Alchemies of Altruism:
Motivations and Opportunities in the Rescue of Jews in
Nazi Europe.
Meir Yaish
Nuffield College, Oxford
And
Federico Varese
Yale University, USA
June 2001
Word count: 9,300
Key Words: Altruism, Holocaust, Situational factors, Motivational
factors, Direct appeal.
Samuel P. Oliner, the director of the ‘Altruistic Personality Project’, has shared the data
he collected on rescuers of Jews during the Nazi occupation of Europe with me. Without
his generous act, this paper would have never been written. We are indebted to Richard
Breen, Hartmut Esser, Geoff Evans, David Firth, Avner Offer, Rachel Reeves, and Luca
Ricolfi, for their useful comments and suggestions. An earlier version of this paper was
presented in the Department of Sociology, Oxford, March, 2000, and the department of
Psychology, Oxford, May 2001. We wish to thank the participants of these seminars for
their valuable comments. The usual disclaimers apply.
1
Alchemies of Altruism:
Motivations and Opportunities in the Rescue of Jews in
Nazi Europe.
Abstract
This paper seeks to explore the underlying mechanisms of altruistic
behaviour – a topic dominated by experimental studies, mainly in the field
of psychology. This paper expands our understandings of the processes
that impinge on such behaviour by focusing on a ‘naturalistic’ study of
altruism in a high-risk situation. It thus concentrates on the rescue of
persecuted Jews in the Holocaust. In explaining altruistic behaviour it is
common to make a distinction between the effect of situational factors and
the effect of motivational factors. Applying a case control sample technique
to the analysis of the APPBI data on those who rescued Jews and those
who did not rescue Jews during the Nazi occupation of Europe (N=510),
we show that situational factors (a direct request for help) substantially
increase the likelihood of a Jew to be rescued. We further show that
motivational factors (defined in this study as a ‘prosocial action
orientation’) are important too in arriving at altruistic behaviour.
However, we also show that motivational and situational factors interact
in generating altruistic behaviour. Thus, we conclude, in arriving at
altruistic behaviour situational factors may be seen to activate
motivational factors. Finally, utilizing nonparametric techniques to the
analysis of the APPBI data, we were able to rule out the hypothesis that
this activation process (i.e., the interaction effect) is governed by a
threshold effect. We show, instead, that this process is linear. A discussion
of the implications of this study concludes the paper.
2
‘No doubt, men are capable of much more unselfish service than they
generally render; and the supreme aim of the economist is to discover how
this latent social asset can be developed more quickly and turned to account
more wisely’ (Marshall, 1890: 9).
1. Introduction
The understanding and explanation of human behaviour and action is
perhaps the ultimate task, and – no doubt – the most ambitious challenge,
for social scientists. Although this challenging task unites the social
sciences, members of the various fields within the discipline approach it –
theoretically and empirically – differently. Thus, for example, the rational
choice paradigm within economics1 would argue that the individual’s
‘desire’ to maximize a utility function provides a plausible and satisfactory
explanation of human behaviour and action. That is, one can argue,
motivations (i.e., a constant preference to maximize utility) can chiefly
account for an observed behaviour. Sociologists, on the other hand, would
often invoke a normative explanation in that context. That is, situational
factors – e.g., the prevailing norm – can chiefly account for the observed
behaviour. The division of the determinants of social behaviour into two
classes – the situation and the personality of the actor – also has a long
tradition in psychology (Staub, 1974: 321).
In this paper we explore the underlying mechanisms of altruistic
behaviour in a high-risk situation. We study rescuers of Jews from Nazi
persecution during the Second World War. Helping behaviour in a highrisk situation, such as the rescue of Jews from Nazi persecution, is a
challenging explanandum for advocates of the situational and the
motivational approaches. As the common theme in the rational choice
paradigm is that the greater the cost the lower the likelihood a person will
Sociologists would refer to this ‘economic model of man’ as being derived from a ‘thin’
(cf. Ferejohn, 1990), or ‘narrow’ (cf. Opp, 1999), definition of rationality. In this paper we
wish to shay away from debating on the definition of rationality – a debate that deserves
a separate paper – and we refer to rationality – and rational choice theory – in its rather
‘narrow’ definition.
1
3
help others (cf. Staub, 1974: 299), how can risking one’s life be
understood? Similarly, when the prevailing norms are survival (i.e., saving
oneself and the family) and conformity to authority (cf. Milgram, 1963),
how can acting contrary to these norms be understood?
The theoretical approach that we adopt in the investigation of the
underlying mechanisms of the rescue of Jews from Nazi persecution
integrates the motivational and the situational explanations. This
approach is influenced by the psychological literature that explores the
ways in which motivational and situational factors interact in generating
behaviour (cf. Fazio, 1986). We show that the altruistic behaviour in the
context of the Holocaust is dominated by a particular situational factor
[i.e., a request to help] (see also, Varese and Yaish, 2000). We argue that a
request to help activates the motivation that subsequently leads to the
observed altruistic behaviour. Our analyses lend support to this
hypothesis as it is shown that the statistical interaction between
motivations (a prosocial action orientation) and situational factors (a
request to help) is significant and positive.
The paper is organised as follows: the next section presents the
theoretical model that we subsequently test empirically. Section three
provides a short review of the literature on the rescue of Jews in Nazioccupied Europe. A discussion of analytical techniques and a presentation
of the data and the variables used in the study follow. Section five
presents the results of the theoretical model that we present in the first
section. Section six concludes the paper.
2. Theoretical Considerations
In a series of experiments, Berkowitz and his associates found that the
greater the dependence of one person on another, the more likely the
latter is to help. Berkowitz and Daniels (1963) suggested, then, that a
norm of social responsibility, which prescribes that people should help
others who are dependent on them, guided this behaviour. Other
4
experimental psychologists (see, Darley and Latané, 1968; Latané, Nida,
and Wilson, 1981; Sober and Wilson, 1998: 255), however, have
emphasised the effect of the situation on helping behaviour. The seminal
work of Darley and Latané (1968), for example, revealed that an
individual’s (bystanders) probability of helping a person in need decreases
as the number of bystanders increases. They explained this tendency in
terms of a ‘diffusion of perceived responsibility’.
Both views would appear to accept that a bystander is more likely
to help someone in need if she is made to believe that she is responsible for
this person.2 Schwartz’s (1977) theory of altruistic behaviour postulates
that a direct appeal for help induces responsibility, which leads
subsequently to altruistic behaviour. Ample evidence exists in support of
the view that a direct appeal induces helping behaviour. Simmons, Klein
and Simmons (1977, quoted in Piliavin and Charng, 1990: 35) find, for
example, that kidney donors were more likely to have been informed in
person of a need for a donor than were non-donors. Freeman (1993), finds
that individuals are more inclined to donate money to charity when asked.
Similarly, a direct request to vote significantly increases voter turnout, as
shown by a field experiment carried out by Gerber and Green (1999). And,
Varese and Yaish (2000), conclude in their study of the rescuers of
persecuted Jews during the Nazi occupation of Europe that being asked is
a significant predictor of helping behaviour. Finally, the most common
reason given for failure to donate to charities, give blood or volunteer time
to worthwhile causes is not having been asked (Piliavin and Charng, 1990:
35). Shaw, Batson and Todd (1994), then, utilised an experiment to show
that, under certain conditions, people avoid placing themselves in a
position where they might be asked for help. They observe ‘empathy
avoidance’ when, before exposure to a person in need, subjects are aware
that they will be asked to help and helping will be costly.
