Preliminary norms are usually established in the same way as regular norms, but before there is enough data in terms of (1) test submissions and (2) usable known scores on other tests by those candidates. The amount of data that constitutes "enough" for a particular test has been raised over time as more data in total became available.