colleagues’ early work on typical and maximum performance (DuBois et al. 1993 ;
Sackett et al. 1988 ) and found that diVerent personality dimensions best predicted
each of the performance constructs. Others have concluded that performance
should be regarded as a dynamic process rather than as a set of static behaviors
that people perform at any given time (e.g. Kozlowski et al. 1999 ; Ployhart and
Hakel 1998 ; Sackett et al. 1988 ).
Regardless of one’s view of performance, these recent advances in theory and
research have emphasized the importance of selecting employees who will be able
to contribute to a wide range of organizational functions, some of which are not
directly related to the tasks for which the employee’s job description holds them
responsible.
- 3 Concerns about Subgroup Representation and Remedies
Reconciling the use of valid selection devices and the desire that the workforce be
representative of societal demographics continues to be of concern among practi-
tioners, researchers, the legal profession, and the public at large (AERA et al. 1999 ;
Barrett and Luecke 2004 ;Grutterv.Bollinger et al. 2003 ; Sharf and Jones 1999 ; SIOP
2003 ). Research on this issue over the last forty years has clariWed several points.
First, the tests that have been examined (most frequently cognitive ability tests) are
not psychometrically biased in that predicted outcomes for protected groups
are not less than similar outcomes predicted for the majority group. Second,
there are large minority–majority group diVerences favoring Caucasians over
African American groups and to a lesser extent Hispanic American groups on
cognitive ability tests (Roth et al. 2001 ) and favoring men over women on physical
ability tests (Hogan 1991 ). Smaller diVerences occur in some instances on other
tests (Bobko et al. 1999 ;Hough 1998 ). Third, various attempts to remove these
subgroup diVerences in cognitive ability may serve to diminish them by a small
amount, but large subgroup diVerences remain and often produce legally deWned
levels of adverse impact on minority groups (Sackett et al. 2001 ).
There have been some new developments in this arena. Statistically, consider-
ation of the impact of reliability and the precision of measurement has resulted
in proposals to band test scores, reXecting the notion that diVerences within
bands are not reliably discriminable. Decisions about test scores within a band
are then made on other bases including ethnic status. An edited book (Aguinis
2004 ) provides a discussion of various approaches to banding including their
mechanisms, the degree to which social values are implicit in these methods of
test use, and their legal status. The impact of banding on minority hiring varies
considerably given the situation and the particular banding remedy employed.
The appropriateness of banding continues to be hotly debated in the scientiWc and
legal communities.
selection decision-making 307