Professional Documents
Culture Documents
To cite this article: Anna Baumert , Constanze Beierlein , Manfred Schmitt , Christoph J. Kemper , Anastassiya Kovaleva ,
Stefan Liebig & Beatrice Rammstedt (2014) Measuring Four Perspectives of Justice Sensitivity With Two Items Each, Journal of
Personality Assessment, 96:3, 380-390, DOI: 10.1080/00223891.2013.836526
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
Journal of Personality Assessment, 96(3), 380–390, 2014
Copyright C Taylor & Francis Group, LLC
ISSN: 0022-3891 print / 1532-7752 online
DOI: 10.1080/00223891.2013.836526
People differ systematically in their vulnerability to injustice. We present two-item scales for the efficient measurement of justice sensitivity
from 4 perspectives (victim, observer, beneficiary, perpetrator). In Study 1 using a quota-based sample of German adults, a latent state–trait
analysis revealed the factorial validity and high reliabilities of the scales. In Study 2 employing a large random sample, we tested for measurement
invariance of the items within the context of our short 2-item scales compared to the original 10-item scales. Multigroup confirmatory factor analyses
confirmed that the validity of the indicators and the internal structure of the assessed constructs did not change across item contexts. In both studies,
correlations with personality dimensions and life satisfaction provide evidence for the validity of our scales. With the presented instrument, future
research can extend scientific knowledge regarding the role of individual differences in reactions to injustice for the explanation of well-being and
physical health.
Feelings of injustice have a great impact on psychological well- scales have been developed to assess each justice sensitivity per-
being and the physical health of individuals. Perceiving one- spective and evidence for their reliability and validity has been
self to be underrewarded can trigger strong negative emotions reported (Schmitt, Baumert, Gollwitzer, & Maes, 2010; Schmitt
and feelings of distress (e.g., Rousseau, Salek, Aubé, & Morin, et al., 2005). To maximize the convergence of the items, the
2009), impair coping, and foster unhealthy behaviors (Green- strength of specific emotional reactions to injustice (i.e., anger,
berg, 2010). This perception can act as a substantial risk factor outrage, and guilt depending on the perspective), intrusiveness
for developing symptoms of insomnia, burnout, and depression of thoughts, and general feelings of strain following experiences
(Greenberg, 2006; Siegrist, 2002). By contrast, feeling fairly and of injustice have been employed as indicators of each of the four
respectfully treated can prevent distress and has been shown to perspectives. The scales’ four-factorial structure has been sup-
be systematically related to well-being and health (e.g., Elo- ported by confirmatory factor analyses (Schmitt et al., 2010).
vainio, Kivimäki, Vahtera, Keltikangas-Järvinen, & Virtanen, The factors share common variance that is assumed to reflect
2003; Greenberg, 2004; Tepper, 2001). Importantly, persons a general concern for justice. They also show distinct patterns
have been found to differ in how readily they perceive situ- of correlations with external criteria. For example, beneficiary
ations to be unjust and how strongly they react to subjective sensitivity, perpetrator sensitivity, and observer sensitivity are
injustice—cognitively, emotionally, and behaviorally (Lovas & positively related to modesty and tender-mindedness as facets
Wolt, 2002; Schmitt, Neumann, & Montada, 1995). These in- of agreeableness (Schmitt et al., 2010) and predict prosocial
dividual differences in justice sensitivity are consistent across tendencies such as solidarity with disadvantaged others (Goll-
situations and stable over time (Schmitt, Gollwitzer, Maes, & witzer, Schmitt, Schalke, Maes, & Baer, 2005) or bystander
Arbach, 2005). In other words, people differ systematically in intervention against norm violations (Lotz, Baumert, Schlösser,
their vulnerability to the adverse consequences of injustice. Gresser, & Fetchenhauer, 2011). By contrast, victim sensitivity
However, the quality of reactions to injustice depends on the was found to be negatively related to facets of agreeableness and
perspective that a person adopts in an unjust situation. Individu- moderately positively to neuroticism (Schmitt et al., 2010) and
als react with distinct emotions and behavioral tendencies if they negative interpersonal feelings, such as jealousy, vengeance, and
perceive themselves to be potential victims of injustice, passive paranoia (Schmitt et al., 2005). Furthermore, victim sensitivity
beneficiaries, active perpetrators, or neutral bystanders (e.g., predicted rather antisocial tendencies, namely a reluctance to
Mikula, Petri, & Tanzer, 1990). Accordingly, justice sensitivity cooperate and invest in common goods (Gollwitzer, Rothmund,
has been differentiated into victim sensitivity, beneficiary sensi- Pfeiffer, & Ensenbach, 2009; Rothmund, Gollwitzer, & Klimmt,
tivity, perpetrator sensitivity, and observer sensitivity. Ten-item 2011) and, thus, appears to reflect a concern for justice for the
self rather than for others, and the fear of being exploited by
interaction partners (Gollwitzer et al., 2005).
