Beruflich Dokumente
Kultur Dokumente
Introduction
Whereas clinically based reports on ospring of Holocaust survivors point to
intergenerational transmission of traumatic experiences, more controlled studies have
not found much psychopathology (Van IJzendoorn et al., 2003), except when secondgeneration subjects were confronted with life-threatening situations (Baider, Peretz,
Hadani, Perry, Avramov, & De Nour, 2000; Solomon, Kotler, & Mikulincer, 1988; see
Van IJzendoorn et al., 2003). Thus, a germane question is whether the trauma has been
passed on to the ospring in the third generation (Bar-On et al., 1998). The question is
of critical importance because of the large scale of the Holocaust. The existence of longterm psychological eects of the Holocaust on the survivors and their ospring still keeps
the scientic and clinical literature divided (Bar-On et al., 1998).
This issue is not only important for the study of the Holocaust. Unfortunately, during
the past century, genocide has not been restricted to World War II. In the International
handbook of multigenerational legacies of trauma (Danieli, 1998), genocides in Armenia,
*Corresponding author. Email: sagi@psy.haifa.ac.il or VANIJZEN@FSW.leidenuniv.nl or
BAKERMANS@FSW.leidenuniv.nl
ISSN 1461-6734 print/ISSN 1469-2988 online
2008 Taylor & Francis
DOI: 10.1080/14616730802113661
http://www.informaworld.com
106
A. Sagi-Schwartz et al.
Cambodia, former Yugoslavia, Rwanda, and Nigeria are mentioned, and these constitute
only a selection. The Holocaust was a unique genocide because of its scale, its almost
industrial design, and its uselessness in political, economic, or military respects (Lacqueur,
2001). Nevertheless, we may learn from the Holocaust survivors and their (grand-)children
about the long-term and intergenerational eects of traumatic experiences, and develop
insights into the fate of survivors and ospring of more recent and future genocidal
catastrophes.
In order to resolve the divergence of the clinical and non-clinical ndings on
intergenerational transmission of trauma, between qualitative and quantitative approaches, and between methodologically more robust versus less robust studies, we
developed a programmatic series of studies to address the issue. We began with our rst
published study (Sagi-Schwartz et al., 2003) in which careful matching of Holocaust
survivors and comparisons was employed to form a research design with three generations,
including 98 families with grandmother, mother, and their infant, who were engaged in
attachment- and trauma-related interviews, questionnaires, and observational procedures.
Holocaust child survivors (now grandmothers) showed severe signs of traumatic stress and
more often lack of resolution of trauma than comparisons, but they were not impaired in
their general adaptation. Also, the traumatic eects did not transmit across second
generation (mothers) and third generation (infants). We concluded that Holocaust child
survivors may have been able to protect their ospring from their war experiences,
although they themselves still suered from the Holocaust.
Because the ndings of our rst study ran against the wide spread belief that Holocaust
survivors do transmit the trauma even across more than one generation, we felt that the eld
might be reluctant to accept our conclusions despite the fact that very robust sampling and
methodology had been employed in our study (for more details, see Sagi-Schwartz et al.,
2003). Therefore our next step was to carry out a series of meta-analyses on 32 samples
involving 4418 second generation participants; the hypothesis was tested that secondary
traumatization in Holocaust survivor families exist. We found that, largely, there is no
evidence for secondary traumatization (Van IJzendoorn et al., 2003). More specically,
secondary traumatization emerged only in studies that consisted of participants who were
recruited through clinical practices or Holocaust survivor support groups and similar
organizations (select samples). In a set of adequately designed studies, with what was
referred to as non-select samples, no evidence for the inuence of the parents traumatic
Holocaust experiences on their children was found. A stress-diathesis model was used to
interpret the remarkable resilience in Holocaust survivors and their ospring (Paris, 2000).
Given the fact that there does not seem to be solid empirical evidence for a higher
prevalence of psychological problems in the second-generation ospring, it may be argued
that no transmission to the third generation should be expected either, especially with nonselect samples. At the same time, studies such as Multigenerational occurrence of
survivor syndrome symptoms in families of Holocaust survivors (Rubenstein, Cutter, &
Templer, 19891990) and The intergenerational transmission of increased anxiety traits
in third-generation Holocaust survivors (Wetter, 1998), characterize many of the existing
reports about the intergenerational eects of the Holocaust. Such far reaching statements
have become ingrained to a large extent in the belief system within the professional as well
as the wider community (not necessarily with sound conceptual or theoretical basis for
such a transmission) and therefore there is a need to set the record straight on the issue of
tertiary traumatization by an integrative review and analysis of all available studies.
Also, because some scholars still argue that intergenerational transmission of profound
traumas might skip a generation, using metaphorically the case of some biologically
107
inherited diseases (Bulmer, 1998), the eld still seems to beg for further systematic
examination of third generation eects, even in light of the absence of evidence-based
second generation eects. A psychological model for transmission to the third generation
might be the idea of more intensive learning and communication of the third generation
about the Holocaust compared to the second generation. The era of the conspiracy of
silence approached an end in the 1980s, and survivors gradually began to be considered
heroes more than lambs led to the slaughter (Bar-On, 1999). Grandchildren were urged
to learn about their grandparents past through interviews for school roots projects. For
the aging grandparents, time was ripe to return to their past and to create a sense of
coherence of their Holocaust experiences (Van der Hal-van Raalte, Van IJzendoorn, &
Bakermans-Kranenburg, in press). Since the last decade of the past century, thousands of
students have participated in Holocaust-centered excursions to Poland, and it is still
unclear what meaning this confrontation with the Holocaust might have (Chaitin, 2000),
but traumatizing eects on at least some of the participants cannot be excluded.
