Editorial
Beyond Validity and Reliability: Meaning-in-Context of Adolescents’
Self-Reports of Sexual Behavior
Some years ago, my colleagues and I were pilot-testing
a self-administered survey item about condom use at most recent
intercourse. The item read ‘‘Was a condom used the last time
you had vaginal sex?’’ The wording was closely adapted from
widely used measures of condom use. Pilot testing went well
until a research assistant showed me a form with ‘‘No’’ marked
as the response. In reply to my query about why this was
a problem, the research assistant pointed to the short sentence
written beneath the item. It said ‘‘The condom was new.’’
I write, read, use, and critique a lot of research questions
related to adolescents’ sexual behavior, and I often think of
this experience while doing so. That young adolescent’s
response to our research inquiry expresses what I’ve come
to consider my penultimate principle in the measurement of
adolescents’ sexual behavior: this is as good as it gets.
‘‘As good as it gets’’ summarizes recognition and accep-
tance of an irresolvable gap between an unknowable truth
and a knowable datum. The unknowable truth is whether
some idealized type of sexual behavior physically occurred.
I say ‘‘idealized’’ because we attempt to measure sex as
specific individual behaviors when it is in fact an integrated
and highly coordinated set of gestures, activities and feelings
that only roughly correspond to our categories. The knowable
datum is an adolescent’s report on the occurrence of that
sexual behavior, if it can be extracted from those complex
gestures, activities and feelings and matched to some key
bit or phrase in the investigator’s question.
The gap between truth and datum haunts investigators and
delights skeptics of the veracity of self-reports of sexual
behavior. One approach to addressing this issue is to assess
test–retest reliability of adolescents’ self-reports. This is the
approach taken by Vanable et al. in this issue [1]. The article
adds to an existing literature that notes, in general, that
measures of adolescents’ self-reported sexual behaviors
have imperfect but satisfactory reliability [2,3]. Moreover,
the paper extends the existing literature by the provision of
reliability data on several other measures relevant to under-
standing of human immunodeficiency virus/sexually trans-
mitted infection risk and protection behaviors, as well as
understanding of the reliability of all of these measures
within the context of an audio computer–assisted interview
(ACASI) data collection format.
The results reported by Vanable et al are particularly inter-
esting in light of a paper by Palen et al published in the Jour-
nal in 2008 [4]. The conclusions of Palen et al, in contrast to
those of Vanable et al, were that reliability of adolescents’
reports about sexual behavior were sufficiently low as to
call into question inferences about their sexual behavior.
One could address the differences between the Vanable et
al and the Palen et al data by a careful analysis of issues tradi-
tional to any evidence-based journal club: the two papers
differ markedly in the linguistic and sociocultural origins of
the sample, in the study design, in the specific questions
asked, and in the survey mode used to ask the questions.
I shall leave this type of close analysis to others. Rather, I
would like to disagree with a point raised by the authors of
both papers—as well as by Lucia O’Sullivan [5] in an edito-
rial accompanying the Palen et al paper—concluding that
ongoing research is required to improve the reliability of
adolescents’ self-reports of sexual behavior. More accu-
rately, I should say that I agree that strenuous efforts toward
accuracy and precision are an explicit responsibility of each
investigator. This is simply part of the rigor, discipline, and
ethics of science. The larger challenge is to search for some
different perspective that allows us to go forward in the
face of irresolvable questions about the reliability of the data.
The perspective I find increasingly important is one that
gives much less emphasis to counting adolescents’ sexual
behaviors and much more emphasis to understanding the
personal and social contexts of their occurrence and the
meanings derived from them [6].
One justification for this different perspective is that it
shifts our attention away from our obsessive categorization
of people as virgins and unvirgins. I use the word ‘‘unvirgin’’
(jocularly similar to the ‘‘undead’’ of vampire mythology) to
highlight the way we consider a person to be profoundly
‘‘changed’’ on the basis of a single, typically brief, sexual
event (typically coitus, but one can be ‘‘virgin’’ in a variety
See Related Article p. 214
1054-139X/09/$ – see front matter Ó 2009 Society for Adolescent Medicine. All rights reserved.
doi:10.1016/j.jadohealth.2008.12.018
Journal of Adolescent Health 44 (2009) 199–200