The Effects of Stress on Young Children's Memory - IDL at Florida ...
The Effects of Stress on Young Children's Memory - IDL at Florida ...
The Effects of Stress on Young Children's Memory - IDL at Florida ...
- TAGS
- infantlab.fiu.edu
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Journal <str<strong>on</strong>g>of</str<strong>on</strong>g> Experimental Psychology: Applied Copyright 1998 by the American Psychological Associ<strong>at</strong>i<strong>on</strong>, Inc.<br />
1998, Vol. 4, No. 4, 308-331 1076-S9SX/98/$3.00<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> <str<strong>on</strong>g>Effects</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Stress</str<strong>on</strong>g> <strong>on</strong> <strong>Young</strong> <strong>Children's</strong> <strong>Memory</strong><br />
for a N<strong>at</strong>ural Disaster<br />
Lorraine E. Bahrick<br />
and Jan<strong>at</strong> Fraser Parker<br />
<strong>Florida</strong> Intern<strong>at</strong>i<strong>on</strong>al University<br />
Mary Levitt<br />
<strong>Florida</strong> Intern<strong>at</strong>i<strong>on</strong>al University<br />
Robyn Fivush<br />
Emory University<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> effects <str<strong>on</strong>g>of</str<strong>on</strong>g> stress <strong>on</strong> children's l<strong>on</strong>g-term memory for a major hurricane<br />
were studied. <str<strong>on</strong>g>Stress</str<strong>on</strong>g> was objectively defined as low, moder<strong>at</strong>e, or high<br />
according to the severity <str<strong>on</strong>g>of</str<strong>on</strong>g> damage to the child's home. One hundred 3-<br />
and 4-year-old children received a structured interview 2-6 m<strong>on</strong>ths<br />
following the hurricane. Older children recalled and elabor<strong>at</strong>ed more than<br />
younger children. Prompted recall was gre<strong>at</strong>er than sp<strong>on</strong>taneous recall.<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g>re was a quadr<strong>at</strong>ic functi<strong>on</strong>, c<strong>on</strong>sistent with an inverted U-shaped curve,<br />
rel<strong>at</strong>ing storm severity with overall as well as sp<strong>on</strong>taneous recall. <str<strong>on</strong>g>The</str<strong>on</strong>g>se<br />
findings can be applied to the effects <str<strong>on</strong>g>of</str<strong>on</strong>g> stress <strong>on</strong> the amount recalled by<br />
children giving retrospective accounts <str<strong>on</strong>g>of</str<strong>on</strong>g> temporally extended, n<strong>at</strong>uralistic<br />
events.<br />
On August 24th, 1992, Hurricane Andrew, a<br />
major Class IV hurricane with 175-mph winds,<br />
devast<strong>at</strong>ed much <str<strong>on</strong>g>of</str<strong>on</strong>g> the densely popul<strong>at</strong>ed area <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
Dade County, <strong>Florida</strong>. <str<strong>on</strong>g>The</str<strong>on</strong>g> storm caused over 20<br />
billi<strong>on</strong> dollars <str<strong>on</strong>g>of</str<strong>on</strong>g> property damage and affected<br />
the lives <str<strong>on</strong>g>of</str<strong>on</strong>g> families and young children for<br />
Lorraine E. Bahrick, Jan<strong>at</strong> Fraser Parker, and Mary<br />
Levitt, Department <str<strong>on</strong>g>of</str<strong>on</strong>g> Psychology, <strong>Florida</strong> Interna-<br />
ti<strong>on</strong>al University; Robyn Fivush, Department <str<strong>on</strong>g>of</str<strong>on</strong>g> Psy-<br />
chology, Emory University.<br />
This research was supported by Grant DBS 9300956<br />
from the N<strong>at</strong>i<strong>on</strong>al Science Found<strong>at</strong>i<strong>on</strong>. We extend<br />
special thanks to Maria Borras, Angela Buffkin, Mi<br />
Yung Chung, Ana D'Costa, Marlisa Dem<strong>on</strong>d, GiUian<br />
Dix<strong>on</strong>, Angela Escobar, Rita Estorino, Laura Ferret,<br />
Brenda Lundy, Julie Pinyero, Rita Soza, and Cristina<br />
Velarde for their assistance in interviewing, transcrib-<br />
ing, coding, and tabul<strong>at</strong>ing d<strong>at</strong>a; to Brenda Lundy for<br />
coordin<strong>at</strong>ing participant recruitment and d<strong>at</strong>a coding;<br />
and to Paulette Johns<strong>on</strong> for her assistance in d<strong>at</strong>a<br />
analyses.<br />
Corresp<strong>on</strong>dence c<strong>on</strong>cerning this article should be<br />
addressed to Lorraine E. Bahrick, Department <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
Psychology, <strong>Florida</strong> Intern<strong>at</strong>i<strong>on</strong>al University, Miami,<br />
<strong>Florida</strong> 33199. Electr<strong>on</strong>ic mail may be sent to<br />
bahrick@fiu.edu.<br />
308<br />
m<strong>on</strong>ths afterward. <str<strong>on</strong>g>The</str<strong>on</strong>g> present investig<strong>at</strong>i<strong>on</strong> as-<br />
sessed memory for events surrounding Hurricane<br />
Andrew in 3- and 4-year-old children. This event<br />
provided a unique forum for addressing many <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
the questi<strong>on</strong>s and issues raised by current devel-<br />
opmental research <strong>on</strong> childhood event memory<br />
and its rel<strong>at</strong>i<strong>on</strong> to stress in a n<strong>at</strong>uralistic, pers<strong>on</strong>-<br />
ally relevant, and highly emoti<strong>on</strong>al event c<strong>on</strong>text.<br />
Recent investig<strong>at</strong>i<strong>on</strong>s <str<strong>on</strong>g>of</str<strong>on</strong>g> childhood memory<br />
have revealed th<strong>at</strong> even quite young children are<br />
able to recall an impressive amount <str<strong>on</strong>g>of</str<strong>on</strong>g> accur<strong>at</strong>e<br />
inform<strong>at</strong>i<strong>on</strong> about events they experienced (see<br />
Fivush, 1993 for a review). Further, memories are<br />
retained over very l<strong>on</strong>g time periods. For ex-<br />
ample, Ham<strong>on</strong>d and Fivush (1991) interviewed<br />
4- to 6-year-old children about a nip to Disney-<br />
world th<strong>at</strong> occurred either 6 or 18 m<strong>on</strong>ths earlier.<br />
Children recalled a gre<strong>at</strong> deal about the event,<br />
approxim<strong>at</strong>ely 40 propositi<strong>on</strong>s, and mothers<br />
judged virtually all the inform<strong>at</strong>i<strong>on</strong> to be accu-<br />
r<strong>at</strong>e. Older children provided more elabor<strong>at</strong>ed<br />
and detailed verbal accounts than younger chil-<br />
dren, but amount <str<strong>on</strong>g>of</str<strong>on</strong>g> propositi<strong>on</strong>al inform<strong>at</strong>i<strong>on</strong><br />
was not different. Other studies have also found<br />
accur<strong>at</strong>e, l<strong>on</strong>g-term verbal recall <str<strong>on</strong>g>of</str<strong>on</strong>g> single, infre-<br />
quent events such as the birth <str<strong>on</strong>g>of</str<strong>on</strong>g> a sibling
(Sheingold & Tenney, 1982), a preschool evac-<br />
u<strong>at</strong>i<strong>on</strong> due to fire (Pillemer, PicarieUo, & Pruett,<br />
1995), trips to a zoo, circus, or birthday party<br />
(Fivush, Gray, & Fromh<str<strong>on</strong>g>of</str<strong>on</strong>g>f, 1987; Todd & Perlmut-<br />
ter, 1980), or a trip to an archaeological museum<br />
(Huds<strong>on</strong> & Fivush, 1991). In these studies, recall was<br />
evident across periods ranging from several m<strong>on</strong>ths<br />
to several years, and in <strong>on</strong>e study, across decades.<br />
A growing body <str<strong>on</strong>g>of</str<strong>on</strong>g> research also indic<strong>at</strong>es th<strong>at</strong><br />
older children recall more inform<strong>at</strong>i<strong>on</strong> than youn-<br />
ger children in resp<strong>on</strong>se to open-ended questi<strong>on</strong>s<br />
but not necessarily to more structured questi<strong>on</strong>s.<br />
<strong>Young</strong>er children rely more <strong>on</strong> specific questi<strong>on</strong>s,<br />
prompts, and structure to facilit<strong>at</strong>e recall and can<br />
sometimes recall as much as somewh<strong>at</strong> older<br />
children when given sufficient prompts (Ham<strong>on</strong>d<br />
& Fivush, 1991; Todd & Perlmutter, 1980; see<br />
Pillemer & White, 1989 for a review). Although<br />
many studies indic<strong>at</strong>e even 1- to 3-year-olds<br />
show remarkably accur<strong>at</strong>e and enduring memo-<br />
ries (Bauer & Fivush, 1992; Bauer & Mandler,<br />
1989; Bauer & Travis, 1993; Fivush, Kuebli, &<br />
Clubb, 1992; Huds<strong>on</strong> & Nels<strong>on</strong>, 1983; Price &<br />
Goodman, 1990; Slackman, Huds<strong>on</strong>, & Fivush,<br />
1986), younger children also produce more errors<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> intrusi<strong>on</strong> than older children (Howe, Courage,<br />
& Peters<strong>on</strong>, 1995) and can be more easily misled<br />
about the details <str<strong>on</strong>g>of</str<strong>on</strong>g> an event they witness (Ceci &<br />
Bruck, 1993; Ceci, Toglia, & Ross, 1987). This is<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> obvious practical importance when evalu<strong>at</strong>ing<br />
the potential for children to provide accur<strong>at</strong>e<br />
testim<strong>on</strong>y in court. Establishing criteria for evi-<br />
dence <str<strong>on</strong>g>of</str<strong>on</strong>g> memory may, however, differ depending<br />
<strong>on</strong> the purpose <str<strong>on</strong>g>of</str<strong>on</strong>g> the investig<strong>at</strong>i<strong>on</strong> (whether it is<br />
designed for gener<strong>at</strong>ing developmental principles<br />
about how the memory system works, such as the<br />
present investig<strong>at</strong>i<strong>on</strong>, versus determining the<br />
reliability and accuracy <str<strong>on</strong>g>of</str<strong>on</strong>g> an eyewitness where<br />
accuracy <str<strong>on</strong>g>of</str<strong>on</strong>g> small details <str<strong>on</strong>g>of</str<strong>on</strong>g> an event may be crucial).<br />
A critical c<strong>on</strong>cern from a forensic perspective<br />
has been the effects <str<strong>on</strong>g>of</str<strong>on</strong>g> stress <strong>on</strong> memory. Good-<br />
man and her colleagues (Goodman, Aman, &<br />
Hirschman, 1987; Goodman, Bottoms, Schwartz-<br />
Kenney, & Rudy, 1991; Goodman, Hirschman,<br />
Hepps, & Rudy, 1991) have argued th<strong>at</strong> children<br />
who experience more arousal during medical<br />
procedures, such as inocul<strong>at</strong>i<strong>on</strong>s, tend to show no<br />
decrement or enhanced memory for the event. In<br />
<strong>on</strong>e study, their memory was superior to the<br />
c<strong>on</strong>trol children who received a washable t<strong>at</strong>oo<br />
(Goodman, Hirschman, et al., 1991, Study 2).<br />
MEMORY FOR A NATURAL DISASTER 309<br />
Other studies (e.g., Howe et al., 1995) found no<br />
rel<strong>at</strong>i<strong>on</strong> between stress (as r<strong>at</strong>ed by parents) and<br />
amount recalled about an emergency room proce-<br />
dure either 3-5 days or 6 m<strong>on</strong>ths l<strong>at</strong>er (see also<br />
Baker-Ward, Gord<strong>on</strong>, Ornstein, Larus, & Clubb,<br />
1993; Howe et al., 1994). In c<strong>on</strong>trast, Peters<br />
(1987, 1991) found th<strong>at</strong> arousal either has no<br />
effect or hinders event memory. Children who<br />
were more highly stressed during a dentist visit,<br />
an inocul<strong>at</strong>i<strong>on</strong>, a fire alarm, or a visit by a<br />
stranger made a gre<strong>at</strong>er number <str<strong>on</strong>g>of</str<strong>on</strong>g> identific<strong>at</strong>i<strong>on</strong><br />
errors in lineups than those who were less<br />
stressed in the c<strong>on</strong>trol c<strong>on</strong>diti<strong>on</strong>, and stress was<br />
corrobor<strong>at</strong>ed by measures such as pulse, blood<br />
pressure, and observ<strong>at</strong>i<strong>on</strong>al r<strong>at</strong>ings. Others have<br />
also documented a neg<strong>at</strong>ive rel<strong>at</strong>i<strong>on</strong> between<br />
stress and memory under some c<strong>on</strong>diti<strong>on</strong>s (Mer-<br />
ritt, Ornstein, & Spicker, 1994; Vandermaas,<br />
Hess, & Baker-Ward, 1993).<br />
One problem inherent in c<strong>on</strong>ducting this re-<br />
search has been the difficulty <str<strong>on</strong>g>of</str<strong>on</strong>g> comparing stress<br />
levels and memory tasks across studies. A variety<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> stress measures have been used and it is<br />
difficult to determine if high stress in <strong>on</strong>e study is<br />
equivalent to high, moder<strong>at</strong>e, or low stress in<br />
another study. This is especially true when subjec-<br />
tive measures <str<strong>on</strong>g>of</str<strong>on</strong>g> stress are used. A further prob-<br />
lem rel<strong>at</strong>es to differences in the type <str<strong>on</strong>g>of</str<strong>on</strong>g> memory<br />
tested. <strong>Memory</strong> has been tested in a variety <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
ways for many aspects <str<strong>on</strong>g>of</str<strong>on</strong>g> the events, ranging<br />
from free recall to identifying individuals in<br />
lineups and for aspects including central versus<br />
peripheral inform<strong>at</strong>i<strong>on</strong> and general knowledge<br />
versus specific details. Finally, it has been diffi-<br />
cult to include a sufficiently wide range <str<strong>on</strong>g>of</str<strong>on</strong>g> stress<br />
such th<strong>at</strong> a complete picture <str<strong>on</strong>g>of</str<strong>on</strong>g> the effects <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
stress <strong>on</strong> memory becomes apparent.<br />
One <str<strong>on</strong>g>of</str<strong>on</strong>g> the most serious problems has been the<br />
difficulty <str<strong>on</strong>g>of</str<strong>on</strong>g> selecting an appropri<strong>at</strong>e c<strong>on</strong>trol<br />
group. Two general approaches have character-<br />
ized the research described above. One approach<br />
has been to cre<strong>at</strong>e a low stress c<strong>on</strong>trol group by<br />
using an event th<strong>at</strong> elicits little stress but is as<br />
comparable as possible to the stressful event in<br />
other ways (Goodman, Hirschman, et al., 1991,<br />
Experiments 1 and 3; Peters, 1997). Alterna-<br />
tively, researchers have chosen a stressful event<br />
for which there is individual variability in the<br />
degree <str<strong>on</strong>g>of</str<strong>on</strong>g> stress experienced and selected the<br />
children who are least anxious as the low stress<br />
c<strong>on</strong>trol group (e.g., Goodman, Hirschman, et al.,
310 BAHRICK, PARKER, FIVUSH, AND LEVrlT<br />
1991, Experiment 2; Howe, Courage, & Peter-<br />
s<strong>on</strong>, 1995). Each approach has advantages and<br />
limit<strong>at</strong>i<strong>on</strong>s. <str<strong>on</strong>g>The</str<strong>on</strong>g> individual difference approach<br />
has the limit<strong>at</strong>i<strong>on</strong> th<strong>at</strong> typically, a fairly narrow<br />
range <str<strong>on</strong>g>of</str<strong>on</strong>g> stress is sampled (a doctor or emergency<br />
room visit is rarely a positive experience). This<br />
restricts the likelihood <str<strong>on</strong>g>of</str<strong>on</strong>g> finding an effect if there<br />
is <strong>on</strong>e present. Further, participants who experi-<br />
ence these events as very stressful versus those<br />
who experience them as <strong>on</strong>ly mildly stressful<br />
may differ a priori in other important ways th<strong>at</strong><br />
may, in turn, be correl<strong>at</strong>ed with memory perfor-<br />
mance. For example, they may have different<br />
temperaments, general anxiety levels, or coping<br />
str<strong>at</strong>egies. <str<strong>on</strong>g>The</str<strong>on</strong>g>se preexisting differences make<br />
any observed differences in memory as a functi<strong>on</strong><br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> stress difficult to interpret. On the other hand,<br />
this approach has the important advantage th<strong>at</strong><br />
the memory test and task are identical across<br />
stress levels, and thus no c<strong>on</strong>founds due to type<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> task arise.<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> disadvantage <str<strong>on</strong>g>of</str<strong>on</strong>g> the approach in which<br />
participants are exposed to a low stress event th<strong>at</strong><br />
is similar to the stressful event is th<strong>at</strong> the two<br />
events themselves differ. This raises the problem<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> c<strong>on</strong>founds due to their lack <str<strong>on</strong>g>of</str<strong>on</strong>g> comparability.<br />
For example, the amount <str<strong>on</strong>g>of</str<strong>on</strong>g> prior knowledge has<br />
an important influence <strong>on</strong> wh<strong>at</strong> and how much<br />
children recall (Bjorklund, 1985; Chi & Ceci,<br />
1987). Thus, if children have a more well-<br />
developed script or general knowledge base about<br />
<strong>on</strong>e event than another, this could enhance amount<br />
recalled by providing better organiz<strong>at</strong>i<strong>on</strong> for<br />
details or by mistaking the general knowledge for<br />
specific event memories (see Clubb, Nida, Mer-<br />
ritt, & Ornstein, 1993). A host <str<strong>on</strong>g>of</str<strong>on</strong>g> other c<strong>on</strong>cerns<br />
arise from the use <str<strong>on</strong>g>of</str<strong>on</strong>g> different events and makes it<br />
difficult to cre<strong>at</strong>e truly comparable memory tests<br />
and events th<strong>at</strong> differ <strong>on</strong>ly in the stress they elicit.<br />
Factors th<strong>at</strong> impact memory performance must be<br />
equ<strong>at</strong>ed across c<strong>on</strong>diti<strong>on</strong>s, such as amount <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
central versus peripheral inform<strong>at</strong>i<strong>on</strong> presented<br />
and assessed, number <str<strong>on</strong>g>of</str<strong>on</strong>g> enabling or causal<br />
rel<strong>at</strong>i<strong>on</strong>s am<strong>on</strong>g comp<strong>on</strong>ents <str<strong>on</strong>g>of</str<strong>on</strong>g> events, and com-<br />
plexity <str<strong>on</strong>g>of</str<strong>on</strong>g> the events themselves. If different<br />
questi<strong>on</strong>s are asked for the different events, then<br />
the degree <str<strong>on</strong>g>of</str<strong>on</strong>g> structure and prompting versus<br />
open-endedness <str<strong>on</strong>g>of</str<strong>on</strong>g> questi<strong>on</strong>s, as well as overall<br />
complexity <str<strong>on</strong>g>of</str<strong>on</strong>g> the questi<strong>on</strong>s, must also be equ<strong>at</strong>ed<br />
(see also Ornstein, Merritt, & Baker-Ward, 1995).<br />
For example, it is not possible to determine<br />
whether receiving an inocul<strong>at</strong>i<strong>on</strong> versus a wash-<br />
able t<strong>at</strong>oo (Goodman, Hirschman, et al., 1991), a<br />
teeth cleaning versus a surgical procedure <strong>at</strong> the<br />
dentist (Vandermaas et al., 1993), or a fire alarm<br />
versus the sound <str<strong>on</strong>g>of</str<strong>on</strong>g> a loud radio (Peters, 1997) is<br />
comparable in complexity, degree <str<strong>on</strong>g>of</str<strong>on</strong>g> general<br />
knowledge, or number <str<strong>on</strong>g>of</str<strong>on</strong>g> causal rel<strong>at</strong>i<strong>on</strong>s. For<br />
these reas<strong>on</strong>s, it is difficult to be sure th<strong>at</strong><br />
observed differences in memory performance are<br />
a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> stress level and are not due to a<br />
priori differences in the memorability <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />
different events or the difficulty <str<strong>on</strong>g>of</str<strong>on</strong>g> the memory<br />
tests. This approach, however, <str<strong>on</strong>g>of</str<strong>on</strong>g>ten allows sam-<br />
piing <str<strong>on</strong>g>of</str<strong>on</strong>g> a much wider range <str<strong>on</strong>g>of</str<strong>on</strong>g> stress than the<br />
former approach. Ultim<strong>at</strong>ely, c<strong>on</strong>verging evi-<br />
dence from a variety <str<strong>on</strong>g>of</str<strong>on</strong>g> studies assessing memory<br />
across different events and settings is likely to<br />
provide us with a picture <str<strong>on</strong>g>of</str<strong>on</strong>g> the memory-stress<br />
rel<strong>at</strong>i<strong>on</strong>ship th<strong>at</strong> is less biased by these factors.<br />
Research <strong>on</strong> the rel<strong>at</strong>i<strong>on</strong> between emoti<strong>on</strong>ality<br />
