Fit line to X% of smallest effects: The rotatable selection lines on the half-normal and normal plot will fit to X% (50% by default) of the smallest effects rather than all unselected effects. Change to 100% to force the program to behave like version 9 or earlier.
Recalculate effects on selection: Automatically “clicks” the recalculate button every time the selected model is changed.
Preselect terms using Bonferroni correction: Automatically selects the obvious model effects. Terms are selected using forward selection with a Bonferroni adjusted alpha in. If no effects are selected automatically, they are simply not big enough to pass the very conservative criteria. Manually select any terms you believe to be part of the model. The automatic select is too conservative to be relied upon, but it gets the ball rolling.
Use normal effects for 2-level: There are two methods to calculate the effect selection plot so for two-level factorial designs. The default standardized effects, and the more general normal effects. Using Normal effects is not recommended.
Separated order statistics for pure error: Checking this box will cause the pure error estimates to be plotted with their own normal distribution separate from the unselected model terms. Our default is to have this off.
Show t-limits: The t-limit and Bonferroni adjusted t-limits can appear on the chart or not.
Maximum ticks: Select how many terms to display across the bottom of the chart. Useful when the number of terms is large.
Click the Defaults button to return to the original installed settings for this screen.