For ceramic floor tiles, directional differences in slip resistance were statistically detectable under dry and soapy conditions but practically negligible, below two percent of the mean coefficient of friction (COF), and they did not improve any predictive model. Slip resistance is a primary engineering control against slip and fall injuries, and measurement standards assess it in several directions but average the readings, implicitly assuming that the surface behaves uniformly with direction. This study tests that assumption on seven commercial tiles for which the COF and six profile-roughness parameters were measured in two orthogonal directions. Longitudinal and transverse results were compared with paired tests and a nested mixed-effects model, and direction-aware and direction-agnostic models spanning k-nearest neighbors, random forest, support vector regression, gradient boosting, and XGBoost were trained under interpolation and leave-one-tile-out protocols. Surface roughness was statistically indistinguishable between directions; the directional effect on COF did not increase with roughness and varied with the shoe sole; and measurement direction ranked last among all predictors under both permutation and SHAP importance, adding nothing even to the strongest gradient-boosting models. While based on seven tile types, these findings indicate that for conventionally manufactured ceramic tiles the two-direction averaging already prescribed by standards yields a representative slip-resistance value, and that single-direction testing may be sufficient; the latter is a proposed simplification that could roughly halve the number of directional measurements in routine testing but that a larger, multi-batch study should confirm before it replaces current practice.

Keywords

Slip resistance, Ceramic tiles, Directional anisotropy, Coefficient of friction, Machine learning, Pedestrian fall safety