Step function technique for estimating data uncertainty in experimental results

✍ Scribed by M. J. Miah

• Publisher: John Wiley and Sons
• Year: 1991
• Tongue: English
• Weight: 574 KB
• Volume: 5
• Category: Article
• ISSN: 0886-9383
• DOI: 10.1002/cem.1180050308

✦ Synopsis

An additive model is used to express the observed value of a sample characteristic as the sum of the true sample characteristic and a value of the data collection error, commonly known as experimental error.

The data uncertainty of the experimental results (or of a survey data set) is defined as the expected squared error. The expected squared error may change with the sample characteristic, e.g. the error moment could be concentration-dependent. The relationship between the error variance and the analyte concentration may not be very distinct. In such a case the data transformation to stabilize the error moments may not be appropriate. A step function is proposed as an alternative way to represent the second moment of the error. The data uncertainty is defined as the weighted average of the step values of the second raw moment of the error, using the appropriate proportions of the routine samples as weights. The data uncertainties associated with the different data collection stages were evaluated by using regional soil survey data.