The relation between expected life estimation and amount of data

Small data sets from e.g. limited tests yield (far) too optimistic impressions of product reliability. This paper explains the background and illustrates this effect numerically.