and memory in adults has yielded similarly<br />
inc<strong>on</strong>sistent findings. Many studies have docu-<br />
mented th<strong>at</strong> emoti<strong>on</strong>al events are remembered<br />
with gre<strong>at</strong>er vividness and detail than neutral<br />
events (Christians<strong>on</strong> & L<str<strong>on</strong>g>of</str<strong>on</strong>g>tus, 1990; Pillemer,<br />
1984; Rubin & Kozin, 1984). In other studies,<br />
emoti<strong>on</strong>ality has undermined memory for detail<br />
(Christians<strong>on</strong> & Nilss<strong>on</strong>, 1984; Clifford & Hol-<br />
lin, 1981; Clifford & Scott, 1978; L<str<strong>on</strong>g>of</str<strong>on</strong>g>tus &<br />
Burns, 1982; see Heuer & Reisberg, 1992 and<br />
Christians<strong>on</strong>, 1992 for reviews). Easterbrook<br />
(1959) proposed th<strong>at</strong> arousal leads to a narrowing<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> <strong>at</strong>tenti<strong>on</strong> and thus emoti<strong>on</strong> may lead to better<br />
memory for central inform<strong>at</strong>i<strong>on</strong> but poorer<br />
memory for peripheral inform<strong>at</strong>i<strong>on</strong>. C<strong>on</strong>sistent<br />
with this hypothesis, several studies found th<strong>at</strong><br />
central details were better remembered in emo-<br />
ti<strong>on</strong>al events than corresp<strong>on</strong>ding details in neu-<br />
tral events (Christians<strong>on</strong> & L<str<strong>on</strong>g>of</str<strong>on</strong>g>tus, 1987, 1990,<br />
1991; Kebeck & Lohaus, 1986). Excepti<strong>on</strong>s to<br />
this p<strong>at</strong>tern have also been noted, however (e.g.,<br />
Heuer & Reisberg, 1990, 1992). Thus, it appears<br />
th<strong>at</strong> we are still far from being able to propose a<br />
clear, integr<strong>at</strong>ive account <str<strong>on</strong>g>of</str<strong>on</strong>g> how affect influences<br />
the memory system.<br />
Hurricane Andrew, a n<strong>at</strong>ural disaster <str<strong>on</strong>g>of</str<strong>on</strong>g> enor-<br />
mous scope, was an extremely stressful event th<strong>at</strong><br />
provided a n<strong>at</strong>ural and opportune forum for<br />
investig<strong>at</strong>ing the rel<strong>at</strong>i<strong>on</strong>ship between memory<br />
and stress. It provided a number <str<strong>on</strong>g>of</str<strong>on</strong>g> excellent<br />
altern<strong>at</strong>ives to many <str<strong>on</strong>g>of</str<strong>on</strong>g> the problems outlined
above. Families were subjected to a range <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
experiences from severe trauma to mild stress<br />
and inc<strong>on</strong>venience. Many residents feared for<br />
their lives and were forced to move from <strong>on</strong>e<br />
room to another to avoid flying glass and debris<br />
as the storm decim<strong>at</strong>ed much <str<strong>on</strong>g>of</str<strong>on</strong>g> their home and<br />
its c<strong>on</strong>tents. Most <str<strong>on</strong>g>of</str<strong>on</strong>g> these families engaged in an<br />
extensive clean up period with no basic services<br />
for weeks and were then l<strong>at</strong>er forced to reloc<strong>at</strong>e<br />
while their homes were rebuilt. <str<strong>on</strong>g>The</str<strong>on</strong>g>y experienced<br />
serious trauma. Other families who lived further<br />
away from the eye <str<strong>on</strong>g>of</str<strong>on</strong>g> the storm experienced a less<br />
severe storm with moder<strong>at</strong>e levels <str<strong>on</strong>g>of</str<strong>on</strong>g> trauma and<br />
a gre<strong>at</strong> deal <str<strong>on</strong>g>of</str<strong>on</strong>g> inc<strong>on</strong>venience. <str<strong>on</strong>g>The</str<strong>on</strong>g>y also received<br />
property damage (typically to the outside <str<strong>on</strong>g>of</str<strong>on</strong>g> their<br />
home, p<strong>at</strong>io, and ro<str<strong>on</strong>g>of</str<strong>on</strong>g>, and there was also w<strong>at</strong>er<br />
damage inside) and engaged in an extensive<br />
cleanup without basic services for several weeks.<br />
Typically, they repaired their homes without<br />
moving out. Those <strong>at</strong> the fringes <str<strong>on</strong>g>of</str<strong>on</strong>g> the storm<br />
experienced little direct stress, virtually no prop-<br />
erty damage, and <strong>on</strong>ly the inc<strong>on</strong>venience <str<strong>on</strong>g>of</str<strong>on</strong>g> a<br />
short period without basic services.<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> hurricane provided an opportunity for<br />
investig<strong>at</strong>ing aspects <str<strong>on</strong>g>of</str<strong>on</strong>g> the memory-stress rela-<br />
ti<strong>on</strong>ship th<strong>at</strong> differ in important ways from those<br />
found in most <str<strong>on</strong>g>of</str<strong>on</strong>g> the prior studies. It provided a<br />
soluti<strong>on</strong> to the problem <str<strong>on</strong>g>of</str<strong>on</strong>g> defining the low stress<br />
c<strong>on</strong>trol group. <str<strong>on</strong>g>The</str<strong>on</strong>g> same event and the same<br />
memory test could be given to children who<br />
experienced different levels <str<strong>on</strong>g>of</str<strong>on</strong>g> stress. All partici-<br />
pants prepared in the same way for the hurricane,<br />
but some experienced an extremely traum<strong>at</strong>ic<br />
event whereas others experienced <strong>on</strong>ly a heavy<br />
rain storm. Sec<strong>on</strong>d, this research has the advan-<br />
tage th<strong>at</strong> stress could be objectively defined<br />
according to the severity <str<strong>on</strong>g>of</str<strong>on</strong>g> the storm in each<br />
child's neighborhood. This objective measure<br />
could also be correl<strong>at</strong>ed with more typical subjec-<br />
tive r<strong>at</strong>ings <str<strong>on</strong>g>of</str<strong>on</strong>g> stress. Third, the hurricane cre<strong>at</strong>ed<br />
a wide range <str<strong>on</strong>g>of</str<strong>on</strong>g> stress, from low to very severe,<br />
thereby allowing us a much broader view <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />
memory-stress rel<strong>at</strong>i<strong>on</strong>ship than is usually pos-<br />
sible. This is particularly important given the<br />
discrepant findings across studies with respect to<br />
the facilit<strong>at</strong>ing versus debilit<strong>at</strong>ing effects <str<strong>on</strong>g>of</str<strong>on</strong>g> stress<br />
<strong>on</strong> memory. Fourth, the differential stress because<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> the storm occurred without regard for social<br />
and ec<strong>on</strong>omic st<strong>at</strong>us. <str<strong>on</strong>g>The</str<strong>on</strong>g> average home value for<br />
families <str<strong>on</strong>g>of</str<strong>on</strong>g> low, moder<strong>at</strong>e, and severe stress did<br />
not differ (see Results and Discussi<strong>on</strong>, Table 1).<br />
MEMORY FOR A NATURAL DISASTER 311<br />
Thus, our approach combines many <str<strong>on</strong>g>of</str<strong>on</strong>g> the advan-<br />
tages <str<strong>on</strong>g>of</str<strong>on</strong>g> the two prior approaches to the rel<strong>at</strong>i<strong>on</strong>-<br />
ship between memory and stress, without being<br />
subject to the major limit<strong>at</strong>i<strong>on</strong>s associ<strong>at</strong>ed with<br />
each.<br />
Prior research <strong>on</strong> effects <str<strong>on</strong>g>of</str<strong>on</strong>g> n<strong>at</strong>ural disasters<br />
and trauma <strong>on</strong> children has focused primarily <strong>on</strong><br />
the n<strong>at</strong>ure <str<strong>on</strong>g>of</str<strong>on</strong>g> the stress and understanding its l<strong>on</strong>g<br />
term effects <strong>on</strong> psychological and social adjust-<br />
ment. Traum<strong>at</strong>ic events investig<strong>at</strong>ed have in-<br />
cluded witnessing a sniper <strong>at</strong>tack (Pynoos et al.,<br />
1987), a kidnapping (Terr, 1987), the murder <str<strong>on</strong>g>of</str<strong>on</strong>g> a<br />
parent (Malmquist, 1986), a dam collapse (Green,<br />
Korol, Grace, Vary, Le<strong>on</strong>ard, Gleser, & Smits<strong>on</strong>-<br />
Cohen, 1991), a severe flood (Earls, Smith,<br />
Reich, & Jung, 1988), and a bushfire (McFarlane,<br />
1987). Research <strong>on</strong> the effects <str<strong>on</strong>g>of</str<strong>on</strong>g> n<strong>at</strong>ural disasters<br />
has highlighted the l<strong>on</strong>g lasting effects <str<strong>on</strong>g>of</str<strong>on</strong>g> these<br />
traumas (e.g., Earls et al., 1988; McFarlane,<br />
1987). L<strong>on</strong>igan, Shann<strong>on</strong>, Taylor, Finch, and<br />
SaUce (1994) reported posttraum<strong>at</strong>ic stress disor-<br />
der (PTSD) symptoms in 5,687 children 3 m<strong>on</strong>ths<br />
after they were exposed to hurricane Hugo.<br />
Similarly, Vernberg, La Greca, Silverman, and<br />
Prinstein (1996) analyzed 568 elementary school<br />
children's reacti<strong>on</strong>s to Hurricane Andrew. More<br />
than 55% showed moder<strong>at</strong>e to very severe levels<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> PTSD symptoms 3 m<strong>on</strong>ths l<strong>at</strong>er.<br />
In spite <str<strong>on</strong>g>of</str<strong>on</strong>g> the recent surge <str<strong>on</strong>g>of</str<strong>on</strong>g> interest in the<br />
rel<strong>at</strong>i<strong>on</strong>ship between stress and memory, n<strong>on</strong>e <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
the research to d<strong>at</strong>e <strong>on</strong> the effects <str<strong>on</strong>g>of</str<strong>on</strong>g> n<strong>at</strong>ural<br />
disasters has focused <strong>on</strong> children's memories for<br />
these disasters. Thus, the present study provides a<br />
bridge between the questi<strong>on</strong>s regarding memory<br />
and stress typically assessed in eyewitness re-<br />
search (in which stressful events such as visits to<br />
the doctor or dentist are assessed) and the n<strong>at</strong>ural<br />
disaster research (in which a much more devast<strong>at</strong>-<br />
ing event with more far reaching c<strong>on</strong>sequences<br />
cre<strong>at</strong>es a much wider range <str<strong>on</strong>g>of</str<strong>on</strong>g> stress).<br />
In the present study, we assessed memory for<br />
hurricane-rel<strong>at</strong>ed events in 3- and 4-year-old<br />
children and their mothers, 2 to 6 m<strong>on</strong>ths follow-<br />
ing the hurricane. We chose to focus <strong>on</strong> preschool-<br />
ers because <str<strong>on</strong>g>of</str<strong>on</strong>g> the c<strong>on</strong>troversy in the liter<strong>at</strong>ure<br />
over the memory abilities <str<strong>on</strong>g>of</str<strong>on</strong>g> children this young.<br />
<str<strong>on</strong>g>Stress</str<strong>on</strong>g> was defined as the degree <str<strong>on</strong>g>of</str<strong>on</strong>g> storm expo-<br />
sure in each child's neighborhood. Storm expo-<br />
sure was coded as high, moder<strong>at</strong>e, or low <strong>on</strong> the<br />
basis <str<strong>on</strong>g>of</str<strong>on</strong>g> the amount and type <str<strong>on</strong>g>of</str<strong>on</strong>g> damage to the<br />
home. Structured interviews were given to chil-
312 BAHRICK, PARKER, FIVUSH, AND LEVITT<br />
dren focusing <strong>on</strong> memory for three time periods:<br />
the prehurricane prepar<strong>at</strong>i<strong>on</strong>s, the hurricane it-<br />
self, and the posthurricane recovery period. Given<br />
the previous liter<strong>at</strong>ure, we assumed th<strong>at</strong> children<br />
would be able to recall the hurricane experience.<br />
More critical, however, was how the level <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
stress might influence the amount and type <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
inform<strong>at</strong>i<strong>on</strong> reported.<br />
Method<br />
Participants<br />
One hundred children aged 3 (n = 39; M =<br />
1,357.3, SD = 133.0 days) and 4 years (n = 61;<br />
M = 1,738.7, SD = 232.3 days) <strong>at</strong> the time <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
testing and their mothers were recruited for<br />
particip<strong>at</strong>i<strong>on</strong> through local preschools. <str<strong>on</strong>g>The</str<strong>on</strong>g>re were<br />
a total <str<strong>on</strong>g>of</str<strong>on</strong>g> 51 boys and 49 girls (21 boys and 18<br />
girls aged 3 years and 30 boys and 31 girls aged 4<br />
years). Only the d<strong>at</strong>a <str<strong>on</strong>g>of</str<strong>on</strong>g> the children are reported<br />
here. Four additi<strong>on</strong>al children were interviewed<br />
but their d<strong>at</strong>a were rejected because the tape<br />
recording was incomprehensible (n = 1), the<br />
equipment failed (n = 2), or the experimenter<br />
failed to follow the interview protocol (n = 1).<br />
Children were recruited from an area spanning a<br />
30-mile distance in Dade County, <strong>Florida</strong>, to<br />
include low (n = 18), moder<strong>at</strong>e (n = 42), and<br />
severe (n = 40) hurricane exposure (see Results<br />
and Discussi<strong>on</strong>, Storm Severity Classific<strong>at</strong>i<strong>on</strong>, for<br />
details). Children were primarily from middle-<br />
class families with incomes between $40,000 and<br />
$80,000, and the mothers and f<strong>at</strong>hers had an<br />
average <str<strong>on</strong>g>of</str<strong>on</strong>g> 15.4 (SD = 2.4) and 15.4 (SD = 2.6)<br />
years <str<strong>on</strong>g>of</str<strong>on</strong>g> educ<strong>at</strong>i<strong>on</strong>, respectively. <str<strong>on</strong>g>The</str<strong>on</strong>g> ethnic back-<br />
ground <str<strong>on</strong>g>of</str<strong>on</strong>g> the children in the sample was 69.4%<br />
White, 20.4% Hispanic, 3.1% African American,<br />
and 7.1% other. All families had we<strong>at</strong>hered the<br />
storm in their own homes.<br />
Child Interview<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> interview (see Appendix) took place in a<br />
single sessi<strong>on</strong> and comprised four parts: ques-<br />
ti<strong>on</strong>s about the storm experience in general, the<br />
prehurricane prepar<strong>at</strong>i<strong>on</strong>s (Time Period 1; T1),<br />
the storm itself (Time Period 2; T2), and the<br />
posthurricane recovery (Time Period 3; T3). <str<strong>on</strong>g>The</str<strong>on</strong>g><br />
interviewer initially asked an open-ended ques-<br />
ti<strong>on</strong> about the general storm experience (e.g.,<br />
"Tell me everything you can remember about the<br />
hurricane?") to elicit as much free recall as<br />
possible. This was followed by n<strong>on</strong>directive<br />
prompts, such as "Wh<strong>at</strong> else?" and "Tell me<br />
more." Next, the interviewer elicited memory<br />
about each <str<strong>on</strong>g>of</str<strong>on</strong>g> the three time periods separ<strong>at</strong>ely.<br />
Present<strong>at</strong>i<strong>on</strong> order for each <str<strong>on</strong>g>of</str<strong>on</strong>g> these parts was<br />
counterbalanced across participants. Each <str<strong>on</strong>g>of</str<strong>on</strong>g> these<br />
parts began with a general open-ended questi<strong>on</strong><br />
and was followed by a standardized series <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
increasingly more specific questi<strong>on</strong>s. (e.g., "Wh<strong>at</strong><br />
did you and your family do to get ready before<br />
the hurricane came?"). When n<strong>on</strong>directive<br />
prompts failed to elicit more inform<strong>at</strong>i<strong>on</strong>, c<strong>at</strong>-<br />
egory prompts were given (e.g., "Wh<strong>at</strong> did your<br />
mommy and daddy do inside and outside the<br />
house to get ready for the hurricane?"). This was<br />
followed by specific item prompts (e.g., "Did<br />
your mommy and daddy do anything to the<br />
windows? Wh<strong>at</strong>?"). <str<strong>on</strong>g>The</str<strong>on</strong>g> interview took approxi-<br />
m<strong>at</strong>ely 30 min. Children then provided a brief<br />
subjective stress r<strong>at</strong>ing. <str<strong>on</strong>g>The</str<strong>on</strong>g>y were assisted in<br />
r<strong>at</strong>ing <strong>on</strong> a 4-point scale how scared versus how<br />
good they and their mother felt during each <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />
three periods <str<strong>on</strong>g>of</str<strong>on</strong>g> the storm.<br />
Parent Questi<strong>on</strong>naire Inform<strong>at</strong>i<strong>on</strong><br />
Mothers completed eight questi<strong>on</strong>naires includ-<br />
ing two measures <str<strong>on</strong>g>of</str<strong>on</strong>g> child behavior and six<br />
designed to measure the mother's stress and<br />
behavior. <str<strong>on</strong>g>The</str<strong>on</strong>g> child measures included a modified<br />
versi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>The</str<strong>on</strong>g> Child Frederick Reacti<strong>on</strong> Index<br />
Form C (Frederick, 1985), c<strong>on</strong>sisting <str<strong>on</strong>g>of</str<strong>on</strong>g> 20 items<br />
designed to assess PTSD symptom<strong>at</strong>ology in<br />
children and the "Child Well-Being Survey,"<br />
c<strong>on</strong>sisting <str<strong>on</strong>g>of</str<strong>on</strong>g> a 35-item questi<strong>on</strong>naire th<strong>at</strong> we<br />
developed to assess changes in ordinary behav-<br />
iors (such as sleeping, e<strong>at</strong>ing, <strong>at</strong>tenti<strong>on</strong>, and<br />
emoti<strong>on</strong>ality) as a result <str<strong>on</strong>g>of</str<strong>on</strong>g> a traum<strong>at</strong>ic event. Six<br />
measures were also included to assess the moth-<br />
er's level <str<strong>on</strong>g>of</str<strong>on</strong>g> stress and changes in her behavior as<br />
a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the hurricane. Because n<strong>on</strong>e <str<strong>on</strong>g>of</str<strong>on</strong>g> these<br />
additi<strong>on</strong>al measures <str<strong>on</strong>g>of</str<strong>on</strong>g> stress for the mother or<br />
child significantly predicted recall (either linearly<br />
or quadr<strong>at</strong>ically), they are not discussed further.<br />
In additi<strong>on</strong>, the mother filled out a questi<strong>on</strong>-<br />
naire assessing demographic inform<strong>at</strong>i<strong>on</strong>, infor-<br />
m<strong>at</strong>i<strong>on</strong> about the c<strong>on</strong>sequences <str<strong>on</strong>g>of</str<strong>on</strong>g> the hurricane
to her home and disrupti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> services, and<br />
rehearsal <str<strong>on</strong>g>of</str<strong>on</strong>g> hurricane-rel<strong>at</strong>ed events. <str<strong>on</strong>g>The</str<strong>on</strong>g> re-<br />
hearsal questi<strong>on</strong>naire required the mother to r<strong>at</strong>e<br />
how <str<strong>on</strong>g>of</str<strong>on</strong>g>ten the family talked about the hurricane<br />
in the presence <str<strong>on</strong>g>of</str<strong>on</strong>g> the child during the first week,<br />
the most recent week, and the period in between.<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> r<strong>at</strong>ings were made <strong>on</strong> a 3-point scale ranging<br />
from 0 (n<strong>on</strong>e) to 1 (<strong>on</strong>ce a day) or 2 (several<br />
times a day) and were broken down according to<br />
time periods (T1, T2, and T3). Four separ<strong>at</strong>e<br />
questi<strong>on</strong>s were asked including, "How <str<strong>on</strong>g>of</str<strong>on</strong>g>ten did<br />
your family talk to your child ..." or " ... in<br />
fr<strong>on</strong>t <str<strong>on</strong>g>of</str<strong>on</strong>g> your child .... " about the prehurricane<br />
prepar<strong>at</strong>i<strong>on</strong>s, the hurricane itself, and the hurri-<br />
cane afterm<strong>at</strong>h? A composite score was obtained<br />
by averaging across all questi<strong>on</strong>s.<br />
Procedure<br />
Testing took place in 23 preschools in Dade<br />
County. Preschools were chosen so th<strong>at</strong> the<br />
envir<strong>on</strong>ment would be neutral with respect to<br />
retrieval cues, given th<strong>at</strong> the homes <str<strong>on</strong>g>of</str<strong>on</strong>g> many <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
the children were still damaged. Following a<br />
warm-up period designed to establish rapport, the<br />
interview began. It typically lasted about 30 min.<br />
Afterwards, the child was given a small toy or a<br />
sticker. Seven different trained interviewers (all<br />
females between the ages <str<strong>on</strong>g>of</str<strong>on</strong>g> 18 and 35) each<br />
tested approxim<strong>at</strong>ely <strong>on</strong>e seventh <str<strong>on</strong>g>of</str<strong>on</strong>g> the children.<br />
