Content uploaded by Norbert Schwarz
Author content
All content in this area was uploaded by Norbert Schwarz on May 09, 2016
Content may be subject to copyright.
www.ssoar.info
Constructing reality and its alternatives : an
inclusion/ exclusion model of assimilation and
contrast effects in social judgment
Schwarz, Norbert; Bless, Herbert
Veröffentlichungsversion / Published Version
Forschungsbericht / research report
Zur Verfügung gestellt in Kooperation mit / provided in cooperation with:
GESIS - Leibniz-Institut für Sozialwissenschaften
Empfohlene Zitierung / Suggested Citation:
Schwarz, Norbert ; Bless, Herbert ; Zentrum für Umfragen, Methoden und Analysen -ZUMA- (Ed.): Constructing reality
and its alternatives : an inclusion/ exclusion model of assimilation and contrast effects in social judgment. Mannheim,
1991 (ZUMA-Arbeitsbericht 1991/05). URN: http://nbn-resolving.de/urn:nbn:de:0168-ssoar-68883
Nutzungsbedingungen:
Dieser Text wird unter einer Deposit-Lizenz (Keine
Weiterverbreitung - keine Bearbeitung) zur Verfügung gestellt.
Gewährt wird ein nicht exklusives, nicht übertragbares,
persönliches und beschränktes Recht auf Nutzung dieses
Dokuments. Dieses Dokument ist ausschließlich für
den persönlichen, nicht-kommerziellen Gebrauch bestimmt.
Auf sämtlichen Kopien dieses Dokuments müssen alle
Urheberrechtshinweise und sonstigen Hinweise auf gesetzlichen
Schutz beibehalten werden. Sie dürfen dieses Dokument
nicht in irgendeiner Weise abändern, noch dürfen Sie
dieses Dokument für öffentliche oder kommerzielle Zwecke
vervielfältigen, öffentlich ausstellen, aufführen, vertreiben oder
anderweitig nutzen.
Mit der Verwendung dieses Dokuments erkennen Sie die
Nutzungsbedingungen an.
Terms of use:
This document is made available under Deposit Licence (No
Redistribution - no modifications). We grant a non-exclusive, non-
transferable, individual and limited right to using this document.
This document is solely intended for your personal, non-
commercial use. All of the copies of this documents must retain
all copyright information and other information regarding legal
protection. You are not allowed to alter this document in any
way, to copy it for public or commercial purposes, to exhibit the
document in public, to perform, distribute or otherwise use the
document in public.
By using this particular document, you accept the above-stated
conditions of use.
ZUMA-ARBEITSBERICHT No. 91/05
The enclosed reprint replaces ZUMA-Arbeitsbericht No. 91/05 by the same authors.
Schwarz, N., & Bless, H. (1992). Constructing reality and its alternatives: Assimilation and
contrast effects in social judgment. In L.L. Martin & A. Tesser (Eds.), The construction
of social judgment (pp. 217-245). Hillsdale, NJ: Erlbaum.
The Construction
of Social Judgments
Edited by
Leonard L. Martin
Abraham Tfesser
University of Georgia
LAWRENCE ERLBAUM ASSOCIATES, PUBLISHERS
1992 Hillsdale, New Jersey Hove and London
218
SCHWARZ AND BLESS
finding that evaluations of moderate stimuli are displaced away from ratings of
extreme stimuli, reflecting a contrast effect. Another group of models focused
on categorization processes (e.g., Herr, Sherman, & Fazio, 1983; Sherif & Hov-
land, 1961; Tajfel, 1959,1981; Tajfel & Wilkes, 1963; Turner, 1987). Accord
ing to these models, assimilation effects are likely to emerge when the target
stimulus and the context stimuli are assigned to the same category, whereas
contrast effects may emerge when they are assigned to different categories.
In this chapter we build on this previous theorizing as well as on recent
research by Barsalou (1987, 1989), Herr et al. (1983), Kahneman and Miller
(1986), and Martin and colleagues (Martin, 1986; Martin, Seta, & Crelia, 1990),
emphasizing the role of categorization processes in the operation of numerous
variables known to elicit assimilation and contrast effects. Clearly, categoriza
tion processes have been addressed in previous models, and much of what we
have to say is not new. Nevertheless, focusing explicitly on the interplay of cog
nitive accessibility and the categorization of accessible information provides a
heuristically fruitful integrative framework for the conceptualization of assimila
tion and contrast effects. Moreover, this focus generates numerous hypotheses
that have not been tested previously and it helps to specify the conditions un
der which each of different processes may result in assimilation or contrast
effects.
INCLUSION/EXCLUSION AND THE EMERGENCE
OF ASSIMILATION AND CONTRAST EFFECTS
Briefly, we assume that individuals who are asked to form a judgment about
some target stimulus first need to retrieve some cognitive representation of
it. In addition, they need to determine some standard of comparison to evaluate
the stimulus. As Kahneman and Miller (1986) suggested, this is frequently a
representation of some “ alternative” state of reality. Both the representation
of reality, that is, of the target stimulus, and of its alternatives are, in part, con
text dependent. Individuals do not retrieve all knowledge that may bear on the
stimulus, nor do they retrieve and use all knowledge that may potentially be
relevant to constructing its alternative. Rather, they rely on the subset of poten
tially relevant information that is most accessible at the time of judgment (see
Bodenhausen & Wyer, 1987; Higgins, 1989; Higgins & Bargh, 1987, for
reviews). Accordingly, their temporary representation of the target stimulus,
as well as their construction of a standard of comparison, includes information
that is chronically accessible, and hence context independent, as well as infor
mation that is only temporarily accessible due to contextual influences (see Bar
salou, 1989).
