Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
D4854-95 – Standard Test Method Technical Guide
The ASTM D4854-95 standard (reapproved 2001) serves as a guide for subcommittees to estimate variability from expected sources in sampling plans that produce variables data. It explains how to quantify contributions from lot sampling units, laboratory sampling units, and specimens to the variation of test results and how to combine these using Analysis of Variance (ANOVA). This guide is applicable to all plans producing variables data regardless of frequency distribution, as noted in Note 1, and is not intended for attribute data which require random specimen selection from all individual items in the lot.
📐 Scope and Significance
This guide provides methods for estimating variability contributions from three primary sources in sampling plans. It covers topics such as sampling plans producing variables data, reducing variability of sampling results, and detailed analysis using ANOVA. Referenced documents include ASTM standards D123 on terminology, D2904 and D4467 on interlaboratory testing for normally and non-normally distributed data, and D4271 on writing sampling statements. The TEX-PAC adjunct software (available through ASTM) can conduct calculations described in the annexes, including cost comparisons of various plans.
🔬 Key Sources of Variability
Understanding variability sources is crucial for reducing overall sampling error. The three identified sources are: lot sampling units, capturing variation between groups within the lot; laboratory sampling units, accounting for variation during subsampling and laboratory preparation; and specimens, reflecting variation from individual test measurements. Accurate estimation of these components is essential for developing robust specifications.
| 🟦 Source | 📏 Description |
|---|---|
| Lot Sampling Units | Variation due to selecting units from the lot |
| Laboratory Sampling Units | Variation from subsampling and laboratory preparation |
| Specimens | Variation from testing individual specimens |
📊 Combining Variability with ANOVA
The Analysis of Variance (ANOVA) is recommended to partition total variance into components of variance for each source. Key concepts include components of variance and degrees of freedom, where degrees of freedom represent the number of independent values in a set. For example, specifying only the average of five observations leaves four degrees of freedom. This partitioning enables estimation of overall sampling plan variability as detailed in Annex A1. A numerical example is provided in Annex A2.
| ⚡ Component | 🎯 Variance Symbol |
|---|---|
| Between Lots | σ²L |
| Within Lot, Between Lab Units | σ²Lab |
| Within Lab Units, Between Specimens | σ²Spec |
By combining these component estimates, subcommittees can evaluate the impact of different sampling plans on test result precision and identify areas for variability reduction.
💡 Tip: Use the ANOVA analysis from Annex A1 to determine which source contributes most to variability, allowing targeted reduction efforts for more cost-effective sampling procedures while maintaining test reliability.
⚠️ Warning: This guide is not applicable to attribute data; it is specifically designed for variables data from staged sampling plans. For attribute data, specimens must be taken at random from all individual items in the lot.
❓ Frequently Asked Questions
🔍 What is the main purpose of ASTM D4854-95?
To estimate variability from expected sources in sampling plans, aiding subcommittees in writing specifications and sampling procedures with quantified variation components.
💡 How are variability estimates from different sources combined?
Using Analysis of Variance (ANOVA) to partition total variance into components associated with lot sampling units, laboratory sampling units, and specimens.
⚡ What types of sampling plans does this guide cover?
All plans that produce variables data, regardless of frequency distribution. It does not cover attribute data plans.
📌 What are the three sources of variability considered?
Lot sampling units, laboratory sampling units, and specimens, which are combined to estimate overall sampling plan variability.