Interviews were tape recorded and then tran-<br />
scribed verb<strong>at</strong>im. After the child's interview was<br />
completed, the mother was l<strong>at</strong>er interviewed<br />
regarding her memory <str<strong>on</strong>g>of</str<strong>on</strong>g> the storm (d<strong>at</strong>a not<br />
presented here) and was then given a hurricane<br />
severity questi<strong>on</strong>naire. This scale was designed<br />
to objectify the degree <str<strong>on</strong>g>of</str<strong>on</strong>g> storm exposure and<br />
served as a basis for classifying the participants<br />
into high, moder<strong>at</strong>e, or low storm severity groups.<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> mothers also completed a questi<strong>on</strong>naire<br />
about the extent <str<strong>on</strong>g>of</str<strong>on</strong>g> damage to their homes and the<br />
n<strong>at</strong>ure and dur<strong>at</strong>i<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> interrupti<strong>on</strong> in basic<br />
services, rehearsal inform<strong>at</strong>i<strong>on</strong> for each time<br />
period separ<strong>at</strong>ely, and demographics.<br />
Coding <str<strong>on</strong>g>of</str<strong>on</strong>g> <strong>Memory</strong> Interviews<br />
<strong>Memory</strong> interviews were coded according to a<br />
propositi<strong>on</strong>al analysis th<strong>at</strong> took place in four<br />
stages.<br />
MEMORY FOR A NATURAL DISASTER 313<br />
Coding for Propositi<strong>on</strong>al Units<br />
and Hurricane Relevance<br />
All inform<strong>at</strong>i<strong>on</strong> was broken down into proposi-<br />
ti<strong>on</strong>al units. Apropositi<strong>on</strong>al unit was defined as a<br />
clause th<strong>at</strong> c<strong>on</strong>tained a subject and a verb, either<br />
explicit or implied. If two subjects were named,<br />
the clause was parsed into two propositi<strong>on</strong>s ("My<br />
brother/and my mother screamed loud"). If two<br />
acti<strong>on</strong> verbs were given for <strong>on</strong>e subject, the c<strong>on</strong>-<br />
tent was also parsed into two propositi<strong>on</strong>s ("My<br />
brother screamed loud/then ran downstairs").<br />
Next, the propositi<strong>on</strong>al units were scored as to<br />
their relevance to the hurricane event. This was<br />
the primary dependent variable. All inform<strong>at</strong>i<strong>on</strong><br />
about and rel<strong>at</strong>ed to the hurricane, getting ready<br />
for it, and the afterm<strong>at</strong>h period were judged<br />
relevant. We also judged as relevant, general<br />
knowledge, opini<strong>on</strong>s, and background informa-<br />
ti<strong>on</strong> th<strong>at</strong> helped in understanding the hurricane<br />
events ("My uncle, who lived in New York,<br />
stayed with us; .... <str<strong>on</strong>g>The</str<strong>on</strong>g> name <str<strong>on</strong>g>of</str<strong>on</strong>g> the hurricane was<br />
Andrew; .... Bryan Norcross told every<strong>on</strong>e wh<strong>at</strong><br />
to do."). Propositi<strong>on</strong>s were deemed irrelevant<br />
under the following c<strong>at</strong>egories: <str<strong>on</strong>g>of</str<strong>on</strong>g>f-topic talk ("I<br />
got a present yesterday."), c<strong>on</strong>vers<strong>at</strong>i<strong>on</strong>al fillers<br />
("You know wh<strong>at</strong>?"), false starts ("I <strong>on</strong>ly.., we<br />
were hiding... "), questi<strong>on</strong>ing the experimenter<br />
("How did you know th<strong>at</strong>?"), uncodable informa-<br />
ti<strong>on</strong> (cannot understand wh<strong>at</strong> the child is saying),<br />
refusals ("Th<strong>at</strong>'s all I'm g<strong>on</strong>na tell you."), repeti-<br />
ti<strong>on</strong>s (the child repe<strong>at</strong>s in gist a prior st<strong>at</strong>ement),<br />
c<strong>on</strong>firm<strong>at</strong>i<strong>on</strong> and denials (e.g., "yes," "nope"),<br />
and st<strong>at</strong>ements corrected l<strong>at</strong>er (child corrects own<br />
st<strong>at</strong>ement in another c<strong>on</strong>vers<strong>at</strong>i<strong>on</strong>al turn). Inter-<br />
r<strong>at</strong>er reliability was established for this stage<br />
am<strong>on</strong>g three independent coders (<strong>on</strong>e primary<br />
and two sec<strong>on</strong>dary) <strong>on</strong> 27 <str<strong>on</strong>g>of</str<strong>on</strong>g> the 100 transcripts.<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> transcripts were system<strong>at</strong>ically selected across<br />
interviewers to represent approxim<strong>at</strong>ely an equal<br />
number <str<strong>on</strong>g>of</str<strong>on</strong>g> 3- and 4-year-olds and children <str<strong>on</strong>g>of</str<strong>on</strong>g> low,<br />
moder<strong>at</strong>e, and high storm exposure. Reliability<br />
coding was d<strong>on</strong>e <strong>on</strong> <strong>on</strong>e quartile (randomly<br />
selected) <str<strong>on</strong>g>of</str<strong>on</strong>g> each transcript. After extensive train-<br />
ing with the coding system, each coder indepen-<br />
dently coded the selected quartile for 27 tran-<br />
scripts. Agreements between the primary and<br />
each sec<strong>on</strong>dary coder were counted when the<br />
phrase was parsed identically and was c<strong>on</strong>sistent<br />
in whether it was identified as hurricane relevant<br />
or irrelevant. <str<strong>on</strong>g>The</str<strong>on</strong>g> percentage <str<strong>on</strong>g>of</str<strong>on</strong>g> agreement be-
314 BAHRICK, PARKER, FIVUSH, AND LEVITT<br />
tween Coder 2 and the primary coder was 89.3%<br />
and between Coder 3 and the primary coder was<br />
90.6%. After establishing this level <str<strong>on</strong>g>of</str<strong>on</strong>g> reliability,<br />
the three coders each coded approxim<strong>at</strong>ely <strong>on</strong>e<br />
third <str<strong>on</strong>g>of</str<strong>on</strong>g> the transcripts independently. Next, all<br />
hurricane-relevant propositi<strong>on</strong>s were coded<br />
according to four subclassific<strong>at</strong>i<strong>on</strong>s (major c<strong>on</strong>-<br />
tent c<strong>at</strong>egories, prepar<strong>at</strong>i<strong>on</strong>--damage-repair, time<br />
period, and type <str<strong>on</strong>g>of</str<strong>on</strong>g> memory) as well as for<br />
elabor<strong>at</strong>i<strong>on</strong>s.<br />
Coding for C<strong>on</strong>tent<br />
All hurricane-relevant propositi<strong>on</strong>s were coded<br />
for c<strong>on</strong>tent according to three major c<strong>at</strong>egories:<br />
acti<strong>on</strong>s, descripti<strong>on</strong>s, or internal st<strong>at</strong>es. Each<br />
propositi<strong>on</strong> was also coded as to whether or not it<br />
was relevant to hurricane prepar<strong>at</strong>i<strong>on</strong>s, damage,<br />
or repair.<br />
Three major c<strong>on</strong>tent c<strong>at</strong>egories. <str<strong>on</strong>g>The</str<strong>on</strong>g>re were<br />
three major c<strong>on</strong>tent c<strong>at</strong>egories: acti<strong>on</strong>s, descrip-<br />
ti<strong>on</strong>s, and internal st<strong>at</strong>es. An acti<strong>on</strong> was defined<br />
as a propositi<strong>on</strong> th<strong>at</strong> had a clear agent <str<strong>on</strong>g>of</str<strong>on</strong>g> acti<strong>on</strong>.<br />
This included physical activity or moti<strong>on</strong>, direct<br />
or implied, and any neg<strong>at</strong>i<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> activity as l<strong>on</strong>g as<br />
the agent <str<strong>on</strong>g>of</str<strong>on</strong>g> acti<strong>on</strong> was the subject (e.g., "My<br />
mommy cooked things <strong>on</strong> the grill; .... It blew the<br />
tiles <str<strong>on</strong>g>of</str<strong>on</strong>g>f my ro<str<strong>on</strong>g>of</str<strong>on</strong>g>"). Most other propositi<strong>on</strong>s<br />
without a clear agent <str<strong>on</strong>g>of</str<strong>on</strong>g> acti<strong>on</strong> were coded as<br />
descripti<strong>on</strong>s. <str<strong>on</strong>g>The</str<strong>on</strong>g>se were defined as propositi<strong>on</strong>s<br />
th<strong>at</strong> menti<strong>on</strong>ed wh<strong>at</strong> things looked like or ap-<br />
peared to be and did not specify an agent. Use <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
verbs such as be, seem, appear, and have typically<br />
characterized descripti<strong>on</strong>s ("<str<strong>on</strong>g>The</str<strong>on</strong>g> pool was all<br />
dirty; .... Abig boom was heard"). When physical<br />
moti<strong>on</strong> was implied but no causal agent was<br />
specified, the propositi<strong>on</strong> was coded as a descrip-<br />
ti<strong>on</strong> ("My house got blown away"). Propositi<strong>on</strong>s<br />
were coded as internal st<strong>at</strong>es if they expressed<br />
emoti<strong>on</strong> or affect (scared, happy, laughing, cry-<br />
ing, or glad) or if they menti<strong>on</strong>ed the following<br />
internal st<strong>at</strong>es: think, hungry, tired, know, dream,<br />
luck~y, have a boo boo, remember, and want ("I<br />
wanted to fix it"). Quotes were also coded but<br />
occurred too infrequently and by <strong>on</strong>ly a few<br />
children and were therefore l<strong>at</strong>er dropped from<br />
all analyses. Interr<strong>at</strong>er reliability was calcul<strong>at</strong>ed<br />
between two trained coders for <strong>on</strong>e quartile <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
each <str<strong>on</strong>g>of</str<strong>on</strong>g> 27 transcripts. Percentage <str<strong>on</strong>g>of</str<strong>on</strong>g> agreement<br />
averaged 95.4%.<br />
Prepar<strong>at</strong>i<strong>on</strong>, damage, and repair c<strong>on</strong>tent.<br />
Each propositi<strong>on</strong> was further c<strong>at</strong>egorized as to<br />
whether it described prepar<strong>at</strong>i<strong>on</strong> ("We bought<br />
lots <str<strong>on</strong>g>of</str<strong>on</strong>g> food"), damage ("<str<strong>on</strong>g>The</str<strong>on</strong>g> trees were all<br />
broken"), or repair and cleanup ("We picked up<br />
the shingles") or whether it was not rel<strong>at</strong>ed to<br />
these activities. Interr<strong>at</strong>er reliability was estab-<br />
lished between the same two coders as before,<br />
and percentage <str<strong>on</strong>g>of</str<strong>on</strong>g> agreement was 96.9%.<br />
Coding for Elabor<strong>at</strong>i<strong>on</strong>s<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> same coders as above also counted all<br />
elabor<strong>at</strong>i<strong>on</strong>s. Elabor<strong>at</strong>i<strong>on</strong>s were defined as any<br />
adjective, adverb, possessive (except pr<strong>on</strong>ouns),<br />
modifying phrase ("near the pool," "by him-<br />
self, .... in the hurricane"), and rel<strong>at</strong>ive clause<br />
(this, th<strong>at</strong>, those, then, there, here, and when). For<br />
example, the propositi<strong>on</strong> "a loud noise sounded<br />
by the car" was coded as having two elabora-<br />
ti<strong>on</strong>s, and the propositi<strong>on</strong> "Nine people <str<strong>on</strong>g>of</str<strong>on</strong>g> my<br />
family stayed in the downstairs b<strong>at</strong>hroom" was<br />
coded as having three elabor<strong>at</strong>i<strong>on</strong>s. If the same<br />
elabor<strong>at</strong>i<strong>on</strong> was repe<strong>at</strong>ed, it was counted <strong>on</strong>ly<br />
<strong>on</strong>ce ("We had a lot, a lot <str<strong>on</strong>g>of</str<strong>on</strong>g> tiles down"). A<br />
given propositi<strong>on</strong> might c<strong>on</strong>tain no elabor<strong>at</strong>i<strong>on</strong>s<br />
or as many as four or five. <str<strong>on</strong>g>The</str<strong>on</strong>g> number <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
elabor<strong>at</strong>i<strong>on</strong>s per propositi<strong>on</strong> was calcul<strong>at</strong>ed for<br />
each child. Interr<strong>at</strong>er reliability was established<br />
as above, and each coder counted the number <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
elabor<strong>at</strong>i<strong>on</strong>s for each propositi<strong>on</strong>. Percentage <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
agreement averaged 99%.<br />
Coding for Time Period and Specificity<br />
All propositi<strong>on</strong>s were also coded for temporal<br />
and structural inform<strong>at</strong>i<strong>on</strong>.<br />
Time period. Each propositi<strong>on</strong> was r<strong>at</strong>ed as<br />
to whether the c<strong>on</strong>tent was relevant to Time<br />
Period 1 (T1), the hurricane prepar<strong>at</strong>i<strong>on</strong> period<br />
("We brought in all the plants"), Time Period 2<br />
(T2), the hurricane itself ("We were very hot in<br />
the closet"), Time Period 3 (T3), the hurricane<br />
recovery period ("and there were no carto<strong>on</strong>s"),<br />
or whether it was unrel<strong>at</strong>ed to a given time period<br />
(T4, "<str<strong>on</strong>g>The</str<strong>on</strong>g> hurricane was called Andrew"). Often,<br />
the coder had to use c<strong>on</strong>text to judge the time<br />
period.<br />
Sp<strong>on</strong>taneous versus cued recall. <str<strong>on</strong>g>The</str<strong>on</strong>g> inter-<br />
view was structured to facilit<strong>at</strong>e investig<strong>at</strong>i<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
sp<strong>on</strong>taneous versus prompted recall because it
was composed <str<strong>on</strong>g>of</str<strong>on</strong>g> open-ended questi<strong>on</strong>s followed<br />
by increasingly more specific prompts. Proposi-<br />
ti<strong>on</strong>s were thus coded according to the specificity<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> the interviewer's questi<strong>on</strong> th<strong>at</strong> elicited the<br />
propositi<strong>on</strong>. Four levels <str<strong>on</strong>g>of</str<strong>on</strong>g> prompting were de-<br />
fined: (1) the general open-ended questi<strong>on</strong> was<br />
defined as the most general level, (2) the tempo-<br />
ral open-ended questi<strong>on</strong>s were defined as the<br />
next most general level, (3) c<strong>at</strong>egory prompts<br />
were defined as more specific, and (4) the most<br />
specific level was a specific item prompt. For the<br />
purpose <str<strong>on</strong>g>of</str<strong>on</strong>g> investig<strong>at</strong>ing sp<strong>on</strong>taneous versus<br />
prompted recall, all propositi<strong>on</strong>s elicited by open-<br />
ended questi<strong>on</strong>s (Levels 1 and 2) were classified<br />
as sp<strong>on</strong>taneous memory, and all propositi<strong>on</strong>s<br />
elicited by c<strong>at</strong>egory and specific item prompts<br />
(Levels 3 and 4) were classified as prompted<br />
recall.<br />
Interr<strong>at</strong>er reliability for time period and level<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> prompting was calcul<strong>at</strong>ed between two new<br />
r<strong>at</strong>ers <strong>on</strong> <strong>on</strong>e quartile <str<strong>on</strong>g>of</str<strong>on</strong>g> 27 interviews as before.<br />
Percentage <str<strong>on</strong>g>of</str<strong>on</strong>g> agreement averaged 93.6% for<br />
time period and 99.6% for level <str<strong>on</strong>g>of</str<strong>on</strong>g> prompting.<br />
Results and Discussi<strong>on</strong><br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> major analyses explored the way in which<br />
the amount recalled about the hurricane was<br />
influenced by stress and child age. <str<strong>on</strong>g>Stress</str<strong>on</strong>g> was<br />
defined as storm severity and classified as high,<br />
moder<strong>at</strong>e, or low (see below). Further, the total<br />
number <str<strong>on</strong>g>of</str<strong>on</strong>g> propositi<strong>on</strong>s gener<strong>at</strong>ed by children<br />
and the degree <str<strong>on</strong>g>of</str<strong>on</strong>g> elabor<strong>at</strong>i<strong>on</strong> provided were<br />
analyzed as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> time period, type <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
recall (sp<strong>on</strong>taneous vs. prompted), and type <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
c<strong>on</strong>tent, as well as age and storm severity.<br />
Finally, the effects <str<strong>on</strong>g>of</str<strong>on</strong>g> factors including rehearsal,<br />
retenti<strong>on</strong> interval, family income, and parental<br />
educ<strong>at</strong>i<strong>on</strong> <strong>on</strong> the main findings were examined.<br />
Storm Severity Classific<strong>at</strong>i<strong>on</strong><br />
Participants were classified into <strong>on</strong>e <str<strong>on</strong>g>of</str<strong>on</strong>g> three<br />
stress levels (high, moder<strong>at</strong>e, or low storm sever-<br />
ity) depending <strong>on</strong> the mother's resp<strong>on</strong>se to a<br />
7-point questi<strong>on</strong>naire regarding storm exposure.<br />
If the storm actually penetr<strong>at</strong>ed the perimeter <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
the home while the family was inside, a r<strong>at</strong>ing <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
high severity was given. <str<strong>on</strong>g>The</str<strong>on</strong>g>se occupants experi-<br />
enced breaking glass, flying objects, and physical<br />
MEMORY FOR A NATURAL DISASTER 315<br />
danger. Most were required to flee from <strong>on</strong>e room<br />
to another to avoid injury. A r<strong>at</strong>ing <str<strong>on</strong>g>of</str<strong>on</strong>g> moder<strong>at</strong>e<br />
severity was given if during the storm, a gre<strong>at</strong><br />
deal <str<strong>on</strong>g>of</str<strong>on</strong>g> property damage occurred, but the perim-<br />
eter <str<strong>on</strong>g>of</str<strong>on</strong>g> the home was left mostly intact during the<br />
storm itself. Although the occupants <str<strong>on</strong>g>of</str<strong>on</strong>g>ten feared<br />
physical harm, there was little actual physical<br />
danger in the home. Much <str<strong>on</strong>g>of</str<strong>on</strong>g> the interior property<br />
damage occurred after the storm as w<strong>at</strong>er c<strong>on</strong>tin-<br />
ued to leak in through holes in the ro<str<strong>on</strong>g>of</str<strong>on</strong>g> and so<br />
forth. Both groups experienced extensive clean-up<br />
periods following the storm. Finally, a r<strong>at</strong>ing <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
low severity was given if families experienced<br />
little or no damage to their homes and, <strong>at</strong> most,<br />
found debris in their yards. It is important to note<br />
th<strong>at</strong> although the r<strong>at</strong>ings <str<strong>on</strong>g>of</str<strong>on</strong>g> stress are objective,<br />
they are also rel<strong>at</strong>ive to the range <str<strong>on</strong>g>of</str<strong>on</strong>g> stress<br />
experienced by children in this study, and thus, it<br />
cannot be determined whether moder<strong>at</strong>e stress in<br />
our study is equivalent to high, moder<strong>at</strong>e, or low<br />
stress in another study. <str<strong>on</strong>g>The</str<strong>on</strong>g>re were approxim<strong>at</strong>ely<br />
equal numbers <str<strong>on</strong>g>of</str<strong>on</strong>g> children who fell into the<br />
moder<strong>at</strong>e (n = 42) and high severity (n = 40)<br />
groups, with fewer in the low severity c<strong>on</strong>diti<strong>on</strong><br />
(n = 18). Table 1 presents the c<strong>on</strong>sequences <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
the storm in terms <str<strong>on</strong>g>of</str<strong>on</strong>g> damage to the home and<br />
c<strong>on</strong>tents and interrupti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> basic services for<br />
children <str<strong>on</strong>g>of</str<strong>on</strong>g> the three stress levels, as well as<br />
analyses <str<strong>on</strong>g>of</str<strong>on</strong>g> variance (ANOVAs) assessing group<br />
differences for each variable. It is important to<br />
notice th<strong>at</strong> the average value <str<strong>on</strong>g>of</str<strong>on</strong>g> the home did not<br />
differ as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> hurricane severity. How-<br />
ever, property damage and days without services<br />
differed significantly across groups in the direc-<br />
ti<strong>on</strong> expected, corrobor<strong>at</strong>ing our hurricane sever-<br />
ity measure.<br />
Accuracy<br />
Although accuracy <str<strong>on</strong>g>of</str<strong>on</strong>g> free recall for such a<br />
subjectively experienced, temporally extended<br />
n<strong>at</strong>ural event cannot be objectively assessed as in<br />