Whether the information that comes to mind results in assimilation or con
trast effects depends on how it is categorized. However, the specific operation
8. ASSIMILATION AND CONTRAST IN JUDGMENT
219
of the categorization process differs somewhat, depending on whether the in
formation that comes to mind is subordinate or superordinate to the target
category. Suppose, for example, that you are asked to evaluate a political party
and a specific politician comes to mind who is a member of that party. In that
case, the politician who comes to mind is subordinate to the target category.
Conversely, however, you may be asked to evaluate this particular politician
and his party membership may come to mind. In that case, information bearing
on his party in general would be superordinate to the target category. In both
cases, the impact of what comes to mind depends on categorization processes,
but the specifics of these categorization processes are somewhat different. Ac
cordingly, we discuss both cases in turn.
Subordinate Context Information and the Evaluation
of Superordinate Targets
If the information that comes to mind is subordinate to the target category, it will
result in an assimilation effect if it is included in the temporary representation
that individuals form of the target category. Thus, thinking of a well-respected
member of a political party is likely to result in more favorable evaluations of
the party as a whole. This simply reflects that the evaluation of a target is based
on the information that is included in the temporary representation that individuals
construct of it.
Empirically, however, assimilation effects due to the inclusion of a given piece
of information can only be observed if the valence of that information is more
extreme than the overall valence of the representation in general. Moreover,
the size of the emerging assimilation effect should depend on the amount of com
peting information: The more information is used in constructing a representa
tion, the smaller should be the impact of any additional piece of information that
is included in the representation.
Information that is excluded from the target category, on the other hand,
may result in contrast effects, although for different, and not mutually exclu
sive, reasons. First, suppose that individuals exclude some positively valenced
information from their representation of the target category. If so, they will base
their judgment on a representation that includes less positive information than
would otherwise be the case, resulting in less positive judgments. Empirically,
this type of contrast effect, which we call a subtraction effect, can only be ob
served if the valence of the excluded information is more extreme than the overall
valence of the representation. Such a subtraction effect does not require any
assumption about a change in the standard of comparison used. Moreover, the
size of subtraction effects should again depend on the amount of information
that is used in constructing a temporary representation: The more information
is included in the representation, the smaller should be the impact of subtract
ing a given piece of information.
220
SCHWARZ AND BLESS
The subtraction assumption discussed here parallels Martin's (1986; see also
Martin et al., 1990) reset assumption, where the emergence of contrast effects
is traced to the exclusion of valenced features from the representation that in
dividuals form of an ambiguously described target. However, we do not assume
that this is the only process that underlies the emergence of contrast effects,
nor would we like to restrict the operation of this process to the encoding of
ambiguous information that was investigated in the research by Martin and col
leagues.
As a second possibility, information that is excluded from the represen
tation of the target stimulus may come to mind when individuals construct
a relevant standard of comparison, and may be used for that purpose. If the
valence of this information is more extreme than the valence of other informa
tion used in constructing the standard, it results in a more extreme standard
of comparison, and hence in more pronounced contrast effects. We assume
that the representations of the target category and of the standard are mu
tually exclusive, and that the same piece of information cannot be used to
represent the target and the standard against which it is evaluated. Accord
ingly, we propose that the exclusion of information from the representation
of the target is a necessary prerequisite for its use in constructing a standard.
Moreover, extreme information that is excluded from the target category
may be used to anchor the response scale, as suggested by Ostrom and Up
shaw's (1968) perspective theory and related models, and may result in contrast
effects for that reason. As an extended debate in social judgment research indi
cates, changes in the standard of comparison used and changes in scale anchor
ing are difficult to distinguish empirically (see Eiser, 1990), and we do not attempt
to do so in this chapter.
Whereas the mere subtraction of information should only affect the evalua
tion of the target category from which this information is excluded, the use of
excluded information in constructing a standard of comparison should also af
fect the evaluation of related stimuli, to which the standard may be relevant.
For example, subtracting a highly respected politician from the representation
formed of his party should result in less favorable evaluations of this particular
party. Using this politician in constructing a relevant standard of comparison,
on the other hand, may also affect the evaluation of other parties, or specific
politicians, to which this standard may be applicable. Moreover, the size of
comparison-based contrast effects should not depend on the amount of infor
mation used in constructing the representation of the target, in contrast to the
subtraction effects discussed previously. Accordingly, assessments of the gener
alization of contrast effects across targets, and the presence or absence of set
size effects, allow for a differentiation of subtraction versus comparison/
anchoring-based contrast effects.