The importance of a direct appeal in generating altruistic behaviour
The difference between the two views can be summarised as follows: whereas according
to the first view, responsibility is a function of dependence, according to the second view,
responsibility is a function of the number of people in a particular situation.
2
5
should be seen in the context of the debate on the motivational vis-à-vis
the situational determinants of altruistic behaviour. To fully understand
(altruistic) behaviour, then, one has to seriously consider the possibility
that individual differences in values and norms (i.e., motivational factors)
interact with situational factors (Schwartz, 1977). That is, a direct appeal
for help is an important element that links the situational factors with the
motivational factors in generating altruistic behaviour.
A handful of scholars have developed theoretical models that take
both motivational and situational factors into account in the explanation
of altruistic – and ultimately all human – behaviour (Staub, 1974;
Schwartz, 1977; Stryker, 1981; Heiner, 1983; Fazio, 1986; 1990; Feather,
1990; Prelec and Herrnstain, 1997). These models can be said to share two
basic properties: (i) in arriving at (altruistic) behaviour people follow a
decision making process; and, (ii) situational factors (e.g., a direct appeal)
activate this decision making process. The main difference between these
models relates to their assumptions about the nature of the decision
making process: whereas several models assume a deliberative, reasoned,
mode (e.g., a cost/benefit calculus), other models opt for a more
spontaneous mode.
From the first category of models, we present two fairly similar
models of (reasoned) behaviour. The first model was initially developed by
Sheldon Stryker (1968; 1981), and then slightly amended by McAdam and
Paulsen (1993). The origin of this model is in the ‘symbolic interaction
theory’ in sociology. The second model, embodied in the psychological
literature, is Shalom H. Schwartz’s (1977) theory of altruistic behaviour.
Stryker’s initial work aimed to provide a set of ‘theoretical
propositions…that approximate a theory’ of human behaviour (1981: 24).
Stryker argues the view that behavioural choices are affected by identity
salience, which is a source of motivations that situational factors activate.
As he puts it: ‘[the self is comprised of] discrete identities [that] exist in a
hierarchy of salience, such that other things equal one can expect
behavioural product to the degree that a given identity ranks high in this
6
hierarchy,’ while, the concept of identity salience ‘may be defined as the
probability, for a given person, of a given identity being invoked in a
variety of situations’ (1968: 560-1). McAdam and Paulsen (1993) make use
of Stryker’s notion of identity salience in their explanation of individuals’
participation in high-risk movement activities. In this context, they argue,
a crucial factor in participating in high-risk movement activities is that
‘the individual must be the object of a recruitment appeal…that succeeds
in creating a positive association between the movement and a highly
salient identity’ (McAdam and Paulsen, 1993: 647). They then add a
further stage to this process in which the recruit is seeking to confirm the
linkage between the movement and the identity. In other words, a process
of deliberation (to seek support and approval) succeeds the activation of a
‘salient identity’. Following this deliberative stage, then, a decision on
involvement in movement activity may be taken.
Similarly, Schwartz (1977: 241), proposed a theory of altruistic
behaviour that includes the following four steps: (i) Activation of
perception of need and responsibility. As mentioned earlier, Schwartz
emphasises the role of a direct appeal in generating these perceptions. As
he puts it: ‘In addition to inducing responsibility, of course, appeals may
promote helping by drawing attention to the existence of a need,
overcoming ambiguity regarding its reality, and pointing to social
expectations for behaviour’ (1977: 249). (ii) Obligation step: norm
construction and a generation of feeling of obligation. (iii) Defence step.
This step is similar to the deliberative step in the previous model;
however, Schwartz makes the cost/benefit calculus very explicit. (iv)
Response step: action or inaction.
The idea that people weight the pros and cons of their intended
action is not novel. A common theme in the rational choice paradigm, for
example, is that people are maximising agents. In the context of helping
behaviour, then, the rational choice paradigm would predict that the
greater the cost the lower the likelihood a person will help others (cf.
Staub, 1974: 299). The rescue of persecuted Jews in Nazi Europe – a
7
potentially very costly (high-risk) behaviour – provides a challenge to this
apparent axiom. The fact is that many Jews were indeed rescued from
persecution in that period. How can one ‘square’ this fact with the rational
choice axiom? Fully committed students of rational choice theory may
argue in this context that rescuers of persecuted Jews in WWII did not
fully perceive the risk they were facing (cf. Opp, 1997). Others may deal
with the benefit component of the maximisation exercise, and argue that
rescuers may have included the persecuted Jew’s welfare into their utility.
In either case, one should arrive at the conclusion that the rational choice
paradigm provides a very problematic, if not a tautological, explanation.3
For a long time now, economists have come to realise that agents
may face situations in which maximisation is a problematic task that may
not even be their optimum strategy (cf. Heiner, 1981). Instead of
maximising, agents may simply follow ‘behavioural rules’ (Heiner, 1981:
561), or ‘principles’ (Prelec and Herrnstein, 1997, and the citations
therein). By definition, a principle does not supplement ordinary cost
benefit analysis, but rather replaces it (Etzioni, 1988). 4 Thus, it is
important to recognise, behaviour may occur without much thought about
expected consequences (Feather, 1990: 187). Having said that, a fully
developed theory of human behaviour has to acknowledge that people may
arrive at a behavioural intention in more than one way. Fazio (1986;
1990), offers such a theoretical framework.
Fazio’s main concern is with the ways in which attitudes guide
behaviour. As in other psychological models, situational factors play a
pivotal role in activating attitudes in Fazio’s model (Fazio, 1990: 92-3).
Fazio claims that a behavioural intention is the outcome of one of two
alternative decision making processes: a deliberative, reasoned, mode, and
We would like to emphasise, again, that we use a narrow definition of the term
rationality (cf. Opp, 1999).
3
It is not surprising, then, that in an interview with Jean Kowalyk Berger, a Ukrainian
villager, about the event that led her to become involved in rescue activities she said that
‘when I saw people being molested, my religious heart whispered to me, ‘Do not kill. Love
others as you love yourself’’ (Block and Drucker, 1992: 237-40, quoted in Geras, 1995: 3031).
4
8
a spontaneous mode. This model, known as the MODE model, postulates
that Motivations and Opportunities are seen as DEterminants of which
processing mode is likely to operate in any given situation. That is, when
both motivation and the opportunity to deliberate exist, the decision
making process will be deliberative. A deliberative mode in this context is
a process by which individuals ‘consider their attitudes toward the
behaviour in question,’ such that ‘these attitudes are computed on the
basis of an examination of the desirability of the likely consequences of the
action’ (1990: 93). Fazio suggests that ‘people reason and deliberate about
their future action in situations that are characterised by fear of
invalidity’ (1990: 93). Alternatively, ‘in situations that are not
characterised by fear of invalidity, or that are so characterised but do not
permit the opportunity for deliberation…behaviour will operate only
through the spontaneous processing mode’ (1990: 93). In this mode, when
a strong association has been established toward the attitude object, an
evaluation about a behavioural intention will be automatically activated
from memory (1990: 93-4).
A similar model was proposed by Feather (1990). However, instead
of confining the determinants of human behaviour to attitudes, he
broadens the analysis to also include values and motivations. Feather
(1990: 184-5) argues that values are more readily activated in situations,
and are organised at a deeper and more abstract level then attitudes.
Thus, Feather’ s model can be said ‘[to link] induced attitudes to
personally held general values and that these positive and negative
attitudes toward objects and events, together with cognitive beliefs or
expectancies, are assumed to influence the action that is taken (1990:
186).
