If the model is run for long enough (or better still, the experiment is repeated often enough), the effects of chance fluctuations average out. (usask.ca)
The risks have been «smoothed» to adjust for chance fluctuations that can occur when there are small numbers of cases especially for cancers such as leukemia and brain cancer. (sciencedaily.com)
When these influences are combined with chance fluctuations due to sampling error — inherent in all surveys, regardless of how carefully they are designed — it is often impossible to tell exactly why surveys yield different results. (educationnext.org)