Lastly, as will be discussed, one may also raise the possibility of domain-specic eects,
namely that third generation eects may emerge in one area of functioning (e.g.,
internalizing or externalizing symptoms) but not in other areas such as attachment.
Therefore, in this review and meta-analysis we wish to shed further light on the
transmission of the Holocaust traumatic experiences, this time to the third generation
ospring.
108
A. Sagi-Schwartz et al.
109
2003; Van IJzendoorn, Grossmann, Grossmann, Joels, Sagi, & Scharf, 1999) infant
attachment security/insecurity (third generation) with the mother (second generation) was
assessed using the Strange Situation procedure (Ainsworth, Blehar, Waters, & Wall, 1978);
no signicant dierences were found between the infantmother attachment status of
children with and without Holocaust background of their grandmothers. In the other two
studies (Goldberg & Wiseman, 2006; Wiseman, 2005), two non-clinical high school-based
samples consisted of late adolescents of Israeli-born parents and from intact families, half
having grandparents who suered the Holocaust and half with no Holocaust background.
In both studies, the Experiences in Close Relationships scale (Brennan, Clark, & Shaver,
1998) was used to assess the adolescents appraisal of their attachment status, with scales
for attachment anxiety and avoidance. In one study, no signicant dierences were
reported on the attachment anxiety scale (Wiseman, 2005), whereas in another study
(Goldberg & Wiseman, 2006) grandchildren of Holocaust survivors reported a higher level
of attachment anxiety than did their non-Holocaust counterparts, but similarly to the rstmentioned study (Wiseman, 2005) no dierences were found on the attachment avoidance
scale. In these two studies also no signicant dierences were found between grandchildren
of Holocaust survivors and comparisons on measures of communication and self-esteem.
Given that the study was carried out during the period of terror bombing attacks in cities
in Israel, the higher attachment anxiety may be interpreted as indicating greater
vulnerability to life-threatening and extremely stressful situations (comparable to Baider
et al., 2000; Solomon et al., 1988). However, this explanation needs to be further explored
in future studies on the third generation of the Holocaust (H. Wiseman, personal
communication, February 2008).
In one more study in Israel with third generation of adolescent grandchildren of
Holocaust survivors and their comparisons, Scharf (2007) dierentiated between families
in which both parents, one parent, or no parent were ospring of Holocaust survivors.
They found that families with both parents being ospring of Holocaust survivors showed
most distress symptoms. The grandchildren in these families reported lower self-esteem,
and their peers rated them as functioning poorer in Israeli military service (Scharf, 2007).
Adolescents with one parent being Holocaust survivor ospring functioned similarly to
the comparisons without any Holocaust background.
Lastly, a large community study with a non-select sample of grandchildren was carried
out in Canada (Sigal & Weinfeld, 1989). In this study, parents rated their children (who
were the third generation children of Holocaust survivors or of a comparison group) using
a 50-item behavior problem checklist. The combined scores did not show a signicant
dierence between the two groups, but on three of the six factors that were extracted from
the questionnaire the grandchildren of Holocaust survivors were functioning better than
the comparisons. In sum, in the six studies with non-select samples the dierences between
ospring of survivors and comparisons were found to be for the most part non-signicant.
Select samples
In the seven studies with select samples also a wide range of measures were used, albeit
more of a clinical nature. The measures were: therapy use, SCL-90, fear, psychopathology,
anxiety, depression, eating problems, partner and parent relationship, and adjustment. All
seven studies were conducted in the USA. Two studies found signicant eects, i.e., the
studies by Huttman (2003) and Wetter (1998) reporting respectively more diculties in
relations with parents and higher rates of anxiety and despression among third generation
ospring with Holocaust background. In another study (Gopen, 2001) assessing
110
A. Sagi-Schwartz et al.
normative functioning, participants who were recruited during a memorial day for the
Holocaust were asked to describe their intimate relationships and family climate; no
signicant dierences emerged between those participants whose grandparents were
survivors of the Holocaust and those who were not.
In a study by Liebenau (1992), focusing on various adjustment measures, the
participants were referred from an organization known as Second Generation of Los
Angeles, and from friends and family members of the investigator. Additionally, the
investigator recruited subjects during the American Gathering of Holocaust Survivors in
Los Angeles. Outcomes in this study were not compared to a control group but were
contrasted to existing norms on self-esteem, locus of control, behavior problems, and
social problems to determine whether they were manifesting signicantly higher or lower
levels of adjustment, testing the hypothesis that individuals of third generation
descendants of Holocaust survivors would demonstrate signicantly less strength and
more problems than the norm. All the hypotheses remained without support.
The remaining three select samples focused on psychological symptoms, psychopathology, and clinical outcomes. In a study carried out by Ganz (2002), young adults
who were grandchildren of Holocaust survivors and comparisons with no Holocaust
background were recruited on the basis of the authors personal network and by
circulating yers in Jewish synagogues in New York City asking to volunteer to the study.
