78 aging, longitudinal studies

among students of aging is that research
designs that collect measurements on the same
persons over time are a particularly valuable
approach to studying the causes and conse
quences of aging. Any research design that
locates and measures events and processes in
time is referred to as longitudinal. There are a
variety of types of longitudinal designs,
including everything from complicated life
history calendars, which go to great lengths
to date events, their timing and duration, on
the one hand, to retrospective life histories
presented in narrative form, on the other
(see Scott & Alwin 1998).
Still, perhaps the most common approach in
the study of aging is the one shot cross sec
tional study in which researchers simply com
pare age groups and from such comparisons
draw inferences about aging. One of the critical
problems with the one shot cross sectional stu
dies is that they confuse the potential effects of
aging with the influences of cohort factors (see
Mason & Fienberg 1985). Persons of a parti
cular age at a given point in time are also
members of the same birth cohort (i.e., persons
born during the same year). Members of a
particular birth cohort share the experience of
the life cycle; that is, they experience birth,
infancy, and childhood, reach adolescence,
grow into early adulthood, and mature into
midlife and old age during the same historical
time. In this sense, members of a birth cohort
share a social history; that is, they experience
the same historical events and the opportu
nities and constraints posed by society at a
given time in history. A persons cohort mem
bership may be thought to index the unique
historical period in which a groups common
experiences are embedded, and their behavior
may have as much to do with their historical
experiences as they do with their age.
It is important to realize that one shot cross
sectional designs are not inherently limited,
especially if they involve the replication of
cross sections over time. The existence of dia
chronic cross sectional data for the same cohorts
can be used as a legitimate basis for separating
the effects of aging and cohort effects under
certain circumstances (see Alwin et al. 2005).
Another way to control for cohort differences is
to study a single cohort over time. Eaton (2002)
provides a strong rationale for studying a single
cohort from conception to death. In this type of
single cohort study, age variation occurs over
time rather than cross sectionally, and this per
mits an explicit focus on within person change.
However, development and aging do not occur
in a historical vacuum, and while studying a
single cohort over time does hold many vari
ables constant, it is difficult to generalize about
processes of aging because of the confounding
influences of aging and history.
Having information on the same persons
across a range of birth cohorts a multiple
cohort longitudinal study design opens up
several possibilities for analyzing the effects of
aging across cohorts. The value of this type of
longitudinal design is borne out by the vast
number of research projects over the past few
decades that locate and measure events and
processes in time (see Young et al. 1991).
Indeed, we have reached a point where there
are several longitudinal data sets that permit
the study of patterns and processes of aging in
different historical and cultural contexts. For
example, in the US, the series of panel sur
veys known as the Health and Retirement
Study (HRS) provides a series of replicated
longitudinal studies of a sequence of birth
cohorts currently and in the future. The first
of these began in 1992 as a panel survey of
persons from cohorts born in 1931 through
1941 and re interviewed biennially since then.
The idea for the HRS derived from a growing
awareness of the inadequacy of data available
from the Retirement History Survey that
began in 1969 and followed a set of cohorts
of men and unmarried women born in 1906
through 1911 for ten years. Basing ones infer
ences about processes of aging, it was argued,
on such a limited spectrum of historical
cohorts had obvious limitations, given, for
example, the growing participation of women
in the labor force and related changes in the
family. The collection of data on health and
other antecedents of work and retirement
decisions for more recent cohorts was viewed
as essential to understanding experiences
related to processes of aging in the more con
temporary social context.
The assessment of change over time is fun
damental to the quantitative study of aging,
and longitudinal designs are vastly superior to
cross sectional studies in their ability to reveal