8. ASSIMILATION AND CONTRAST IN JUDGMENT
221
Superordinate Context Information and the Evaluation
of Subordinate Targets
Similarly, context information that bears on a superordinate category will result
in assimilation effects in the evaluation of a subordinate target if the target is
included in the superordinate category. Thus, recalling a politician's party mem
bership will result in more favorable evaluations when we like his party than
when we dislike it. This reflects the fact that inclusion of his party membership
in our cognitive representation of the target politician adds a positively evaluat
ed feature. Moreover, his inclusion in the superordinate category does allow
the derivation of other features, such as his standing on various issues, reflect
ing that ‘ ‘categorization of an object licenses inductive inferences about that ob
ject” (Smith, 1990, p. 35).
Note, however, that the inclusion process discussed here is more complex
than in the case of subordinate context information and superordinate targets:
We assume that inclusion of a subordinate target in a superordinate category
allows the derivation of features from our knowledge about the superordinate
category. These features, as well as the category membership, are then in
cluded in the temporary representation of the target, which serves as a basis
of judgment. Of course, this is a key assumption of many current theories of
stereotyping (see, for example, Allport, 1954; Brewer, 1988; Fiske & Neu-
berg, 1990; Hamilton, 1981), and we will not elaborate on this point beyond
noting its compatibility with the general approach offered here.
Excluding the target from the superordinate category, on the other hand,
again allows for the emergence of contrast effects. These contrast effects may
again reflect that a valenced feature is subtracted from the temporary represen
tation of the target (subtraction effect), or that information bearing on the su
perordinate category is used in constructing a relevant standard of comparison
or scale anchor
(comparison effect). As discussed previously, subtraction-based
contrast effects should be limited to the evaluation of the specific target, whereas
comparison-based contrast effects may generalize to other targets for whom
the standard of comparison may be relevant.
Summary
In summary, the key assumptions of the present model hold that the inclusion
of a given piece of information in the temporary representation of the target
category is a necessary prerequisite for the emergence of an assimilation ef
fect, whereas the exclusion of a given piece of information from that represen
tation is a necessary prerequisite for the emergence of contrast effects. Whether
222
SCHWARZ AND BLESS
the emerging contrast effect is limited to the evaluation of the target stimulus,
or generalizes to related stimuli, depends on whether the excluded information
is merely subtracted from the data base used to evaluate the target, or is used
in constructing a standard of comparison, or a scale anchor, which may be ap
plicable to related stimuli as well. In addition, we assume that the default opera
tion is to include easily accessible information in the representation of the target
category and that exclusion needs to be triggered by salient features of the task
or its context, an issue to which we will return later.
The general inclusion/exclusion approach offered here provides a heuristi-
cally fruitful framework for the conceptualization of many variables that are known
to moderate the emergence of assimilation and contrast effects. Before we
review some of these variables in more detail, however, we report on two studies
in which the inclusion or exclusion of context information was directly manipu
lated, using these studies to elaborate on our basic assumptions.
Direct Manipulations of Categorization:
The Varying Impact of a Politician On
Evaluations of His Party
Suppose that you are asked to provide a general evaluation of the politicians
of the Christian Democratic Party of the Federal Republic of Germany. To do
so, you may either retrieve a previously formed judgment from memory, or
you may form a judgment on the spot (Strack & Martin, 1987). If a previously
formed judgment can be retrieved, the current context is unlikely to exert much
influence. Hence, we will not address this special case. If you need to form a
judgment on the spot, you presumably need to retrieve some representation
of the politicians of the Christian Democratic Party. Assuming some basic familiar
ity with German politics, this representation is likely to include some chronical
ly accessible information, for example, that it is a conservative party, and that
Chancellor Kohl is one of its prominent members. In addition to such context-
independent information, the representation may include some information that
only comes to mind under specific circumstances, for example, because it was
needed to answer a preceding question (Schwarz & Strack, 1991a; Strack &
Martin, 1987). Under which conditions this context-dependent information results
in assimilation or contrast effects on your general evaluation of politicians of the
Christian Democratic Party is of key interest in the present chapter.
One of the most highly regarded members of the Christian Democratic Party
is Richard von Weizsäcker, who currently serves as president of the Federal
Republic of Germany. He has been a member of this party for several decades,
but the office of president requires that he no longer actively participate in party
politics. The president, as the representative figurehead of the Federal Repub
lic of Germany, is supposed to take a neutral stand on party issues, much as
8. ASSIMILATION AND CONTRAST IN JUDGMENT
223
the Queen in the United Kingdom. This rendered him particularly suitable for
the present experiment, in which we exploited this ambiguity of his status as
a party member (Schwarz & Bless, 1990, Experiment 1).
Specifically, we asked subjects a number of political knowledge questions.
In one condition, they were asked to recall the party of which “ Richard von
Weizsäcker has been a member for more than 20 years.” Answering “ Chris
tian Democratic Party” should make it more likely that the highly respected
Richard von Weizsäcker is included in the temporary representation that sub
jects form of politicians of the Christian Democratic Party when they are later
asked to evaluate this group. Accordingly, it was expected that they would evalu
ate the Christian Democratic Party more favorably when they were asked the
party membership question than when they were not.
