1. Home
  2. Power for REML

Power for REML

This menu allows you to perform a power analysis for a fixed term or contrast in a REML analysis using the VPOWER procedure.

Fixed term to calculate the power for

The dropdown list contains the fixed terms in the REML model. Select the term you want to calculate the power for.

Display

This controls what output is printed from the analysis:

Power The estimated power
Number of failures The number of bootstrap samples whose analysis failed to converge
Monitoring Monitoring information, showing the progress of the bootstrap sampling

Type of test

Specifies the type of test. You can select the following types:

F ratio Use the F ratio with an F test to assess the significance of the term
Wald Use the Wald statistic with a chi-square test to assess the significance of the term
Two sided Use a contrast with a two-sided test to assess if the term differs from zero
Greater than Use a contrast with a one-sided test to assess if the term is greater than zero
Less than Use a contrast with a one-sided test to assess if the term is less than zero
Equivalence Use a contrast with a one-sided test to assess if the term is not significantly different from zero. For example, if the treatments have a difference c, the null hypothesis that the treatments are not equivalent is that either: c ≤ – size of difference or c) ≥ – size of difference
Non-inferiority Use a contrast with a one-sided test to assess if the term does not differ significantly from zero. For example, if the treatments have a difference c, to demonstrate non-inferiority of treatment 1 compared to treatment 2, the null hypothesis becomes: c ≥ -size of difference which represents a simple one-sided t-test

Note that apart from types F ratio and Wald, you need to specify the contrast between treatment means that you are examining with the Define Contrast dialog.

Significance level (alpha)%

Significance level as a percentage (0-100) at which the response is required to be detected; default 5%.

Size of the difference to be detected

Size of the difference between the treatment effects that is to be detected.

Specific critical value

If this is selected, then the value provided in the field will be used as the critical value for the test statistic (F ratio, Wald or t). If this is not specified the critical value will obtained from the usual F, chi-square or t-distribution, according to the type of test and estimated degrees of freedom.

Number of samples

This option specifies the number of bootstrap samples that are performed, default 99 (as well as the “null” sample where the data keep their original values).

Number of retries

This option specifies the number of extra samples that are performed when the REML model does not converge for a bootstrap sample, default 99.

Seed

Specifies the seed for the random number generator used to make the bootstrap samples; default 0 continues from the previous generation or (if none) initializes the seed automatically.

Save

Use this option to save the calculated power. After selecting the Power box, you need to type the identifier of a scalar into the In: field.

See also

Updated on April 1, 2019

Was this article helpful?