Researchers can attempt to statistically correct for non-response bias, but this method relies on making assumptions about the individuals who did not participate. For instance, a correction might assume that non-responders are demographically or behaviorally similar to late responders. Because these underlying assumptions may be incorrect, statistical correction is generally considered less reliable than proactively minimizing the number of non-responders.