No dierences between the groups were found on psychological symptoms checklist and
on the use of therapy. In another study testing for multigenerational occurrence of
psychopathological symptoms in families of Holocaust survivors, third generation adult
participants with Holocaust background were recruited through Jewish organizations and
the comparisons were friends and acquaintances referred by the participants who belonged
to the Holocaust group (Rubenstein et al., 19891990). The authors inferred a
transmission eect on the basis of various reported indicators of psychopathology. Our
own re-computations showed a signicant third generation eect for aggression, but a
non-signicant overall eect based on the combined psychopathology measures used in
the study.
The last select study (Zelman, 1997) examined eating problems among third generation
female participants ranging from age 14 to age 25. These grandchildren of Holocaust
survivors were recruited from the records of the United States Holocaust Memorial
Museum in Washington, DC, and their comparisons without Holocaust background were
recruited from an undergraduate pool of students at the University of Hartford. The two
groups were not found to dier on eating problems.
In all, the set of select samples suggests a mixed picture of ndings with both signicant
and non-signicant dierences between grandchildren of Holocaust survivors and
comparisons without Holocaust background. Moreover, a rather wide range of measures
and assessments were used in these studies, ranging from aggression to eating problems,
anxiety, and attachment. Such a variety of measures may represent three global domains
of internalizing symptoms, externalizing symptoms, and attachment patterns. In the metaanalysis, we will dierentiate between these dimensions.
A meta-analytic approach
In a research domain with conicting results, the method of narrative review may be
insucient to create a coherent picture (Cooper & Hedges, 1994). Counting studies with
supporting versus falsifying evidence might be an inadequate approach to reconciling
diverging study outcomes, as it does not take into account the highly heterogeneous
111
quality and size of the studies (Kellermann, 2001). A quantitative analysis and synthesis of
the tertiary traumatization literature is needed to come to a more denite conclusion.
Through meta-analysis, it is possible to compute the average eect size across studies, and
to explain dierences in outcome between studies on the basis of study characteristics
(Mullen, 1989; Rosenthal, 1991, 1995). Thus, in order to overcome the clinical-science
gap, which leads to inconsistent conclusions, we present meta-analytic ndings that are
based on all existing studies with third generation which meet the basic criteria for
inclusion in a meta-analysis. Given the lack of a clear and adequate conceptual model that
may endorse the existence of tertiary traumatization and given the mixture of ndings in
prior research, we left the question of transmission to the third generation open without
assuming a specic hypothesis in comparing third generation ospring of Holocaust
survivors with those of non-Holocaust survivors. We also explored whether any
dierences emerged in the areas of externalizing, internalizing, and attachment issues.
Methods
Data collection
Pertinent studies were collected systematically, using three dierent search strategies
(Mullen, 1989; Rosenthal, 1991). First, PsychInfo, Medline, and PILOTS (a comprehensive database on posttraumatic stress) were searched with keywords Holocaust, third
generation, grandchildren, and survivors. Second, the references of the collected
papers, books, and book chapters were searched for relevant Holocaust studies. Third,
some recent narrative reviews were used as a source for relevant papers, in particular the
exhaustive review of Kellermann (2001).
Holocaust survivors spent the war in Nazi-occupied Europe, either in concentration/
labor camps, or in various hideaway shelters, being adopted by gentile families, or using
a combination of escape and survival strategies. More often than not, they lost parents and
other family members. Second and third generation Holocaust survivors have become
accepted terms to refer to ospring of Holocaust survivors (Solomon, 1998). For obvious
reasons, the number of studies with third generation ospring is much smaller than that on
rst and second generation. Thus, our selection criteria were rather broad, in order to
include as many Holocaust studies as possible, regardless of research design and platform
of publication. We included any formal platform such as journals, books, PhD
dissertations, conference presentations, and nal reports to grant foundations.
The most important criterion was that the study should contain at least one
comparison group, and that it presented data to derive the pertinent meta-analytic
statistics from. As with our meta-analytic study on secondary traumatization (Van
IJzendoorn et al., 2003), the idea was to test empirically also with third generation the
inuence of design features, and therefore not to exclude any quantitative study on a priori
grounds (Rosenthal, 1995). Case studies and qualitative publications were excluded
because they do not t into a meta-analysis paradigm. In Tables 1 and 2 we listed all
existing publications, including case studies and qualitative studies, so as to inform the
reader about all available work in the eld. As a result of our search, 23 papers were
identied, of which 13 met the set of criteria for inclusion in the meta-analysis. For an
overview, see Table 1 for studies not included in the meta-analysis and Tables 2 and 3 for
studies included in the meta-analysis. All but one of the studies suitable for a meta-analytic
analysis were non-clinical, so that it was impossible to test the moderating role of clinical
status in our meta-analyses. We therefore decided that ndings would be more robust if we
kept the meta-analysis more homogenous by including only non-clinical samples.
112
Table 1.
A. Sagi-Schwartz et al.
Studies on third generation Holocaust survivors not included in the meta-analysis.
