Psychological Assessment Copyright 1995 by the American Psychological Association, Inc.
1995, Vol. 7, No. 3,309-319 1040-3590/95/$3.00
Constructing Validity: Basic Issues in Objective Scale Development
Lee Anna Clark and David Watson
The University of Iowa
A primary goal of scale development is to create a valid measure of an underlying construct. We
discuss theoretical principles, practical issues, and pragmatic decisions to help developers maximize
the construct validity of scales and subscales. First, it is essential to begin with a clear conceptualiza-
tion of the target construct. Moreover, the content of the initial item pool should be overinclusive
and item wording needs careful attention. Next, the item pool should be tested, along with variables
that assess closely related constructs, on a heterogeneous sample representing the entire range of the
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
target population. Finally, in selecting scale items, the goal is unidimensionality rather than internal
This document is copyrighted by the American Psychological Association or one of its allied publishers.
consistency; this means that virtually all interitem correlations should be moderate in magnitude.
Factor analysis can play a crucial role in ensuring the unidimensionality and discriminant validity
of scales.
Scale development remains a growth industry within psychol- signed to be completed by clinicians, parents, teachers, spouses,
ogy. A PsycLIT database survey of articles published in the 6- peers, and so forth.
year period from 1989 through 1994 revealed 1,726 articles Before proceeding further, it is interesting to examine the new
with the key words "test construction" or "scale development" measures comprising our Psychological Assessment sample. This
published in English-language journals, 270 in other-language examination sample offers a glimpse at why scale development
journals, and 552 doctoral dissertations. During this same pe- continues unabated, as well as the nature of the unmet needs these
riod (i.e., beginning with its inception), 50 articles addressing scale developers are seeking to fill. First, not surprisingly given this
scale development or test construction were published in Psy- journal's focus, more than half (61%) of the scales assess some
chological Assessment alone. The majority of these articles re- aspect of psychopathology, personality, or adjustment. The next
ported the development of one or more new measures (82%); most common categories are measures of attitudes and interper-
most of the rest presented new scales derived from an existing sonal relations (20% and 15%, respectively). The remaining scales
instrument (10%). We use these 41 scale-development articles assess a miscellany of behaviors, abilities, response validity,
as a reference set for our discussion. Clearly, despite the criti- trauma experience, and so forth. In all categories, most new scales
cism leveled at psychological testing in recent years, assessment apparently tap relatively narrow constructs, such as suicidality,
retains a central role within the field. fear of intimacy, postpartum adjustment, drug-use expectancies,
Given that test construction remains a thriving activity, it is or parent-teenager relations, that have a focused range of utility.
worthwhile to reconsider the scale development process period- However, the extent to which the score variance of such scales is,
ically to maintain and enhance the quality of this enterprise. in fact, attributable to the named target construct is an important
The goal of this article is to articulate some basic principles that issue that we will consider.
we believe anyone developing a scale should know and follow.
Many of these principles have been stated before, but we repeat
them here both because they are sufficiently important to bear
The Centrality of Psychological Measurement
repetition and because a review of the recent literature indicates It has become axiomatic that (publishable) assessment in-
that they are still not universally honored. struments are supposed to be reliable and valid; indeed, every
We focus on verbally mediated measures; thus, for example, article in the Psychological Assessment set addresses these qual-
we do not address the development of behavioral observation ities. However, it appears that many test developers do not fully
scales. Moreover, our primary focus is on self-report measures, appreciate the complexity of these concepts. As this article is
because these constitute the majority (67%) of our reference being prepared, the Standards for Educational and Psychologi-
sample. Nonetheless, most of the basic principles we articulate cal Testing (American Psychological Association, 1985) are un-
are applicable to interview-based measures and rating scales de- dergoing intensive review and revision for the first time in a de-
cade. Strong and conflicting pressures regarding the Standards'
revision are being brought to bear on the Joint Committee on
Lee Anna Clark and David Watson, Department of Psychology, The
the Standards for Educational and Psychological Testing by di-
University of Iowa.
We wish to acknowledge the helpful comments of Jane Loevinger on verse groups, and major changes in the Standards are expected.
a previous version of this article. Whatever else it may do, however, the Joint Committee intends
Correspondence concerning this article should be addressed to Lee to emphasize the centrality of construct validity in testing even
Anna Clark, Department of Psychology, The University of Iowa, E11 more than in previous versions, according to Co-Chair C. D.
Seashore Hall, Iowa City, Iowa 52242-1407. Electronic mail may be Spielberger (personal communication, February 15, 1995).
sent via Internet to [email protected]. And yet, widespread misunderstanding remains regarding pre-
309
