Statistical Boundary
Quality control thresholds calculated without assuming a specific underlying distribution for the data define a reliable method for evaluating textile properties. These non-parametric tolerance limits are particularly useful when the data is skewed or contains outliers that would invalidate normal distribution statistics. A mill uses this approach to set realistic pass or fail criteria for irregular properties like tear strength or pilling resistance.
Sample Size
Order statistics and the binomial distribution provide the framework for determining these boundaries. Because non-parametric tolerance limits rely on the actual rank of the data points, they generally require larger sample sizes to achieve the same confidence level as parametric methods. This ensures that the resulting interval covers a minimum proportion of the population with high certainty.
Application Area
Raw fiber testing often yields data sets with heavy tails or multiple peaks. In such cases, non-parametric tolerance limits prevent the manufacturer from setting unrealistic targets based on a false assumption of symmetry. This method protects the producer from excessive rejections while maintaining a reliable standard for the buyer.
Reliability Measure
Calculation of these limits involves identifying the extreme values within a set of observations. Using non-parametric tolerance limits allows for a more defensible assessment of product consistency in complex textile systems.