To recapitulate the above discussion, then, the logic inherent in the
rational choice paradigm does not always guide behaviour. That is, in
arriving at a behavioural intention individuals are not always maximising
agents. Situational factors may induce individuals to act in accordance
with embedded norms, a salient personality, and so forth. The implication
9
of this discussion to our problem – i.e., the rescue of persecuted Jews in
the Holocaust – is fairly straightforward.
In the following analysis, then, we seriously consider the possibility
that in arriving at a helping intention rescuers of Jews in the Holocaust
may have opted for a non-deliberative decision-making mode. The above
discussion also emphasises that situational factors activate motivational
factors in arriving at this intention. Thus, in predicting helping behaviour,
we expect a positive interaction effect between situational and
motivational factors. Finally, in the literature two competing hypotheses
exist about the way in which this interaction effect may trigger altruistic
behaviour. The first hypothesis suggests that this interaction effect is
linear. This can be claimed based on Stryker’s (1981) ‘theory’ that suggests
that the more salient a ‘personality’ (e.g., values, motivations,
orientations, dispositions, attitudes, etc.) is the more readily available it is
for activation. Against this background, Schwartz (1977: 242) argues that
the activation of helping behaviour may not be linear. Instead, he
postulates that this process is governed by a threshold effect. Before we
embark on the analysis, however, we summarise the results of a number
of studies that explore the rescue of persecuted Jews in the Holocaust.
3. The Rescue of Jews in the Holocaust
London and his associates initiated (perhaps) the first systematic study of
rescuers of Jews in the Holocaust. The objective of this study was to
compare personality characteristics between rescuers, bystanders, and
known collaborators with the Nazis. As this study was never completed,
London (1970), reported the results of the analysis of its pilot study
(carried out in the 1960s), comprising of 27 rescuers and 42 rescued
individuals who had emigrated from Europe since 1945. In this report,
London contends that ‘a zest for adventure and the working of chance both
were important in the initiation of rescue behaviour…’ (p. 249). Although,
as London acknowledged, the lacunae in these data are vast to even
10
speculate about the generality of the results, London brings to the fore the
importance of situational and motivational (personality) factors in the
explanation of altruistic behaviour.
Nevertheless, influential sociologists invoke the ‘motivational’
explanation for the rescue of Jews during the Nazi occupation of Europe in
WWII. Thus, for example, the actions of the Danes who saved Jews have
been described as deriving from ‘clear convictions […] in accord with the
inner truth of man’s own rational nature, as well as in accordance with the
fundamental law of God: “thou shalt love thy neighbor as thyself”’
(Merton, 1971: 167, quoted in Gross, 1997: 128). Similarly, the altruistic
behaviour of the French citizens of Le Chambon provides the background
for a motivational explanation for Jon Elster (1989). In this small village
in southern France, inspired by a Protestant pastor, Andre Trocmé, the
villagers provided asylum for a large number of German Jews at great risk
to themselves and under constant surveillance by the Vichy government
and, later, the German army. In Elster’s view, rescuers did not consider
the consequences of their actions. Rather, they acted because they were
motivated by a moral principle: ‘Never turn away anyone who needs help’’
(Elster, 1989: 193).5
The majority of the research on this unique altruistic behaviour,
however, went on outside the realms of sociology. Thus, for example,
Oliner and Oliner (1988) use the label ‘altruist personality’ in their
extensive study of rescuers in Nazi Europe. They interviewed 231 gentiles
(non-Jews) who saved Jews, and 126 non-rescuers matched on age, sex,
education, and geographic location during the war. Oliner and Oliner link
a variety of psycho-social conditions to the ‘altruist personality’ and
conclude that rescuers had a capacity for ‘extensive relationships’, defined
as a ‘stronger sense of attachment to others and their feeling of
In the context of our theoretical discussion on decision-making processes, it is possible
to argue that Merton and Elster would subscribe to the view that in arriving at altruistic
behaviour people may follow principles and rules instead of a deliberative process.
However, these authors may still over -estimate the effects of motivations in this process.
5
11
responsibility for the welfare of others, including those outside their
immediate familial or communal circle’ (1988: 249).6
Monroe, Barton and Klingemann (1990) base their study on a
sample of thirteen rescuers of Jews, an unidentified number of
entrepreneurs, and five ordinary Europeans who lived in Nazi-occupied
Europe but did not participate in rescuing Jews. They describe the
altruism of rescuers in terms of ‘self-identity’, by which individuals
perceive themselves ‘as one with all humankind’, an identity which
reaches beyond group affiliation, mere empathy and calculation of
expected utility. For rescuers, the concept of a cost/benefit calculus was
‘meaningless’. On the contrary, they were motivated by the ‘shared
perception of a common humanity’ (1990: 117; see also Monroe, 1991 and
1996).7
Geras (1995), the author of a philosophical critique of Richard
Rorty, discusses the ‘Righteous Among the Nations’. In a detailed review
of the existing literature on the rescue of Jews in Nazi Europe, he
examines whether rescuing behaviour was associated with gender, class
status, political affiliation, religion and other personal characteristics, and
prior acquaintance with Jews. Geras concludes from this review that none
of the above mentioned ‘sociological factors’ were good predictors of
altruistic behaviour towards Jews in Nazi Europe. On the contrary, he
argues that people were moved by a sense of belonging to ‘human kind’. A
‘universalistic moral outlook’ motivated helping behaviour (1995: 36).8
Interestingly, this notion of ‘attachment’ runs contrary to London’s conclusion that
‘…the experience of social marginality gave people the impetus to continue their rescue
activities.’ (1970: 249).
6
As indicated in our theoretical discussion, we also think that the decision making
process in these, and perhaps other, occasions is fairly ‘free’ from the cost/benefit calculus
(cf. Heiner, 1983; Fazio, 1986; Prelec and Herrnstein, 1997). However, in that discussion
we present an alternative view of the decision making process; a view that does not
consider motivations to be the only explanation.
7
As far as motivations are the focus of the analysis, Geras’s position does not differ
fundamentally from that of Richard Rorty (1989). They differ only in the unit of
reference. Rorty argues that helpers were moved by feelings of psychological attachment
to members of smaller groups, such as ‘comrade in the movement’, ‘fellow Bocce player’,
or ‘fellow Milanese’, rather than a sense of belonging to humanity.
8
12
Although the majority of the studies cited above embrace the
motivational explanation for altruistic behaviour, they do not turn a blind
eye to situational factors. These studies, moreover, acknowledged that
rescuers were aware of the costs, or at least the risk, involved in helping
Jews. However, this awareness did not impinge on their decision to rescue;
it simply made rescuers more cautious (see Monroe et al. 1990: 108; Oliner
and Oliner, 1988: 126-7). Oliner and Oliner (1988: 271-2), for example,
recognise that material opportunities, information and other factors
outside the scope of the individual rescuer played a role in explaining
rescue activity, but their work stops short of offering an account of the
nature and importance of such factors.
A step in that direction, however, is provided by Gross (1997), an
advocate of rational choice theory. His study is based on historical records
and memoirs, and a survey of 175 French and Dutch rescuers who
operated in Le Chambon, several small villages in the Cevennes region in
southern France, and in Niewlande, Holland. In this study Gross contrasts
the presence of both motivational and situational factors with instances
where situational factors were missing, such as the case of the JapaneseAmericans who were deported during WWII in the United States (Gross,
1997: 129; see also Gross, 1993). Gross, then, argues from a rational choice
perspective that in explaining altruistic behaviour ‘moral motivations
must be considered in conjunction with non-moral motivations, situational
factors, and mobilization contexts’ (1997: 129).
