Tensile capacity of finger joint connections - Considering censored data

When testing finger joint connections (FJ) in tension only a limited number of the tested specimens fail in the FJ. If the failure occurs outside the FJ the actual tensile strength of the FJ is known to be equal or larger to the maximal stresses observed in the experimental investigation; so-called censored data. Based on an experimental investigations were the tensile strength of 40 FJ were investigated the influence of considering censored data in the analysis is discussed. The investigation is performed on two sub-samples. One sub-sample with large variation of the test results and one sub-sample with relative small variations. For both sub-samples a significant error due to the wrong consideration of censored data is observed. The tensile strength of FJ is underestimated when neglecting censored test results or considering them like non-censored data. In general also the variability is underestimated. For one sub-sample the reduced variability even results in an overestimation of the 5%-value. The present paper highlights the importance of considering censored information when performing data analysis.