2.3.2 Search for data errors

A data error in a statistical context means that the sample (measured quantity) does not belong to the parent population for this value with the given distribution. Therefore, the estimation of the unknowns of the block cannot be unbiased.

Here we are considering gross data errors, large and small. Accidental errors and formal errors of data input, e.g. a missing point name are not considered.

BINGO supports gross-error detection in three ways. Internal test values are computed for all three methods which mark doubtful observations in the output list by one to four stars ("*" to "****").

Observations with four stars are the most questionable and will therefore be investigated first for gross errors.