some studies <str<strong>on</strong>g>of</str<strong>on</strong>g> memory and stress in which<br />
records are available (e.g., visits to the doctor or<br />
dentist; Goodman et al., 1987; Goodman, Hirsch-<br />
man, et al., 1991), we nevertheless asked 25% <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
the mothers to judge their child's accuracy from a<br />
transcript <str<strong>on</strong>g>of</str<strong>on</strong>g> the child's interview. Mothers were<br />
asked to mark any c<strong>on</strong>tent the child reported th<strong>at</strong><br />
was not completely true and to explain fully.<br />
Only 9 <str<strong>on</strong>g>of</str<strong>on</strong>g> the 25 transcripts were returned (many
316 BAHRICK, PARKER, FIVUSH, AND LEVITT<br />
o~<br />
r~<br />
t~ eq t~<br />
t",l oO '~-<br />
,.~m ~ ~- ~ ~<br />
._.i" "-"-<br />
families were understandably more c<strong>on</strong>cerned<br />
with rebuilding their homes)• Still, these 9 tran-<br />
scripts can give us a general idea <str<strong>on</strong>g>of</str<strong>on</strong>g> the difficulty<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> judging accuracy and the type <str<strong>on</strong>g>of</str<strong>on</strong>g> errors<br />
mothers noted• Overall, the vast majority <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
propositi<strong>on</strong>s were deemed accur<strong>at</strong>e. <str<strong>on</strong>g>The</str<strong>on</strong>g> mean<br />
number <str<strong>on</strong>g>of</str<strong>on</strong>g> propositi<strong>on</strong>s c<strong>on</strong>taining errors was<br />
<strong>on</strong>ly .033 per child, with a range <str<strong>on</strong>g>of</str<strong>on</strong>g> .013 to .078.<br />
We developed a rough classific<strong>at</strong>i<strong>on</strong> system for<br />
describing the range and types <str<strong>on</strong>g>of</str<strong>on</strong>g> errors children<br />
typically made. Many <str<strong>on</strong>g>of</str<strong>on</strong>g> the items marked inaccu-<br />
r<strong>at</strong>e by parents nevertheless reflected memories<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> events th<strong>at</strong> actually occurred. <str<strong>on</strong>g>The</str<strong>on</strong>g>y included<br />
three c<strong>at</strong>egories: (1) time distorti<strong>on</strong> (n = 3 <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />
9 children; e.g., c<strong>on</strong>fused events before with after<br />
the storm), (2) inaccur<strong>at</strong>e detail or part <str<strong>on</strong>g>of</str<strong>on</strong>g> a<br />
memory (n = 7; e.g., "W<strong>at</strong>er came in the fr<strong>on</strong>t<br />
door window," but it was another window), and<br />
(3) exagger<strong>at</strong>i<strong>on</strong> (n = 5; e.g., "Every window<br />
broke"). For the purposes <str<strong>on</strong>g>of</str<strong>on</strong>g> our study, we are not<br />
particularly c<strong>on</strong>cerned with these types <str<strong>on</strong>g>of</str<strong>on</strong>g> errors<br />
because they reflect memories <str<strong>on</strong>g>of</str<strong>on</strong>g> actual events.<br />
Other items marked as inaccur<strong>at</strong>e by mothers<br />
were c<strong>at</strong>egorized as (4) self-c<strong>on</strong>tradicti<strong>on</strong>s (n = 2)<br />
or (5) illogical or impossible events (n = 3; e.g.,<br />
"<str<strong>on</strong>g>The</str<strong>on</strong>g> soup was chasing us"). <str<strong>on</strong>g>The</str<strong>on</strong>g>se, too, may not<br />
be errors in memory as much as errors in<br />
expressi<strong>on</strong> or thought.<br />
Finally, three c<strong>at</strong>egories reflected some degree<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> memory distorti<strong>on</strong> or inaccuracy and are more<br />
relevant to our purpose <str<strong>on</strong>g>of</str<strong>on</strong>g> assessing true memo-<br />
ties. <str<strong>on</strong>g>The</str<strong>on</strong>g>y included (6) inaccur<strong>at</strong>e facts (n = 6;<br />
e.g., "Daddy went into the guest room," but he<br />
actually could not get in), (7) wishful thinking<br />
(n = 3; e.g., "I got a lot <str<strong>on</strong>g>of</str<strong>on</strong>g> Kool Aid"), and (8)<br />
complete fabric<strong>at</strong>i<strong>on</strong>s (n = 2). However, for the<br />
inaccur<strong>at</strong>e facts, it was <str<strong>on</strong>g>of</str<strong>on</strong>g>ten not clear whether<br />
the child was actually in a positi<strong>on</strong> to know the<br />
inform<strong>at</strong>i<strong>on</strong> requested or not. <str<strong>on</strong>g>The</str<strong>on</strong>g>se errors were<br />
more <str<strong>on</strong>g>of</str<strong>on</strong>g>ten made in resp<strong>on</strong>se to a specific ques-<br />
ti<strong>on</strong> asked by the interviewer. Finally, complete<br />
fabric<strong>at</strong>i<strong>on</strong>s were rare. One child described cook-<br />
ing marshmallows, and another fabric<strong>at</strong>ed events<br />
including his grandf<strong>at</strong>her in the closet (who was<br />
not present) and "the skelet<strong>on</strong>s and their b<strong>on</strong>es<br />
broke•" <str<strong>on</strong>g>The</str<strong>on</strong>g>se examples highlight the complexity<br />
and difficulty <str<strong>on</strong>g>of</str<strong>on</strong>g> meaningfully assessing the accu-<br />
racy <str<strong>on</strong>g>of</str<strong>on</strong>g> free recall <str<strong>on</strong>g>of</str<strong>on</strong>g> autobiographical informa-<br />
ti<strong>on</strong> in young children. Further, it is not clear th<strong>at</strong><br />
the mother should be the ultim<strong>at</strong>e standard against
which to judge the child's memory for such a<br />
subjective, temporally extended event.<br />
Amount Recalled as a Functi<strong>on</strong><br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> Age and <str<strong>on</strong>g>Stress</str<strong>on</strong>g><br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> number <str<strong>on</strong>g>of</str<strong>on</strong>g> hurricane-relevant propositi<strong>on</strong>s<br />
gener<strong>at</strong>ed by children as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> age and<br />
storm severity is shown in Table 2. <str<strong>on</strong>g>The</str<strong>on</strong>g>y ranged<br />
from 21 to 554, with an overall (unweighted)<br />
mean <str<strong>on</strong>g>of</str<strong>on</strong>g> 153.6 (SD = 93.8). In general, children<br />
spoke a gre<strong>at</strong> deal about the hurricane rel<strong>at</strong>ed<br />
events. Because the distributi<strong>on</strong> was skewed, log<br />
transforms were performed and the log (base 10)<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> the total number <str<strong>on</strong>g>of</str<strong>on</strong>g> propositi<strong>on</strong>s gener<strong>at</strong>ed<br />
served as our primary dependent measure (see<br />
Table 2).<br />
Is there a rel<strong>at</strong>i<strong>on</strong> between stress and amount<br />
recalled and is it affected by age? An ANOVA<br />
was c<strong>on</strong>ducted with child age (3 or 4 years) and<br />
storm severity (high, moder<strong>at</strong>e, and low) <strong>on</strong> the<br />
log <str<strong>on</strong>g>of</str<strong>on</strong>g> the total number <str<strong>on</strong>g>of</str<strong>on</strong>g> hurricane-relevant<br />
propositi<strong>on</strong>s recalled. Results indic<strong>at</strong>ed a main<br />
effect <str<strong>on</strong>g>of</str<strong>on</strong>g> child age with 4-year-olds recalling<br />
significantly more than 3-year-olds, F(1, 94) =<br />
10.32, p -- .002 (see means in Table 2). This<br />
difference occurs even though developmental<br />
differences in the language producti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> 3- and<br />
4-year-olds were partly minimized by excluding<br />
elabor<strong>at</strong>i<strong>on</strong>s from this measure (such as the "big,<br />
bad" storm) and by evalu<strong>at</strong>ing elabor<strong>at</strong>i<strong>on</strong>s sepa-<br />
r<strong>at</strong>ely (see below). Of even gre<strong>at</strong>er interest, a<br />
significant main effect <str<strong>on</strong>g>of</str<strong>on</strong>g> hurricane severity was<br />
found, F(2, 94) = 3.14, p = .048. Trend analyses<br />
were performed to assess the n<strong>at</strong>ure <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />
rel<strong>at</strong>i<strong>on</strong>ship between stress and memory. Results<br />
indic<strong>at</strong>ed a significant quadr<strong>at</strong>ic trend, F(1, 94) =<br />
5.84, p = .018, rel<strong>at</strong>ing the amount recalled and<br />
stress (as shown in Figure 1) and no linear trend,<br />
F(1, 94) = 1.46, p > .1. <str<strong>on</strong>g>The</str<strong>on</strong>g> moder<strong>at</strong>e severity<br />
group recalled the most about the hurricane,<br />
whereas those in the low and high severity groups<br />
recalled the least.<br />
Additi<strong>on</strong>al analyses were c<strong>on</strong>ducted for further<br />
evalu<strong>at</strong>i<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the n<strong>at</strong>ure <str<strong>on</strong>g>of</str<strong>on</strong>g> the quadr<strong>at</strong>ic trend.<br />
Given th<strong>at</strong> the points defining the curve were<br />
somewh<strong>at</strong> arbitrarily demarc<strong>at</strong>ed (they represent<br />
an ordinal r<strong>at</strong>her than an interval scale), it was not<br />
necessarily expected th<strong>at</strong> recall from adjacent<br />
points would differ from <strong>on</strong>e another. R<strong>at</strong>her, we<br />
were more interested in the shape <str<strong>on</strong>g>of</str<strong>on</strong>g> the curve.<br />
MEMORY FOR A NATURAL DISASTER 317<br />
¢q<br />
• ° .<br />
¢qt'qcq<br />
t~ cc'l t'~ 1<br />
¢qo~lr~<br />
¢q~OQ<br />
¢qCqCq<br />
,..~¢q'~l"<br />
-.~.-.<br />
t'q Cq t",l<br />
~ea~<br />
O0 ¢¢)<br />
• ° .<br />
,.-~¢qCq<br />
tr~o~<br />
¢D<br />
g<br />
2~<br />
-d<br />
-d
318 BAHRICK, PARKER, FIVUSH, AND LEVrIT<br />
0<br />
0<br />
em<br />
e~<br />
o<br />
o<br />
2.2-<br />
2.1-<br />
2.0-<br />
i'<br />
/<br />
2.01<br />
/<br />
/<br />
2.19<br />
9<br />
l<br />
j, ',<br />
!<br />
b 2.10<br />
Low Moder<strong>at</strong>e High<br />
Storm Severity<br />
Figure 1. Mean log number <str<strong>on</strong>g>of</str<strong>on</strong>g> propositi<strong>on</strong>s recalled<br />
as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> storm severity.<br />
Did retenti<strong>on</strong> increase and then decrease as a<br />
functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> stress, or did it increase and then<br />
remain stable? Thus, we c<strong>on</strong>ducted analyses to<br />
evalu<strong>at</strong>e the significance <str<strong>on</strong>g>of</str<strong>on</strong>g> the linear increase<br />
between recall <str<strong>on</strong>g>of</str<strong>on</strong>g> the low and moder<strong>at</strong>e groups<br />
and the significance <str<strong>on</strong>g>of</str<strong>on</strong>g> the decrease from the<br />
moder<strong>at</strong>e to high storm severity groups. Results<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> linear trend tests (<strong>on</strong>e-tailed) indic<strong>at</strong>ed th<strong>at</strong> the<br />
increase in recall from low to moder<strong>at</strong>e storm<br />
severity groups was significant, F(1, 94) = 5.86,<br />
p = .009, and the decrease in recall from the<br />
moder<strong>at</strong>e to high storm severity groups was<br />
marginally significant, F(1, 94) = 2.19, p = .071.<br />
Thus, our d<strong>at</strong>a appear to be most c<strong>on</strong>sistent<br />
with an inverted U-shaped functi<strong>on</strong> rel<strong>at</strong>ing the<br />
amount recalled and stress in young children.<br />
Apparently, when a broad range <str<strong>on</strong>g>of</str<strong>on</strong>g> stress is<br />
sampled, recall first increases and then decreases<br />
as stress intensifies. Further analyses were under-<br />
taken to determine the impact <str<strong>on</strong>g>of</str<strong>on</strong>g> sec<strong>on</strong>dary<br />
variables including type <str<strong>on</strong>g>of</str<strong>on</strong>g> recall and time period<br />
<strong>on</strong> this functi<strong>on</strong>.<br />
Amount Recalled as a Functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> Time<br />
Period and Type <str<strong>on</strong>g>of</str<strong>on</strong>g> Recall<br />
We c<strong>on</strong>ducted more detailed analyses to exam-<br />
ine memory as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> time period<br />
(T1 = prehurricane prepar<strong>at</strong>i<strong>on</strong>s, T2 = the hurri-<br />
cane itself, and T3 = posthurricane recovery pe-<br />
riod) and type <str<strong>on</strong>g>of</str<strong>on</strong>g> recall (sp<strong>on</strong>taneous vs.<br />
prompted). Most, but not all, propositi<strong>on</strong>s could<br />
be classified into <strong>on</strong>e <str<strong>on</strong>g>of</str<strong>on</strong>g> the three time periods.<br />
Occasi<strong>on</strong>ally, children supplied relevant informa-<br />
ti<strong>on</strong> th<strong>at</strong> was not specific to a time period such as<br />
facts, opini<strong>on</strong>s, or <strong>at</strong>titudes, and these were<br />
classified as unrel<strong>at</strong>ed to a time period (T4). <str<strong>on</strong>g>The</str<strong>on</strong>g><br />
mean number <str<strong>on</strong>g>of</str<strong>on</strong>g> propositi<strong>on</strong>s classified into <strong>on</strong>e<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> the three time periods was 143.24, compared<br />
with 153.6 total propositi<strong>on</strong>s gener<strong>at</strong>ed about<br />
hurricane-rel<strong>at</strong>ed events.<br />
All propositi<strong>on</strong>s were also classified as either<br />
sp<strong>on</strong>taneous or prompted, depending <strong>on</strong> whether<br />
they had been elicited by open-ended questi<strong>on</strong>s<br />
versus c<strong>at</strong>egory or specific prompts. Propositi<strong>on</strong>s<br />
elicited by the four open-ended questi<strong>on</strong>s were<br />
coded as sp<strong>on</strong>taneous recall. Those elicited by<br />
c<strong>at</strong>egory prompts and by specific item prompts<br />
were coded as prompted recall (see Procedure<br />
secti<strong>on</strong> for details).<br />
Did Amount Recalled Differ Across Time<br />
Periods and for Sp<strong>on</strong>taneous Versus<br />
Prompted Memories ?<br />
A four-way ANOVA with child age and storm<br />
severity as between-subjects factors and time<br />
period and type <str<strong>on</strong>g>of</str<strong>on</strong>g> recall as within-subjects<br />
factors was c<strong>on</strong>ducted <strong>on</strong> the log <str<strong>on</strong>g>of</str<strong>on</strong>g> the total<br />
propositi<strong>on</strong>s. Results indic<strong>at</strong>ed significant main<br />
effects <str<strong>on</strong>g>of</str<strong>on</strong>g> child age and storm severity as before<br />
and highly significant main effects <str<strong>on</strong>g>of</str<strong>on</strong>g> time period,<br />
F(2, 188) = 139.7, p < .001 (with untransformed<br />
means <str<strong>on</strong>g>of</str<strong>on</strong>g> 22.20, 63.18, and 57.86 for T1, T2, and<br />
T3, respectively), and type <str<strong>on</strong>g>of</str<strong>on</strong>g> recall, F(1, 94) =<br />
425.52, p < .001 (with untransformed means <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
33.28 and 109.96 for sp<strong>on</strong>taneous and prompted<br />
recall, respectively.) Thus, significantly more<br />
inform<strong>at</strong>i<strong>on</strong> was elicited by prompted than by<br />
open-ended questi<strong>on</strong>s. Only two interacti<strong>on</strong>s<br />
reached significance: time period and storm sever-<br />
ity, F(4, 188) = 3.26, p = .014, and time period<br />
and type <str<strong>on</strong>g>of</str<strong>on</strong>g> recall, F(2, 188) = 14.68, p < .001,<br />
and they are examined further below. Because
child age did not interact with any variables,<br />
further discussi<strong>on</strong>s will collapse across child age.<br />
In Wh<strong>at</strong> Way Did Recall for Events<br />
Surrounding Each <str<strong>on</strong>g>of</str<strong>on</strong>g> the Three<br />
Time Periods Differ?<br />
We c<strong>on</strong>ducted post hoe tests comparing the log<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> the propositi<strong>on</strong>s recalled across the three time<br />
periods. Results indic<strong>at</strong>ed th<strong>at</strong> children recalled<br />
significantly more propositi<strong>on</strong>s for T2 and T3<br />
than for T1, (p < .05, Tukey's h<strong>on</strong>estly signifi-<br />
cant difference). In fact, children recalled more<br />
than 2.5 times as many propositi<strong>on</strong>s about the<br />
hurricane itself than the prepar<strong>at</strong>i<strong>on</strong> period. <str<strong>on</strong>g>The</str<strong>on</strong>g>se<br />
differences should be viewed in the c<strong>on</strong>text <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />
n<strong>at</strong>ure and the extent <str<strong>on</strong>g>of</str<strong>on</strong>g> the three temporal events.<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> prepar<strong>at</strong>i<strong>on</strong> period was approxim<strong>at</strong>ely 2 days<br />
l<strong>on</strong>g and quite similar across children <str<strong>on</strong>g>of</str<strong>on</strong>g> different<br />
stress levels, the hurricane itself was approxi-<br />
m<strong>at</strong>ely 8 h l<strong>on</strong>g, and the afterm<strong>at</strong>h ranged from a<br />
few days to many weeks. Although the hurricane<br />
itself was the shortest period, much more was<br />
recalled about this period than the prepar<strong>at</strong>i<strong>on</strong><br />
period. Figure 2 displays the mean log number <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
propositi<strong>on</strong>s recalled for each time period as a<br />
functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> storm severity. To help interpret the<br />
interacti<strong>on</strong> between time period and severity,<br />
results were broken down according to time<br />
period and trend analyses were performed. A<br />
significant quadr<strong>at</strong>ic functi<strong>on</strong> was found across<br />
severity levels for memories surrounding T3,<br />
F(1, 94) = 8.21, p = .005. Results <str<strong>on</strong>g>of</str<strong>on</strong>g> linear trend<br />
analyses also indic<strong>at</strong>ed th<strong>at</strong> both the increase in<br />
recall from low to moder<strong>at</strong>e storm severity c<strong>on</strong>di-<br />
ti<strong>on</strong>s and the decrease from moder<strong>at</strong>e to high<br />
storm severity c<strong>on</strong>diti<strong>on</strong>s were significant, F(1,<br />
94) = 7.07, p = .0005 and F(1, 94) = 4.07, p =<br />
.023, respectively. A marginally significant qua-<br />
dr<strong>at</strong>ic functi<strong>on</strong> was also found for memories<br />
surrounding T1, F(1, 94) = 2.88, p = .09.<br />
Results <str<strong>on</strong>g>of</str<strong>on</strong>g> linear trend analyses revealed th<strong>at</strong> the<br />
increase in recall from the low to moder<strong>at</strong>e storm<br />
severity c<strong>on</strong>diti<strong>on</strong>s was not significant, however,<br />
the decrease from moder<strong>at</strong>e to high storm sever-<br />
ity was significant, F(1, 94) = 4.70, p = .016.<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> quadr<strong>at</strong>ic functi<strong>on</strong> for memories surrounding<br />
T2, however, was not significant, F(1, 94) =<br />
1.89, p > .1. <strong>Memory</strong> for this time period was<br />
also more c<strong>on</strong>founded by extraneous factors th<strong>at</strong><br />
differed as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> storm severity than<br />
MEMORY FOR A NATURAL DISASTER 319<br />
1.8-<br />
1.7-<br />
• -~ 1.6-<br />
1.5-<br />
• -, 1.4-<br />
~ 1.3-<br />
~ 1.2-<br />
I.I-<br />
1.77 1.76<br />
/<br />
/;/ .....<br />
"..<br />
,.<br />
Tlrae<br />
Period<br />
o,/°j ",. 2<br />
// f ""..,<br />
• / .,,.'/ "'-.,. 1i6 3<br />
1.61/'/ ,."<br />
A" ,o"<br />
/,<br />
Time<br />
Period<br />
/<br />
1.54.."<br />
3<br />
1.25. -"<br />
e"<br />
1.33<br />
%%<br />
",•1.15<br />
• Time<br />
Period<br />
1<br />
I I I<br />
Low Moder<strong>at</strong>e High<br />
Storm Severity<br />
Figure 2. Mean log number <str<strong>on</strong>g>of</str<strong>on</strong>g> propositi<strong>on</strong>s recalled<br />
for each time period as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> storm severity.<br />
memories for the other two periods. For example,<br />
in the low and moder<strong>at</strong>e storm severity c<strong>on</strong>di-<br />
ti<strong>on</strong>s, children were more likely to be able to<br />
sleep through porti<strong>on</strong>s <str<strong>on</strong>g>of</str<strong>on</strong>g> the storm (44% and<br />
24%, respectively), than in the high severity<br />
c<strong>on</strong>diti<strong>on</strong> (5%), and in the high severity c<strong>on</strong>di-<br />