Author
Non-clinical samples
Bar-On and Gilad
(1994)
Chaitin (2002)
Chaitin (2000)
Hogman (1998)
Jurkowitz (1996)1
Clinical samples
Sigal et al. (1988)2
Bienstock (1989)3
Berger-Reiss (1997)
Rosenthal and
Rosenthal (1980)
Winship and
Knowles (1996)
Gender
Country
of residence
Outcome
120
1
Female
Israel
Mixed
Mixed
Mixed
Mixed
Israel
Israel
USA
USA
Mixed
Female
Female
Male
Canada
USA
USA
USA
Clinical referrals/Quantitative
Parents behavior (PBI)/Quantitative
Clinical impressions
Clinical impressions
Mixed
UK
Clinical impressions
12
9
7
91
162
127
30
1
1
3
Coding system
A coding system was used to rate every Holocaust study on design, sample, and
measurement characteristics (see Table 2). We coded sample size (grandchildren of
Holocaust survivors and comparison group in the study) and recruitment as design
characteristics. Studies were coded as non-select when participants were randomly
sampled, e.g., from several neighborhoods or a population registry, or when the entire
Jewish population of a certain country was involved. They were coded as select when
samples were recruited through, e.g., Holocaust survivor meetings, personal contacts, or
advertisements. We also coded gender of the samples (male, female, or mixed). Eleven
studies were represented by both males and females, whereas one sample consisted of allmales only and one sample of all-female only. Moreover, we coded whether the sample
was clinical or non-clinical, and we registered the current country of residence of the
sample (Israel, Canada, and USA).
We coded an overall type of outcome, namely adjustment, which was dened broadly,
and involved positive indicators such as general adjustment, attachment security, self
esteem, partner relationship and family climate, close relationship, coping, and negative
indicators like anxiety/depression, aggression, eating problems, clinical referrals,
psychopathology. All these constructs were measured with standard and valid
instruments, commonly used for the assessment of mental health, posttraumatic stress,
coping, adaptation, and maladaptation. Furthermore, we distinguished three categories of
functioning, relating to externalizing, internalizing, and attachment issues. When
outcomes in more than one domain of functioning were reported within a study, we
selected the outcome that most unambiguously reected one of the three domains of
functioning, and at the same time led to the largest numbers of eect sizes in the various
domains (resulting in k 3 studies for externalizing, k 5 studies for internalizing, and
k 5 studies for attachment, see Table 3). As an example, Rubenstein et al. s (19891990)
study provided a combined outcome for psychopathology that was used in the
Mixed
Mixed
60
72
74
36
51
74
29
396
Huttman (2003)
Liebenau (1992)1
Rubinstein et al. (19891990)
Wetter (1998)
Zelman (1997)
Total select samples
Mixed
67
616
Wiseman (2005)
Total non-select samples
Select samples
Ganz (2002)
Gopen (2001)
USA
USA
USA
USA
USA
USA
USA
Israel
Canada
Therapy use/SCL-90/Fear
Partner relationship and attachment
to parent
Close relationship
Adjustment
Psychopathology
Anxiety/depression
Eating problems
Outcomes are not compared to a control group but contrasted to existing norms.
Homogeneous set of outcomes.
Mixed
Mixed
Mixed
Mixed
Female
Mixed
118
Israel
70.62
70.71
0.27
0.73
0.24
0.07
0.11
70.16
70.02
0.032
70.36
0.03
Male
79
(70.99
(72.14
(70.30
(0.27
(70.54
(70.22
*
*
*
*
*
*
7.25)
0.72)
0.84)
1.19)
1.01)
0.35)
(70.19 * 0.41)
(70.63 * 0.31)
(70.51 * 0.47)
(70.11 * 0.17)
(70.73 * 0.02)
(70.23 * 0.29)
(70.15 * 0.65)
(70.15 * 0.47)
(70.33 * 0.47)
0.25
0.16
0.07
Aggression
Attachment/communication/self-esteem
Attachment security
Mixed
Mixed
Mixed
97
160
95
Israel
Israel
Israel
(70.14 * 0.23)
0.04
Outcome
1012
Country
of residence
Total samples
Non-select samples
Bachar et al. (1994)
Goldberg and Wiseman (2006)
Sagi-Schwartz et al. (2003)
Van IJzendoorn et al. (1999)
Scharf (2007)
Gender
95% CI
General adaptation of third generation Holocaust ospring in non-clinical studies suitable for meta-analysis.
d
Author
Table 2.
.001
.33
.35
.002
.55
.65
.83
.51
.24
.69
.06
.82
.22
.32
.73
.66
114
A. Sagi-Schwartz et al.
Country
Gender of residence Outcome
97 Mixed
118 Mixed
51 Mixed
0.25
70.15 * 0.65
.22
Canada
Conduct
problems
70.30
70.67 * 0.07
.11
USA
Aggression
0.68
0.02 * 1.34
.04
0.15
7 0.32 * 0.61
.54
Peer-reported 70.06
distress and
functioning
Fear
70.06
Self esteem
71.20
Anxiety/
0.73
depression
Eating
0.24
problems
0.12
70.51 * 0.39
.81
70.58 * 0.46
72.63 * 0.23
0.27 * 1.19
.82
.10
.002
70.54 * 1.01
.55
70.29 * 0.53
.56
Ganz (2002)
Liebenau (1992)1
Wetter (1998)
60 Mixed
36 Mixed
74 Mixed
USA
USA
USA
Zelman (1997)
29 Female
USA
Total
Aggression
Israel
Huttman (2003)
95% CI
Israel
79 Male
Total
Attachment
Goldberg and
Wiseman
(2006)
Sagi-Schwartz
et al. (2003)
Van IJzendoorn
et al. (1999)
Wiseman (2005)
Gopen (2001)
160 Mixed
Israel
Attachment
0.25
70.06 * 0.57
.12
95 Mixed
Israel
Attachment
security
0.07
70.33 * 0.47
.73
67 Mixed
72 Mixed
Israel
USA
70.65 * 0.33
70.70 * 0.25
.52
.36
74 Mixed
USA
Attachment
70.16
Attachment
70.22
to parent
70.63
Aective
relationship
to parent
70.032
.73
Contrast for eect sizes of the various domains: Qbetween 1.03, p .60.
