Blocking Factors

A blocking factor is a factor which enters a model only as a main effect. Usually the column in the design contains -1 to represent one level (such as Machine A) and +1 to represent a second level (Machine B). When the -1s and +1s are placed wisely in that column, the blocking factor’s effect will not affect the other column’s effects; that is, the blocking factor will be orthogonal to all other terms in the model. In this way, the two machines may be used to speed up the experiment without damaging the other b’s in the model. A factor which can interact with the other factors must not be used as a blocking factor.

You can request that one blocking factor column (the x1x2x3 column) be added to a 3 factor interaction design if you have no interest in computing the three factor interaction. Up to 5 blocking factors be added to interaction designs with 4 or more factors (namely, three factor interactions and/or the four factor interaction). Blocking factors can be combined: you can use one operator on one machine testing those results on one instrument when the blocking column contains a minus; and the other operator, other machine, other instrument when the blocking column has a plus. A factor can be a blocking factor only if it does not interact with any other factor.