ti<strong>on</strong>, there were more distinctive events during<br />
the storm itself (the family typically moved from<br />
<strong>on</strong>e room to another to avoid flying objects). This<br />
resulted in more potentially recallable salient<br />
events and more time awake for children in the<br />
high severity c<strong>at</strong>egory and may have <str<strong>on</strong>g>of</str<strong>on</strong>g>fset the<br />
tendency for recall to decrease with high levels <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
stress. In c<strong>on</strong>trast, during T1, children all experi-<br />
enced the same kind <str<strong>on</strong>g>of</str<strong>on</strong>g> activities regardless <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />
level <str<strong>on</strong>g>of</str<strong>on</strong>g> hurricane severity because <strong>at</strong> th<strong>at</strong> time, it<br />
was not known where the storm would make<br />
landfall. During T3, children in the low, moder-<br />
<strong>at</strong>e, and high severity groups may also have had<br />
increasingly more hurricane-rel<strong>at</strong>ed activities and<br />
thus more potentially recallable m<strong>at</strong>erial. For<br />
example, children who experienced high storm<br />
severity had the largest number <str<strong>on</strong>g>of</str<strong>on</strong>g> days without
320 BAHRICK, PARKER, FIVUSH, AND LEVITT<br />
electricity, running w<strong>at</strong>er, or teleph<strong>on</strong>e, cre<strong>at</strong>ing<br />
unusual circumstances for a l<strong>on</strong>ger period <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
time. <str<strong>on</strong>g>The</str<strong>on</strong>g> inverted O-shaped functi<strong>on</strong>s, both<br />
overall and particularly for the posthurricane<br />
recovery period, are thus striking given the<br />
expect<strong>at</strong>i<strong>on</strong> th<strong>at</strong> the high storm severity group<br />
had more potentially recallable m<strong>at</strong>erial, yet<br />
recall was the most <strong>at</strong>tenu<strong>at</strong>ed.<br />
How Do Age and <str<strong>on</strong>g>Stress</str<strong>on</strong>g> Affect Sp<strong>on</strong>taneous<br />
Versus Prompted Memories ?<br />
Figure 3 displays the mean log number <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
propositi<strong>on</strong>s for open-ended versus prompted<br />
questi<strong>on</strong>s as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> storm severity. In a<br />
three-way ANOVA with age, severity c<strong>at</strong>egory,<br />
and type <str<strong>on</strong>g>of</str<strong>on</strong>g> recall, results revealed main effects <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
child age, storm severity, and type <str<strong>on</strong>g>of</str<strong>on</strong>g> recall as<br />
before, as well as the significant quadr<strong>at</strong>ic trend<br />
rel<strong>at</strong>ing amount recalled with storm severity.<br />
However, there was also a significant interacti<strong>on</strong><br />
between storm severity entered as a quadr<strong>at</strong>ic and<br />
type <str<strong>on</strong>g>of</str<strong>on</strong>g> recall, F(1, 94) = 4.53, p = .036 (see<br />
Figure 3). <str<strong>on</strong>g>The</str<strong>on</strong>g>se effects led us to examine and<br />
2.1-<br />
2.0-<br />
1.9"<br />
1.8'<br />
O<br />
~'~ 1.7-<br />
~ 1.6-<br />
~ 1.5-<br />
~ 1.4-<br />
1.3<br />
1.27<br />
1.11<br />
1.11,"<br />
o"<br />
/<br />
f-<br />
f-<br />
t •<br />
t<br />
i<br />
2.04<br />
.//- -''~'- .......... !~9<br />
1.47<br />
s t "-.<br />
i i i<br />
Low Moder<strong>at</strong>e High<br />
Storm Severity<br />
Prompted<br />
"-. 1.33<br />
"o Sp<strong>on</strong>taneous<br />
Figure 3. Mean log number <str<strong>on</strong>g>of</str<strong>on</strong>g> propositi<strong>on</strong>s recalled<br />
for sp<strong>on</strong>taneous versus prompted memory as a func-<br />
ti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> storm severity.<br />
characterize in more detail the n<strong>at</strong>ure <str<strong>on</strong>g>of</str<strong>on</strong>g> memory<br />
under each type <str<strong>on</strong>g>of</str<strong>on</strong>g> recall c<strong>on</strong>diti<strong>on</strong> separ<strong>at</strong>ely. It<br />
was <str<strong>on</strong>g>of</str<strong>on</strong>g> particular interest to determine under wh<strong>at</strong><br />
type <str<strong>on</strong>g>of</str<strong>on</strong>g> interview c<strong>on</strong>diti<strong>on</strong>s children's memories<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> an event were most clearly differenti<strong>at</strong>ed as a<br />
functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> stress. Separ<strong>at</strong>e analyses were c<strong>on</strong>-<br />
ducted examining the log <str<strong>on</strong>g>of</str<strong>on</strong>g> the number <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
propositi<strong>on</strong>s gener<strong>at</strong>ed by open-ended versus<br />
prompted questi<strong>on</strong>s as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> age and<br />
storm severity.<br />
Sp<strong>on</strong>taneous <strong>Memory</strong>: Wh<strong>at</strong> Is the Rel<strong>at</strong>i<strong>on</strong><br />
Between <str<strong>on</strong>g>Stress</str<strong>on</strong>g> and Sp<strong>on</strong>taneous Recall ?<br />
Analyses for the sp<strong>on</strong>taneous memories taken<br />
separ<strong>at</strong>ely indic<strong>at</strong>ed a robust quadr<strong>at</strong>ic trend<br />
rel<strong>at</strong>ing the amount recalled with hurricane sever-<br />
ity, F(1, 94) = 7.91, p = .006 (see Figure 3).<br />
Further analyses revealed a significant linear<br />
increase in recall from the low to moder<strong>at</strong>e storm<br />
severity groups, F(1, 94) = 9.07, p = .002, and a<br />
marginally significant decrease from the moder-<br />
<strong>at</strong>e to the high storm severity groups, F(1, 94)<br />
2.18, p = .072. As before, there were also<br />
significant main effects <str<strong>on</strong>g>of</str<strong>on</strong>g> child age, F(1, 94) =<br />
6.53, p = .012, and hurricane severity, F(2, 94) =<br />
4.61, p = .012. When each time period was<br />
examined separ<strong>at</strong>ely, analyses indic<strong>at</strong>ed a signifi-<br />
cant main effect <str<strong>on</strong>g>of</str<strong>on</strong>g> storm severity for T3, F(2,<br />
94) = 4.09, p = .02, and a significant quadr<strong>at</strong>ic<br />
functi<strong>on</strong> rel<strong>at</strong>ing sp<strong>on</strong>taneous recall and storm<br />
severity, F(1, 94) = 8.18, p = .005. Results <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />
linear trend analyses again revealed a significant<br />
increase in recall from the low to the moder<strong>at</strong>e<br />
storm severity groups and a significant decrease<br />
in recall from the moder<strong>at</strong>e to the high storm<br />
severity groups, F(1, 94) = 6.06, p = .008, and<br />
F(1, 94) = 5.94, p = .013, respectively. <str<strong>on</strong>g>The</str<strong>on</strong>g><br />
quadr<strong>at</strong>ic functi<strong>on</strong> also approached significance<br />
for sp<strong>on</strong>taneous recall <str<strong>on</strong>g>of</str<strong>on</strong>g> T2 events, F(1, 94) =<br />
3.59, p = .06. <str<strong>on</strong>g>The</str<strong>on</strong>g>se effects replic<strong>at</strong>e and magnify<br />
those found for the overall results in which<br />
children in the moder<strong>at</strong>e severity c<strong>on</strong>diti<strong>on</strong> re-<br />
called the most about the event as a whole and<br />
especially about the posthurricane recovery pe-<br />
riod. When children's memory was not prompted,<br />
the amount recalled about the storm and espe-<br />
cially the recovery period also tended to be<br />
gre<strong>at</strong>est as a result <str<strong>on</strong>g>of</str<strong>on</strong>g> moder<strong>at</strong>e stress and <strong>at</strong>tenu-<br />
<strong>at</strong>ed as a result <str<strong>on</strong>g>of</str<strong>on</strong>g> high stress. Thus, the rel<strong>at</strong>i<strong>on</strong>-<br />
ship between sp<strong>on</strong>taneous recall and stress was
most c<strong>on</strong>sistent with an inverted U-shaped<br />
functi<strong>on</strong>.<br />
Prompted <strong>Memory</strong>: Wh<strong>at</strong> Is the Rel<strong>at</strong>i<strong>on</strong><br />
Between <str<strong>on</strong>g>Stress</str<strong>on</strong>g> and Prompted Recall?<br />
Results <str<strong>on</strong>g>of</str<strong>on</strong>g> analyses for prompted memories<br />
were not as robust as those for the sp<strong>on</strong>taneous<br />
memories. Although the main effect <str<strong>on</strong>g>of</str<strong>on</strong>g> child age<br />
was still evident when the prompted memories<br />
were c<strong>on</strong>sidered al<strong>on</strong>e, F(1, 94) = 10.60, p =<br />
.002, the main effect <str<strong>on</strong>g>of</str<strong>on</strong>g> hurricane severity was no<br />
l<strong>on</strong>ger apparent, F(2, 94) = 2.18,p > .1. Further,<br />
the quadr<strong>at</strong>ic trend rel<strong>at</strong>ing amount recalled with<br />
hurricane severity approached but no l<strong>on</strong>ger<br />
reached significance, F(1, 94) = 3.61, p = .06<br />
(see Figure 3). When prompted recall was exam-<br />
ined for each time period separ<strong>at</strong>ely; there were<br />
no significant main effects <str<strong>on</strong>g>of</str<strong>on</strong>g> storm severity (all<br />
ps > .05), but there was <strong>on</strong>e significant quadr<strong>at</strong>ic<br />
trend as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> severity for T3, as before,<br />
F(1, 94) = 5.07, p = .03. Results <str<strong>on</strong>g>of</str<strong>on</strong>g> linear trend<br />
analyses revealed a significant linear increase<br />
from the low to moder<strong>at</strong>e storm severity groups,<br />
F(1, 94) = 4.75, p = .016, but <strong>on</strong>ly a marginally<br />
significant decrease from the moder<strong>at</strong>e to the<br />
high storm severity c<strong>on</strong>diti<strong>on</strong>s, F(1, 94) = 2.17,<br />
p = .07. Thus, overall, the analyses from the<br />
prompted memories show a p<strong>at</strong>tern th<strong>at</strong> is similar<br />
to those <str<strong>on</strong>g>of</str<strong>on</strong>g> the sp<strong>on</strong>taneous and total memories,<br />
but effects <str<strong>on</strong>g>of</str<strong>on</strong>g> storm severity are <strong>at</strong>tenu<strong>at</strong>ed. <str<strong>on</strong>g>The</str<strong>on</strong>g><br />
main effect <str<strong>on</strong>g>of</str<strong>on</strong>g> storm severity and the quadr<strong>at</strong>ic<br />
trend found for the total recall measure were no<br />
l<strong>on</strong>ger observed when prompted recall was c<strong>on</strong>-<br />
sidered al<strong>on</strong>e, even though prompting elicited<br />
more than 2.5 times as much recall as open-ended<br />
questi<strong>on</strong>s. In c<strong>on</strong>trast, these effects were most<br />
MEMORY FOR A NATURAL DISASTER 321<br />
evident when sp<strong>on</strong>taneous recall was examined<br />
separ<strong>at</strong>ely. Apparently, <strong>on</strong>ce sp<strong>on</strong>taneous ac-<br />
counts have been given by children, trying to<br />
elicit more memory by using c<strong>at</strong>egory prompts or<br />
specific cues results in a gre<strong>at</strong> deal more recall<br />
about the event for all stress levels and may<br />
diminish differences in amount recalled am<strong>on</strong>g<br />
children who experienced different levels <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
stress. Thus, by allowing children to report their<br />
memory in a free-recall c<strong>on</strong>text, differences in<br />
amount recalled as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> stress may be<br />
maximized. By prompting memory, recall will be<br />
enhanced, especially for children who experi-<br />
enced more severe stress.<br />
Rel<strong>at</strong>i<strong>on</strong>s Am<strong>on</strong>g Recall Retenti<strong>on</strong> Interval<br />
Rehearsal and Socioec<strong>on</strong>omic St<strong>at</strong>us (SES)<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> effects observed in the above analyses<br />
were reexamined in the c<strong>on</strong>text <str<strong>on</strong>g>of</str<strong>on</strong>g> a number <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
covari<strong>at</strong>es to determine whether they signifi-<br />
cantly qualified the main results including any<br />
potential differences across severity groups as a<br />
functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> SES (family income, parent's educa-<br />
ti<strong>on</strong>, and home value), rehearsal <str<strong>on</strong>g>of</str<strong>on</strong>g> events sur-<br />
rounding the hurricane, and retenti<strong>on</strong> interval.<br />
Could SES Predict the Rel<strong>at</strong>i<strong>on</strong> Between<br />
<str<strong>on</strong>g>Stress</str<strong>on</strong>g> and Recall?<br />
Table 3 depicts the mean family income and<br />
number <str<strong>on</strong>g>of</str<strong>on</strong>g> years <str<strong>on</strong>g>of</str<strong>on</strong>g> educ<strong>at</strong>i<strong>on</strong> for the mother and<br />
f<strong>at</strong>her as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> storm severity. N<strong>on</strong>e <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
these measures significantly predicted amount<br />
recalled when used as covari<strong>at</strong>es, F(1, 91) =<br />
Table 3<br />
Socioec<strong>on</strong>omic St<strong>at</strong>us: Means and Standard Devi<strong>at</strong>i<strong>on</strong>s for Family Income<br />
and Number <str<strong>on</strong>g>of</str<strong>on</strong>g> Years <str<strong>on</strong>g>of</str<strong>on</strong>g> Educ<strong>at</strong>i<strong>on</strong> as a Functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> Storm Severity<br />
Low Moder<strong>at</strong>e High<br />
Income and (n = 18) (n = 41) (n = 39) Overall<br />
educ<strong>at</strong>i<strong>on</strong> M SD M SD M SD M SD<br />
Family income 5.19 1.44 5.10 2.03 5.68 1.33 5.32 1.57<br />
F<strong>at</strong>her's educ<strong>at</strong>i<strong>on</strong> 14.98 2.45 16.59 a 2.73 14.56 2.66 15.38 2.62<br />
Mother's educ<strong>at</strong>i<strong>on</strong> 15.33 2.57 15.90 2.05 15.00 2.67 15.41 2.44<br />
Note. For income, 1 = under $10,000, 2 = $10,000-$19,999, 3 = $20,000-$29,000, 4 =<br />
$30,000-$39,000, 5 = $40,000-$59,000, 6 = $60,O90-$79,000, and 7 = Over $80,000.<br />
an = 40.
322 BAHRICK, PARKER, FIVUSH, AND LEVITT<br />
0.52, p > .1, F(1, 91) = 1.84, p > .1, and F(1,<br />
90) = 0.68, p > .1, for income, mother's<br />
educ<strong>at</strong>i<strong>on</strong>, and f<strong>at</strong>her's educ<strong>at</strong>i<strong>on</strong>, respectively.<br />
Further, the main effects <str<strong>on</strong>g>of</str<strong>on</strong>g> the child's age, storm<br />
severity, and type <str<strong>on</strong>g>of</str<strong>on</strong>g> recall, as well as the<br />
quadr<strong>at</strong>ic functi<strong>on</strong>s rel<strong>at</strong>ing storm severity and<br />
amount recalled were all significant even after the<br />
means were adjusted for the effects <str<strong>on</strong>g>of</str<strong>on</strong>g> each <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
these variables (all ps < .05). Further, the value<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> the family home (see Table 1) was also used as<br />
a covari<strong>at</strong>e and did not predict recall, F(3, 72) =<br />
0.58,p > .1.<br />
Could Rehearsal Predict the Quadr<strong>at</strong>ic<br />
Rel<strong>at</strong>i<strong>on</strong> Between <str<strong>on</strong>g>Stress</str<strong>on</strong>g> and Recall?<br />
Might the quadr<strong>at</strong>ic functi<strong>on</strong> rel<strong>at</strong>ing amount<br />
recalled and stress be medi<strong>at</strong>ed by differential<br />
rehearsal across the storm severity groups? Per-<br />
haps families who experienced moder<strong>at</strong>e storm<br />
severity discussed the hurricane-rel<strong>at</strong>ed events<br />
more than families who experienced high or low<br />
storm severity. It is important to recall th<strong>at</strong> our<br />
questi<strong>on</strong>naire assessed rehearsal <strong>on</strong> a 3-point<br />
scale ranging from 0 (n<strong>on</strong>e) to 1 (<strong>on</strong>ce a day) or 2<br />
(several times a day) for a number <str<strong>on</strong>g>of</str<strong>on</strong>g> different<br />
questi<strong>on</strong>s and time periods, and a composite<br />
score was derived. <str<strong>on</strong>g>The</str<strong>on</strong>g> mean rehearsal r<strong>at</strong>ing<br />
across all participants was 1.5 (SD = 0.40). It<br />
was 1.33, 1.55, and 1.57 for the low, moder<strong>at</strong>e,<br />
and high storm severity groups, respectively.<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g>re was a significant positive correl<strong>at</strong>i<strong>on</strong> be-<br />
tween amount <str<strong>on</strong>g>of</str<strong>on</strong>g> rehearsal and storm severity<br />
r<strong>at</strong>ing (r = .39, p < .01). Families who experi-<br />
enced a more severe hurricane tended to talk<br />
about the hurricane more. When the composite<br />
rehearsal score was used as a covari<strong>at</strong>e in an<br />
analysis with the child's age, storm severity, and<br />
type <str<strong>on</strong>g>of</str<strong>on</strong>g> recall as main factors, rehearsal was not a<br />
significant predictor <str<strong>on</strong>g>of</str<strong>on</strong>g> amount recalled, F(1,<br />
84) = 0.04, p >. 1. Further, all effects previously<br />
found for child age, storm severity entered as a<br />
quadr<strong>at</strong>ic, and type <str<strong>on</strong>g>of</str<strong>on</strong>g> recall were still significant<br />
even after adjusting the means for rehearsal, F(1,<br />
84) = 8.8, p = .004, F(1, 84) = 7.2, p = .009,<br />
and F(1, 85) = 274.24, p < .001, respectively.<br />
Thus, amount recalled as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> stress, age,<br />
and type <str<strong>on</strong>g>of</str<strong>on</strong>g> recall cannot be explained by differen-<br />
tial rehearsal. Rehearsal was also examined in<br />
more detail for the three time periods separ<strong>at</strong>ely<br />
to determine whether gre<strong>at</strong>er rehearsal could<br />
account for the superior memory for T2 and T3<br />
rel<strong>at</strong>ive to T1. <str<strong>on</strong>g>The</str<strong>on</strong>g> mean amounts <str<strong>on</strong>g>of</str<strong>on</strong>g> rehearsal for<br />
T1, T2, and T3 were 1.02, 1.75, and 1.68,<br />
respectively. Rehearsal was a significant predic-<br />
tor <str<strong>on</strong>g>of</str<strong>on</strong>g> differential recall across the three time<br />
periods, F(1,167) = 4.36, p = .038, however, the<br />
main effect <str<strong>on</strong>g>of</str<strong>on</strong>g> time was still evident even after the<br />
means were adjusted for effects <str<strong>on</strong>g>of</str<strong>on</strong>g> rehearsal, F(2,<br />
167) = 38.34,p = < .001.<br />
Could Retenti<strong>on</strong> Interval Predict<br />
the Quadr<strong>at</strong>ic Rel<strong>at</strong>i<strong>on</strong> Between<br />
<str<strong>on</strong>g>Stress</str<strong>on</strong>g> and Recall?<br />
It is important to recall th<strong>at</strong> children were<br />
interviewed between 60 and 179 days (M = 102.7,<br />
SD = 24.02) following the hurricane, and thus<br />
we did not expect to observe the typical neg<strong>at</strong>ive<br />
correl<strong>at</strong>i<strong>on</strong> between retenti<strong>on</strong> interval and amount<br />
recalled found for short delays. When retenti<strong>on</strong><br />
interval was entered as a covari<strong>at</strong>e, it was not a<br />
significant predictor <str<strong>on</strong>g>of</str<strong>on</strong>g> the amount recalled, F(1,<br />
93) = 0.65, p > .1. Further, all main effects <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
age, type <str<strong>on</strong>g>of</str<strong>on</strong>g> recall, and the quadr<strong>at</strong>ic functi<strong>on</strong><br />
were still significant when the means were ad-<br />
justed for retenti<strong>on</strong> interval (all ps < .05).<br />
How Was the Retenti<strong>on</strong> Interval Rel<strong>at</strong>ed<br />
to the Amount Recalled?<br />
A correl<strong>at</strong>i<strong>on</strong> between retenti<strong>on</strong> interval and<br />
the log <str<strong>on</strong>g>of</str<strong>on</strong>g> the total propositi<strong>on</strong>s recalled revealed<br />
a n<strong>on</strong>significant correl<strong>at</strong>i<strong>on</strong> (r =. 15). Thus, there<br />
was no linear rel<strong>at</strong>i<strong>on</strong> between retenti<strong>on</strong> interval<br />
and amount recalled. Further, when children were<br />
classified into three retenti<strong>on</strong> interval groups <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
approxim<strong>at</strong>ely equal number, short (60-92 days),<br />
