Bias (systematic error)

Bias is a systematic or non-random inaccuracy and is one of the most troublesome difficulties in using numerical data. Biases may be associated with incorrectly calibrated instruments, e.g. a faulty electrode or syringe, or with experimental manipulations, e.g. decomposition of chemical compound on storage. Bias in measurement can also be subjective, or personal, e.g. an experimenter's preconceived ideas about an 'expected' result.

Bias can be minimized by using a carefully standardized procedure, with fully calibrated instruments. Investigate bias in 'trial runs' by measuring a variable in several different ways, to see if the same result is obtained.

To avoid personal bias, 'blind' measurements should be made where the identity of each sample is unknown to the operator, e.g. use a coding system.