The data supported this hypothesis, as shown in the first row of Table 8.1.
Specifically, including Richard von Weizsäcker in the category increased the
evaluation of Christian Democratic politicians as a group, relative to a condition
in which no question about Richard von Weizsäcker was asked.
In another condition of the same study, however, subjects were asked which
office Richard von Weizsäcker holds “that sets him aside from party politics.”
Answering this question should exclude Richard von Weizsäcker from the
category of Christian Democratic Party politicians. If so, it was expected to result
in lower evaluations of Christian Democratic politicians in general. This was again
the case, as shown in the first row of Table 8.1.
So far, these findings indicate that asking a preceding question increased the
cognitive accessibility of the information that was used to answer it. This, in
turn, increased the likelihood that this information came to mind when respon
dents were later asked another question to which it was relevant. How this easily
accessible information affected the judgment, however, depended on whether
it was included in the temporary representation that subjects formed of the tar
get category or not. If Richard von Weizsäcker was assigned to the target
category, that is, the Christian Democrats in the preceding example, he was
included in the data base considered in making the general judgment. This
TABLE 8.1
Evaluation of Political Parties as a Function of the Inclusion or Exclusion
of a Highly Respected Politician
Preceding Question About Richard von Weizsäcker
Target Party Membership
None
Presidency
Christian Democrats
6.5
5.2
3.4
Social Democrats
6.3
6.3
6.2
Note: n = 19 to 25 per condition. 1 = unfavorable; 11 = very favorable opinion about politi
cians of the respective party in general. Adapted from Schwarz and Bless (1990, Experment 1).
224 SCHWARZ AND BLESS
resulted in an assimilation effect, relative to a control group in which no ques
tion about Richard von Weizsäcker was asked.
If Richard von Weizsäcker was excluded from the target category, however,
a contrast effect emerged, again relative to a condition in which no question
about Richard von Weizsäcker was asked. This may reflect either of two process
es. On the one hand, Richard von Weizsäcker and his party membership may
have been chronically accessible for some subjects who were not asked a ques
tion about him. If so, the assimilation effect would reflect that the party mem
bership question increased the number of subjects who included Richard von
Weizsäcker in their temporary representation, whereas the presidency ques
tion decreased the number of subjects who did so. In that case, the obtained
contrast effect would reflect the subtraction of Richard von Weizsäcker from
the data base used. On the other hand, subjects may not only have excluded
Richard von Weizsäcker from their representation of politicians of the Christian
Democratic Party, but may also have used him in constructing a standard of
comparison, or a scale anchor, against which politicians of the Christian
Democratic Party in general were evaluated.
We can distinguish both possibilities by assessing the generalization of the
obtained contrast effect across different stimuli to which the standard of com
parison, or the scale anchor, may be relevant. If the obtained contrast effect
solely reflects the exclusion of Richard von Weizsäcker from the representa
tion of the target category politicians of the Christian Democratic Party, the
presidency question should only affect the evaluation of this party. If respon
dents used Richard von Weizsäcker in constructing a standard of comparison
or a relevant scale anchor, on the other hand, the obtained contrast effect should
generalize to the evaluation of politicians of other parties, such as the Social
Democratic Party, as well. In this case, we may expect contrast effects to
emerge in response to both questions about Richard von Weizsäcker, because
the presidency as well as the party membership question should bring this
respected politician to mind without allowing his inclusion in subjects' temporary
representation of the target category politicians of the Social Democratic Party,
of which he has never been a member.
To explore this possibility, other subjects of the aforementioned study
(Schwarz & Bless, 1990, Experiment 1) were exposed to the same questions
but were asked to provide a general evaluation of politicians of the Social
Democratic Party. As shown in the second row of Table 8.1, neither of the ques
tions about Richard von Weizsäcker affected subjects’ evaluations of politicians
of the Social Democratic Party. This suggests that the contrast effect obtained
on the evaluation of Christian Democratic politicians reflected a subtraction ef
fect, rather than a change in the standard of comparison, or scale anchor, used.
In summary, directly manipulating the categorization of Richard von
Weizsäcker by different knowledge questions provided clear support for the
emergence of assimilation effects as a function of his inclusion in respondents’
8. ASSIMILATION AND CONTRAST IN JUDGMENT
225
representations of the target category, and for the emergence of a subtraction-
based contrast effect as a function of his exclusion from the target category.
The Impact of Category Width:
Scandals and Trust in Politicians
If the emergence of assimilation and contrast effects is determined by subjects'
categorizations of highly accessible information, as the preceding experiment
suggests, all variables that influence the inclusion or exclusion of information
from a target category may be expected to moderate the impact of highly ac
cessible information. Whereas we manipulated subjects’ categorizations of con
text information in the previous study by means of direct questions, a particularly
relevant variable in social judgment research may be the width of the target
category. In principle, a given piece of information should be more likely to be
included in a category, the wider the respective category is, but more likely
to be excluded from the category, the narrower it is. The next study bears on
this assumption and extends the inclusion/exclusion logic by demonstrating
changes in the standard of comparison or scale anchor, as indicated by a gener
alization of contrast effects across different targets.
