Effect of re-sampling methods of original data sets on the determination of confidence intervals for the subpopulation size of <i>egfp</i>-expressing cells in stationary phase cultures of <i>P. knackmussii</i> B13 strain 2399 (single copy P_inR<i>-egfp</i>) grown on 3CBA.

<p>(A) Bootstrapping of original data sets (re-sampling with replacement). Methods <i>Default</i> and <i>Manual</i> were repeated 20 times with manual intervention of the slope line determination. Methods <i>Boxplot1.5</i> and <i>Boxplot3</i> use 500 automatically re-sampled data sets. 95% confidence intervals (red, dotted lines) were calculated assuming a normal distribution of the results (mean±SD×1.96). (B) same data as in (A) but re-sampled subpopulation size determinations plotted as Q-Q plots. Note the normal distribution of the results.