Statistical analysis of experimental data

Fluorescence measurements from different donors and experimental repetitions formed the data samples that we analyzed using statistical methods. The samples were tested for normality using the Jarque–Bera test (Jarque and Bera 1987). Normal samples were compared using the unpaired t test, and non-normal samples were compared using the rank-sum test (Glantz 2005). In cases of multiple comparisons, we used the Bonferroni correction for P value adjustment (Glantz 2005). P ≤ 0.05 was considered statistically significant. To fully utilize the available data in our sample comparisons between experimental conditions, we pooled data points from different donors and different repetitions—corresponding to a single experimental condition—into a single data sample. The main general issue with data pooling is the possibility of non-independence for distinct data points (Jenkins 2002). However, in our case, distinct clotting events could be regarded as statistically independent, because, due to our experimental design, the occurrence of a clotting event did not affect the probabilities of other such events.