Suppose, for example, that subjects are induced to think about politicians
who were involved in a specific political scandal, and are subsequently asked
to evaluate the trustworthiness of politicians in general. According to the present
model, the politicians involved in the scandal are members of the superordinate
category politicians and are therefore likely to be included in subjects’ temporary
representations of that category. If so, subjects’ evaluations of the trustworthi
ness of politicians in general should decrease, reflecting an assimilation effect.
Suppose, however, that subjects are not asked to evaluate the trustworthi
ness of politicians in general, but the trustworthiness of a specific politician, Mr.
Joe Doe, who was not involved in the scandal. We may assume that in evaluat
ing a specific person, this person makes up a category by him or herself (cf.
Wyer & Srull, 1989). If so, the politicians who were involved in the scandal should
not be included in subjects' temporary representations of Joe Doe. Neverthe
less, the scandal-ridden politicians should be highly accessible in memory and
may come to mind when subjects are asked to evaluate Joe Doe. If so, they
may be used in constructing a standard against which Joe Doe is evaluated, or
they may be used to anchor the response scale. In either case, Joe Doe should
be evaluated as particularly trustworthy by comparison, reflecting a contrast
effect. Thus, the present model predicts that thinking about politicians who were
involved in a scandal may decrease judgments of the trustworthiness of politi
cians in general, but may increase judgments of the trustworthiness of specific
exemplars of the category, provided that they were not involved in the scandal.
To test this implication of the inclusion/exclusion model, we (Schwarz & Bless,
226
SCHWARZ AND BLESS
in press) asked subjects to recall the names of some politicians who were in
volved in a recent political scandal in West Germany, either before or after they
answered the dependent variables. The scandal used in this study was the so-
called "Barschel Scandal,” which bears some resemblance to the Watergate
scandal in the United States. All subjects were able to provide the names of
at least two participants. Subsequently, some subjects were asked to evaluate
the trustworthiness of politicians in the Federal Republic of Germany in gener
al. As shown in the first row of Table 8.2, thinking about the Barschel scandal
resulted in decreased judgments of the trustworthiness of German politicians
in general. This assimilation effect presumably reflects that subjects included
the politicians who were involved in the scandal in their representation of Ger
man politicians in general.
Other subjects, however, were asked to evaluate the trustworthiness of three
specific politicians, whom pretests had shown to be not particularly trustwor
thy to begin with, although they were not involved in the scandal under study.
As shown in the second row of Table 8.2, thinking about the Barschel scandal
increased judgments of trustworthiness of these specific politicians. This con
trast effect presumably reflects that subjects used the easily accessible politi
cians who were involved in the scandal in constructing a standard of comparison
or a relevant scale anchor.
Note that this contrast effect cannot be accounted for on the basis of a mere
subtraction process. The information that was primed by the scandal questions
was presumably never part of the subjects' representations of the specific poli
ticians they had to evaluate. Hence, the contrast effect obtained here presuma
bly reflects the use of the recalled politicians in constructing a relevant standard
of comparison or scale anchor. This information could only be used in constructing
the standard, however, when it was not perceived to bear on the respective
target category in the first place. As a result, we found that naming politicians
who were involved in a scandal resulted in assimilation effects on the evaluation
of a wide category that allowed the inclusion of these politicians. However, the
same priming task resulted in contrast effects on the evaluation of specific poli
ticians, reflecting that the primed politicians could not be included in the specific
TABLE 8.2
Evaluation of the Trustworthiness of Politicians in General and of
Three Exemplars as a Function of Thinking About a Scandal
Scandal Question
Target Not Asked
Asked
Politicians in General
5.0
3.4
Specific Exemplars
4.9
5.6
Note: n = 8 per condition. 1
= not at aD trustworthy; 11 =
very trustworthy. Adapted from
Schwarz and Bless (in press).
8. ASSIMILATION AND CONTRAST IN JUDGMENT
227
categories made up by those persons, and were hence available for the con
struction of a relevant standard or anchor.
These findings indicate that the same information may affect related judg
ments in opposite directions, depending on whether the respective target
category invites the inclusion or the exclusion of the information that comes
to mind. Accordingly, it comes as no surprise that political scandals are typically
accompanied by attempts to channel the public’s categorization of scandal-related
information (see Ebbighausen & Neckel, 1989, for discussions of scandal manage
ment). To the extent that individual politicians, or groups of politicians, can dis
sociate themselves from the scandal, they may actually benefit from the
misdemeanor of their peers, although the impact on the perception of the profes
sion as a whole is likely to be negative.
Subtraction Versus Comparison:
The Role of Salient Dimensions
So far, we have seen some evidence for the operation of subtraction as well
as comparison or anchoring processes under exclusion conditions. Under which
conditions, however, is each of these processes likely to operate? We propose
that the mere accessibility of an extreme stimulus is unlikely to elicit compari
son or anchoring processes, unless the stimulus brings the relevant dimension
of judgment to mind. Accordingly, thinking about some stimulus only influences
the evaluation of subsequent stimuli by means of comparison or anchoring
processes if the stimulus is linked to the dimension of judgment. If the stimulus
is thought about with regard to some other dimension, it is unlikely to be used
as a standard or scale anchor. With regard to the preceding studies, this sug
gests that thinking about a political scandal was likely to bring the dimension
of trustworthiness to mind. On the other hand, thinking about Richard von
Weizsacker’s party membership or office may have been less likely to bring
the evaluative dimension to mind that was relevant to subsequent judgments
of the Social Democrats.
