+ Reply to Thread
Results 1 to 4 of 4

Thread: Sample size in equivalence trial?

  1. #1
    Points: 9, Level: 1
    Level completed: 17%, Points required for next Level: 41

    Posts
    2
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Sample size in equivalence trial?




    Hey all,

    For my thesis I need to calculate the sample size for an equivalence trial with four groups (the score on four equivalent forms in order to prove they are indeed equivalent). The study is done before with only one of the four forms, they found a mean = 19.8, SD = 2.0, tolerance = 1SD. I use an ANOVA to see whether they're significantly different.

    Unfortunately, all I can find is programs for two groups or formulas I don't understand like this http://www.mddionline.com/article/de...ng-equivalence and this http://www.biostat.uzh.ch/teaching/m...quivalence.pdf (slide 17)..

    For example the following:

    n=2((Z1-a+Z1-b/2)^2)/((Delta-SD)^2)

    This doesn't help me a lot, as I have no clue how to get Z(1-a)..

    Anyone who could help me please?
    Thanks a lot!!

  2. #2
    TS Contributor
    Points: 12,501, Level: 73
    Level completed: 13%, Points required for next Level: 349
    rogojel's Avatar
    Location
    I work in Europe, live in Hungary
    Posts
    1,491
    Thanks
    162
    Thanked 334 Times in 314 Posts

    Re: Sample size in equivalence trial?

    hi,
    if you use R you might want to look at the daewr package which has a function FPower1 for sample sizes of one way ANOVA. Other SWs will have their own variants.

    regards

  3. #3
    TS Contributor
    Points: 12,501, Level: 73
    Level completed: 13%, Points required for next Level: 349
    rogojel's Avatar
    Location
    I work in Europe, live in Hungary
    Posts
    1,491
    Thanks
    162
    Thanked 334 Times in 314 Posts

    Re: Sample size in equivalence trial?

    I ran the function and assuming that the tolerance = 1SD is the smallest difference between the average of two groups where you will call them different the results look like this:

    Code: 
    > power=Fpower1(alpha=0.05, nlev=4, nreps=2:25, Delta=2, sigma=2)
    > power
          alpha nlev nreps Delta sigma      power
     [1,]  0.05    4     2     2     2 0.07785485
     [2,]  0.05    4     3     2     2 0.11297917
     [3,]  0.05    4     4     2     2 0.15072795
     [4,]  0.05    4     5     2     2 0.19041991
     [5,]  0.05    4     6     2     2 0.23147661
     [6,]  0.05    4     7     2     2 0.27335307
     [7,]  0.05    4     8     2     2 0.31554916
     [8,]  0.05    4     9     2     2 0.35761767
     [9,]  0.05    4    10     2     2 0.39916772
    [10,]  0.05    4    11     2     2 0.43986506
    [11,]  0.05    4    12     2     2 0.47943046
    [12,]  0.05    4    13     2     2 0.51763677
    [13,]  0.05    4    14     2     2 0.55430516
    [14,]  0.05    4    15     2     2 0.58930083
    [15,]  0.05    4    16     2     2 0.62252853
    [16,]  0.05    4    17     2     2 0.65392793
    [17,]  0.05    4    18     2     2 0.68346914
    [18,]  0.05    4    19     2     2 0.71114838
    [19,]  0.05    4    20     2     2 0.73698386
    [20,]  0.05    4    21     2     2 0.76101206
    [21,]  0.05    4    22     2     2 0.78328422
    [22,]  0.05    4    23     2     2 0.80386327
    [23,]  0.05    4    24     2     2 0.82282102
    [24,]  0.05    4    25     2     2 0.84023576
    So, you would need about 20-25 measurements per group to be reasonably sure you did not miss an effect that is important.

    I hope this helps

  4. #4
    Points: 9, Level: 1
    Level completed: 17%, Points required for next Level: 41

    Posts
    2
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Sample size in equivalence trial?


    Thanks a lot, this is great! Just what I was looking for!

+ Reply to Thread

           




Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats