Littell et al. (2006) use the data here to introduce analysis of split plot designs using mixed models. Twelve silicon wafers were randomly selected from a lot, and were randomly assigned to four different processing modes. Resistance on the chips was measured in four different positions (four different chips) on each wafer. Mode of processing and position of chips were fixed factors, while wafer was a random effect. The experimental units with respect to process are the wafers. The experimental units with respect to position are individual chips. Thus the wafer is the whole plot, whereas the positions (chips) are split plot units
data(Semiconductor)
The dataframe contains four columns:
Resistance
The response variable of interest. Measured in ohms.
Process
The explanatory variable of interest. The type of process used to create the computer chips. A factor with 4 levels.
Wafer
The whole plot containing four chips. There were four wafers tested, i.e. four levels, 1,2,3,4
.
Chip
Position on the wafer. These are split plots within the whole plots. Four levels: 1,2,3,4
.