intermedi<strong>at</strong>e (93-108 days), and l<strong>on</strong>g (109-179<br />
days), those with the shortest interval recalled the<br />
most (M = 2.21), but those with the l<strong>on</strong>gest<br />
interval recalled the next most (M = 2.13), and<br />
those with the intermedi<strong>at</strong>e interval recalled the<br />
least (M = 1.80). <str<strong>on</strong>g>The</str<strong>on</strong>g> analyses above indic<strong>at</strong>e<br />
th<strong>at</strong> the retenti<strong>on</strong> interval did not qualify the<br />
observed differences in recall, nor was retenti<strong>on</strong><br />
interval rel<strong>at</strong>ed to recall in a linear way, most<br />
likely because all intervals were l<strong>on</strong>g, occurring<br />
after the initial forgetting leveled <str<strong>on</strong>g>of</str<strong>on</strong>g>f.<br />
Taken together, the results <str<strong>on</strong>g>of</str<strong>on</strong>g> the analyses <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
covariance indic<strong>at</strong>e th<strong>at</strong> n<strong>on</strong>e <str<strong>on</strong>g>of</str<strong>on</strong>g> the covari<strong>at</strong>es,<br />
including four measures <str<strong>on</strong>g>of</str<strong>on</strong>g> SES, rehearsal, or
etenti<strong>on</strong> interval, significantly predicted the<br />
amount recalled by children about the hurricane-<br />
rel<strong>at</strong>ed events.<br />
Rel<strong>at</strong>i<strong>on</strong> Between <str<strong>on</strong>g>Stress</str<strong>on</strong>g> and Elabor<strong>at</strong>i<strong>on</strong>s<br />
Are Memories <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Stress</str<strong>on</strong>g>ful Events More<br />
or Less Elabor<strong>at</strong>ed Than Memories<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> Less <str<strong>on</strong>g>Stress</str<strong>on</strong>g>ful Events?<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> mean number <str<strong>on</strong>g>of</str<strong>on</strong>g> elabor<strong>at</strong>i<strong>on</strong>s per proposi-<br />
ti<strong>on</strong> served as the dependent variable in these<br />
analyses. Overall, children gener<strong>at</strong>ed 0.88<br />
(SD = 0.26) elabor<strong>at</strong>i<strong>on</strong>s per propositi<strong>on</strong>, with<br />
an average <str<strong>on</strong>g>of</str<strong>on</strong>g> 0.81 (SD = 0.23) for 3-year-olds<br />
and 0.95 (SD = 0.29) for 4-year olds. An ANOVA<br />
with child age and storm severity as main factors<br />
revealed a significant main effect <str<strong>on</strong>g>of</str<strong>on</strong>g> age,<br />
F(1, 94) = 4.86, p = .03, and no effect <str<strong>on</strong>g>of</str<strong>on</strong>g> storm<br />
severity or interacti<strong>on</strong> (ps > .1). Four-year-olds<br />
produced significantly more elabor<strong>at</strong>i<strong>on</strong>s than<br />
3-year-olds. Further, when storm severity was<br />
entered as a quadr<strong>at</strong>ic, there was no significant<br />
effect <strong>on</strong> elabor<strong>at</strong>i<strong>on</strong>s, F(1, 94) = 1.52,p > .1. In<br />
general, recall was not more elabor<strong>at</strong>ed for chil-<br />
dren with moder<strong>at</strong>e storm exposure than those<br />
with low or high exposure.<br />
Did Children Elabor<strong>at</strong>e More When<br />
Describing Events Th<strong>at</strong> Occurred During<br />
One Time Period Over Another?<br />
Results <str<strong>on</strong>g>of</str<strong>on</strong>g> a three-way ANOVA with time as a<br />
repe<strong>at</strong>ed measures factor revealed a significant<br />
main effect <str<strong>on</strong>g>of</str<strong>on</strong>g> time period, F(2, 184) = 13.03,<br />
p < .001, with means <str<strong>on</strong>g>of</str<strong>on</strong>g> .72, .85, and .96 for T1,<br />
T2, and T3, respectively. Apparently, children<br />
gener<strong>at</strong>ed more elabor<strong>at</strong>i<strong>on</strong>s when they were<br />
recalling events surrounding the hurricane itself<br />
and the afterm<strong>at</strong>h as compared with the prepara-<br />
ti<strong>on</strong> period.<br />
How Is Elabor<strong>at</strong>i<strong>on</strong> Rel<strong>at</strong>ed to Sp<strong>on</strong>taneous<br />
Versus Prompted Recall?<br />
A three-way ANOVA was c<strong>on</strong>ducted with type<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> recall as a factor al<strong>on</strong>g with child age and<br />
storm severity. Results indic<strong>at</strong>ed no significant<br />
main effect <str<strong>on</strong>g>of</str<strong>on</strong>g> type <str<strong>on</strong>g>of</str<strong>on</strong>g> recall, F(1, 93) = 1.11,p ><br />
.1, or interacti<strong>on</strong>s. However, when storm severity<br />
was entered as a quadr<strong>at</strong>ic, there was a margin-<br />
MEMORY FOR A NATURAL DISASTER 323<br />
ally significant effect, F(1, 93) = 3.16, p = .08.<br />
This quadr<strong>at</strong>ic functi<strong>on</strong> was significant when<br />
sp<strong>on</strong>taneous recall was examined separ<strong>at</strong>ely, F(1,<br />
93) = 4.89, p = .029, and was not significant for<br />
prompted recall (p >. 1). Thus, when resp<strong>on</strong>ding<br />
to open-ended questi<strong>on</strong>s, children tended to elabo-<br />
r<strong>at</strong>e more under the moder<strong>at</strong>e storm severity<br />
c<strong>on</strong>diti<strong>on</strong>s and less under the high and low<br />
severity c<strong>on</strong>diti<strong>on</strong>s.<br />
Taken together, these analyses indic<strong>at</strong>e th<strong>at</strong><br />
4-year-olds produced more elabor<strong>at</strong>i<strong>on</strong>s than<br />
3-year-olds and th<strong>at</strong> children elabor<strong>at</strong>ed most<br />
when they were describing the hurricane after-<br />
m<strong>at</strong>h and least when they were describing the<br />
prepar<strong>at</strong>i<strong>on</strong> period. Overall, the degree <str<strong>on</strong>g>of</str<strong>on</strong>g> elabora-<br />
ti<strong>on</strong> was not system<strong>at</strong>ically rel<strong>at</strong>ed to storm<br />
severity. However, under certain c<strong>on</strong>diti<strong>on</strong>s, espe-<br />
cially for sp<strong>on</strong>taneous recall, there was a qua-<br />
dr<strong>at</strong>ic functi<strong>on</strong> rel<strong>at</strong>ing the degree <str<strong>on</strong>g>of</str<strong>on</strong>g> elabor<strong>at</strong>i<strong>on</strong><br />
with storm severity.<br />
<strong>Memory</strong> C<strong>on</strong>tent as a Functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> Age<br />
and <str<strong>on</strong>g>Stress</str<strong>on</strong>g><br />
Wh<strong>at</strong> Kind <str<strong>on</strong>g>of</str<strong>on</strong>g> lnform<strong>at</strong>i<strong>on</strong> Is Recalled<br />
by Children and Does This Differ<br />
as a Functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> Age and <str<strong>on</strong>g>Stress</str<strong>on</strong>g> ?<br />
Table 4 displays the mean number <str<strong>on</strong>g>of</str<strong>on</strong>g> proposi-<br />
ti<strong>on</strong>s classified into each c<strong>on</strong>tent c<strong>at</strong>egory (ac-<br />
ti<strong>on</strong>s, descripti<strong>on</strong>s, and internal st<strong>at</strong>es). Children<br />
gener<strong>at</strong>ed a mean <str<strong>on</strong>g>of</str<strong>on</strong>g> 51.92 descripti<strong>on</strong>s, 86.16<br />
acti<strong>on</strong>s, and 13.18 internal st<strong>at</strong>es. A three-way<br />
ANOVA with child age and storm severity as<br />
between-subjects factors and c<strong>on</strong>tent c<strong>at</strong>egory as<br />
a within-subjects factor was c<strong>on</strong>ducted <strong>on</strong> the log<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> the total hurricane-relevant propositi<strong>on</strong>s. Re-<br />
suits indic<strong>at</strong>ed significant main effects <str<strong>on</strong>g>of</str<strong>on</strong>g> child<br />
age, F(1, 94) = 10.97, p = .001, and storm<br />
severity entered as a quadr<strong>at</strong>ic, F(1, 94) = 5.51,<br />
p = .021, as in previous analyses, and a signifi-<br />
cant main effect <str<strong>on</strong>g>of</str<strong>on</strong>g> c<strong>on</strong>tent c<strong>at</strong>egory, F(2, 188) --<br />
677.99, p < .0001. Post hoc analyses indic<strong>at</strong>ed<br />
th<strong>at</strong> children produced significantly more acti<strong>on</strong>s<br />
than descripti<strong>on</strong>s (p < .05) than internal st<strong>at</strong>es<br />
(p < .05). However, there were no interacti<strong>on</strong>s<br />
between c<strong>on</strong>tent c<strong>at</strong>egory and child age or be-<br />
tween c<strong>on</strong>tent c<strong>at</strong>egory and storm severity<br />
(ps > .1). Apparently, age and stress level had<br />
little influence <strong>on</strong> the rel<strong>at</strong>ive amounts <str<strong>on</strong>g>of</str<strong>on</strong>g> acti<strong>on</strong>s,
324 BAHRICK, PARKER, FIVUSH, AND LEVITr<br />
Table 4<br />
<strong>Memory</strong> C<strong>on</strong>tent: Mean Number <str<strong>on</strong>g>of</str<strong>on</strong>g> Propositi<strong>on</strong>s, Standard Devi<strong>at</strong>i<strong>on</strong>s, and Log Transforms (Base 10)<br />
for the Means as a Functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> Storm Severity and C<strong>on</strong>tent C<strong>at</strong>egory<br />
C<strong>on</strong>tent<br />
Low Moder<strong>at</strong>e High Overall<br />
c<strong>at</strong>egory M SD log M SD log M SD log M SD log<br />
Descripti<strong>on</strong>s 45.95 35.11 1.54 57.79 34.37 1.70 52.00 25.46 1.67 51.92 31.95 1.64<br />
Acti<strong>on</strong>s 79.71 68.90 1.77 101.04 63.64 1.94 77.73 47.24 1.83 86.16 60.63 1.85<br />
Internal st<strong>at</strong>es 11.43 9.91 0.96 14.97 8.95 1.14 13.02 9.17 1.06 13.14 9.35 1.05<br />
Note. <str<strong>on</strong>g>The</str<strong>on</strong>g> marginal means shown here were calcul<strong>at</strong>ed by weighting the cells equally to be c<strong>on</strong>sistent with the analyses <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
variance used.<br />
descripti<strong>on</strong>s, and internal st<strong>at</strong>es gener<strong>at</strong>ed by<br />
children.<br />
Did Memories About Hurricane<br />
Prepar<strong>at</strong>i<strong>on</strong>s, Damage, and Repair<br />
Vary as a Functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>Stress</str<strong>on</strong>g>?<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> c<strong>on</strong>tent <str<strong>on</strong>g>of</str<strong>on</strong>g> children's recall was classified<br />
as rel<strong>at</strong>ed or unrel<strong>at</strong>ed to hurricane prepar<strong>at</strong>i<strong>on</strong>,<br />
damage, or repair and recovery (PDR). Table 5<br />
displays the mean number <str<strong>on</strong>g>of</str<strong>on</strong>g> propositi<strong>on</strong>s re-<br />
called by 3- versus 4-year-olds about hurricane<br />
PDR as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> storm severity. Did children<br />
who were more severely stressed talk less about<br />
the direct effects <str<strong>on</strong>g>of</str<strong>on</strong>g> the storm (PDR)? Was<br />
discussi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> this central c<strong>on</strong>tent rel<strong>at</strong>ed to stress<br />
by a quadr<strong>at</strong>ic functi<strong>on</strong>? A three-way ANOVA<br />
with age, storm severity, and c<strong>on</strong>tent (PDR vs.<br />
other c<strong>on</strong>tent) was c<strong>on</strong>ducted <strong>on</strong> the log <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />
total hurricane propositi<strong>on</strong>s recalled. In additi<strong>on</strong><br />
to the main effects <str<strong>on</strong>g>of</str<strong>on</strong>g> age and storm severity<br />
observed previously, children showed a main<br />
effect <str<strong>on</strong>g>of</str<strong>on</strong>g> PDR c<strong>on</strong>tent, F(1, 94) = 256.40, p <<br />
.001. <str<strong>on</strong>g>The</str<strong>on</strong>g>re was significantly more inform<strong>at</strong>i<strong>on</strong><br />
recalled about m<strong>at</strong>erial unrel<strong>at</strong>ed to PDR<br />
(M = 115.3) than m<strong>at</strong>erial rel<strong>at</strong>ed to this c<strong>on</strong>tent<br />
(M = 38.1). Of gre<strong>at</strong>er interest were the two<br />
interacti<strong>on</strong>s th<strong>at</strong> emerged. One was between type<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> c<strong>on</strong>tent and storm severity, F(1, 94) = 7.13,<br />
p = .001. It reflects the fact th<strong>at</strong> although there<br />
was a significant quadr<strong>at</strong>ic functi<strong>on</strong> rel<strong>at</strong>ing storm<br />
severity and recall for both PDR and n<strong>on</strong>PDR<br />
c<strong>on</strong>tent, F(1, 94) = 4.96,p = .028, and F(1, 94) =<br />
5.24, p = .024, respectively, there was also a<br />
str<strong>on</strong>g linear comp<strong>on</strong>ent for recall <str<strong>on</strong>g>of</str<strong>on</strong>g> PDR c<strong>on</strong>-<br />
tent, F(1, 95) = 11.74, p = .001. <str<strong>on</strong>g>The</str<strong>on</strong>g>re was no<br />
evidence <str<strong>on</strong>g>of</str<strong>on</strong>g> a linear comp<strong>on</strong>ent for n<strong>on</strong>PDR<br />
c<strong>on</strong>tent. This effect is better understood by the<br />
Table 5<br />
<strong>Memory</strong> C<strong>on</strong>tent: Mean Number <str<strong>on</strong>g>of</str<strong>on</strong>g> Propositi<strong>on</strong>s, Standard Devi<strong>at</strong>i<strong>on</strong>s, and Log Transforms (Base 10)<br />
for Mean Recall About Hurricane Prepar<strong>at</strong>i<strong>on</strong>, Damage, or Repair (PDR) Versus Other C<strong>on</strong>tent<br />
as a Functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> Age and Storm Severity<br />
Storm severity<br />
Low Moder<strong>at</strong>e High Overall<br />
C<strong>on</strong>tent (n = 18) (n = 42) (n = 40) (N = 100)<br />
and age M SD log M SD log M SD log M SD log<br />
PDR C<strong>on</strong>tent<br />
3years(n = 39) 28.71 39.80 1.15 32.58 18.37<br />
4years(n = 61) 33.09 32.61 1.30 53.48 31.43<br />
Overall(N= 100) 30.90 36.38 1.23 43.03 22.22<br />
Other C<strong>on</strong>tent<br />
3 years (n = 39) 75.29 52.56 1.78 116.90 52.04<br />
4years(n = 61) 141.55 102.65 2.06 151.48 106.70<br />
Overall(N= 100) 108.42 81.56 1.92 134.19 83.94<br />
1.45 40.54 21.82 1.56 33.94 28.32 1.39<br />
1.65 40.15 19.92 1.56 42.24 28.57 1.50<br />
1.55 40.34 20.89 1.56 38.09 27.41 1.45<br />
2.02 75.92 41.04 1.82 89.37 48.84 1.87<br />
2.10 130.52 78.34 2.04 141.18 96.71 2.07<br />
2.06 103.22 62.54 1.93 115.28 76.61 1.97<br />
Note: <str<strong>on</strong>g>The</str<strong>on</strong>g> log transforms may result in numerical differences with respect to the observed means. <str<strong>on</strong>g>The</str<strong>on</strong>g> marginal means shown<br />
here were calcul<strong>at</strong>ed by weighting the cells equally to be c<strong>on</strong>sistent with the analyses <str<strong>on</strong>g>of</str<strong>on</strong>g> variance.
other interacti<strong>on</strong> th<strong>at</strong> emerged, a three-way inter-<br />
acti<strong>on</strong> between type <str<strong>on</strong>g>of</str<strong>on</strong>g> c<strong>on</strong>tent, storm severity,<br />
and child age, F(1, 94) = 3.26, p = .04.<br />
Three-year-olds showed an increasing linear func-<br />
ti<strong>on</strong> for recall <str<strong>on</strong>g>of</str<strong>on</strong>g> PDR c<strong>on</strong>tent as storm severity<br />
increased, F(1, 36) = 7.21, p = .011, whereas<br />
4-year-olds showed the typical quadr<strong>at</strong>ic func-<br />
ti<strong>on</strong>, F(1, 58) = 5.66, p = .021.<br />
Thus, for memories rel<strong>at</strong>ed to PDR, <strong>on</strong>ly the<br />
4-year-olds showed a quadr<strong>at</strong>ic functi<strong>on</strong> rel<strong>at</strong>ing<br />
recall and stress. Three-year-olds showed an<br />
increasing tendency to talk about the PDR c<strong>on</strong>-<br />
tent as hurricane severity increased. Although<br />
these findings were not designed a priori to be<br />
relevant to the liter<strong>at</strong>ure <strong>on</strong> <strong>at</strong>tenti<strong>on</strong>al narrowing<br />
and stress (e.g., Easterbrook, 1959), they do<br />
suggest th<strong>at</strong> there are age differences in the<br />
p<strong>at</strong>terns <str<strong>on</strong>g>of</str<strong>on</strong>g> recall across stress levels for different<br />
types <str<strong>on</strong>g>of</str<strong>on</strong>g> memory c<strong>on</strong>tent. If PDR c<strong>on</strong>tent is<br />
c<strong>on</strong>ceptualized as exclusively central informa-<br />
ti<strong>on</strong>, then the present results suggest some evi-<br />
dence <str<strong>on</strong>g>of</str<strong>on</strong>g> <strong>at</strong>tenti<strong>on</strong>al narrowing with increasing<br />
stress for 3-year-olds but not 4-year-olds. Th<strong>at</strong> is,<br />
they recalled more about hurricane PDR as storm<br />
severity increased, and less about c<strong>on</strong>tent unre-<br />
l<strong>at</strong>ed to prepar<strong>at</strong>i<strong>on</strong>, damage, and repair. It should<br />
be noted, however, th<strong>at</strong>, the remaining (n<strong>on</strong>PDR)<br />
c<strong>on</strong>tent cannot be c<strong>on</strong>sidered peripheral because<br />
it included highly relevant inform<strong>at</strong>i<strong>on</strong> about<br />
other aspects <str<strong>on</strong>g>of</str<strong>on</strong>g> the storm such as the sounds it<br />
made, the people who were present, wh<strong>at</strong> they<br />
did and said, feelings about the storm, and so<br />
forth, as well as peripheral inform<strong>at</strong>i<strong>on</strong>. In the<br />
liter<strong>at</strong>ure, definiti<strong>on</strong>s <str<strong>on</strong>g>of</str<strong>on</strong>g> central inform<strong>at</strong>i<strong>on</strong> vary<br />
widely, ranging from "weap<strong>on</strong> focus," to m<strong>at</strong>e-<br />
rial rel<strong>at</strong>ed to the critical event, to m<strong>at</strong>erial th<strong>at</strong> is<br />
centrally loc<strong>at</strong>ed in the visual field in photos,<br />
slides, and videotaped present<strong>at</strong>i<strong>on</strong>s. A more<br />
objective definiti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> central versus peripheral<br />
inform<strong>at</strong>i<strong>on</strong> is needed if meaningful comparis<strong>on</strong>s<br />
are to be made across studies.<br />
C<strong>on</strong>clusi<strong>on</strong>s<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> results <str<strong>on</strong>g>of</str<strong>on</strong>g> this study present remarkable<br />
evidence <str<strong>on</strong>g>of</str<strong>on</strong>g> l<strong>on</strong>g, detailed accounts <str<strong>on</strong>g>of</str<strong>on</strong>g> a stressful<br />
experience th<strong>at</strong> occurred 2 to 6 m<strong>on</strong>ths earlier.<br />
Three and 4-year-old children recalled an average<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> 153 propositi<strong>on</strong>s about Hurricane Andrew.<br />
This is far gre<strong>at</strong>er than the amount <str<strong>on</strong>g>of</str<strong>on</strong>g> inform<strong>at</strong>i<strong>on</strong><br />
reported in prior studies in which young children<br />
MEMORY FOR A NATURAL DISASTER 325<br />
recounted pers<strong>on</strong>al events after lengthy delays<br />
(e.g., Fivush et al., 1987; Hamm<strong>on</strong>d & Fivush,<br />
1991; Ornstein et al., 1995). This may be due in<br />
part to the dur<strong>at</strong>i<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the hurricane event and the<br />
amount <str<strong>on</strong>g>of</str<strong>on</strong>g> discussi<strong>on</strong> about it. <str<strong>on</strong>g>The</str<strong>on</strong>g> hurricane<br />
event was temporally extended, lasting weeks<br />
from prepar<strong>at</strong>i<strong>on</strong> to posthurricane recovery, espe-<br />
cially for the moder<strong>at</strong>e and high stress groups,<br />
and it was a topic <str<strong>on</strong>g>of</str<strong>on</strong>g> discussi<strong>on</strong> for weeks<br />
afterwards. In fact, families reported th<strong>at</strong> they<br />
discussed the event in fr<strong>on</strong>t <str<strong>on</strong>g>of</str<strong>on</strong>g> their child between<br />
<strong>on</strong>ce a day and several times a day across the 2- to<br />
6-m<strong>on</strong>th retenti<strong>on</strong> interval. Thus, rehearsal may<br />
in part account for the impressive amount <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
recall. Still, these results provide clear evidence<br />
th<strong>at</strong> very young children report a gre<strong>at</strong> deal <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
inform<strong>at</strong>i<strong>on</strong> about highly emoti<strong>on</strong>al and stressful<br />
experiences. Further, although we were unable to<br />
address the issue <str<strong>on</strong>g>of</str<strong>on</strong>g> accuracy objectively, the<br />
