Non Parametric Statisticas Methods
Interested parties are invited to join in the development of a new proposed standard being developed by ASTM International Committee E11 on Quality and Statistics. WK32565, Guide for the Use of Non-Parametric Statistical Methods, is under the jurisdiction of Subcommittee E11.10 on Sampling/Statistics.
According to Stephen Luko, statistician, Hamilton Sundstrand, Shared Engineering Services and an E11 member, non-parametric techniques are used by statisticians in many situations, particularly where the nature of the distribution from which the sample arose is uncertain.
"The methods described in the proposed standard do not depend on underlying distribution and are therefore attractive alternatives to the usually normal distribution based methods," says Luko. "Other situations occur in which one needs to forecast or predict where the next observation will fall, using a small sample that may also be covered by non parametric techniques. These techniques provide yet another set of tools for special situations that may arise in the broader community of users of ASTM standards."
Luko says non parametric methods are particularly useful to anyone using data analysis tools, particularly in the following common situations::
Luko says the task group developing WK32565 is particularly interested in user applications involving smaller sample sizes and how this situation is currently handled.
CONTACT Technical Information: Stephen N. Luko, Hamilton Sundstrand, Shared Engineering Services • Windsor Locks, Conn. • Phone: 860-654-3447 O ASTM Staff: Christine Sierk • Phone: 610-832-9728 O Upcoming Meeting: Oct. 26-28 • ASTM Headquarters, West Conshohocken, Pa.