In line with this assumption, we observed in one of our studies that compari
son or anchoring effects only emerged when highly accessible context-dependent
information was linked to the dimension of judgment (Schwarz, Miinkel, & Hipp
ier, 1990). Specifically, we asked subjects to rate how “typically German” a
number of different beverages are, namely wine, coffee, and milk. Before they
made this judgment, some subjects were asked to estimate the caloric content
of a glass of vodka, or of a glass of beer, respectively. Other subjects, however,
were asked to estimate how frequently Germans drink vodka or beer.
Both questions should increase the accessibility of vodka or beer in memory.
However, only the frequency-of-consumption question is related to the typical
ity dimension, whereas the caloric content question is not. If it is sufficient that
an extreme stimulus comes to mind, both questions should result in contrast
228
SCHWARZ AND BLESS
effects on subsequent typicality ratings. On the other hand, if the emergence
of comparison or anchoring effects requires that the extreme stimulus is linked
to the relevant judgmental dimension, contrast effects should only emerge when
subjects estimate the frequency of consumption, but not when they estimate
the caloric content.
Empirically, this was the case. Table 8.3 shows the mean ratings of the “ Ger-
manicness” of wine, coffee, and milk as a function of the preceding questions.
When subjects estimated the frequency of consumption, they rated all bever
ages as more typically German after thinking about vodka than after thinking
about beer. Estimating the caloric content of vodka or beer, on the other hand,
did not affect their ratings. Accordingly, we concluded that the emergence of
comparison or anchoring effects requires that the context-dependent informa
tion be linked to the relevant dimension of judgment. Otherwise, it may not
be considered when individuals construct a standard of comparison or select
a scale anchor.
This suggests that we may only see contrast effects that generalize across
various target categories when respondents think about the excluded informa
tion with regard to the respective dimension of judgment. If the excluded infor
mation is thought about with regard to some other dimension, it may still result
in contrast effects, but only by means of a subtraction process. Accordingly,
the contrast effects that emerge under this condition should be limited to the
evaluation of the category from which the information was excluded in the first
place, as was the case in the Weizsäcker study. Such a subtraction effect could
not be observed in the present study, however, because the extreme bever
ages were not part of the target categories to begin with. Most obviously, more
research is needed to test these conjectures.
VARIABLES THAT DETERMINE INCLUSION
AND EXCLUSION: CONJECTURES AND FINDINGS
If the emergence of assimilation and contrast effects is a function of categoriza
tion processes, then any variable that influences categorization can presumably
elicit assimilation or contrast effects. In the following sections we review the
TABLE 8.3
Contrast Effects as a Function of the Dimension Tapped by Preceding Questions
Preceding Question
Vodka
Context Stim ulus
Beer
Consumption
5.4 4.4
Caloric Content
4.4
4.5
Note: n = 25 to 27 per cell; 9 = "very typical.” The mean of ratings of three beverages (milk,
wine, and coffee) is given. Adapted from Schwarz, Munkel, and Hippier (1990).
8. ASSIMILATION AND CONTRAST IN JUDGMENT
229
operation of a number of different variables that have been shown to moderate
the emergence of assimilation and contrast effects, emphasizing their impact
on inclusion/exclusion processes.
Representativeness
One of the key variables that determines the inclusion or exclusion of informa
tion is the perceived representativeness of the information for the target
category. Information that is not representative for the target category is likely
to be excluded, and is therefore likely to result in contrast effects. A number
of diverse studies bear on this prediction and we review only a few of them.
Temporal Distance of E vents. Strack, Schwarz, and Gschneidinger
(1985, Experiment 1) investigated the emergence of assimilation and contrast
effects on judgments of current life-satisfaction as a function of the representa
tiveness of specific life-events for the target category my life now. (See also
Tversky & Griffin, 1991, for conceptual replications, and Schwarz & Strack,
1991b for a general discussion of life-satisfaction judgments.) In the Strack et
al. study, some subjects were asked to think about their present life and to write
down three events that were either particularly positive and pleasant or partic
ularly negative and unpleasant. This was done under the pretext of collecting
life-events for a life-event inventory, and the dependent variables, among them
happiness and satisfaction, were said to be being assessed in order to “find the
best response scales” for that instrument. As shown in Table 8.4, subjects who
had previously been induced to think about positive aspects of their present life
described themselves as happier and more satisfied with their life-as-a-whole
than subjects who had been induced to think about negative aspects. Presuma
bly, this assimilation effect reflects that subjects included the recent life-events
that they thought about in their temporary representation of the target category
my life now.
Other subjects, however, had to recall events that had occurred several years
ago. These events are no longer representative for one's life now, but bear
TABLE 8.4
Subjective Well-Being: The Impact of Valence of Event and Extendure
Valence of Event
Extendure
Positive
Negative
Present
8.9
7.1
Past 7.5
8.5
Note: Mean score of happiness and satisfaction questions, range is 1 to 11, with higher values
indicating reports of higher well-being. Adapted from Strack, Schwarz, and Gschneidinger (1985,
Experiment 1).