little d<strong>at</strong>a we were able to collect showed th<strong>at</strong><br />
mothers judged their children's reports to be<br />
highly accur<strong>at</strong>e.<br />
In this research, we were able to resolve a<br />
number <str<strong>on</strong>g>of</str<strong>on</strong>g> problems typically inherent in c<strong>on</strong>duct-<br />
ing n<strong>at</strong>uralistic studies <strong>on</strong> memory and stress.<br />
<str<strong>on</strong>g>Stress</str<strong>on</strong>g> was objectively defined in terms <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />
degree <str<strong>on</strong>g>of</str<strong>on</strong>g> hurricane severity and children were<br />
classified into three stress levels <strong>on</strong> this basis.<br />
Thus, the problem <str<strong>on</strong>g>of</str<strong>on</strong>g> obtaining an appropri<strong>at</strong>e<br />
low stress c<strong>on</strong>trol group was resolved by includ-<br />
ing families who prepared for the hurricane but<br />
experienced <strong>on</strong>ly a typical storm. Further, be-<br />
cause the hurricane provided an opportune and<br />
objective way <str<strong>on</strong>g>of</str<strong>on</strong>g> measuring stress, we did not<br />
have to rely <strong>on</strong> subjective r<strong>at</strong>ings given by young<br />
children. In additi<strong>on</strong>, our results are not affected<br />
by differences across stress levels in the memory<br />
test because a single test was given <str<strong>on</strong>g>of</str<strong>on</strong>g> a distinct<br />
event th<strong>at</strong> varied in intensity. Finally, there were<br />
no system<strong>at</strong>ic individual differences across stress<br />
groups because assignment to groups was not<br />
system<strong>at</strong>ic, and prior knowledge <str<strong>on</strong>g>of</str<strong>on</strong>g> the event was<br />
minimal and not rel<strong>at</strong>ed to stress level.<br />
In this c<strong>on</strong>text, our research uncovered a<br />
quadr<strong>at</strong>ic rel<strong>at</strong>i<strong>on</strong> between stress and memory.<br />
Children who experienced a moder<strong>at</strong>e level <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
stress recalled the most inform<strong>at</strong>i<strong>on</strong> whereas<br />
those who experienced high or low stress recalled<br />
the least. Overall, recall showed a significant<br />
linear increase from the low to moder<strong>at</strong>e c<strong>on</strong>di-<br />
ti<strong>on</strong>s and a marginally significant decrease from
326 BAHRICK, PARKER, FIVUSH, AND LEVITT<br />
the moder<strong>at</strong>e to high storm severity c<strong>on</strong>diti<strong>on</strong>s.<br />
When memory for the three parts <str<strong>on</strong>g>of</str<strong>on</strong>g> the storm<br />
(the hurricane prepar<strong>at</strong>i<strong>on</strong> period, the storm itself,<br />
and the recovery period) was examined sepa-<br />
r<strong>at</strong>ely, we found th<strong>at</strong> the quadr<strong>at</strong>ic functi<strong>on</strong> was<br />
significant and most pr<strong>on</strong>ounced for memories<br />
surrounding the recovery period, marginally sig-<br />
nificant for memories <str<strong>on</strong>g>of</str<strong>on</strong>g> the prepar<strong>at</strong>i<strong>on</strong> period,<br />
and not significant for memories <str<strong>on</strong>g>of</str<strong>on</strong>g> the hurricane<br />
itself. This lack <str<strong>on</strong>g>of</str<strong>on</strong>g> significance for memories <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
the storm itself may have occurred because for<br />
this period, recall was more c<strong>on</strong>founded by<br />
extraneous factors than for the other time periods.<br />
For example, potential decreases in memory from<br />
the moder<strong>at</strong>e to the high storm severity groups<br />
may have been <str<strong>on</strong>g>of</str<strong>on</strong>g>fset by the increase in the<br />
amount <str<strong>on</strong>g>of</str<strong>on</strong>g> hurricane-rel<strong>at</strong>ed activities during the<br />
storm itself for the high severity group (moving<br />
from room to room to avoid flying objects) and<br />
by the fact more children in the moder<strong>at</strong>e and low<br />
severity groups were able to sleep through por-<br />
ti<strong>on</strong>s <str<strong>on</strong>g>of</str<strong>on</strong>g> the storm. This would result in more<br />
potentially recallable m<strong>at</strong>erial during the storm<br />
itself for children in the high storm severity<br />
c<strong>on</strong>diti<strong>on</strong>. Moreover, analyses indic<strong>at</strong>ed th<strong>at</strong> for<br />
the two time periods where the quadr<strong>at</strong>ic functi<strong>on</strong><br />
was observed, there was a significant linear decrease<br />
in recall from the moder<strong>at</strong>e to the high storm<br />
severity c<strong>on</strong>diti<strong>on</strong>s, and for memory <str<strong>on</strong>g>of</str<strong>on</strong>g> the recov-<br />
ery period, a significant linear increase in recall<br />
from the low to the moder<strong>at</strong>e storm severity<br />
c<strong>on</strong>diti<strong>on</strong>s. Taken together, the findings <str<strong>on</strong>g>of</str<strong>on</strong>g> a<br />
quadr<strong>at</strong>ic functi<strong>on</strong> rel<strong>at</strong>ing amount recalled and<br />
storm severity appear to be most c<strong>on</strong>sistent with<br />
an inverted O-shaped curve where memory first<br />
increases and then decreases as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> stress.<br />
When sp<strong>on</strong>taneous recall was examined sepa-<br />
r<strong>at</strong>ely, the quadr<strong>at</strong>ic functi<strong>on</strong> was most pro-<br />
nounced. Children were most likely to report<br />
differing amounts <str<strong>on</strong>g>of</str<strong>on</strong>g> m<strong>at</strong>erial as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
stress when they answered open-ended questi<strong>on</strong>s.<br />
Here, too, the linear increase in recall from low to<br />
moder<strong>at</strong>e storm severity c<strong>on</strong>diti<strong>on</strong>s was signifi-<br />
cant and the linear decrease from moder<strong>at</strong>e to<br />
high storm severity was marginally significant.<br />
Similar to results <str<strong>on</strong>g>of</str<strong>on</strong>g> overall recall, the quadr<strong>at</strong>ic<br />
functi<strong>on</strong> and the inverted 0-shaped rel<strong>at</strong>i<strong>on</strong> be-<br />
tween recall and stress was most evident in the<br />
sp<strong>on</strong>taneous memories for the posthurricane re-<br />
covery period. <str<strong>on</strong>g>The</str<strong>on</strong>g> amount recalled showed both<br />
a significant linear increase and then a significant<br />
decrease from low to moder<strong>at</strong>e to high storm<br />
severity groups. Thus, results <str<strong>on</strong>g>of</str<strong>on</strong>g> sp<strong>on</strong>taneous<br />
recall parallel and magnify the effects found for<br />
overall recall. <str<strong>on</strong>g>The</str<strong>on</strong>g> quadr<strong>at</strong>ic functi<strong>on</strong> also ap-<br />
proached significance for prompted recall but<br />
was not as robust.<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> quadr<strong>at</strong>ic functi<strong>on</strong> was also observed<br />
under a variety <str<strong>on</strong>g>of</str<strong>on</strong>g> other c<strong>on</strong>diti<strong>on</strong>s and for both 3-<br />
and 4-year-old children. Although the degree <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
memory elabor<strong>at</strong>i<strong>on</strong> was not system<strong>at</strong>ically re-<br />
l<strong>at</strong>ed to stress when all memories were c<strong>on</strong>sid-<br />
ered together, when sp<strong>on</strong>taneous memory was<br />
c<strong>on</strong>sidered al<strong>on</strong>e, the degree <str<strong>on</strong>g>of</str<strong>on</strong>g> elabor<strong>at</strong>i<strong>on</strong> was<br />
rel<strong>at</strong>ed to stress by a quadr<strong>at</strong>ic functi<strong>on</strong>. In<br />
additi<strong>on</strong>, the amount recalled was rel<strong>at</strong>ed to stress<br />
by a quadr<strong>at</strong>ic functi<strong>on</strong> regardless <str<strong>on</strong>g>of</str<strong>on</strong>g> the c<strong>on</strong>tent<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> the children's recollecti<strong>on</strong>s. When recall <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
m<strong>at</strong>erial rel<strong>at</strong>ed to the most stressful aspects <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
the event (PDR) was examined separ<strong>at</strong>ely, the<br />
quadr<strong>at</strong>ic functi<strong>on</strong> was found. This functi<strong>on</strong> was<br />
also obtained for recall <str<strong>on</strong>g>of</str<strong>on</strong>g> m<strong>at</strong>erial unrel<strong>at</strong>ed to<br />
PDR, and thus did not appear to depend <strong>on</strong> the<br />
centrality <str<strong>on</strong>g>of</str<strong>on</strong>g> the c<strong>on</strong>tent. Finally, the curvilinear<br />
functi<strong>on</strong> was observed regardless <str<strong>on</strong>g>of</str<strong>on</strong>g> whether the<br />
c<strong>on</strong>tent <str<strong>on</strong>g>of</str<strong>on</strong>g> children's memories included acti<strong>on</strong>s,<br />
descripti<strong>on</strong>s, or internal st<strong>at</strong>es. Thus, the qua-<br />
dr<strong>at</strong>ic functi<strong>on</strong> rel<strong>at</strong>ing memory and stress was<br />
robust and was observed for children <str<strong>on</strong>g>of</str<strong>on</strong>g> both ages<br />
under a variety <str<strong>on</strong>g>of</str<strong>on</strong>g> c<strong>on</strong>diti<strong>on</strong>s. Further, the qua-<br />
dr<strong>at</strong>ic rel<strong>at</strong>i<strong>on</strong> between recall and stress was not<br />
qualified by the family's SES (mother's or fa-<br />
ther's educ<strong>at</strong>i<strong>on</strong>, value <str<strong>on</strong>g>of</str<strong>on</strong>g> the home, or family<br />
income), the retenti<strong>on</strong> interval, or the amount <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
rehearsal. In fact, the rehearsal <str<strong>on</strong>g>of</str<strong>on</strong>g> storm-rel<strong>at</strong>ed<br />
events was gre<strong>at</strong>est for the families who experi-<br />
enced the gre<strong>at</strong>est storm severity. Th<strong>at</strong> recall was<br />
<strong>at</strong>tenu<strong>at</strong>ed for the high stress group despite the<br />
gre<strong>at</strong>er level <str<strong>on</strong>g>of</str<strong>on</strong>g> rehearsal also <strong>at</strong>tests to the<br />
robustness <str<strong>on</strong>g>of</str<strong>on</strong>g> the curvilinear rel<strong>at</strong>i<strong>on</strong> between<br />
memory and stress.<br />
For most measures, our research revealed th<strong>at</strong><br />
the p<strong>at</strong>terns <str<strong>on</strong>g>of</str<strong>on</strong>g> recall were similar across age, and<br />
4-year-olds recalled c<strong>on</strong>sistently more informa-<br />
ti<strong>on</strong> than 3-year-olds. At both ages, children<br />
showed a curvilinear rel<strong>at</strong>i<strong>on</strong> between amount <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
recall and stress, remembering the most under the<br />
moder<strong>at</strong>e storm severity c<strong>on</strong>diti<strong>on</strong>. At both ages,<br />
they also remembered more acti<strong>on</strong>s, than descrip-<br />
ti<strong>on</strong>s, than internal st<strong>at</strong>es, and they recalled more<br />
about the hurricane afterm<strong>at</strong>h and the hurricane<br />
itself than they did about the hurricane prepara-
ti<strong>on</strong> period. It is interesting th<strong>at</strong> prompting<br />
memory with c<strong>at</strong>egory or specific prompts led to<br />
more than a tw<str<strong>on</strong>g>of</str<strong>on</strong>g>old increase in the number <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
propositi<strong>on</strong>s reported over sp<strong>on</strong>taneous memory,<br />
and this was true <strong>at</strong> both ages as well. One <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />
few excepti<strong>on</strong>s to the parallel p<strong>at</strong>tern across age<br />
was the recall <str<strong>on</strong>g>of</str<strong>on</strong>g> inform<strong>at</strong>i<strong>on</strong> rel<strong>at</strong>ed to the more<br />
stressful aspects <str<strong>on</strong>g>of</str<strong>on</strong>g> the event, hurricane PDR.<br />
Overall, there was a significant quadr<strong>at</strong>ic func-<br />
ti<strong>on</strong> rel<strong>at</strong>ing stress and amount recalled. How-<br />
ever, when recall was broken down by age, the<br />
4-year-olds showed a quadr<strong>at</strong>ic functi<strong>on</strong> and the<br />
3-year-olds showed an increasing linear functi<strong>on</strong>.<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> d<strong>at</strong>a <str<strong>on</strong>g>of</str<strong>on</strong>g> the 3-year-olds were most c<strong>on</strong>sistent<br />
with an <strong>at</strong>tenti<strong>on</strong>al narrowing hypothesis.<br />
This research, thus, provides important and<br />
practical inform<strong>at</strong>i<strong>on</strong> about the n<strong>at</strong>ure <str<strong>on</strong>g>of</str<strong>on</strong>g> memory<br />
for highly stressful experiences. <str<strong>on</strong>g>The</str<strong>on</strong>g> stress gener-<br />
<strong>at</strong>ed by the hurricane was clearly more extreme<br />
than can be studied in experimentally c<strong>on</strong>trolled<br />
or labor<strong>at</strong>ory studies <str<strong>on</strong>g>of</str<strong>on</strong>g> stress. <str<strong>on</strong>g>The</str<strong>on</strong>g> type <str<strong>on</strong>g>of</str<strong>on</strong>g> trauma<br />
gener<strong>at</strong>ed by a hurricane is extended over time<br />
and diverse, including both physical and psycho-<br />
logical stress. Families in the high severity c<strong>on</strong>di-<br />
ti<strong>on</strong> feared physical injury, experienced unpredict-<br />
ability and helplessness during the storm, and<br />
experienced an extended period <str<strong>on</strong>g>of</str<strong>on</strong>g> physical and<br />
emoti<strong>on</strong>al hardship for several days prior to and<br />
several weeks after the storm. Thus, the results <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
this research can be applied to our understanding<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> children's retrospective reports <str<strong>on</strong>g>of</str<strong>on</strong>g> temporally<br />
extended, stressful events th<strong>at</strong> occur outside the<br />
labor<strong>at</strong>ory setting. <str<strong>on</strong>g>The</str<strong>on</strong>g>y highlight the type and<br />
amount <str<strong>on</strong>g>of</str<strong>on</strong>g> m<strong>at</strong>erial presented by children experi-<br />
encing different levels <str<strong>on</strong>g>of</str<strong>on</strong>g> stress and how this<br />
changes with prompting versus open-ended ques-<br />
ti<strong>on</strong>ing and as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> centrality <str<strong>on</strong>g>of</str<strong>on</strong>g> informa-<br />
ti<strong>on</strong>. <str<strong>on</strong>g>The</str<strong>on</strong>g>se findings can educ<strong>at</strong>e forensic inter-<br />
viewers as to the most effective approach to use<br />
as a functi<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> the child's stress. Further, to<br />
better evalu<strong>at</strong>e the trade-<str<strong>on</strong>g>of</str<strong>on</strong>g>fs between open-<br />
ended versus prompted questi<strong>on</strong>ing as a functi<strong>on</strong><br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> stress, future research must assess the impact<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> these variables <strong>on</strong> the accuracy <str<strong>on</strong>g>of</str<strong>on</strong>g> the informa-<br />
ti<strong>on</strong> reported.<br />
In c<strong>on</strong>trast with prior research <strong>on</strong> the rel<strong>at</strong>i<strong>on</strong><br />
between memory and stress, the range <str<strong>on</strong>g>of</str<strong>on</strong>g> stress<br />
experienced by children in our study is most<br />
likely more broad than th<strong>at</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> most lab and field<br />
studies <str<strong>on</strong>g>of</str<strong>on</strong>g> event memory. During the hurricane,<br />
some children actually feared for their lives and<br />
MEMORY FOR A NATURAL DISASTER 327<br />
pers<strong>on</strong>al safety whereas others experienced <strong>on</strong>ly<br />
a normal storm. Prior research has <str<strong>on</strong>g>of</str<strong>on</strong>g>ten classi-<br />
fied children into <strong>on</strong>ly two levels <str<strong>on</strong>g>of</str<strong>on</strong>g> stress or<br />
assessed individual differences in stress to a<br />
single event, restricting the range <str<strong>on</strong>g>of</str<strong>on</strong>g> stress<br />
sampled. Some studies have found th<strong>at</strong> stress<br />
does not impact memory or can even enhance<br />
memory (e.g., Goodman et al., 1987; Goodman,<br />
Hirschman, et al., 1991), whereas others have<br />
found th<strong>at</strong> it impairs memory (e.g., Peters, 1987,<br />
1991). By classifying children according to three<br />
levels <str<strong>on</strong>g>of</str<strong>on</strong>g> stress, we were able to identify a<br />
quadr<strong>at</strong>ic rel<strong>at</strong>i<strong>on</strong> between memory and stress<br />
most c<strong>on</strong>sistent with an inverted U-shaped func-<br />
ti<strong>on</strong> th<strong>at</strong> cannot be revealed by assessing memory<br />
across <strong>on</strong>ly two stress levels or by sampling a<br />
narrow range <str<strong>on</strong>g>of</str<strong>on</strong>g> stress.<br />
In summary, children reported a gre<strong>at</strong> deal <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
inform<strong>at</strong>i<strong>on</strong> about a stressful event and the level<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> stress influenced the amount recalled several<br />
m<strong>on</strong>ths l<strong>at</strong>er. However, to assert th<strong>at</strong> moder<strong>at</strong>e<br />
levels <str<strong>on</strong>g>of</str<strong>on</strong>g> stress enhance memory and high levels<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> stress disrupt memory would be an oversimpli-<br />
fic<strong>at</strong>i<strong>on</strong>. It is apparent th<strong>at</strong> this highly emoti<strong>on</strong>al<br />
and important event elicited an impressive amount<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g> recall, even by those in the high storm severity<br />
c<strong>on</strong>diti<strong>on</strong>, rel<strong>at</strong>ive to studies <str<strong>on</strong>g>of</str<strong>on</strong>g> other events<br />
recounted by children. Further, the fact th<strong>at</strong><br />
children under all stress levels recounted much<br />
more inform<strong>at</strong>i<strong>on</strong> after being prompted points out<br />
th<strong>at</strong> additi<strong>on</strong>al memories <str<strong>on</strong>g>of</str<strong>on</strong>g> even the most stress-<br />
ful events can be elicited by prompting. Thus, the<br />
finding <str<strong>on</strong>g>of</str<strong>on</strong>g> a quadr<strong>at</strong>ic rel<strong>at</strong>i<strong>on</strong> between memory<br />
and stress should be viewed in the c<strong>on</strong>text <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />
large number <str<strong>on</strong>g>of</str<strong>on</strong>g> memories reported by children<br />
about a highly emoti<strong>on</strong>al and temporally ex-<br />
tended, n<strong>at</strong>uralistic event.<br />
References<br />
Baker-Ward, L., Gord<strong>on</strong>, B. N., Ornstein, P. A., Larus,<br />
D. M., & Clubb, P. A. (1993). <strong>Young</strong> children's<br />
l<strong>on</strong>g-term retenti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> a pedi<strong>at</strong>ric examin<strong>at</strong>i<strong>on</strong>.<br />
Child Development, 64, 1519-1533.<br />
Bauer, P. J., & Fivush, R. (1992). C<strong>on</strong>structing event<br />
represent<strong>at</strong>i<strong>on</strong>s: Building <strong>on</strong> a found<strong>at</strong>i<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> varia-<br />
ti<strong>on</strong> and enabling rel<strong>at</strong>i<strong>on</strong>s. Cognitive Development,<br />
7, 381-401.<br />
Bauer, P. J., & Mandler, J. M. (1989). One thing<br />
follows another: <str<strong>on</strong>g>Effects</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> temporal structure <strong>on</strong> 1-<br />
to 2-year-olds' recall <str<strong>on</strong>g>of</str<strong>on</strong>g> events. Developmental<br />
Psychology, 25, 197-206.
