of different age groups, at the same point in time.
Alongitudinal designfollows the same subjects over
time. The basic format of these two approaches is
shown in Figure 4-3. In this example,row aillus-
trates the longitudinal design andcolumn bshows
the cross-sectional design.
Cross-sectional approaches are correlational
because the investigator cannot manipulate age nor
can participants be assigned to different age groups.
Because there are different participants in each age
group, we cannot assume that the outcome of the
study reflects age changes; it only reflects differences
among the age groups employed. These differences
could be due to the eras in which participants were
raised rather than age per se. For example, a group
of 65-year-olds might show up as more frugal than a
group of 35-year-olds. Does this mean that advanc-
ing age promotes frugality? Perhaps. But it might
simply reflect the historical circumstance that the
65-year-olds were raised during a different time
when money was very hard to come by.
Longitudinal studies are those in which we col-
lect data on the same people over time. Such designs
allow us to gain insight into how behavior or mental
processes change with age. In the interpretive sense,
longitudinal studies enable investigators to better
speculate about time-order relationships among
factors that vary together. They also help reduce the
third-variable problem that so often arises in correla-
tional studies. For example, suppose we know that
states of depression come and go over the years. If
depression is responsible for the correlation between
significant weight loss and decreased self-confidence,
then both weight loss and decline in self-confidence
should vary along with depressive states.
There are, of course, many variations in cross-
sectional and longitudinal designs. In the case of lon-
gitudinal studies, however, the main problems are
practical ones. Such studies are costly to carry out,
and they require great patience and continuity of
leadership in the research program. Sometimes,
too, researchers must live with design mistakes
made years earlier or put up with outmoded research
and assessment methods. Finally, participants some-
times drop out of longitudinal studies, and one must
demonstrate that the retained participants are repre-
sentative of those that started the study.The Experimental Method
To determine cause–effect relationships among
events, we must use experimental methods. Consider
the following hypothetical study, an expansion of a
pilot study by Naylor et al. (2007), that examined1890
1895
1900
1905
1910
1915
1920
1925Time of
measurement65
60
55
50
45
40
35
30195570
65
60
55
50
45
40
35196075
70
65
60
55
50
45
40196580
75
70
65
60
55
50
45197085
80
75
70
65
60
55
501975Birth date Age
baF I G U R E 4-3 Cross-sectional and longitudinal designs.
SOURCE: From Woodruff-Pak, Psychology and Aging (p. 32). Copyright © 1988. Reprinted by permission of Diana S. Woodruff-Pak.
106 CHAPTER 4