Recalibrating sample size, non-normal distribution
March 21, 2007 at 12:13 pm #23895
I used the formula n=(1.96S/delta)sq. to determine sample size for a data collection effort collecting continuous (time) data, having estimated S as 7 (no historical data).
As the data came in it soon became obvious that the distribution was not normal, but positively skewed; I fitted a Weibull distribution model with shape 0.83, scale 6.64.
How do I ‘recalibrate’ my estimate of sample size, it seems I don’t need anything like as many samples as I thought, looking at the histogram, but there is now no standard deviation figure from this non-normal distribution to plug into the original formula I used?
any guidance appreciated.0
The forum ‘Europe’ is closed to new topics and replies.