328 BAHRICK, PARKER, FIVUSH, AND LEVITT<br />
Bauer, P. J., & Travis, L. L. (1993). <str<strong>on</strong>g>The</str<strong>on</strong>g> fabric <str<strong>on</strong>g>of</str<strong>on</strong>g> an<br />
event: Different sources <str<strong>on</strong>g>of</str<strong>on</strong>g> temporal invariance<br />
differentially affect 24-m<strong>on</strong>th-olds' recall. Cogni-<br />
tive Development, 8, 319-341.<br />
Bjorkhnd, D. E (1985). <str<strong>on</strong>g>The</str<strong>on</strong>g> role <str<strong>on</strong>g>of</str<strong>on</strong>g> c<strong>on</strong>ceptual<br />
knowledge in the development <str<strong>on</strong>g>of</str<strong>on</strong>g> organiz<strong>at</strong>i<strong>on</strong> in<br />
children's memory. In C. J. Brainerd & M. Pressley<br />
(Eds.), Basic processes in memory development (pp.<br />
103-142). New York: Springer-Verlag.<br />
Ceci, S. J., & Bruck, M. (1993). <str<strong>on</strong>g>The</str<strong>on</strong>g> suggestibility <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
the child witness: An historical review and synthe-<br />
sis. Psychological Bulletin, 113, 403--439.<br />
Ceci, S. J., Toglia, M. P., & Ross, D. E (1987).<br />
Children 's eyewitness memory. New York: Springer-<br />
Verlag.<br />
Chi, M. T. H., & Ceci, S. J. (1987). C<strong>on</strong>tent knowl-<br />
edge: Its role, represent<strong>at</strong>i<strong>on</strong>, and restructuring in<br />
memory development. In H. W. Reese (Ed.), Ad-<br />
vances in child development and behavior, 20, (pp.<br />
91-142). Orlando, FL: Academic Press.<br />
Christians<strong>on</strong>, S.-A. (1992). Emoti<strong>on</strong>al stress and eye-<br />
witness memory: A critical review. Psychological<br />
Bulletin, 112, 284-309.<br />
Christians<strong>on</strong>, S.-A., & L<str<strong>on</strong>g>of</str<strong>on</strong>g>tus, E. E (1987). <strong>Memory</strong><br />
for traum<strong>at</strong>ic events. Applied Cognitive Psychology,<br />
1, 225-239.<br />
Christians<strong>on</strong>, S.-A., & L<str<strong>on</strong>g>of</str<strong>on</strong>g>tus, E. E (1990). Some<br />
characteristics <str<strong>on</strong>g>of</str<strong>on</strong>g> people's traum<strong>at</strong>ic memories. Bul-<br />
letin <str<strong>on</strong>g>of</str<strong>on</strong>g> the Psych<strong>on</strong>omic Society, 28, 195-198.<br />
Christians<strong>on</strong>, S.-A., & L<str<strong>on</strong>g>of</str<strong>on</strong>g>tus, E. E (1991). Remember-<br />
ing emoti<strong>on</strong>al events: <str<strong>on</strong>g>The</str<strong>on</strong>g> f<strong>at</strong>e <str<strong>on</strong>g>of</str<strong>on</strong>g> detailed informa-<br />
ti<strong>on</strong>. Emoti<strong>on</strong> and Cogniti<strong>on</strong>, 5, 81-108.<br />
Christians<strong>on</strong>, S.-A., & Nilss<strong>on</strong>, L. G. (1984). Func-<br />
ti<strong>on</strong>al amnesia as induced by a psychological trauma.<br />
<strong>Memory</strong> and Cogniti<strong>on</strong>, 12, 142-155.<br />
Clifford, B., & Hollin, C. (1981). <str<strong>on</strong>g>Effects</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> the type <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
incident and the number <str<strong>on</strong>g>of</str<strong>on</strong>g> perpelr<strong>at</strong>ors <strong>on</strong> eyewit-<br />
ness memory. Journal <str<strong>on</strong>g>of</str<strong>on</strong>g> Applied Psychology, 66,<br />
364-370.<br />
Clifford, B., & Scott, J. (1978). Individual and situ-<br />
<strong>at</strong>i<strong>on</strong>al factors in eyewitness testim<strong>on</strong>y. Journal <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
Applied Psychology, 63, 352-359.<br />
Clubb, P. A., Nida, R. E., Merritt, K., & Ornstein, P. A.<br />
(1993). Visiting the doctor: <strong>Children's</strong> knowledge<br />
and memory. Cognitive Development, 8, 361-372.<br />
Earls, F., Smith, E., Reich, W., & Jung, K. G. (1988).<br />
Investig<strong>at</strong>ing psychop<strong>at</strong>hological c<strong>on</strong>sequences <str<strong>on</strong>g>of</str<strong>on</strong>g> a<br />
disaster in children: A pilot study incorpor<strong>at</strong>ing a<br />
structure diagnostic interview. Journal <str<strong>on</strong>g>of</str<strong>on</strong>g> the Ameri-<br />
can Academy <str<strong>on</strong>g>of</str<strong>on</strong>g> Child and Adolescent Psychi<strong>at</strong>ry,<br />
27, 90-95.<br />
Easterbrook, J. A. (1959). <str<strong>on</strong>g>The</str<strong>on</strong>g> effect <str<strong>on</strong>g>of</str<strong>on</strong>g> emoti<strong>on</strong> <strong>on</strong> the<br />
utiliz<strong>at</strong>i<strong>on</strong> and organiz<strong>at</strong>i<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> behavior. Psychologi-<br />
cal Review, 66, 183-201.<br />
Fivush, R. (1993). Developmental perspectives <strong>on</strong><br />
autobiographical recall. In G. Goodman & R.<br />
Bottoms (Eds.), ChiM victims, chiM witnesses:<br />
Understanding and improving testim<strong>on</strong>y (pp. 1-24).<br />
New York: Guilford Press.<br />
Fivush, R., Gray, J., & Fromh<str<strong>on</strong>g>of</str<strong>on</strong>g>f, F. (1987). Two-year-<br />
olds talk about the past. Cognitive Development, 2,<br />
393-409.<br />
Fivush, R., Kuebli, J., & Clubb, P. A. (1992). <str<strong>on</strong>g>The</str<strong>on</strong>g><br />
structure <str<strong>on</strong>g>of</str<strong>on</strong>g> events and event represent<strong>at</strong>i<strong>on</strong>s: A<br />
developmental analysis. Child Development, 63,<br />
188-201.<br />
Frederick, C. J. (1985). Children traum<strong>at</strong>ized by<br />
c<strong>at</strong>astrophic situ<strong>at</strong>i<strong>on</strong>s. In S. Eth & R. S. Pynoos<br />
(Eds.), Posttraum<strong>at</strong>ic stress disorders in children<br />
(pp. 73-99). Washingt<strong>on</strong>, DC: American Psychi<strong>at</strong>-<br />
ric Press.<br />
Goodman, G. S., Aman, C., & Hirschman, J. (1987).<br />
Child sexual and physical abuse: <strong>Children's</strong> testi-<br />
m<strong>on</strong>y. In S. J. Ceci, M. P. Toglia, & D. E Ross<br />
(Eds.), <strong>Children's</strong> eyewitness memory (pp. 1-23).<br />
New York: Springer-Vedag.<br />
Goodman, G. S., Bottoms, B. L., Schwartz-Kenney,<br />
B. M., & Rudy, L. (1991). <strong>Children's</strong> testim<strong>on</strong>y<br />
about a stressful event: Improving children's reports.<br />
Journal <str<strong>on</strong>g>of</str<strong>on</strong>g> Narr<strong>at</strong>ive and Life History, 1, 69-99.<br />
Goodman, G. S., Hirschman, J. E., Hepps, D., & Rudy,<br />
L. (1991). <strong>Children's</strong> memory for stressful events.<br />
Merrill Palmer Quarterly, 37, 109-157.<br />
Green, B. L., Korol, M., Grace, M. C., Vary, M. G.,<br />
Le<strong>on</strong>ard, A. C., Gleser, G. C., & Smits<strong>on</strong>-Cohen, S.<br />
(1991). Children and disaster: Gender and parental<br />
effects <strong>on</strong> PTSD symptoms. Journal <str<strong>on</strong>g>of</str<strong>on</strong>g> the Ameri-<br />
can Academy <str<strong>on</strong>g>of</str<strong>on</strong>g> Child and Adolescent Psychi<strong>at</strong>ry,<br />
30, 945-951.<br />
Ham<strong>on</strong>d, N. R., & Fivush, R. (1991). Memories <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
Mickey Mouse: <strong>Young</strong> children recount their trip to<br />
Disneyworld. Cognitive Development, 6, 433--448.<br />
Heuer, F., & Reisberg, D. (1990). Vivid memories <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
emoti<strong>on</strong>al events: <str<strong>on</strong>g>The</str<strong>on</strong>g> accuracy <str<strong>on</strong>g>of</str<strong>on</strong>g> remembered<br />
minutiae. <strong>Memory</strong> and Cogniti<strong>on</strong>, 18, 496-506.<br />
Heuer, F., & Reisberg, D. (1992). Emoti<strong>on</strong>, arousal,<br />
and memory for detail. In S. Christians<strong>on</strong> (Ed.), <str<strong>on</strong>g>The</str<strong>on</strong>g><br />
handbook <str<strong>on</strong>g>of</str<strong>on</strong>g> emoti<strong>on</strong> and memory: Research and<br />
theory (pp. 151-180). Hillsdale, NJ: Erlbaum.<br />
Howe, M. L., Courage, M. L., & Peters<strong>on</strong>, C. (1994).<br />
How can I remember when 'T' wasn't there:<br />
L<strong>on</strong>g-term retenti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> traum<strong>at</strong>ic experiences and
emergence <str<strong>on</strong>g>of</str<strong>on</strong>g> the cognitive self. C<strong>on</strong>sciousness and<br />
Cogniti<strong>on</strong>, 3, 327-355.<br />
Howe, M. L., Courage, M. L., & Peters<strong>on</strong>, C. (1995).<br />
Intrusi<strong>on</strong>s in preschoolers' recall <str<strong>on</strong>g>of</str<strong>on</strong>g> traum<strong>at</strong>ic childhood<br />
events. Psych<strong>on</strong>omic Bulletin & Review, 2, 130-134.<br />
Huds<strong>on</strong>, J. A., & Fivush, R. (1991). As time goes by:<br />
Sixth graders remember a kindergarten experience.<br />
Applied Cognitive Psychology, 5, 346-360.<br />
Huds<strong>on</strong>, J. A., & Nels<strong>on</strong>, K. (1983). <str<strong>on</strong>g>Effects</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> script<br />
structure <strong>on</strong> children's story recall. Developmental<br />
Psychology, 19, 625-635.<br />
Kebeck, G., & Lohaus, A. (1986). Effect <str<strong>on</strong>g>of</str<strong>on</strong>g> emoti<strong>on</strong>al<br />
arousal <strong>on</strong> free recall <str<strong>on</strong>g>of</str<strong>on</strong>g> complex m<strong>at</strong>erial. Percep-<br />
tual and Motor Skills, 63, 461--462.<br />
L<str<strong>on</strong>g>of</str<strong>on</strong>g>tus, E. E, & Bums, T. E. (1982). Mental shock can<br />
produce retrograde amnesia. <strong>Memory</strong> and Cogni-<br />
ti<strong>on</strong>, 10, 318-323.<br />
L<strong>on</strong>igan, C. J., Shann<strong>on</strong>, M. P., Taylor, C. M., Finch,<br />
A. J., Jr., & Sallee, E R. (1994). Children exposed<br />
to disaster: II. Risk factors for the development <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
post-traum<strong>at</strong>ic symptom<strong>at</strong>ology. Journal <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />
American Academy <str<strong>on</strong>g>of</str<strong>on</strong>g> Child and Adolescent Psy-<br />
chi<strong>at</strong>ry, 33, 94-105.<br />
Malmquist, C. P. (1986). Children who witness paren-<br />
tal murder: Post-traum<strong>at</strong>ic aspects. Journal <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />
American Academy <str<strong>on</strong>g>of</str<strong>on</strong>g> Child Psychi<strong>at</strong>ry, 25, 320-325.<br />
McFarlane, A. C. (1987). Posttraum<strong>at</strong>ic phenomena in<br />
a l<strong>on</strong>gitudinal study <str<strong>on</strong>g>of</str<strong>on</strong>g> children following a n<strong>at</strong>ural<br />
disaster. Journal <str<strong>on</strong>g>of</str<strong>on</strong>g> the American Academy <str<strong>on</strong>g>of</str<strong>on</strong>g> Child<br />
and Adolescent Psychi<strong>at</strong>ry, 26, 764-769.<br />
Merritt, K., Omstein, P. A., & Spicker, B. (1994).<br />
<strong>Children's</strong> memory for a salient medical procedure:<br />
Implic<strong>at</strong>i<strong>on</strong>s for testim<strong>on</strong>y. Pedi<strong>at</strong>rics, 94, 17-23.<br />
Ornstein, P. A., Merritt, K., & Baker-Ward, L. (1995,<br />
July). <strong>Children's</strong> recollecti<strong>on</strong>s <str<strong>on</strong>g>of</str<strong>on</strong>g> medical experi-<br />
ences: Exploring the linkage between stress and<br />
memory. In J. E Parker (Chair), <str<strong>on</strong>g>Effects</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g> stress and<br />
arousal up<strong>on</strong> children's memories. Symposium c<strong>on</strong>-<br />
ducted <strong>at</strong> the first biennial meeting <str<strong>on</strong>g>of</str<strong>on</strong>g> the Society<br />
for Applied Research in <strong>Memory</strong> and Cogniti<strong>on</strong>,<br />
Vancouver, British Columbia, Canada.<br />
Peters, D. P. (1987). <str<strong>on</strong>g>The</str<strong>on</strong>g> impact <str<strong>on</strong>g>of</str<strong>on</strong>g> n<strong>at</strong>urally occurring<br />
stress <strong>on</strong> children's memory. In S. J. Ceci, M. P.<br />
Toglia, & D. E Ross (Eds.), <strong>Children's</strong> eyewitness<br />
memory (pp. 122-141). New York: Springer-Verlag.<br />
Peters, D. P. (1991). <str<strong>on</strong>g>The</str<strong>on</strong>g> influence <str<strong>on</strong>g>of</str<strong>on</strong>g> stress and<br />
arousal <strong>on</strong> the child witness. In J. L. Doris (Ed.),<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> suggestibility <str<strong>on</strong>g>of</str<strong>on</strong>g> children's recollecti<strong>on</strong>s (pp.<br />
60-76). Washingt<strong>on</strong>, DC: American Psychological<br />
Associ<strong>at</strong>i<strong>on</strong>.<br />
MEMORY FOR A NATURAL DISASTER 329<br />
(Appendix follows <strong>on</strong> next page)<br />
Peters, D. E (1997). <str<strong>on</strong>g>Stress</str<strong>on</strong>g>, arousal, and children's<br />
eyewitness memory. In N. L. Stein, E A. Ornstein,<br />
B. Tversky, & C. J. Brainerd (Eds.), <strong>Memory</strong> for<br />
everyday and emoti<strong>on</strong>al events (pp. 351-370).<br />
Hillsdale, NJ: Erlbaum.<br />
Pillemer, D. B. (1984). Flashbulb memories <str<strong>on</strong>g>of</str<strong>on</strong>g> the<br />
assassin<strong>at</strong>i<strong>on</strong> <strong>at</strong>tempt <str<strong>on</strong>g>of</str<strong>on</strong>g> President Reagan. Cogni-<br />
ti<strong>on</strong>, 16, 63-80.<br />
Pillemer, D. B., Picariello, M., & Pruett, J. C. (1995).<br />
Very l<strong>on</strong>g-term memories <str<strong>on</strong>g>of</str<strong>on</strong>g> a salient childhood<br />
event. Applied Cognitive Psychology, 8, 95-106.<br />
Pillemer, D. B., & White, S. H. (1989). Childhood<br />
events recalled by children and adults. In H. W.<br />
Reese (Ed.), Advances in child development and<br />
behavior: Vol. 21 (pp. 297-340). Orlando, FL:<br />
Academic Press.<br />
Price, D. W. W., & Goodman, G. S. (1990). Visiting<br />
the wizard: <strong>Children's</strong> memory <str<strong>on</strong>g>of</str<strong>on</strong>g> a recurring event.<br />
Child Development, 61, 664--680.<br />
Pynoos, R. S., Fredrick, C., Nader, K., Arroyo, W.,<br />
Eth, S., Nunez, W., Steinberg, A., & Fairbanks, L.<br />
(1987). Life thre<strong>at</strong> and posttraum<strong>at</strong>ic stress in<br />
school age children. Archives <str<strong>on</strong>g>of</str<strong>on</strong>g> General Psychia-<br />
try, 44, 1057-1063.<br />
Rubin, D. C., & Kozin, M. (1984). Vivid memories.<br />
Cogniti<strong>on</strong>, 16, 81-95.<br />
Sheingold, K., & Tenney, Y. J. (1982). <strong>Memory</strong> for a<br />
salient childhood event. In U. Neisser (Ed.), <strong>Memory</strong><br />
observed (pp. 201-212). San Francisco: Freeman.<br />
Slackman, E. A., Huds<strong>on</strong>, J. A., & Fivush, R. (1986).<br />
Acti<strong>on</strong>s, actors, links, and goals: <str<strong>on</strong>g>The</str<strong>on</strong>g> structure <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
children's event represent<strong>at</strong>i<strong>on</strong>s. In K. Nels<strong>on</strong> (Ed.),<br />
Event knowledge: Structure and functi<strong>on</strong> in develop-<br />
ment (pp. 47-69). Hillsdale, NJ: Erlbaum.<br />
Terr, L. C. (1987). Childhood psychic trauma. In J. D.<br />
Noshpitz (Ed.), Basic handbook <str<strong>on</strong>g>of</str<strong>on</strong>g> child psychi<strong>at</strong>ry,<br />
Vol. 5 (pp. 262-272). New York: Basic Books.<br />
Todd, C., & Perlmutter, M. (1980). Reality recalled by<br />
preschool children. In M. Perlmutter (Ed.), Chil-<br />
dren's memory: Vol. 10. New directi<strong>on</strong>s for child<br />
development (pp. 69-85). San Francisco: Jossey-Bass.<br />
Vandermaas, M. O., Hess, T. M., & Baker-Ward, L.<br />
(1993). Does anxiety affect children's reports <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
memory for a stressful event? Applied Cognitive<br />
Psychology, 7, 109-127.<br />
Vernberg, E. M., La Greca, A. M., Silverman, W. K., &<br />
Prinstein, M. J. (1996). Predicti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> posttraum<strong>at</strong>ic<br />
stress symptoms in children after Hurricane An-<br />
drew. Journal <str<strong>on</strong>g>of</str<strong>on</strong>g> Abnormal Psychology, 105, 237-<br />
248.
330 BAHRICK, PARKER, FIVUSH, AND LEVITT<br />
Open-ended general questi<strong>on</strong><br />
"Remember the really big hurricane we had<br />
here? I'm really interested in wh<strong>at</strong> kids like you<br />
remember about th<strong>at</strong> hurricane. I've already<br />
talked to some kids about wh<strong>at</strong> happened and<br />
they told me all kinds <str<strong>on</strong>g>of</str<strong>on</strong>g> interesting things. I<br />
bet you remember a lot about the hurricane, too.<br />
Can you think really hard about the hurricane<br />
and tell me everything th<strong>at</strong> you can remember<br />
about it?" (Follow up with n<strong>on</strong>directive prompts<br />
[NDP]; "Wh<strong>at</strong> else?," "Anything more?,"<br />
"Anything else?," "Wh<strong>at</strong>?," and "Any other<br />
things?")<br />
"You really remembered a lot about wh<strong>at</strong> hap-<br />
pened. Th<strong>at</strong>'s gre<strong>at</strong>!"<br />
Prehurricane prepar<strong>at</strong>i<strong>on</strong>s<br />
"Can you tell me wh<strong>at</strong> you and your family did to<br />
get ready before the hurricane came? Think about<br />
all the things th<strong>at</strong> you and your mommy and<br />
daddy did to get ready for the hurricane. Tell me<br />
everything you can. Wh<strong>at</strong> was the very first thing<br />
th<strong>at</strong> your family did?" (NDP)<br />
(If the child did not menti<strong>on</strong> the following items,<br />
the specific questi<strong>on</strong>s were asked and were<br />
followed by NDPs.)<br />
"Wh<strong>at</strong> did your mommy and daddy do outside<br />
the house?"<br />
"Did they bring anything inside? Plants? Lawn<br />
furniture?"<br />
"Wh<strong>at</strong> did your mommy and daddy do inside the<br />
house? Wh<strong>at</strong> about the rugs? Furniture? W<strong>at</strong>er?"<br />
"Wh<strong>at</strong> special things did your mommy and<br />
daddy buy to get ready for the hurricane?" (NDP)<br />
"Wh<strong>at</strong> about groceries?"<br />
"Did you help your family get ready for the<br />
hurricane? Wh<strong>at</strong> did you do?"<br />
"Was anybody <strong>at</strong> your house before the storm<br />
besides who usually lives there? Did they help<br />
you get ready?"<br />
"Wh<strong>at</strong> was the most scary thing about getting<br />
ready for the hurricane?"<br />
"Wh<strong>at</strong> was the most fun thing about getting<br />
ready for the hurricane?"<br />
Appendix<br />
Hurricane <strong>Memory</strong> Interview<br />
<str<strong>on</strong>g>The</str<strong>on</strong>g> hurricane itself<br />
"Wh<strong>at</strong> do you remember about wh<strong>at</strong> happened<br />
when the hurricane finally came? Tell me every-<br />
thing you can." (NDP) "And then wh<strong>at</strong>?"<br />
(If the child did not menti<strong>on</strong> the following items,<br />
the specific questi<strong>on</strong>s were asked and were<br />
followed by NDPs.)<br />
"Were you awake during any <str<strong>on</strong>g>of</str<strong>on</strong>g> the hurricane?<br />
Wh<strong>at</strong> part? Tell me about it."<br />
"Where did you stay in the house?"<br />
"Where did your mommy and daddy stay in the<br />
house?"<br />
"Did you hide or put something over you?"<br />
"Wh<strong>at</strong> was it like in the house?"<br />
Lights? Candles? Flashlights? Radio? TV? Wa-<br />
ter? Rainw<strong>at</strong>er? Anything break or fall?<br />
"Wh<strong>at</strong> was it like outside during the hurricane?"<br />
"Could you see anything?" Wind? Noises? Wh<strong>at</strong><br />
were the noises from?<br />
"Was anybody <strong>at</strong> your house during the hurricane<br />
besides who usually lives there?"<br />
"Wh<strong>at</strong> was the scariest thing during the hurri-<br />
cane?"<br />
"Wh<strong>at</strong> was the most fun thing during the hurri-<br />
cane?"<br />
Posthurricane recovery period<br />
"Okay, now I want you to tell me all the things<br />
you can remember from after the hurricane was<br />
over. Tell me everything th<strong>at</strong> you can think <str<strong>on</strong>g>of</str<strong>on</strong>g>."<br />
(If the child did not menti<strong>on</strong> the following items,<br />
the specific questi<strong>on</strong>s were asked and were<br />
followed by NDPs.)<br />
"Wh<strong>at</strong> did the inside <str<strong>on</strong>g>of</str<strong>on</strong>g> your house look like?"<br />
Rainw<strong>at</strong>er? Carpets wet? Windows? Ro<str<strong>on</strong>g>of</str<strong>on</strong>g>?. Any-<br />
thing broken?<br />
"Now try to remember wh<strong>at</strong> is was like when you<br />
first looked outside and the hurricane was over.<br />
Wh<strong>at</strong> did it look like? Wh<strong>at</strong> was in your yard th<strong>at</strong><br />
was different?"<br />
Trees? Junk?<br />
"Wh<strong>at</strong> did your mommy and daddy do?"<br />
"Did you help with anything? Wh<strong>at</strong> did you do?"<br />
"Think about wh<strong>at</strong> happened those first few days<br />
after the hurricane. Tell me wh<strong>at</strong> th<strong>at</strong> was like.<br />
Think <str<strong>on</strong>g>of</str<strong>on</strong>g> all the things th<strong>at</strong> were different."
Food? Cooking? Drinking? Ice? Refriger<strong>at</strong>or?<br />
B<strong>at</strong>h time? Air c<strong>on</strong>diti<strong>on</strong>ing? TV and movies?<br />
"Did you stay in your own house?" If no,<br />
"Where did you go?" If yes, "Did some<strong>on</strong>e<br />
come to stay with you?"<br />
"Wh<strong>at</strong> was the scariest thing after the hurricane<br />
was over?"<br />
MEMORY FOR A NATURAL DISASTER 331<br />
"Wh<strong>at</strong> was the most fun thing after the hurricane<br />
was over?"<br />
Received October 28, 1996<br />
Revisi<strong>on</strong> received January 23, 1998<br />
Accepted January 30, 1998 •