1
Outcomes are not compared to a control group but contrasted to existing norms.
2
Homogeneous set of outcomes.
meta-analysis on general adjustment, and an outcome for aggression (as part of the
psychopathology outcome) that was included in the externalizing category of the metaanalysis exploring the dierences in functioning in the three domains.
Data analysis
Because the studies included in this series of meta-analyses reported various statistics, the
outcomes of all studies were re-computed with Mullens (1989) Advanced BASIC Metaanalysis program, and transformed into Cohens d (the standardized dierence in means
115
between the third generation of Holocaust survivors and comparisons). In several cases,
we had to compute the eect sizes on the basis of means and standard deviations provided
in the study report. When more than one outcome was reported (e.g., Wiseman, 2005),
they were meta-analytically combined into one eect size for adjustment, that is Cohens d.
Moreover, an overall eect size for general adjustment based on available positive and
negative indicators for adjustment was computed for each study in order to avoid counting
a study or participant more than once.
The resulting set of eect sizes were inserted into Borenstein, Rothstein, and Cohens
(2000) Comprehensive Meta-Analysis (CMA) program that computed xed as well as
random eect model parameters. CMA also computed condence intervals around the
point estimate of an eect size. Because the leading hypothesis in this area of research is
that the grandchildren of Holocaust survivors would be less well-adapted, we used this
directed hypothesis and present here the 95% condence boundaries of the point estimates
for the eect sizes (with alpha set at .05; see Tables 2 and 3). Thus, for meta-analytic
purposes, we assigned positive signs to those eects that indicated the presence of a third
generation eect (i.e., negative outcomes for the third generation compared to
comparisons without Holocaust background), and negative signs to eects that indicated
better outcomes for the Holocaust third generation ospring.
Signicance tests and moderator analyses in xed eects models are based on the
assumption that dierences between studies leading to dierences in eects are not
random, and that in principle the set of study eect sizes is homogeneous at the population
level. Signicance testing is based on the total number of subjects, but generalization is
restricted to other participants that might have been included in the same studies of the
meta-analysis (Rosenthal, 1995). Statistical inferences may be regarded as applying only to
the specic set of studies at hand (Hedges, 1994). In random eects models this
assumption is not made (Hedges & Olkin, 1985), and they allow for the possibility that
each separate study has its own population parameter. In random eects models
signicance testing is based only on the total number of studies and generalization is to the
population of studies from which the current set of studies was drawn (Rosenthal, 1995). It
has been argued that random eects models more adequately mirror the heterogeneity in
behavioral studies, and use non-inated alpha levels when the requirement of homogeneity
has not been met (Hunter & Schmidt, 2000). In our meta-analyses, some data sets were
heterogeneous. In those cases, the random eects model parameters (signicance,
condence intervals) were presented (see Tables 2 and 3).
The current analyses included 13 studies, including 1012 participants (grandchildren of
Holocaust survivors and comparisons). The sample sizes ranged from 29 (Zelman, 1997)
to 160 (Goldberg & Wiseman, 2006). For each of the 13 samples, the standardized
dierence between the Holocaust and comparison group was computed (Cohens d;
Mullen, 1989). For each sample, Fisher Z was computed as an equivalent to the
correlation coecient r (see Mullen, 1989). No outlying eect sizes were identied in the
set of Holocaust studies on the basis of standardized z-values larger than 3.29 or smaller
than 73.29 (p 5 .001; Tabachnick & Fidell, 2001).
Results
Does tertiary traumatization exist?
In the total set of 13 samples on 1012 families, we did not nd a signicant dierence in
psychological well-being and adaptation between the third generation Holocaust survivors
and their comparisons. The size of the combined eect was a Cohens d of 0.04 (p .66),
116
A. Sagi-Schwartz et al.
condence interval 70.14 * 0.23 (see Table 2). Based on the overall eect size for general
adjustment in each sample, the third-generation Holocaust survivors did not dier
signicantly from the comparisons.
Are study results associated with the type of recruitment of participants?
When the set of studies was divided in sub-sets with select and non-select samples, we
found basically similar outcomes. The seven select samples, including 616 subjects, showed
a small but non-signicant combined eect size of d 0.07 (p .65) for general
adjustment. The six non-select studies with a more adequate recruitment of participants
also showed a non-signicant eect size of d 0.03 (p .69) in a homogeneous set of
outcomes. Thus, both studies with non-select and with select samples failed to show a
tertiary traumatization eect, that is, third generation Holocaust survivors did not show
less well-being or adaptation than did the comparisons.
Does tertiary traumatization exist in specic domains of functioning?
