General QR decomposition of the observation matrix is used to solve identification functions to evaluate identification results of every parameter in parallel mechanism calibrations. A relationship between measured information and identification results is obtained by analyzing numerous matrix transforms and QR decompositions. When distributions of measurement error are determined, random distributions of iden- tification result disturbances (IRDs) can be obtained from this relationship as a function of measurement er- rors. Then the ranges of the IRDs can be effectively estimated, even if true parameter values are unknown. An optimization index based on IRD estimate is presented to select measurement configurations to achieve smaller IRDs. Two simulation examples were carried out with different modes and calibration methods. The results show that the method is effective and that the optimization index is useful. Some regular parameter identification problems can be explained by the IRD estimates. General QR decomposition of the observation matrix is ​​used to solve identification functions to evaluate identification results of every parameter in parallel mechanism calibrations. A relationship between measured information and identification results is obtained by analyzing numerous matrix transforms and QR decompositions. When distributions of measurement error are determined, random distributions of iden- tification result disturbances (IRDs) can be obtained from this relationship as a function of measurement er- rors. Then the ranges of the IRDs can be effectively estimated, even if true parameter values ​​are unknown. based on IRD estimate is presented to select measurement configurations to achieve smaller IRDs. Two results are shown that the method is effective and that the optimization index is useful. be explained by the IR D estimates.