Morphometric parameters are porosity, osteon density and osteon diameters. The damage parameters crack density (#1mm2), crack length (mm/mm2 ), and crack angle.
The primary goal of the statistical analyses of the damage measured per region was to compare specific damage amongst to the three holes types (foramen, drilled hole, or control hole). All statistical analyses were performed with commercially available software (StatView; SAS Institute; Cary, North Carolina). The confounding effects were proximity to each hole (near or far) and applied strain level (5,000 tpg or 8,000 [tpg). The types of physical damage, the hole type, and the two confounding effects were accounted for by performing a full interaction analysis of variance (ANOVA). If no significance resulted in the interaction, the strain level factor was eliminated from the model and the ANOVA repeated with only the hole type-proximity interaction active. Consequently, if no significance resulted, then the proximity factor was eliminated, leaving only the hole type factor. At any stage of the analysis, if a significant ANOVA resulted, Fisher's protected least significant difference tests were performed to determine if significant individual comparisons existed. For all analyses, significance was assumed to exist for P < 0.05.
To perform statistical analyses on the cracks orientation, another database was constructed in which the experimental unit was each individual crack. Angles were measured for black wispy mesocracks (BWM), diffused stained mesocracks (DSM), and bundles of microcracks (BM). All angles were measured with respect to the neutral axis
(NA) in each section; four levels were chosen, 0', 30', 60' and 90'. By limiting the alternative of angles to only four possible outcomes, the response variable (angle), becomes a nominal response variable. Statistical analyses on the orientation of the linear