Figure 3.

Variation of mean strain differences across the four replicate experiments in the conventional (red) and the mini-experiment design (blue). Shown are the four outcome measures, which yielded a significant ‘strain-by-replicate experiment’-interaction effect (i.e. impaired reproducibility) in one of the experimental designs. Notably, we found significant ‘strain-by-replicate experiment’-interaction effects only in the conventional design. Shown are mean strain differences in (a) ‘faecal corticosterone metabolite concentrations’ in comparison 1 ‘C57BL/6J–DBA/2N’. (b) ‘faecal corticosterone metabolite concentrations’ in comparison 4 ‘DBA/2N–BALB/cN’. (c) ‘proportion of 'active' home cage behaviour observations’ in comparison 5 ‘DBA/2N–B6D2F1N’ and (d) ‘Open Field centre time’ in comparison 6 ‘BALB/cN–B6D2F1N’. The black dashed line and the shaded area indicate the overall mean strain difference of this parameter and its corresponding 95% confidence interval (CI₉₅). The black solid line reflects a null effect. Dots and vertical dashed lines reflect the mean strain differences and corresponding CI₉₅ of the four replicate experiments in each design.