Because convenience and non-convenience samples did not dier in combined eect sizes,
we examined dierences between studies addressing externalizing, internalizing, and
attachment problems across the total set of studies. Table 3 presents the outcomes and the
combined eect sizes for the three categories of functioning (k 3 studies for
externalizing, k 5 studies for internalizing, and k 5 studies for attachment). Although
the number of study outcomes in the externalizing category was small, we decided to
explore the dierences in combined eect sizes between these three domains. For
externalizing, internalizing, and attachment the combined eect sizes were d 0.15,
d 0.12, and d 70.03, respectively (see Table 3). These combined eect sizes for the
three domains were non-signicant, and the contrast between the three sets was not
signicant either (Q(2) 1.03, p .60), see Table 3. Tertiary traumatization could not be
detected in the three domains of functioning separately or combined into the overarching
category of adaptation. It should be noted that our test of dierences between domains
was only exploratory because of the small number of study outcomes in the externalizing
category.
Discussion and conclusions
Our narrative review of studies on the transmission of trauma to the third generation
ospring of Holocaust survivors showed a mixed picture of signicant and non-signicant
dierences between grandchildren of Holocaust survivors and comparisons without
Holocaust background. In a meta-analysis of 13 studies, involving over 1000 participants,
no evidence for tertiary traumatization was found, whether or not participants were
recruited in a select or non-select manner, and whether or not we divided the study
outcomes into three possible domains of traumatization, that is externalizing, internalizing, or attachment issues.
We should not overlook the limited evidence that is presented for specic
traumatization eects in clinical (Sigal et al., 1988) or particularly vulnerable populations,
evidence that might be lost or discounted in the meta-analysis. For example, in order to
increase statistical power, we had to merge the subgroups as reported by Scharf (2007).
Although in that study grandchildren with both parents raised by Holocaust survivors
showed more adaptation problems, the combined groups (with one and two parents raised
117
118
A. Sagi-Schwartz et al.
PTSD through their genes (Goldberg, True, Eisen, & Henderson, 1990) as well as through
myriads of other personality, social, and chance factors. This genetic protection against
PTSD might have been transmitted to the next generations. Children of Holocaust
survivors, therefore, may not have been especially sensitive to potentially traumatic events
or reports arising from their parents eorts to deal with the Holocaust atrocities. We are
currently exploring the genetic component more directly by collecting DNA in our Israeli
study, in a search for vulnerability genes (Sagi-Schwartz et al., 2003).
The last factor in a stress-diathesis model concerns the presence or absence of social
support to cope with trauma afterwards. We proposed that after World War II the
survivors were forced to nd meaning in helping to build up a new society, everywhere
they could, in Europe and Northern America, but also, and maybe especially, in Israel
(Frankl, 1984). The newly founded State of Israel, to which many survivors emigrated
after the war, and also the success of many Jews throughout the Western world served as
symbols of the ultimate failure of the Final Solution (Solomon, 1998). Ever since the
establishment of Israel, various memorials in Israel and subsequently also outside Israel
have been erected to commemorate the victims of the Holocaust and to support survivors
and their families in working through the traumatic memories of the past, all of which
might have served as important social support for Holocaust survivors.
Unfortunately, due to the relatively small number of pertinent studies it was not
possible to test country dierences meta-analytically (but see Van IJzendoorn et al., 2003).
Similarly, no separate results were available in the various studies for males versus females
and the variability in this set of studies was insucient to test whether gender was a
moderator. The search for moderators and mediators would have been very important,
though, if the database in the dierent studies included in the meta-analyses would have
allowed for such analyses.
In sum, our series of meta-analyses show that second generation as well as third
generation ospring of Holocaust survivors are, in general, well adapted. The provocative
idea that the intergenerational transmission of the trauma might skip a generation and
thus may emerge in the third generation was not substantiated in our meta-analyses. Such
a potential sleeper eect does not appear to be the case, the youngest generation seems
also to develop in a normal way. Protective factors in the ospring or in their environment
may have lessened the impact of the rst generations trauma. More plausibly, studies on
the second-generation and third-generations psychological consequences of the Holocaust
indicate a remarkable resilience of traumatized survivors in their parental roles, even when
they personally may be traumatized profoundly.
Clinically, every case is a unique constellation of etiological factors responsible for
current suering, and the Holocaust experiences of the earlier generation might be one of
those factors even though on the level of the population secondary and tertiary
traumatization are not the rule. The absence of tertiary traumatization, however, should
sensitize clinicians working with (second or) third generation ospring of Holocaust
survivors to the possibility that their clients may be stimulated to search for the roots of
their problems in other directions besides the Holocaust experience of their grandparents.
Acknowledgement
Abraham Sagi-Schwartz was supported by the Mary Main Visiting Professional Chair at the Centre
for Child and family studies, Leiden University, the Netherlands. Marinus van IJzendoorn and
Marian Bakermans-Kranenburg were supported by research awards from the Netherlands
Organization for Scientic Research (NWO SPINOZA prize and VIDI grant no. 452-04-306,
respectively).
119
References*
References marked with an asterisk indicate studies included in the meta-analysis.
Ainsworth, M.D.S., Blehar, M.C., Waters, E., & Wall, S. (1978). Patterns of attachment: A
psychological study of the Strange Situation. Hillsdale, NJ: Erlbaum.
