13 14 The study data were analysed using a non-parametric bootstrapping technique.
Differences in means and proportions of baseline sample characteristics were analysed using t-test with bootstrapped standard errors (4,000 replications) and χ-test, respectively.
Costs were analysed using random coefficient models and bootstrap techniques.
To estimate the level of support, 1,000 bootstrap replicates were analysed using RaXML [ 99].
In such circumstances economic data have been analysed using methods such as the nonparametric bootstrap [ 5], first proposed by Efron [ 6].
If available, plasma samples from multiple visits, considered as experimentally "bootstrapped" samples, of the same SJIA patient at different disease states were also collected for confirmatory analyses using bootstrapping.
Nodal support for the MP and ML analyses was assessed using bootstrap analyses with 1000 replications.
Support for the parsimony analyses was generated using bootstrap analysis (BS) with 1000 bootstrap pseudo-replicates [ 18].
Clade support was assessed using bootstrap analyses and Bremer support analyses as described above.
The parsimony trees were generated using bootstrap analyses with 10,000 replicates.
