Re: Conflicting Claims : : The claims of two papers, which are published in the May edition of NDTnet, conflict. The paper entitled "The Selection of an NDT Method..." (http://www.ndt.net/article/v04n05/rihar/rihar.htm) claims that "The selection of an NDT method is usually the concern of NDT laboratories, which know the NDT methods and their feasibility in detecting defects" but the paper entitled "Erratic Measure" (http://www.ndt.net/article/v04n05/oldberg/oldberg.htm) claims that all of the various methods for detecting defects violate an axiom of probability theory.
: : If the claim of the latter paper is correct, a caveat must be placed the claim of the former paper. The caveat is that the ability of NDT laboratories to know the feasibility of detecting defects is severely restricted.
: thanks for bringing up this important subject.
: First of all, I would like to point to forum messages of Feb 99 in which
: the paper entitled "Erratic Measure" was discussed intensively between you and Robert A. Day.
: Especially new readers may brows this thread at
: http://www.ndt.net/wshop/forum/messages/642.htm .
: Thanks to the ASME copyright agreement we are pleased having the full-text in this May issue available,
: and are able to discuss this paper much better now.
: For my opinion your concerns seems being very theoretical.
: If NDT should not be used at all, what practically alternative do we have?
: You mentioned "restricted knowledge of NDT laboratories",
: this is life as usual we have to deal with.
: That does not mean we should accept this weakness,
: moreover we must use control systems that help to prevent any error
: that may rise by unprofessional usage of NDT methods.
Your posting raises some important issues; a portion of these issues are ethical rather than scientific. I propose discussing the essentially ethical issue of what should be done in light of thefindings of "Erratic Measure" in a separate thread, to be entitled "Ethics and Verifiability." The thread entitled "Conflicting Claims," then, would be restricted to the essentially scientific issues of 1) is the Unit Measure axiom of probability theory is violated by NDT and 2) how NDT can be reconfigured to preserve Unit Measure.
Your posting raises issue 2) and touches on issue 1). I respond on issue 2) that NDT can be reconfigured to preserve Unit Measure and thus to be suseptible to statistical validation. However, this will require changes in some of NDT's tests. In particular, it will require changes in NDT's defect detection methods.
In order for it to preserve Unit Measure, it is sufficient for a test to meet all of the following requirements:
o Testing generates a sequence of events, each of which is either a positive or a negative.
o The set of such events relates one-to-one to a set of objects; this set is a partition of the material which is inspected. The elements of this set are called "sampling units" in statistics.
o A definitive test is identified, by which each positive or negative may be determined to be true or false. This test is performed on a randomized sample of the sampling units. The data facilitate estimation of the probabilities of a true positive, false positive, true negative and false negative.
In referring readers to my dialogue with Robert A. Day, you have touched on issue 1), for he objects to the argument presented by "Erratic Measure" on the grounds that he does not understand it. Without details on his state of mind, it is difficult for me to repond, other than to speculate that his need is for material which is tutorial in the concepts surrounding the statistical validation of testing. The authors of "Erratic Measure" could not provide much such material while remaining the 6 page limit which was imposed upon them. However, we have been able to provide a good deal more tutorial material in presenting the paper orally. Our audiences for these oralpresentations have seemed to leave with an understanding of the paper. Perhaps this result can be achieved for a wider audience through the use of your forum.
Finally, I wish to address your comment that "...your concerns seems being very theoretical." It would be more accurate to describe the concern of "Erratic Measure" as "scientific" rather than "theoretical." At issue is whether theories of the reliability of NDT which are expressed in terms of probabilities are empirically verifiable over the range of the associated sample space. If they are not empirically verifiable, these theories lie outside science. "Erratic Measure" claims to demonstrate that they are not empirically verifiable.