Although Gross’s treatment of the rescue of Jews is a major step
forward, his study has two main shortcomings. First, Gross focuses only
on cases of collective help, thereby failing to offer insights on individual
acts of helping. Instances of individual acts of rescue occurred alongside
collective efforts. Even within Gross’s sample of rescuers, mainly drawn
from individuals involved in collective rescue activities, instances of
individual acts of rescue are of significant number: twenty-one percent of
the French respondents said that ‘no one’ had organised their rescue
activities (Gross, 1997: 140, Table 5.2). Secondly, and perhaps more
13
importantly, Gross’s data lack variation in the dependent variable. This
weakness is due to the well-known danger of sampling on the dependent
variable.9 It may well be that some Europeans with similar characteristics
to the 175 individuals in Gross’s sample did not help Jews in those dark
days. Can we then claim a relation of causality between any of these
characteristics and rescue activity?
A remedy to these shortcomings is provided in Varese and Yaish’s
(2000) study. These authors studied the determinants of helping Jews
from Nazi persecution in World War II by analysing Oliner and Oliner’s
data. To these data they were able to apply the case control sample
techniques, and showed that being asked to help is the most significant
factor in predicting helping behaviour. Elster (2000) viewed this finding in
the context of situational factors that impinge on behaviour. That is, and
as also Varese and Yaish suggest, the request for help provides individuals
with the opportunity to act. However, Varese and Yaish (2000) further
postulate that being asked to help may trigger such action. Thus, they
hypothesise that given the individual’s disposition – willingness to help –
being asked (situational factors) puts this disposition to a test, the result
of which is the observed behaviour.
This hypothesis coincides very well with our current theoretical
discussion on the determinants of human action and behaviour. In what
follows, we put this hypothesis to an empirical test. Before we embark on
the analysis, however, we present the data and the analytical strategy of
our investigation.
4. Data and Methodology
Our study is based on a secondary analysis of data collected by The
Altruistic Personality and Prosocial Behaviour Institute (APPBI), which
were first analysed by Oliner and Oliner (1988). The data as we received
See McAdam and Paulsen’s (1993) discussion of this problem in relation to studies on
joining voluntary movements.
9
14
them from APPBI contain a sample of 346 identified rescuers of Jews,10
and a sample of 164 individuals who lived in Nazi Europe during WWII
but were not identified as Jewish rescuers (N=510). These data are
slightly different from the set used by Oliner and Oliner (1988: appendices
A, B), and we were unable to reconstruct their analyses with complete
accuracy. In what follows, we refer to the data as we received them from
APPBI in October 1996.
Studying the rescuers of Jews during the Nazi occupation of Europe
is best seen as the study of rare events since the dependent variable
(rescuing Jews) would not be easily identified in a random sample of men
and women who lived in Europe during the WWII period. A solution to
this problem can be achieved by the use of retrospective samples – known
also as case-control samples – (see Agresti, 1990; 1996; Lacy, 1997;
Manski, 1995; Manski and Lerman, 1977; Xie and Manski, 1989). In the
collection of the APPBI data, Oliner and Oliner (1988) followed this
sampling method.
The Oliners’ first task was to identify a sample of altruistic
individuals who helped Jews during the War period – i.e., the case. For
Oliner and Oliner, behaviour is characterised as altruistic when: ‘(1) it is
directed toward helping another; (2) it involves a high risk or sacrifice to
the actor; (3) it is accompanied by no external reward; (4) it is voluntary’
(1988: 6). The majority of rescuers (95%) were sampled from the Yad
Vashem list of ‘Righteous Among the Nations,’ which included at the time
of the collection of the data approximately 5,500 rescuers. 11 However,
individuals were not randomly sampled from that list; they were selected
so that the entire sample would be as diverse as possible in terms of age,
The Yad Vashem institute in Jerusalem undertook the identification process. The Yad
Vashem is an Israeli agency established in 1953, seeking to identify and give due
recognition to rescuers of Jews during Nazi rule in Europe. Over the years, it has
certified more than six thousand people as ‘Righteous Among the Nations’ (Oliner and
Oliner, 1988: 262). It should be noted that the number of Yad Vashem rescuers per
country does not correspond to the number of Jews rescued per country.
10
Today the Yad Vashem list of ‘ Righteous Among the Nations’ includes 16,542 identified
rescuers (http://www.yad-vashem.org.il/righteous/index.html).
11
15
socioeconomic class, country of origin, as well as other factors (Oliner and
Oliner, 1988: 263). The other five per cent in this category consisted of
individuals whose names were obtained from rescuees interviewed by the
project (Oliner and Oliner, 1988: 262). As mentioned above, the APPBI
data we analyse include 346 individuals that meet these criteria of
altruistic behaviour.
The second task was to identify a sample of individuals who did not
help Jews during the War period – i.e., the control. Oliner and Oliner
defined a non-rescuer as ‘a person neither on Yad Vashem list nor verified
by our project as a rescuer living in Nazi occupied Europe during the War’
(1988: 263). Again, these individuals were not randomly sampled from the
entire universe. Furthermore, the case and the control samples did not
share the property of matched case control samples (cf. Agresti, 1990;
1996). Instead, non-rescuers were selected so that no statistically
significant differences between the rescuers and the non-rescuers existed
in relation to age, sex, education, and geographical location during the
War period (1988: 263). 12 As noted earlier, the data we analyse include
164 non-rescuers. However, when the non-rescuers were interviewed, it
became apparent that they were not homogeneous on the dependent
variable; that is, with respect to helping behaviour. Some 40 per cent of
these ‘non-rescuers’ claimed to ‘have done something out of the ordinary to
help people during the War period’.
To recapitulate the above, then, the APPBI data we analyse
(N=510) are made up of two samples that consist of three sub-populations:
(i) identified rescuers (N=346); (ii) self reported rescuers (N=67); and (iii)
non-rescuers (N=97). Oliner and Oliner approached these data by
analysing the three sub-populations separately (1988: 264). An alternative
approach to analysing these data would be to include in the case sample
both the identified rescuers and the self reported rescuers
(N=346+67=413), and to assign to the control sample only those who did
In the non-rescuer sample the mean average age is four years lower than in the rescuer
sample.
12
16
not help anyone during the War period (N=97). In our analysis we decided
to adopt the second approach for two reasons. First, we have no reason to
doubt the claim of self-reported rescuers. 13 The only appreciable difference
between the two ‘altruist’ populations is that self reported rescuers were
not – in 1988 – certified by the Yad Vashem authority (a number of them
were indeed certified later). Second, consolidating the two sub-populations
has the advantage of increasing the total N in our analysis.
As mentioned above, the most appropriate method to adopt in the
analysis of these data is the case-control samples. This method requires us
to apply logistic regression to the analysis of these data, which is based on
odds ratios, in order to estimate the retrospective effects of the
independent variables on the response variable. Since we do not have
information on the true marginal distribution of the response variable in
the population, we are unable to weight the data. This would cause some
problems in interpreting the intercept in our models. However, the
coefficients for the effects of the independent variables are interpretable.
That is to say, we lose the predictive power of the models, but we can learn
about the effects of different variables on the outcome we are interested in.
We turn next to present the variables we used in the analysis.