230
SCHWARZ AND BLESS
on a previous period of one’s life. Accordingly, they should be excluded from
the target category my life now, and may serve as standards of comparison,
resulting in contrast effects on current life-satisfaction. The data supported this
prediction. Subjects who thought about negative past events reported higher
current well-being than respondents who thought about positive past events,
reversing the previously obtained effect of recent events.
These findings are consistent with recent research in autobiographical memory
that suggests that life-events are organized in terms of life-time periods, often
referred to as extendures (see Cohen, 1989, for a review). If the recalled event
is included in the representation of the extendure that is to be evaluated, it results
in assimilation effects. If it is excluded from this extendure, it is likely to trigger
contrast effects. Accordingly, we may expect that future research into the con
struction of autobiographical extendures, and the variables that determine the
boundaries of extendures, and the variables that determine the boundaries of
extendures, will bear directly on the impact of previous life-events on judgments
of current well-being (see Clark & Collins, in press, for a related discussion).
Consistent with the assumption that the impact of life-events on judgments
of well-being depends on the boundaries used in constructing autobiographical
extendures, subsequent research by Strack, Schwarz, and Nebel (1987) demon
strated that it is not the temporal distance of the event per se that moderates
the use of accessible information about one’s life, but rather the subjective per
ception of whether the event one thinks about pertains to one’s current condi
tions of living or to a different extendure of one’s life. Specifically, students were
asked to describe either a positive or a negative event that they expected to
occur in “five years from now.” For half of the sample, a major role transition
was emphasized that would occur in the meantime, namely leaving university
and entering the job market. As major role transitions of this type are known
to mark the boundaries of autobiographical extendures (Cohen, 1989), this
manipulation should increase the probability that respondents assign the expected
event to a different phase of their life. Accordingly, they should be likely to use
the expected event in constructing a standard of comparison. The results sup
ported this reasoning. When the role transition was not emphasized, subjects
reported higher happiness and life-satisfaction when they had to describe posi
tive rather than negative expectations. When the role transition
was empha
sized, this pattern was reversed, and subjects reported higher well-being after
thinking about negative rather than positive future expectations. Again, these
findings suggest that easily accessible information elicits assimilation effects if
it is included in the temporary representation of the target category, but results
in contrast effects if it is excluded from that category.
Feature Overlap. In a well-known study, Herr, Sherman, and Fazio (1983)
asked subjects to rate ambiguous stimuli (e.g., a fictitious animal) in the con
text of moderate or extreme related stimuli. They observed assimilation effects
8. ASSIMILATION AND CONTRAST IN JUDGMENT
231
in the ratings of ambiguous stimuli when they were presented in the context
of moderate ones, but contrast effects when they were presented in the con
text of extreme ones. Later research by Herr (1986), using a social category,
replicated these findings. Herr (1986) concluded that “to the extent that a com
parison of features of the activated category and the target stimulus results in
matching or overlap, a judgment of category membership should occur” (p.
1107), eliciting an assimilation effect. On the other hand, if the overlap is in
sufficient, thus constituting an exclusion relationship, “the priming exemplars
serve as standards of comparison” (Herr, 1986, p. 1107), resulting in a con
trast effect.
In a related vein, Seta, Martin, and Capehart (1979) observed assimilation
effects in attractiveness ratings of two target persons, who showed differential
agreement with the subject's attitudes, when the targets shared a salient fea
ture (namely, their college major), but contrast effects when they did not, in
line with the predictions of Tajfel’s (1959) accentuation theory. They concluded
that the probability of obtaining an assimilation effect “is increased as the per
ceived commonality between two individuals is increased. Further, the proba
bility of obtaining a contrast effect is increased as the perceived commonality
between two individuals is decreased” (Seta et al., 1979, p. 406), reflecting
the impact of category membership (i.e., inclusion/exclusion) decisions.
Unitary Versus Composite Categories. Following Sherif and Hovland's
(1961) hypothesis that the emergence of assimilation and contrast effects de
pends on the distance between a behavior and its referent distribution, Hilton
and von Hippel (1990, Experiment 1) induced different expectations about the
distribution of pathological behaviors at different hospitals. In line with research
by Manis and colleagues (e.g., Manis & Paskewitz, 1984; Manis, Paskewitz,
& Cotler, 1986), they observed that ambiguous behaviors were assimilated to
the stereotype of the respective hospital when they were consistent with ex
pectations, and hence representative of the behaviors associated with patients
of that hospital. However the ambiguous behaviors were contrasted to the
stereotype when they were inconsistent with expectations. Thus, an assimila
tion effect emerged when the target behavior could be included in the superor
dinate hospital category, whereas a contrast effect emerged when the target
behavior was excluded from that category.
