ISO 3534 Statistical Terminology
ISO 3534 Statistical Terminology
ISO 3534 Statistical Terminology
net/publication/238047645
CITATIONS READS
0 4,408
2 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Stephen N. Luko on 18 July 2016.
Quality Engineering
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/lqen20
To cite this article: Stephen N. Luko & Mark E. Johnson (2012): Statistical Standards and ISO, Part 2—Terminology, Quality
Engineering, 24:2, 346-353
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to
anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should
be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims,
proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in
connection with or arising out of the use of this material.
Quality Engineering, 24:346–353, 2012
Copyright # Taylor & Francis Group, LLC
ISSN: 0898-2112 print=1532-4222 online
DOI: 10.1080/08982112.2012.654437
INTRODUCTION
Terminology and nomenclature are the raison d’être for ISO TC69=SC1.
To decipher this opening sentence, a terminology document covering
French phrases and International Standards acronyms would prove useful,
yielding a definition of raison d’eˆtre (the reason for existence) and a trans-
lation of ISO TC69=SC1 (International Standards Organization Technical
Committee 69 on applications of statistical methods=Subcommittee One).
Terminology standards document the ‘‘official’’ definitions of terms and
symbols to facilitate the production and understanding of the standards that
include them. Because some international standards are noted in contractual
relationships, explicit and precise definitions are essential to facilitate legal
proceedings (avoiding contentious discussions involving terminology).
The creation and development of international statistical terminology
standards follows the same elaborate, multistage approval process as other
statistical standards that was recently reviewed in Part One of this series of
review articles on ISO=TC 69 standards (Boulanger and Johnson 2011). In
brief, standards can require years of development, voting in stages from a
new work item, to sundry committee drafts, draft international standards,
Edited by Stephen N. Luko.
and final draft international standards. Having withstood the consensus-
Address correspondence to Stephen
building and review process, published terminology documents are recog-
N. Luko, 1 Hamilton Road, Hamilton nized as normative references by other statistical standards published under
Sundstrand Corporation, Windsor TC69.
Locks, CT 06096, USA. E-mail:
stephen.luko@hs.utc.com
346
TC 69=SC1 presides over three published termin- to Quality Engineering professionals. Other compre-
ology standards having a common designation of hensive terminology surveys include works by
ISO 3534. The three currently published parts with Dodge, Cox, Commenges and Davison (2006); Ever-
their latest publication dates are itt (2006); Freund and Williams (1991); Marriot
(1990); Tietjen (1986); and Upton and Cook (2008).
. ISO 3534-1:2006—Statistics—Vocabulary and sym- Statisticians in general and statistical standards wri-
bols—Part 1: General statistical terms and terms ters in particular are not naturally lexicographers.
used in probability Familiarity with a multitude of definitions used in stat-
. ISO 3534-2:2006—Statistics—Vocabulary and istics and probability is an asset from a subject matter
symbols—Part 2: Applied statistics perspective but does not automatically translate to
. ISO 3534-3:1999—Statistics—Vocabulary and sym- the capability of producing viable definitions that will
bols—Part 3: Design of experiments pass an international muster. In an analogous sense,
individuals who have completed many surveys are
This series of documents is dynamic in that revi- not necessarily adept at constructing excellent survey
sions occur on an ongoing basis. Revisions to instruments. Fortunately, ISO Technical Committee
Parts 1 and 2 are not yet officially registered work 37 on terminology and other language and content
items, although preliminary planning for these efforts resources has published standards governing termin-
is underway. Part 3 has been revised through mul- ology documents that SC1 has fully implemented.
Downloaded by [Stephen N. Luko] at 10:35 26 March 2012
tiple committee draft stages and is at the second draft These guidelines when followed produce clean defi-
international standard voting stage (closed on Janu- nitions that do not suffer common difficulties. The
ary 2, 2012). Pending accommodation of comments following are such recommendations and warnings:
and corrections, the document could be published
late in 2012. A new fourth standard on Survey Sam- 1. Use concept diagrams to identify the interrelation-
pling to be designated ISO 3534-4 is at the initial ships among terms and to provide a logical frame-
committee draft stage, for which a planned work for all the terms defined in the standard.
publication date in 2013 is projected. 2. Abide by the substitution principle.
This article begins with the development of termin- 3. Restrict one concept per term.
ology and nomenclature documents, following the 4. Avoid definitions in the negative.
advice of ISO TC37, which is charged with standar- 5. Use singular form for terms that are nouns.
dizing terminology standards. TC37 champions the 6. Avoid peripheral qualifiers not critical to the
use of concept diagrams that can delineate the inter- definition.
relationship of terms and concepts, with an emphasis 7. Avoid circular definitions.
on subordinate relationships. The next three sections
concern the three published ISO 3534 parts in turn. An example of a concept diagram is given in Figure 1,
The final technical section deals with the aforemen- adapted from a more complicated one in the
tioned survey sampling standard under development proposed ISO-3534-4. The term error of estimation
and describes other SC1 collaborations within TC69 is partitioned into two subordinate terms and
and across other ISO technical committees.