The dependent, or response, variable ALTRUIST corresponds to
the case (RESCUE=1 and E9a=1) and the control (RESCUE? 1 and E9a? 1)
samples, as we have explained above. Since we have already shown in a
previous study that a request to help is the most important determinant of
altruistic behaviour in these data (Varese and Yaish, 2000), and since our
theoretical model suggests that asking might have triggered that
behaviour (in the sense that a request to help [situational factor] might
‘activate’ an existing altruistic orientation [motivations]), we consider only
two independent variables in the analysis. The first – ASKED –
corresponds to the situation (opportunity), and the second – PROSOCIAL
We have nevertheless repeated our analyses without the ‘self-reported altruists’ and
found that the pool of self-reported rescuers did not alter the final result we report; that
is, our decision to put together self-reported and ‘identified’ rescuers has not affected our
results. We are willing to supply the relevant analysis to those who might be interested.
13
17
ACTION ORIENTATION – corresponds to the personality of the
individual (i.e., motivation or disposition).
The variable Asked distinguishes between those who were asked to
help (ASKED=1), and those who were not asked to help (ASKED=0). This
variable is constructed separately for the case and the control samples.
Those who did behave altruistically during the War (case) were asked to
report (E27): ‘How did you become involved in this first activity? Did you
initiate it yourself or did someone ask for your help?’ Those who did not
behave altruistically during the War (control) were asked to report (E40):
‘Was there ever a time during the war that you were asked to help
somebody and had to say no?’ 14 We restrict our analysis to the first
involvement in helping activity (or the first rejection of a request for such
involvement), because a path-dependent process can account for
subsequent behaviour.15
Our motivational variable is a Prosocial Action Orientation
score. This variable was first constructed from the APPBI data by Oliner
and Oliner (1988), and it represents one of three factors that emerged
from a factor analysis on 42 personality items relating to present time
(F14: 1-42; see, Oliner and Oliner, 1988: 317). We followed Oliner and
Oliner in constructing this variable (see Appendix A below). 16 It is
As in every survey, the issue of the validity of the answers applies here. This issue is
even more significant in the case of individuals who might be unwilling to admit having
being asked and refused to help fellow human beings in danger. However, the wording of
the question by Oliner and Oliner enables the respondent to admit to not having helped
with minimal loss of face. Indeed, the survey was able to identify a number of people who
were both asked and replied in the negative. Furthermore, it was able to establish that
most people had to be asked in order to help Jews (Oliner and Oliner, 1988; Varese and
Yaish, 2000). In other words, rescuers did not offer their help spontaneously, an
admission which might also be unwelcome. Setting aside the general question
surrounding the reliability of survey data, the validity of the results presented below
seems to us to be worth taking seriously.
14
The psychological literature discusses in this context the ‘foot in the door’ effect
(Freedman and Fraser, 1966). Although not completely comparable with the path
dependence process that we refer to, it emphasises that people are likely to take a large
and consequential action if they are initially induced to take a small and nonconsequential action (cf. Ross and Nisbett, 1991: 50-1).
15
We have constructed this measure in various other ways, all of which did not produce
significantly different results. The results of these analyses can be obtained from the first
author on request.
16
18
important to emphasise the potential bias that this variable may produce
in the analysis.
Since our Prosocial Action Orientation measure is the product of a
factor analysis on 42 personality items relating to present time, we make a
rather strong assumption about the consistency and stability of an
individual’s personality characteristics over time. Thus, we assume that
the ‘prosocial action orientation’ measure, which considers present day
attitudes, is a valid and accurate measure of an individual’s prosocial
action orientation during the Second World War. This issue is even more
significant in our analysis because the event of helping a distressed Jew
during the War – or having experienced the hardship of this period – may
have affected one’s attitudes and orientations. We cannot resolve this
problem, nor can we estimate the bias it may generate in our analysis.
However, the psychological literature, which has confronted this issue
extensively, suggests that attitudes and orientations are developed early
in life, and after these impressionable years attitudes and orientations are
fairly stable and consistent (Searing et al., 1976; Sears, 1981; 1983). The
Bennington study, for example, supports this claim, as it is concluded that
‘after some early period of influence and change, attitudes become
crystallized and increasingly stable with age’ (Alwin, Cohen, and
Newcomb, 1991: 264).17 Having made these preliminary clarifications, we
can embark on the analysis.
5. Results
The point of departure in our analysis is the claim that a direct appeal for
help triggers altruistic behaviour (cf. Schwartz, 1977). In the context of
the Holocaust, for example, Varese and Yaish (2000) have recently shown
the importance of a direct appeal for altruistic behaviour by analysing the
It follows from the above that young individuals (in the War period) may be a greater
source of bias in our analysis. Thus, we repeated the analysis excluding those under 21
years of age in 1940, and found no significantly differently results. These results can be
obtained from the first author on request.
17
19
APPBI data. Thus, a cross tabulation of the variables Asked by Altruist
(see Table 1) shows that two-thirds (237/359=66%) of the rescuers were
asked to help, and only one third initiated their action. Moreover, nearly
all of those who were asked to help Jews did so (237/247=96%), while a
request to give help increased the likelihood of helping others by a factor
of two compared to help that was initiated without a request
(237/122=1.94).
TABLE 1 ABOUT HERE
An appeal for help, we have argued, initiates two simultaneous processes.
It may guide behaviour by activating a salient personality (or orientation)
of the individual, while at the same time, it triggers a decision making
process about the response (i.e., behaviour) to this appeal. In arriving at a
decision, however, people may follow a spontaneous or a deliberative
mode. It is the latter part of the argument that we first confront.
The notion that rescuers of persecuted Jews may have followed a
spontaneous mode of decision-making process concerning their rescue
activity would appear to have gained some support from the APPBI data.
Thus, for example, despite the fact that some fifty percent of those who
helped Jews claimed that by so doing they had taken an extreme risk to
themselves and their family, the majority of them arrived at this decision
instantly (within minutes) – without even consulting anyone. 18 That is not
to say, however, that risk did not impinge on the decision to help, as it
emerged that about eighty percent of those who did not help after being
asked claimed to have been facing extreme risk. 19
Amongst those who helped after being asked (minimum N=147), 75% made the
decision within minutes, 77% did not consult anyone about this decision, 46% claimed to
have taken extreme risk to themselves and 54% claimed to have taken extreme risk to
their family. Amongst those who helped without being asked (minimum N=91), 80%
made the decision within minutes, 80% did not consult anyone about this decision, 51%
claimed to have taken extreme risk to themselves and 44% claimed to have taken
extreme risk to their family.
18
However, we are unable to carry out a more robust analysis of the effect of perceived
risk on helping behaviour because we lack information on the perceived risk of those who
did not help and were not asked to help.
19
20
It is important to note that in arriving at a helping behaviour
different people may follow different modes of the decision-making
process. That is, an appeal for help does not necessarily invoke the
spontaneous mode of the decision-making process. We cannot rule out the
possibility that when some people were asked to help they began a
deliberative process that resulted in a decision to help. We simply argue
that this may not be the only way in which people arrived at this
decision.20 We also argue that, regardless of the decision mode, being
asked to help may guide behaviour by activating a salient personality (or
orientation). This being the case, then, we would expect to find a positive
interaction effect between a request for help and the individual’s
personality. This hypothesis, then, is the focus of the next analysis.
