two tests, the use of reference distributions, and blocking and randomization techniques to reduce unavoidable sources of variability. These techniques are demonstrated using plant and laboratory data.