I__. - International Military Testing Association

-..
additional level of speculative subjectivity. Subjective items
would appear to increase the chance of self-deception. Although
subjective corroboration from others is feasible, subjective
items are never objectively verifiable, and hence the chance for
social desirability faking is increased.
Conversely, a number of biodata researchers have made
frequent use of interpretive items. In some studies, subjective
items have actually been shown to have higher predictive
validities than objective ones. An advantage to subjective items
that address self-perceptions is that they can better focus on
unitary theoretical constructs. By contrast, performance of
objective behaviors is often determined by multiple causes and
dispositions, making it difficult to isolate the role of any one.
Barge (1987) has provided evidence that homogeneous items,
tapping a single disposition or tendency, are more predictive
than heterogeneous items such as school or work performance.
Construct-based items are also easier to use to develop
rationally-based biodata scales. It would thus appear that the
use of some subjective items may provide some countervailing
advantages as well.
Discrete versus Summarv Actions. Methodologically, it may
be preferable to focus on discrete actions, dealing with a
single, unique behavior (e.g. age when received driver's
license), as opposed to summary responses (e.g. average time
spent studying). Responses to discrete items only require memory
retrieval, while summary items also require computation or
estimation, thus increasing the chance of inaccuracy. However,
the above preference for discrete actions would obtain only when
the event is unique or singularly memorable. With a regularly
performed behavior, summary recall could be more realistic and
accurate than recall of a single, arbitrarily chosen instance.
Verifiable. A verifiable item is an item that can be
corroborated from an independent source. Item verifiability thus
goes beyond both the external event and objective criteria. The
optimal source of verification is archival data, such as school
transcripts or,work records. Alternatively, the testimony of
knowledgeable persons, such as a teacher, employer, or coach, is
also considered verification by most researchers. Asher (1972)
and Stricker (1987) have advocated exclusive use of verifiable
items, though others utilize non-verifiable items, and some
advocate interleaving verifiable and non-verifiable items
(Mumford et al., 1990).
One reason to use verifiable items is to reduce social
desirability faking and outright falsification. However,
Shaffer, Saunders, and Owens (1986) have shown that social
desirability distortion is not a serious concern with biodata.
Previous research on false or inaccurate responding to verifiable
biodata items has shown mixed results (Cascio, 1975; Goldstein,
1971) which may be due partly to methodological factors (Mumford
& Owens, 1987). Merely warning respondents that answers will be
288