We return to the APPBI data in order to test this hypothesis. Table
2 presents four logistic regression models on the dependent variable:
helping vis-à-vis not helping. Model I assesses the log-odds effects of our
‘prosocial orientation action’ variable on helping behaviour. As one might
expect (see, Oliner and Oliner, 1988), this model reveals that the higher is
one’s prosocial orientation, the more likely this person to help Jews. In
model II we then assess the log-odds effects of being asked on helping
behaviour. As expected (see table 1), this model reveals that asking for
help is positively and statistically significantly associated with helping
behaviour. Respondents who were asked to help were about 16 times more
likely to help (e2.79=16.3) compared to respondents who were not asked.21
Model III then estimates the effects of both prosocial orientation and being
asked to help on helping behaviour. It is shown that each of the two
In other words, since we are not committed to the rational choice explanation we do not
need to make strong assumptions about ‘costs and benefits’ that may have affected an
individual’s decision to help Jews in the context of the Holocaust.
20
In a similar analysis with the APPBI data, Varese and Yaish included in this model a
variety of independent variables (demographic and opportunity) and arrived at a similar
conclusion (2000: 322). In that analysis they were not only able to show that these
demographic and opportunity variables have relatively weak effects, but also that the
vast majority of them are not statistically significant (Varese and Yaish, 2000: 321, Table
4). In light of this, we do not think that it is necessary to ‘control’ for these covariates in
our analysis.
21
21
factors (i.e., motivational and situational, respectively) exerts an
independent, and relatively stable (compared with models I and II) effect
on helping behaviour. It is also apparent from this analysis that being
asked is a more powerful predictor of that behaviour.22
TABLE 2 ABOUT HERE
To what extent, then, might asking for help be a subtle way of inducing
the receiver of a request to say ‘yes’? To rephrase this question in light of
our theoretical discussion from above would be to say that a direct appeal
might guide behaviour by activating a salient personality (or orientation)
of the individual. This being the case, we would expect an interaction
effect between our two independent variables – asked and prosocial action
orientation. Model IV, then, provides an empirical test of this hypothesis.
It can be seen that allowing for an interaction between asked and
prosocial action orientation improves the overall fit of the model (relative
to model III) significantly (the deviance is reduced by more than 6 points
for one additional parameter used by the interaction effect), and the Cox
and Snell R 2 is the highest achieved. More importantly, however, the
interaction effect is positive and statistically significant. Thus, a request
for help increases the likelihood of helping Jews by a factor of 30
(e3.392=29.7), but a request for help from someone who is also prosocially
oriented increases this likelihood even more (by a further factor of 3
[e1.023=2.78] for every unit increase in prosocial orientation). We interpret
this positive interaction effect as a confirmation of our hypothesis that a
direct appeal activates a pre-existing disposition (orientation) to engage in
helping activity.
It is important to know, next, if this interaction effect is accurately
estimated by the linear function. In the theoretical discussion it was
Varese and Yaish (2000) argue that a selection process governs the identification of
potential helpers who may then be asked to help. That is, persecuted Jews did not ask for
help at random. One of the implications of this is that the effect of being asked is biased,
and hence very strong.
22
22
argued (cf. Stryker, 1981) that the more salient a personality is the more
likely a situational factor (e.g., a direct appeal) is to activate this
personality. Thus, it is also implied that the interaction between
situational and motivational factors should be accurately estimated by the
linear function. Against this background, Schwartz (1977: 242) argues
that the activation of helping behaviour may be governed by a threshold
effect. Thus, it is expected that the interaction between situational and
motivational factors should not be accurately estimated by the linear
function. Our final analysis assesses this hypothesis.
In the following, then, we explore the interaction model by applying
nonparametric techniques to the analysis of the APPBI data. More
specifically, we draw on the S-Plus’s ‘Modern Regression Module’
(Venables and Ripley, 1994: ch. 10). These regression methods do not
necessarily use non-linear parameterisation, but they do allow non-linear
functions of the independent variables to be chosen by the procedures
(1994: 247). Thus, for example, in our original logistic regression models
(see Table 2) there are n observational units, each of which records a
random variable Yi with a binomial (ni ,pi) distribution, and (pi) is
determined by
p
log( p ) = α + ∑ β
j =1
j
X
j
The procedure GAM (generalised additive models) in S-Plus then replaces
the linear function
β X
j
j
by a non-linear function, to get
p
log( p ) = α + ∑
j =1
f (X
j
j
)
Since it will not be useful to allow an arbitrary function fj , it will be useful
to think about it as a smooth function. In what follows, we use the spline
smoothing function. Having made these preliminary clarifications, we can
embark on the analysis.
In examining the interaction effect, we apply a generalized additive
model to each category of the variable ASKED. That is, in the first model
we predict altruistic behaviour as a smoothed function of prosocial action
23
orientation only for those who were asked to help, while in the second
model the same procedure is followed but only for those who were not
asked to help. Figure 1 presents the smoothed functions of the predictor
prosocial action orientation, separately for the two sub-samples. Table 3
then indicates that neither of these smoothed functions is statistically
significantly different from the linear functions (the nonparametric chisquare is reduced by little over 5 points for 3 nonparametric d.f.). The
conclusion from the above analyses is that the interaction terms in our
model are linear. In other words, we find no evidence in our data that
would suggest a threshold effect in the activation of motivational factors.
FIGURE 1 ABOUT HERE
TABLE 3 ABOUT HERE
6. Discussion
This paper is, partially, about the good in society. It explores the
mechanisms under which those with good will and those in need are
matched. As the citation in the introduction to this paper indicates, this
matching process is far from being optimal, and thus the observed acts of
altruism do not account for the potential acts of altruism human beings
are capable of. Our task, however, did not stop at this point. We thus took
Marshall’s comment very seriously, and sought to show how the observed
acts of altruism might reach this potential.
Our analysis has shown that in the context of the Holocaust, a
direct appeal for help is an important factor in the explanation of altruistic
behaviour. We then argued that a request for help not only provides
individuals with the opportunity to help, but it also triggers such
behaviour. In support of this claim, our analysis has shown that a request
for help interacts with prosocial action orientations (motivations) in
24
generating altruistic behaviour. Thus, in the context of the debate about
the determinants of altruistic behaviour, between the ‘situationalist’ and
the ‘motivationalist’ views, our analysis lends support to an integrative
view that postulates that situational factors activate motivational factors
(cf. Schwartz, 1977, and Stryker, 1981, among others).
To recapitulate the arguments from above, rescuers of persecuted
minorities – such as the Jews in Nazi occupied Europe – face at least one
dilemma. They might be willing to help but be uncertain how to go about
rescuing. Being asked to help may solve this dilemma, by providing
individuals with the opportunity to act in accordance with their
motivations (see also, Varese and Yaish, 2000). That is, a direct appeal
activates pre-existing motivations. Altruistic behaviour, however, does not
necessarily follow from this activation process. What does follow from this
activation process is a decision making process, the end result of which
may be helping persecuted minorities.
This paper, then, is also partly about this decision making process.
We have discussed earlier two different modes of the decision-making
process: a deliberative mode, and a spontaneous mode (cf. Fazio, 1986;
1990; Feather, 1990). We then showed that in helping Jews in the
Holocaust, some rescuers, at least, have made a spontaneous decision
(after motivations were activated by a request to help). This mode of
decision-making process runs contrary to the deliberative mode, and
implies that for some individuals ‘the concept of a cost benefit calculus was
‘meaningless’’ (Monroe, Barton and Klingemann, 1990: 117). As the cost
benefit calculus is an essential element in the rational choice paradigm,
we would like to discuss next the implication of our study to this
paradigm.
