216 Developmental Psychology
century is to discover which aspects of behavior are likely to
be altered by environmental events at specific points in de-
velopment and which aspects remain more plastic and open
to influence across wide spans of development.
In terms of units of analysis, researchers have begun to
conceptualize the unit of analysis as dyads within the family
system, such as the parent–child dyad, the husband–wife
dyad, and the sibling dyad (Belsky, 1984; Cowan & McHale,
1996; Parke, 1988). Moreover, units beyond the dyad have
been recognized as important as well. Several researchers
have recently begun to investigate triads (Hinde & Stevenson-
Hinde, 1988; Kreppner, 1988) as well as the family as units of
analysis (Dickstein et al., 1998).
At present, this shift toward units beyond the individual is
evident in cognitive as well as social development and is due,
in part, to the revival of interest in Vygotskian theory. Lev
Vygotsky (1896–1934), a Russian psychologist, championed
the view that mental functioning is a kind of action that may
be exercised by individuals or by dyads or larger groups
(Wertsch, 1991). His view was one in which mind is under-
stood as “extending beyond the skin.” “Mind, cognition and
memory...are understood not as attributes or properties of
individuals but as functions that may be carried out intermen-
tally or intramentally” (Wertsch & Tulviste, 1992, p. 549).
Such terms as socially shared cognition(Resnick, Levine, &
Teasley, 1991), socially distributed cognition (Hutchins,
1991), and collaborative problem solving(Rogoff, 1990) re-
flect the increasing awareness that cognition can be a social
as well as an individual enterprise.
In terms of methods, variety best describes the contempo-
rary period. In the 1990s, the use of longitudinal designs in-
creased markedly, motivated in part by an increased interest
in issues of developmental stability and change. Two types of
longitudinal studies are evident. Short-term longitudinal
studies, in which a particular issue is traced over a short time
period of a few months to a year, are currently popular
(Clarke-Stewart, Gruber, & Fitzgerald, 1994; Hetherington &
Clingempeel, 1992). These studies are of value for detecting
short-term stability or for tracking development across a time
period of assumed rapid change in an emerging developmen-
tal process or structure.
Other longitudinal studies have a long-term character and
have continued from infancy through childhood and into
adolescence. For example, a number of investigators have
followed families from infancy to the preadolescent or ado-
lescent years (Sameroff, 1994; Sroufe, 1996). This strategy
has permitted a more definitive evaluation of a variety of the-
oretical issues, especially those concerning the effect of early
experience, including the role of sensitive and critical periods
on later development. Nonetheless, because of the expense
and difficulty of longitudinal research, cross-sectional de-
signs still predominate among developmental investigations.
Often researchers will use both strategies, and, in an area that
is not yet well developed either theoretically or empirically,
cross-sectional studies often precede longitudinal pursuit of
an issue. A commitment to multiple design strategies rather
than a near-exclusive reliance on a single design is character-
istic of the current area.
In terms of experimental designs, a greater openness to
multiple strategies is evident. Laboratory-based experimen-
tal studies and field-based experimental investigations
coexist with nonexperimental observational field studies.
Data-collection strategies come in a variety of forms as well.
In spite of its less than stellar history, the self-report measure
has reentered our methodological repertoire; parent, teacher,
and peer reports are now commonly used. Another note-
worthy trend reflects in part the openness of researchers to
multimethod strategies as opposed to strict adherence to one
approach. Observational methods are widely used along with
verbal reports. Evidence, not just speculation, may be driving
the field to this new openness to a wide range of methods.
Some researchers have found that ratings of behavior yield
better prediction of later social behavior (Bakeman & Brown,
1980) and later cognitive assessments (Jay & Farran, 1981)
than do more microanalytic and more expensive measures of
parent–child interaction.
Finally, our sampling methods have come of age. Shifts in
awareness of the importance of sampling have led to an in-
crease in use of large representative national samples in de-
velopmental research. Although this has typically been the
domain of sociologists and survey researchers, in the early
1990s, developmentalists have shown an increased aware-
ness of the potential value of supplementing their usual
small-sample strategies with these large-sample approaches.
One prominent example is the use of the National Longitudi-
nal Study of Youth (NLSY) for the examination of develop-
mental issues, including divorce, achievement, and day care
(Brooks-Gunn, Phelps, & Elder, 1991). These surveys have
several advantages, including a large number of subjects,
more representative samples, a multifaceted range of vari-
ables, and longitudinal designs. In turn, these characteristics
permit testing of more complex models of development that
require large numbers of subjects. In addition, these studies
allow examination of connections across content-based do-
mains as well as encouraging interdisciplinary cooperation.
Finally, they permit testing of the cultural generality of the
models.
Newer, more innovative approaches that combine levels
of sampling are becoming increasingly common as well. As a
supplement to a large-scale survey approach, researchers are