TERMINOLOGY AS A TECHNICAL
DISCIPLINE
Many excellent sources for terminology in Prob-
ability and Statistics exist. For example ASTM’s
E456 is a terminology standard developed using a
similar process to ISO standards. The general ASTM
overview article was published in this journal during
2010 (see Ulman and Luko). ASQ’s Glossary and
Tables for Statistical Quality Control is more targeted FIGURE 1 Sample concept diagram.
plot of the response variable versus a predictor variable so care must also be given to the notes and examples
with identified settings of a second predictor variable is so that they can be understood in multiple cultures.
a poor definition. Likewise, circular definitions are not Bellos (2011) described many challenges in translat-
helpful (e.g., random sample is a sample that is ran- ing less technical material as found in the examples.
dom). Circular definitions could be avoided by employ- In terms of purely technical difficulties, the various
ing well-understood nontechnical words in some cases. published dictionaries do not provide panaceas for
Implicit in this discussion is that any existing pub- the standards writers. For example, bias is defined
lished statistical dictionaries are not themselves suit- sometimes in the context of estimator bias and some-
able to serve as normative references on terminology times relative to slanted designs. Incorrect definitions
acceptance sampling, and bulk sampling are exten- definition references the term being used back to
sively covered. This standard has been updated mul- its definition.
tiple times on a roughly 5- to 10-year cycle. There are
so many terms included that this standard is . 1.2.1 population
approaching the need to be split into more manage- (reference) totality of items (1.2.11) under con-
able documents. TC 37 recommends a range of sideration
50–200 concepts in a single terminology document. NOTE 1: A population can be real and finite or
The organization of Part 2 is given in the following hypothetical and infinite.
outline: NOTE 2: Extended sampling (1.3.1) from a finite
real population can give rise to the generation
1. Data generation and collection of actual relative frequencies or frequency dis-
1.1 Systems of reference values for characteristics tributions (2.5.1). Alternatively, or arising from
1.2 Sources of data this, a theoretical model of the hypothetical
1.3 Types of sampling population based on probability distributions
2. Statistical process management can be derived. This enables predictions to
2.1 General process-related concepts be made.
2.2 Variation related concepts NOTE 3: A population can be the result of an
2.3 Control-related charts ongoing process that may include future out-
2.4 Control chart components put.
2.5 Fundamental terms related to process perfor- NOTE 4: A population can consist of distinguish-
mance and process capability able objects or bulk material.
2.6 Process performance hmeasured datai . 3.3.2 bias
2.7 Process capability hmeasured datai
3. Specifications, values, and test results difference between the expectation of a test result
3.1 Specification-related concepts (3.4.1) or measurement result (3.4.2) and a true
3.2 Determination of characteristics and quanti- value (3.2.5).
ties NOTE 1: Bias is the total systematic error as con-
3.3 Properties of test and measurement methods trasted to random error. There may be one or
3.4 Properties of test and measurement results more systematic error components contributing
3.5 Capability of detection to the bias. A larger systematic difference from
4. Inspection and general acceptance sampling the true value is reflected by a larger bias value.
annex outlining the methodology used to develop incomplete block design design.
vocabulary (identical to the same annex used with
Parts 1 and 2). Annex C contains two checklists iden-
TABLE 3c ISO 3534—Part 3, design of experiments, terms
tifying key items for use in designing experiments. related to methods of analysis
Annex D provides a description of experimental
graphical method
design from a systems design perspective. main effects plot
The general terms are shown in Table 3a; terms interaction plot
related to type of experiments in Table 3b and terms quantile plot of effects
related to analysis in Table 3c. residual plot
method of least squares
regression analysis
analysis of variance
TABLE 3a ISO 3534—Part 3, design of experiments, general fixed effects analysis of variance
terms random effects analysis of variance
mixed model analysis of variance
experiment curvature analysis of covariance
model contrast
response variable (output variable) orthogonal contrast
predictor variable experimental unit
Factor block
SURVEY SAMPLING AND OTHER
residual error (error term) blocking ACTIVITIES
residual designed experiment
variance component experimental design The proposed standard ISO 3534-4 on survey sam-
pure random error (pure error) experimental plan pling is in its formative stages; international experts
misspecification error randomization have worked on this document for just a few years.
design region (design space) orthogonal array The official launch of this effort via a new work item
factor level degrees of freedom
(subject to balloting and the cooperation of at least
run (experimental treatment) one factor experiment
factor effect two-factor experiment
five member countries) was deliberately held up
main effect k-factor experiment until a complete committee draft was completed.
dispersion effect replication Once launched, the ISO Secretariat following the
interaction cube point Directives maintains a strict clock vigil to assure that
confounded effect star point progress is made in a timely fashion. Part 4 consists
confounding centre point
of one section containing general terms and a second
Alias rotatability
section on terms related to estimation.