The common wisdom embodied in the rational choice paradigm is
that people are maximising agents. Thus, in the context of helping
behaviour the rational choice paradigm would predict that the greater the
cost the lower the likelihood a person will help others (cf. Staub, 1974:
299). This prediction is fairly confirmed in the context of the Holocaust, as
25
it is known that about six million Jews who lived in Nazi occupied Europe
were not rescued. In other words, one can safely maintain that a large
number (if not the majority) of the non-Jews did not rescue their fellow
Jews from the Holocaust. However, the fact that many Jews were indeed
rescued from persecution in that period is a puzzle to advocates of this
approach. To ‘successfully’ apply the rational choice explanation to these
puzzling occurrences, then, requires one to ‘adjust’ the assumptions about
the costs or the benefits that are involved in rescuing Jews in that period.
Thus, for example, one has to either speculate that rescuers of persecuted
Jews in WWII did not fully perceive the risk they were facing (cf. Opp,
1997), or that they may have included into their utility the persecuted
Jew’s welfare.
Either adjustment would make the refutation of the rational choice
explanation impossible. A theory, however, must meet the falsification
criterion (cf. Goldthorpe, 2000). This being the case, then, one has to
arrive at the conclusion that the rational choice explanation has its
limitations. That is, although rational choice theory can explain why some
people did not help Jews in the Holocaust, it cannot explain the opposite.
The rational choice theory, it must be said, confronts similar difficulties in
the explanation of other types of behaviours. Thus, for example, rational
choice theory provides a very plausible explanation of voting behaviour,
though it fails almost completely to explain why people vote at all.
26
References
Alwin, D.F., Cohen, R.L. and Newcomb, T.M. 1991. Political Attitudes over
the Life Span. The Bennington Women after Fifty Years. Wisconsin:
The University of Wisconsin Press.
Agresti, A. 1990. Categorical Data Analysis. New-York: John Wiley and
Sons.
Agresti, A. 1996. An Introduction to Categorical Data Analysis. New-York:
John Wiley and Sons.
Berkowitz, L. and Daniels, L.R. 1963. “Responsibility and Dependence”.
Journal of Abnormal and Social Psychology, 66: 429-36.
Block, G. and Drucker, M. 1992. Rescuers. Portraits of Moral Courage in
the Holocaust. New York: Holmes and Meier.
Darley, J. M. and Latane, B. 1968. “Bystanders Intervention in
Emergencies: Diffusion in Responsibility”. Journal of Personality
and Social Psychology 8: 377-83.
Elster, J. 1989. The Cement of Society. Cambridge: Cambridge University
Press.
Elster, J. 2000. “Rational choice history: A case of excessive ambition”.
American Political Science Review, 94 (3): 685-695.
Etzioni, A. 1988. The Moral Dimension: Toward a New Economics. New
York: Free Press.
Fazio, H.R. 1986. “How do attitudes guide behaviour”. In Handbook of
Motivation and Cognition. Foundations of Social Behaviour, Vol. 1,
eds. R.M. Sorrentino and E.T. Higgins, pp. 204-243. New York: John
Willy & Sons.
Fazio, H.R. 1990. “Multiple processes by which attitudes guide behaviour:
the MODE model as an integrative framework”. In Advances in
Experimental Social Psychology, Vol. 23, ed. M.P. Zanna, pp. 75109. New York: Academic Press.
Feather, N.T. 1990. “Bridging the gap between values and action”. In
Handbook of Motivation and Cognition. Foundations of Social
Behaviour, Vol. 2, eds. E.T. Higgins and R.M. Sorrentino, pp. 151192. New York: The Guilford Press.
Ferejohn, J. 1990. “Rationality and Interpretation. Parliamentary
Elections in Early Stuart England”. In K.R. Monroe (ed.) The
Economic Approach to Politics. A Critical Reassessment of the
Theory of Rational Action, New York: Harper-Collins, pp. 279-305.
Freedman, J.L. and Fraser, S.C. 1966. “Compliance without pressure: the
foot-in-the-door technique”. Journal of Personality and Social
Psychology, 4: 195-202.
27
Freeman, R.B. 1993. “Give to Charity? - Well. Since you Asked”.
Unpublished paper. LSE conference on the economics and
psychology of happiness and fairness. 4-5 November.
Geras, N. 1995. Solidarity in the Conversation of Humankind. London:
Verso.
Gerber, A. and Green D. 1999. ‘The Effects of Canvassing, Phone Calls
and Direct Mail on Voter Turnout: A field experiment’. Unpublished
manuscript, Institution for Social and Policy Studies at Yale
University.
Goldthorpe, J.H. 2000. On Sociology. Number, Narratives, and the
Integration of Research in Theory. Oxford: Oxford University Press.
Gross, M.L. 1993. “The Collective Dimension of Political Morality”.
Political Studies 42: 40-61.
Gross, M.L. 1997. Ethics and Activism. The theory and practice of political
morality. Cambridge: Cambridge University Press.
Heiner, R.A. 1983. “The Origin of Predictable Behaviour”. American
Economic Review, 73(4): 560-595.
Lacy, M.G. 1997. “Efficiently Studying Rare Events: Case-control Methods
for Sociologists.” Sociological Perspectives 40: 129-45.
Latané, B., Nida, S.A. and Wilson, D.W. 1981. “The Effects of Group Size
on Helping Behavior”. In Altruism and Helping Behavior: Social
Personality and Developmental Perspectives, eds J.P. Rushton and
R.M. Sorrentino, pp. 287-313. Hillsdale, N.J.: Lawrence Erlabaum
Associates.
London, P. 1970. “The Rescuers: Motivational Hypotheses about
Christians who Saved Jews from the Nazis”. In Altruism and
Helping Behaviour, eds. J. Macaulay and L. Berkowitz, pp. 241-250.
New York: Academic Press.
Manski, C. F. 1995. Identification Problems in the Social Sciences.
Cambridge: Harvard University Press.
Manski, C. F. and Lerman. S. R. 1977. “Estimation of Choice Probabilities
from Choice-based Samples.” Econometrica 45: 1977-1989.
Marshall, A. 1890. Principles of Economics. London: Macmillan.
McAdam, D. and Paulsen R. 1993. ”Specifying the Relationship between
Social Ties and Activism”. American Journal of Sociology 99: 64067.
Merton, T. 1971. The Non-violent Alternative. New York: Farrar. Strauss
and Giroux.
Milgram, S. 1963. “A Behavioural Study of Obedience”. Journal of
Abdnormal and Psychology, 67: 371-78.
28
Monroe, K., M.C. Barton and U. Klingemann. 1990. “Altruism and the
Theory of Rational Action: Rescuers of Jews in Nazi Europe”. Ethics
101: 103-22.
Monroe, K. 1991. “John Donne’s People: Explaining Altruism through
Cognitive Frameworks”. Journal of Politics 53: 394-433.
Monroe, K.R. 1996. The Heart of Altruism. Perception of a Common
Humanity. Princeton: Princeton University Press.
Oliner, S.P. and Oliner P.M. 1988. The Altruistic Personality. New York:
The Free Press.
Opp, K.-D. 1997. “Can Identity Theory Better Explain the Rescue of Jews
in Nazi Europe than Rational Action Theory?” Research in Social
Movements, Conflict and Change 20: 223-53.
Opp, K.-D. 1999. “Contending Conceptions of the Theory of Rational
Action”. Journal of Theoretical Politics 11 (2): 171-202.
