TABLE 4.
Effect of three, six, and more than seven replicates on reproducibility
| Replicate no. tested | Replicate F value (p)a |
|---|---|
| 1, 2, and 3 | 0.71 (0.49) |
| 4, 5, and 6 | 0.19 (0.83) |
| 1, 2, 3, 4, 5, and 6 | 0.39 (0.86) |
| 1, 2, 3, 4, 5, 6, and >7 | 0.19 (1.00) |
| 1-3 vs 4-6 vs ≥7 | 0.07 (0.93) |
| 1-3 vs 4-6 | 0.91 (0.98) |
| 1 and 6 vs 2, 3, and 4 | 1.56 (0.21) |
| 1 and 6 vs 2, 3, and 4 vs >7 | 0.91 (0.40) |
| 1, 2, 3, 4, 5, and 6 vs >7 | 0.06 (0.81) |
a
The F value is the test statistic used in GLM analysis of variance; it is the ratio of the replicate mean square to the replicate error mean square. The null hypothesis was that there was no effect of replicate (1st, 2nd, 3rd, etc.) on measures. This was rejected where the F ratio is large and P is <0.05. The effect of replicate was not significant for the comparisons presented when P was <0.05 (effect significant).