TS 13217

TS 13217 is a standard practice for dealing with outlying observations in samples and for testing their statistical significance. In data analysis, an outlier is a result that departs markedly from the overall pattern of the sample. Such values may arise from measurement error, data mix-ups, unusual process conditions, or a genuinely rare but valid variation. Without a disciplined approach, outliers can distort averages, spread measures, and any decision based on the data.

Statistical handling of suspect results

The subject of this standard is the statistical basis for deciding whether one observation, or a small set of observations, should be regarded as outlying. A key point in general statistical practice is that not every unusual value is wrong, and not every unexpected result should be discarded automatically. Significance testing helps judge whether a value is consistent with the rest of the sample or is unlikely to have occurred as part of the same data pattern. This is especially relevant in laboratory measurements, quality control, and the interpretation of experimental results.

Key concepts in general use

• Outlying observation: A single value, or a few values, that lie noticeably far from the general behaviour of the sample.
• Statistical significance: The idea that a deviation is unlikely to be explained by random variation alone.
• Sample data: Measurements taken from a sample intended to represent a larger population.
• Decision support: A structured way to judge whether suspect results should be retained, verified, or rejected.