Piliavin, J.A. and Charng. H-W. 1990. “Altruism: A Review of Recent
Theory and Research”. Annual Review of Sociology 16: 27-65.
Prelec. D. and Herrnstein, R.J. 1997. “Preferences or Principles:
Alternative Guidelines for Choice”. In The Matching Law. Papers in
Psychology and Economics, eds. H. Rachlin and D.I. Laibson, pp.
293-312. New York: Russell Sage Foundation, and Cambridge,
Mass.: Harvard University Press.
Rorty, R. 1989. Contingency, Irony, and Solidarity, Cambridge: Cambridge
University Press.
Ross, L. and Nisbett, R.E. 1991. The Person and the Situation. Perspectives
of Social Psychology. Boston, Mass.: McGraw-Hill.
Searing, D.D., Wright, G. and Rabinowitz, G. 1976. “The Primacy
Principle: Attitude Change and Political Socialisation”. British
Journal of Political Science, 6: 83-113.
Sears, D. 1981. “Life-stage effects on attitude change, especially among
the elderly”. In Aging: Social Change, eds. S.B. Kiesler, J.N.
Morgan, and V.K. Oppenheimer, pp. 183-204. New York: Academic
Press.
Sears, D. 1983. “On the persistence of early political dispositions: The
roles of attitude object and life stage.” In Review of personality and
social psychology, ed. L. Wheeler, Vol. 4: 79-116. Beverly Hills: Sage
Shaw, L., Batson C. and Todd, M. 1994. “Empathy avoidance: Forestalling
feeling for another in order to escape the motivational
consequences”. Journal of Personality and Social Psychology 65:
879-887.
Schwartz, S.H. 1977. “Normative Influences on Altruism”. In Advances in
Experimental Social Psychology, Vol. 10, ed. L. Berkowitz, pp. 222279. New York: Academic Press.
29
Simmons, R.G., Klein, S.D., Simmons, R.L. 1977. The Gift of Life: The
Social and Psychological Impact of Organ Transplantation. New
York: Wiley.
Sober E. and Wilson, D.S. 1998. Unto Others. The Evolution and
Psychology of Unselfish Behavior. Cambridge, Mass.: Harvard
University Press.
Staub, E. 1974. “Helping a Distressed Person: Social, Personality and
Stimulus Determinants”. In Advances in Experimental Social
Psychology, vol. 7, ed. L. Berkowitz, pp. 293-341. New York:
Academic Press.
Stryker, S. 1968. “Identity Salience and Role Performance: The Relevance
of Symbolic Interaction Theory for Family Research”. Journal of
Marriage and the Family, (November): 558-565.
Stryker, S. 1981. “Symbolic Interactionism: Themes and Variations”. In
Social Psychology: Sociological Perspectives, eds M. Rosenberg, and
R.H. Turner, pp. 3-29. New York: Basic.
Varese, F. and Yaish, M. 2000. “The importance of being asked: The rescue
of Jews in Nazi Europe”. Rationality and Society, 12 (3): 307-334.
Venables, W.N. and Ripley B.D. 1994. Modern Applied Statistics With SPlus. New York: Springer-Verlag.
Xie, Y. and Manski, C. F. 1989. “The Logit Model and Response-Based
Samples.” Sociological Methods and Research 17: 283-302.
30
Table 1: Cross Tabulation of the Variables Asked by Altruist (N=450)
Altruist
Asked
Yes
No
Total
Yes
237
122
359
No
10
81
91
Total
247
203
450
Table 2: Logistic Regression on the Variable Altruist (standard error in parentheses)
Model I
Model II
Model III
Model IV
Asked
-.-
2.790
(0.373)
2.840
(0.377)
3.392
(0.524)
Prosocial
orientation
0.321
0.409
0.196
(0.162)
(0.179)
Independent
variables
-.-
(0.139)
(Orientation*asked)
-.-
Constant
-.-
1.333
(0.126)
0.327
(0.151)
-.0.354
(0.154)
1.023
(0.410)
0.338
(0.152)
Fit statistics
Deviance
402
318
312
305
d.f.
390
390
389
388
R2 (Cox & Snell)
N
0.013
0.204
392
392
0.217
392
0.230
392
Table 3: Non-Parametric Tests for the Effect of Prosocial Orientation on Altruist
Effect
Npar d.f.
Npar Chisq
p-value (chi)
Were asked to help:
Spline (orientation)
3
5.384
0.142
Were not asked to help:
Spline (orientation)
3
5.140
0.365
31
Figure 1: The Smoothed Prosocial Orientation Effects (log-odds), as Derived
from the Generalised Additive Models
2
0
-2
-4
Spline (Prosocial Orientation)
4
Were not Asked
-2
-1
0
Prosocial Orientation
1
2
0
-2
-4
-6
Spline (Prosocial Orientation)
Were Asked
-3
-2
-1
Prosocial Orientation
0
1
Notes: The dashed lines indicate plus and minus two pointwise standard deviations.
32
Appendix A
The Prosocial Action Orientation Variable
In the APPBI survey respondents were asked to indicate, on a five-point
scale, the degree to which they agree with 42 personality questions that
were related to the present time (1=strongly agree—5=strongly disagree).
Using a factor analysis technique to these 42 personality items, Oliner and
Oliner (1988) have identified a Prosocial Action Orientation factor. The
values of the Prosocial Action Orientation variable in our analysis are the
products of the same items. That is, we selected into a factor analysis
procedure (principle component) the items that Oliner and Oliner have
already identified (see, Oliner and Oliner, 1988: 317, Table 7.16). This
analysis yielded three factors, the first of which explained about 26 per
cent of the variance. The remaining two factors, then, contributed about
nine additional per cent each. We then reapplied to these data a factor
analysis, but restricted the number of factors to one (this factor explained
about 25 per cent). Table A1 below presents the items that were included
in this analysis and the loading of each item on this factor.
Table A1: The personality items in the Prosocial Action Orientation factor
Q. No.
F14: 2
F14: 3
F14: 7
F14: 8
F14: 10
F14: 12
F14: 21
F14: 26
F14: 28
F14: 29
F14: 35
F14: 36
Personality item
Every person should give time for the good of the
country
I feel I am a person of worth at least on an equal
basis with others
I can not feel good if others around me feel sad
The feelings of people in books affect me
I get very upset when I see an animal in pain
It upsets me to see helpless people
I get angry when I see someone hurt
The words of a song can move me deeply
I feel very bad when I have failed to finish
something I promised I would do
I get very involved with my friends’ problems
If it is worth starting, it is worth finishing
Seeing people cry upsets me
N
Loading
492
.352
490
491
487
490
491
488
489
.301
.515
.464
.495
.476
.494
.374
486
489
484
481
.431
.463
.477
.452
The saved scores of the one factor procedure are then the values of the
Prosocial Action Orientation variable in our analysis.
33
Table A2: Descriptive Statistics for Prosocial Action Orientation
Variable Name
Altruist
Asked
Mean
s.e
N
No
0.239
0.959
89
Yes
-0.059
0.834
361
No
0.049
0.847
181
Yes
0.036
0.885
217
In Table A2 we present the means (and standard errors) of the prosocial
action orientation scores of selected sub populations in the APPBI data.
Note, however, that the higher the value is the weaker the prosocial action
orientation (see the five-point scale). Thus, in the final analysis we
multiplied the prosocial action orientation score by –1, so that higher
values will indicate a stronger prosocial action orientation.