Table 1 |
||||
|
Sample SD based on a pilot study with n = 50 experiments in 5 omission proportions for a desired confidence α ≥ 95% and a desired Error E ≤ 1%. |
||||
|
Omission Proportion |
SD |
A*2 |
Required Experiments |
KS test p-value |
|
|
||||
|
5% |
0.140 |
1.09 |
749 |
0.068 |
|
10% |
0.096 |
0.565 |
352 |
|
|
15% |
0.081 |
0.733 |
252 |
|
|
20% |
0.070 |
0.505 |
187 |
|
|
25% |
0.054 |
0.300 |
113 |
|
|
30% |
0.052 |
0.419 |
103 |
|
|
35% |
0.044 |
0.716 |
75 |
|
|
40% |
0.044 |
0.360 |
74 |
|
|
45% |
0.041 |
0.438 |
66 |
|
|
50% |
0.039 |
0.150 |
59 |
|
|
|
||||
|
The Anderson-Darling values, adjusted for sample size (A*2) greater than 0.752 rejects the null-hypothesis that sample is normally distributed. For the 5% omission proportion, a truncated normal distribution was tested using the KS test which gave a p-value > 5%. |
||||
|
Tsafnat et al. BMC Bioinformatics 2009 10:281 doi:10.1186/1471-2105-10-281 |
||||
