Design-Relevant Strength Inference with Small Samples for Bamboo Composite and Recycled Plastic

Early-stage development of recycled wood–plastic composites (WPCs) often relies on very small specimen sets with imperfect geometry, which limits the usefulness of conventional hypothesis testing for engineering design. This study re-analyses ultra-small bending and compression tests on recycled high-density polyethylene (HDPE) and a 70:30 HDPE: bamboo WPC to produce design-oriented metrics based on effect sizes rather than p-values. Three WPC and one plastic cylindrical rods (14 mm diameter, 360 mm span) were tested in three-point bending under sequential static loads of 100, 400, and 600 g, with mid-span displacement as a proxy for flexural stiffness, while three WPC and three plastic cylinders (14 mm diameter) were tested in axial compression to failure to obtain breaking load and compressive stress. Complete specimen-level results are reported in Tables 1 and 2, with group medians, interquartile ranges, and effect sizes summarized in Tables 3 and 4 and Figs. 1 and 2. The bamboo–HDPE WPC shows consistently lower mid-span displacement than recycled plastic at all loads, with median reductions of about 48%, 58%, and 43% at 100 g, 400 g, and 600 g, respectively, and higher median compressive stress (14.94 MPa vs. 12.34 MPa). The Hodges–Lehmann median difference of 1.95 MPa (95% bootstrap confidence interval 1.30–3.25 MPa) and a large standardized effect size (Hedges’ g ≈ 1.77) illustrate how transparent specimen-level reporting and effect-size-focused summaries can turn tiny, batch-specific tests into cautious yet usable evidence for preliminary sizing and material down-selection.