*Bachar, E., Cale, M., Eisenberg, J., & Dasberg, H. (1994). Aggression expression in grandchildren
of Holocaust survivors: A comparative study. Israeli Journal of Psychiatry and Related
Disciplines, 31, 4147.
Baider, L., Peretz, T., Hadani, P.E., Perry, S., Avramov, R., & De Nour, A.K. (2000). Transmission
of response to trauma? Second-generation Holocaust survivors reaction to cancer. American
Journal of Psychiatry, 157, 904910.
Bar-On, D. (1999). Israeli society between the culture of death and the culture of life. In K. Nader,
N. Dbrow, & B.H. Stamm (Eds.), Honoring dierences: Cultural issues in the treatment of trauma
and loss (pp. 211233). Philadelphia: Brunner/Mazel.
Bar-On, D., & Gilad, N. (1994). To rebuild life: A narrative analysis of three generations of an Israeli
Holocaust survivors family. In A. Lieblich & R. Josselson (Eds.), Exploring identity and gender:
The narrative study of lives: Vol. 2 (pp. 83112). Thousand Oaks, CA: Sage.
Bar-On, D., Eland, J., Kleber, R.J., Krell, R., Moore, Y., Sagi, A., et al. (1998). Multigenerational
perspectives an coping with the Holocaust experience: An attachment perspective for
understanding the developmental sequelae of trauma across generations. International Journal
of Behavioral Development, 22, 315338.
Bauman, Z. (1989). Modernity and the Holocaust. Cambridge: Polity Press.
Berger-Reiss, D. (1997). Generations after the Holocaust: Multigenerational transmission of trauma.
In B.S. Mark & J.A. Incorvaia (Eds.), The handbook of infant, child, and adolescent psychotherapy.
Vol. 2: New directions in integrative treatment (pp. 209219). New York: Aronson.
Bienstock, B.E. (1989). Daughters and granddaughters of female concentration camp survivors:
Mother-daughter relationships. Unpublished doctoral dissertation, Florida Institute of Technology, Florida.
Borenstein, M., Rothstein, D., & Cohen, J. (2000). Comprehensive meta-analysis. A computer
program for research synthesis. Englewood, NJ: Biostat.
Brennan, K.A., Clark, C.L., & Shaver, P.R. (1998). Self-report measurement of adult romantic
attachment: An integrative overview. In J.A. Simpson & W.S. Rholes (Eds.), Attachment theory
and close relationships (pp. 4676). New York: Guilford.
Bulmer, M. (1998). Galtons law of ancestral heredity. Heredity, 81, 579585.
Chaitin, J. (2000). Facing the Holocaust in generations of families of survivors. The case of partial
relevance and interpersonal values. Contemporary Family Therapy, 22, 289313.
Chaitin, J. (2002). Issues and interpersonal values among three generations in families of Holocaust
survivors. Journal of Social and Personal Relationships, 19, 385408.
Cooper, H., & Hedges, L.V. (Eds.). (1994). The handbook of research synthesis. New York: Russell
Sage Foundation.
Danieli, Y.E. (1998). International handbook of multigenerational legacies of trauma. New York:
Plenum.
Frankl, V.E. (1984). Mans search for meaning. An introduction to logotherapy (3rd revised and
enlarged edition). New York: Simon & Schuster.
*Ganz, E.T. (2002). Intergenerational transmission of trauma: Grandchildren of Holocaust survivors.
Unpublished doctoral dissertation, Adelphi University, New York.
*Goldberg, A., & Wiseman, H. (2006, July). Attachment styles, parental bonding, parental monitoring
and risk taking behavior in adolescents: The case of the third generation of Holocaust survivors.
Poster session presented at the International Association for Relationship Research Conference,
Rethymno, Crete, Greece.
Goldberg, J., True, W.R., Eisen, S.A., & Henderson, W.G. (1990). A twin study of the eects of the
Vietnam War on posttraumatic stress disorder. JAMA, 263, 12271232.
*Gopen, A. (2001). The aftermath of the Holocaust trauma across family generations: Family
environment, relationship environment and empathy of the third generation. Unpublished doctoral
dissertation, New School University, New York.
Hedges, L.V. (1994). Fixed eects models. In H. Cooper & L.V. Hedges (Eds.), The handbook of
research synthesis (pp. 285299). New York: Russell Sage Foundation.
Hedges, L.V., & Olkin, I. (1985). Statistical methods for meta-analysis. San Diego, CA: Academic
Press.
120
A. Sagi-Schwartz et al.
Hogman, F. (1998). Trauma and identity through two generations of the Holocaust. Psychoanalytic
Review, 85, 573578.
Hunter, J.E., & Schmidt, F.L. (2000). Fixed eects versus random eects meta-analysis models:
Implications for cumulative research knowledge. International Journal of Selection and
Assessment, 8, 275292.
*Huttman, J.P. (2003). The intergenerational eects of the Holocaust on patterns of attachment in the
grandchildren of survivors. Unpublished doctoral dissertation, Alliant International University,
Fresno, California.
Jurkowitz, S.W. (1996). Transgenerational transmission of depression, shame and guilt in Holocaust
families: An examination of three generations. Unpublished doctoral dissertation, California
School of Professional Psychology, Los Angeles, CA.
Kellermann, N.P.F. (2001). Transmission of Holocaust trauma. An integrative view. Psychiatry:
Interpersonal and Biological Processes, 64, 256267.
