s-news
[Top] [All Lists]

Bug in s+seqtrial ?

To: <s-news@lists.biostat.wustl.edu>
Subject: Bug in s+seqtrial ?
From: Abdulkadir Hussein <hussein@math.ualberta.ca>
Date: Sun, 25 Aug 2002 14:36:19 -0600 (MDT)
Hi, to whoever is using s+seqtrial, the following code gives wrong
scientific representation for the power (supposed to be "e-1"):

seqDesign(prob.model = "normal", arms = 2, log.transform = F,
        null.hypothesis = c(105, 105), alt.hypothesis = 95, variance
         = c(324, 324), ratio = c(1, 1), nbr.analyses = 1, sample.size
         = 138, test.type = "two.sided", power = "calculate", alpha =
        0.025, epsilon = c(1, 1), design.family = seqScale(scaleType
         = "X"), P = c(1, 1, 1, 1), R = c(0, 0, 0, 0), A = c(0, 0, 0,
        0), display.scale = seqScale(scaleType = "X"))

PROBABILITY MODEL and HYPOTHESES:
   Two arm study of normally distributed response variable
   Theta is difference in means (Treatment - Comparison)
   Two-sided hypothesis test:
           Null hypothesis : Theta =   0    (size  = 0.05    )
    Alternative hypothesis : Theta < -10    (power = 8.797e-8)
   (Fixed sample test)

STOPPING BOUNDARIES: Sample Mean scale
                          a       b      c      d
    Time 1 (N= 138) -6.0064 -6.0064 6.0064 6.0064




Regards


<Prev in Thread] Current Thread [Next in Thread>