In a second experiment, Hilton and von Hippel (1990, Experiment 2) direct
ly manipulated subjects’ opportunities to recategorize a target behavior that was
inconsistent with expectations. Specifically, they attributed all behaviors either
to a random group of people whose names began with a letter in the same half
of the alphabet, to a family, or to an individual, assuming that subjects would
expect most consistency among the behaviors of an individual and least con
sistency among the behaviors shown by a random group of people. As expect
ed, ambiguous behaviors were assimilated to the induced expectations when
232
SCHWARZ AND BLESS
an individual was the alleged source of all behaviors, but were contrasted to
expectations when a random group was given as the alleged source. Apparent
ly, attributing all behaviors to the same individual ensured that the ambiguous
behavior was included in the category constituted by the unambiguous behaviors,
resulting in an assimilation effect. Attributing the behaviors to a diverse group
of people, on the other hand, allowed the exclusion of the ambiguous behaviors,
resulting in a contrast effect. As the authors note: "To the extent that recategori-
zation is a viable alternative, the pressure to assimilate unexpected behaviors
should diminish” (Hilton & von Hippel, 1990, p. 445).
In more general terms, it may be assumed that unitary categories (such as
specific individuals) are less likely to allow for the inclusion of discrepant infor
mation than composite categories (such as random groups of individuals), reflect
ing assumptions about the higher degree of variation that may be observed in
the latter case. If so, judgments of unitary target categories should be more
likely to show contrast effects, whereas composite categories should be more
likely to show assimilation effects.
In summary, the studies reviewed in this section indicate that the perceived
representativeness of a given piece of information for the respective target
category determines its inclusion in, or exclusion from, that category, and hence
the emergence of assimilation or contrast effects.
Category Width
Closely related to the impact of perceived representativeness is the issue of
category width. The wider a category is, the more likely it becomes that a given
piece of information may be included. The study on the impact of political scan
dals on judgments of trustworthiness (Schwarz & Bless, 1990), reviewed previ
ously, supported this assumption. For the domain of person perception, this
suggests, for example, that information about a specific group member is likely
to result in assimilation effects on the evaluation of the group in general, reflecting
the inclusion relationship constituted by group membership. Such an assimila
tion effect should not be obtained, on the other hand, if the individual member
is so distinct from the group that he or she is excluded from the category, or
if the individual member is assigned to a different category (see Rothbart &
John, 1985), as discussed previously in the context of the Herr et al. (1983)
and Hilton and von Hippel (1990) studies.
On the other hand, thinking about an individual should be likely to result in
contrast effects on the evaluation of other individuals, reflecting that an individual
person is likely to make up a category by him- or herself, constituting an exclu
sion relationship. Assimilation effects would only be predicted if some higher-
order category is identified to which both individuals can be assigned, and if this
higher-order category serves as a basis of judgment (see Fiske & Neuberg,
1990).
8. ASSIMILATION AND CONTRAST IN JUDGMENT 233
According to the present model, any variable that influences category width
should also influence the emergence of assimilation and contrast effects. Whereas
the preceding examples reflected the nature of the respective category itself,
category width may also be influenced by individual difference variables, such
as the degree of differentiation at which the respective content domain is
represented in the judge’s knowledge system. Theoretically, more differentiated
knowledge systems are composed of a larger set of more specific categories
than less differentiated knowledge systems. For example, Rosch, Mervis, Gray,
Johnson, and Boes-Braem (1976) observed that experts identified objects at
a lower level of abstraction than novices, reflecting a shift in the level of basic
categories. Whereas a novice may, for example, identify all airplanes as air
planes, an airplane mechanic in their study distinguished between many types
of different airplanes, reflecting the use of more, and more narrowly defined,
categories in the domain of his expertise. The use of more specific categories,
however, implies that a given piece of information may only be included in the
representation of one specific category, but may be excluded from many others.
Accordingly, we may expect contrast effects to be more likely to emerge in
judgments made by experts than in judgments made by novices.
Another variable that is likely to influence category width is an individual’s
affective state at the time of judgment. In several studies, individuals in an elat
ed mood were found to use wider categories than individuals in a depressed
mood (see Isen, 1987; Schwarz, 1990, for reviews). For example, items that
were not generally considered good exemplars of a category (e.g., cane as a
member of the category clothing) were more likely to be assigned to that category
by individuals in an elated rather than a nonmanipulated mood (Isen & Daub-
man, 1984). This suggests that the emergence of assimilation effects may be
more likely under elated moods, whereas the emergence of contrast effects may
be more likely under depressed moods. To our knowledge, data bearing on the
impact of moods and expert status are not yet available.
Presentation and Judgment Order
So far, we have considered variables that are inherent to the presented infor
mation, such as the primed information's representativeness for the target
category or the perceived inclusiveness of the target category itself. However,
the categorization of stimuli may also be affected by more fortuitous aspects
of the experimental procedures used, such as the order in which stimuli are
presented or judgments are assessed.
For example, Wedell, Parducci, and Geiselman (1987) asked subjects to rate
the attractiveness of faces that were either presented successively or in pairs.
When the faces were presented successively, the same face was rated more
favorably when presented in the context of less attractive faces, reflecting a
contrast effect. When the faces were presented in pairs, however, the same
234 SCHWARZ AND BLESS
face was rated less favorably when presented simultaneously with a less attrac
tive