Lacqueur, W.E. (2001). The Holocaust encyclopedia. New Haven and London: Yale University
Press.
*Liebenau, K.P. (1992). A comparison of third generation descendants of Holocaust survivors scores
with the norms on self-esteem, locus-of-control, behavioral and social problems. Unpublished
doctoral dissertation, California School of Professional Psychology, Fresno, CA.
Lipsey, M.W., & Wilson, D.B. (2001). Practical meta-analysis. Applied social research methods series:
Vol. 49. Thousand Oaks, CA: Sage.
Mullen, B. (1989). Advanced basic meta-analysis. Hillsdale, NJ: Erlbaum.
Paris, J. (2000). Predispositions, personality traits, and posttraumatic stress disorder. Harvard
Review of Psychiatry, 8, 175183.
Rosenthal, P.A., & Rosenthal, S. (1980). Holocaust eect in the third generation: Child of another
time. American Journal Psychotherapy, 34, 572580.
Rosenthal, R. (1991). Meta-analytic procedures for social research. Beverly Hills, CA: Sage.
Rosenthal, R. (1995). Writing meta-analytic reviews. Psychological Bulletin, 118, 183192.
*Rubenstein, I., Cutter, F., & Templer, D.I. (19891990). Multigenerational occurrence of survivor
syndrome symptoms in families of Holocaust survivors. Omega: Journal of Death and Dying, 20,
239244.
*Sagi, A., Grossmann, K.E., Joels, T., Grossmann, K., Scharf, M., & Van IJzendoorn, M.H. (1999,
April). The Holocaust Child Survivors Study: Attachment across Generations, Part I: Some
Theoretical and Methodological Considerations. Paper presented in a symposium On the
Developmental Sequelae of Catastrophic Holocaust Childhood Experiences across Generations at the Biennial Meeting of the Society for Research in Child Development, Albuquerque,
NM.
Sagi-Schwartz, A., van IJzendoorn, M.H., Grossmann, K.E., Joels, T., Grossmann, K., Scharf, M.,
et al. (2003). Attachment and traumatic stress in female Holocaust child survivors and their
daughters. American Journal of Psychiatry, 160, 10861092.
*Scharf, M. (2007). Long-term eects of trauma: Psychosocial functioning of the second and third
generation of Holocaust survivors. Development and Psychopathology, 19, 603622.
Schwartz, S., Dohrenwend, B.P., & Levav, I. (1994). Non genetic familial transmission of psychiatric
disorders? Evidence from children of Holocaust survivors. Journal of Health and Social Behavior,
35, 385402.
*Sigal, J.J., & Weinfeld, M. (1989). Trauma and rebirth: Intergenerational eects of the Holocaust.
New York: Praeger.
Sigal, J.J., DiNicola, V.F., & Buonvino, M. (1988). Grandchildren of survivors: Can negative eects
of prolonged exposure to excessive stress be observed two generations later? Canadian Journal
Psychiatry, 33, 207212.
Solomon, Z. (1998). Transgenerational eects of the Holocaust. In Y.E. Danieli (Ed.), International
handbook of multigenerational legacies of trauma (pp. 6983). New York: Plenum.
Solomon, Z., Kotler, M., & Mikulincer, M. (1988). Combat-related posttraumatic stress disorder
among 2nd-generation Holocaust survivors. Preliminary ndings. American Journal of
Psychiatry, 145, 865868.
Tabachnick, B.G., & Fidell, L.S. (2001). Using multivariate statistics. New York: Harper & Row.
Van der Hal-Van Raalte, E.A.M., Van IJzendoorn, M.H., & BakermansKranenburg, M.J.
(in press). Sense of coherence moderates late eects of early childhood Holocaust exposure.
Journal of Clinical Psychology.
121
Van IJzendoorn, M.H., Bakermans-Kranenburg, M.J., & Sagi-Schwartz, A. (2003). Are children of
Holocaust survivors less well-adapted? No meta-analytic evidence for secondary traumatization.
Journal of Traumatic Stress, 16, 459469.
*Van IJzendoorn, M.H., Grossmann, K.E., Grossmann, K., Joels, T., Sagi, A., & Scharf, M. (1999,
April). The Holocaust child survivors study: attachment across generations, part II: some empirical
data. Paper presented in a symposium On the Developmental Sequelae of Catastrophic
Holocaust Childhood Experiences across Generations at the Biennial Meeting of the Society for
Research in Child Development, Albuquerque, NM.
*Wetter, M.G. (1998). The intergenerational transmission of increased anxiety traits in thirdgeneration Holocaust survivors. Unpublished doctoral dissertation, Pepperdine University,
Malibu, California.
Winship, G., & Knowles, J. (1996). The transgenerational impact of cultural trauma: Linking
phenomena in treatment of third generation survivors of the Holocaust. British Journal
Psychotherapy, 13, 259266.
*Wiseman, H. (2005). The experience of parenting adolescents among second generation Holocaust
survivors: recollections of experiences with their parents and current parentadolescent relationship.
Final Scientic Research Report: Israel Foundations Trustees Grant (20022004).
*Zelman, E.K. (1997). The transgenerational eects of the Nazi Holocaust experience on eating
attitudes and behaviors. Unpublished doctoral dissertation, University of Hartford, West
Hartford, CT.