Log in
RegisterPowerTOST_0.9-11 on the way [Design Issues]
Dear Ben,
Shit happens. It is a bug
.
In the functions power.TOST(), power2.TOST() and power.noninf() the robust degrees of freedom are used although the argument robust=FALSE by default. Setting (falsely !) robust=TRUE will give the correct answer using the non-robust df's:
This bug affects all higher order crossover designs. But not the sample size estimation.
Sorry for any inconvenience. A bug fix version is on the way via CRAN.
❝ Side note and question regarding PowerTOST:
❝ ...
❝ Is that a bug in PowerTOST?
Shit happens. It is a bug
.In the functions power.TOST(), power2.TOST() and power.noninf() the robust degrees of freedom are used although the argument robust=FALSE by default. Setting (falsely !) robust=TRUE will give the correct answer using the non-robust df's:
power.TOST(CV=0.2, design="3x3", n=18, robust=T)
[1] 0.8089486
power.TOST(CV=0.2, design="3x6x3", n=18, robust=T)
[1] 0.8089486This bug affects all higher order crossover designs. But not the sample size estimation.
Sorry for any inconvenience. A bug fix version is on the way via CRAN.
—
Regards,
Detlew
Regards,
Detlew
Complete thread:
- 3x3x3 vs 3x6x3 reloaded Ben 2012-08-04 13:46 [Design Issues]
![[Mix]](https://static.bebac.at/img/mix.png "open in Mix view")
- PowerTOST_0.9-11 on the wayd_labes 2012-08-07 09:13
- Carry over issues d_labes 2012-08-07 19:14
- Carry over issues ElMaestro 2012-08-08 01:02
- Carry over issues d_labes 2012-08-08 07:59
- Carry over issues Ben 2012-08-09 19:22
- Carry over issues ElMaestro 2012-08-08 01:02
- Correction Ben 2012-08-19 12:54
Ing. Helmut Schütz