Home / DOE / Analyze / Evaluate Design / Options
Evaluate Design Options¶
Every control in the Evaluate Design dialog. There is one page and no tabs.
The dialog opens whenever the command is run, even in a workbook that holds a single design sheet, because the power values still have to be supplied.
The header reads:
Calculates the variance inflation factors, the alias matrix, the column correlations, the leverage of each run, the power of each model term, and the D, A and G efficiency measures of the selected design sheet, and writes them to a new worksheet.
Select the design sheet to analyze¶
A list of the design sheets in the active workbook. The active sheet is first, then the rest in workbook order, and the first entry is selected when the dialog opens.
One sheet must be selected before OK is live:
Pick the design sheet to evaluate.
In a workbook with no design sheet the dialog does not open at all. A message appears instead:
There are no design sheets in the active workbook.
Power Analysis¶
Include power analysis¶
A check box, ticked when the dialog opens.
Clearing it greys the three boxes below and their labels, stops them being checked, and leaves the two power blocks off the worksheet altogether. A value in a greyed box cannot hold the dialog shut, so an entry that was refused stops blocking OK the moment the analysis is switched off.
Under the section heading:
Power is the probability that this design will correctly flag a model term as statistically significant when the term's true coefficient is as large as the value you enter in Anticipated coefficient below.
Anticipated coefficient¶
Default 1. Any number other than zero, positive or negative; only its size enters the power.
Enter the anticipated coefficient, which for coded linear effects is half of the term's effect, the average change in the response from a factor's low setting to its high setting.
The anticipated coefficient must be a number other than zero.
Standard deviation¶
Default 1. Must be greater than zero.
This is the expected run-to-run standard deviation of the response itself, not of any input factor. Get it from repeated runs at the same settings, from historical process data, or from a pilot study.
The standard deviation must be a number greater than zero.
Significance level¶
Default 0.05. Must be strictly between 0 and 1; neither endpoint is accepted.
A significance level of 0.05 is most common, but adjust it to your acceptable level of risk.
The significance level must be a number between 0 and 1, excluding both.
How the three boxes are checked¶
They are checked in the order they appear, and the first refusal is the one stated. A box holding text that cannot be used leaves the last accepted values standing, so the report is always computed from a stated set of numbers.
The three appear on the worksheet as the first three rows of the power block, written exactly as they were typed, so the table says what it was computed from.
The footer¶
Left to right as they are drawn:
| Button | What it does |
|---|---|
| Help | opens this tool's help |
| OK | evaluates the selected design and writes the worksheet. Greyed while a sheet is not selected, or while the power analysis is included and one of its three boxes is refused |
| Cancel | closes the dialog without evaluating |
Messages after OK¶
The dialog closes first, then the worksheet is written, so a problem is reported after it has gone.
The design sheet "name" could not be read: reason
The design sheet "name" could not be used as a design: reason
The design sheet "name" could not be measured: reason
The worksheet could not be written: reason
The worksheet could not be written.