In mechanical characterization methods, the mechanical properties of multiple material test specimens are measured to determine the probabilistic characteristics of the material's properties through statistical inference. Several of these methods require the measurement of deformations, and to do so, they rely on local strain measuring techniques, such as bonded strain gages and extensometers. In this study, we show that for non-crimp fabric laminates, local strain measurements acting as proxies of global laminate strain contain a random strain measurement error. Furthermore, we demonstrate that this strain measurement error can significantly reduce the accuracy of characterization methodologies for non-crimp fabric laminates. The strain measurement error pollutes the mechanical property measurements on laminate test specimens, leading to inaccurate statistical inferences. Because the strain measurement error is random, the inferences regarding the mechanical properties may occasionally be conservative or non-conservative with respect to the inference that would have been made if there was no strain measurement error. The results presented in this study are of importance because over-conservative mechanical properties can lead to unnecessarily heavy structures, and non-conservative ones may lead to unsafe structures, endangering life property and the environment. Both scenarios were discussed along with their likelihood and possible consequences.