Received August 27, 2012; Revised May 23, 2013.
Highlighting the usefulness of this construct, research has
Address correspondence to Anna Baumert, Department of Psychology, Uni- shown that the justice sensitivity perspectives cannot be re-
versity of Koblenz-Landau, Fortstrasse 7, 76829 Landau, Germany; Email: duced to general personality factors (Schmitt et al., 2005) or
baumert@uni-landau.de to a combination of personality facets (Schmitt et al., 2010).
380
JUSTICE SENSITIVITY SHORT SCALES 381
Additionally, there is consistent evidence for the predictive for the future comparison of results obtained with different test
value of justice sensitivity for reactions to injustice above forms.
and beyond alternative constructs such as trait anger or self-
assertiveness (e.g., Mohiyeddini & Schmitt, 1997). The first STUDY 1: RELIABILITY AND VALIDITY OF THE JUSTICE
empirical studies conducted in this area have shown the impor- SENSITIVITY SHORT SCALES
tance of justice sensitivity for explaining health-related justice Method
effects. For example, Schmitt and Dörfel (1999) showed that Sample. Participants were recruited in 2010 by the certified
the psychosomatic well-being of people high in victim sen- poll agency BIK MARPLAN based on quotas representing pop-
sitivity was more strongly impaired by perceived unfairness ulation distributions of age, gender, education, and residency in
at the workplace than the well-being of people low in victim the 16 federal states of Germany. At Time 1, 536 persons partic-
sensitivity. Hessler, Pretsch, Hillert, and Schmitt (2012) found ipated (52.5% women). Ages ranged between 18 and 88 years
that teachers who suffer from burnout symptoms score signif- (M = 47.22, SD = 15.15). Regarding education, 23.6% indi-
icantly higher on victim and observer sensitivity than healthy cated that they had graduated from high school (in Germany
teachers. called Abitur, with a minimum of 12 years of school atten-
Beyond these findings, scientific knowledge on justice sensi- dance), 30.4% from intermediate secondary school (in Germany
tivity as a risk factor for psychological well-being and physical called Realschule, with a minimum of 10 years of school atten-
health is still scarce. This is at least partially due to the fact dance), and 44.7% from secondary school (in Germany called
that research on justice sensitivity as a personal determinant Hauptschule, with a minimum of 9 years of school attendance).
Downloaded by [University of Connecticut] at 21:16 11 October 2014
of reactions to injustice has been conducted predominantly in At Time 2 (approximately 6 weeks later), 338 persons partic-
the laboratory or has relied on convenience samples and mostly ipated (i.e., a dropout rate of 37%). People who participated at
on student samples. Therefore, it is highly desirable for future both measurement occasions did not differ from those who par-
research to include justice sensitivity scales in large-scale sur- ticipated only at Time 1 in age, t(534) = 1.02, p = .31, d = .09;
veys of the general population as well as in clinical studies. In gender, χ 2(1) = 0.07, p = .79, d = .02, or level of education,
both domains, there is a strong demand for very short measures χ 2(3) = 1.25, p = .74, d = .10.
due to time and space limitations in surveys and screenings.
Accounting for this demand, ultrashort measures have been de- Material. For all measures, we aggregated the items by cal-
veloped, for example, for the Big Five with two items per factor culating individual mean values. Note that response options var-
(Rammstedt & John, 2007), or for social phobia with six items ied across instruments because scale ranges were employed as
(Fergus, Valentiner, McGrath, Gier-Lonsway, & Kim, 2012). originally published to secure comparability of our results with
These measures have demonstrated good psychometric charac- those of previous studies using the respective instruments. How-
teristics despite their brevity. ever, to minimize inconsistencies, all response scales started
Accordingly, in this article, our first goal was to develop with 1 (disagree strongly) at the lower end.
ultrashort scales for the economic assessment of justice sensi-
tivity without substantial losses in reliability and validity. As Justice Sensitivity Short Scales: Two items were selected
a second goal, we determined the item stability of the two se- from each of the original 10-item scales based on the following
lected items within the context of the short scales compared criteria. To achieve maximum content validity, we selected items
to the context of the original scales. In future research on jus- that covered two of the three theoretical indicators employed in
tice sensitivity, the question could well arise as to whether re- the original scales: the strength of emotional reactions (i.e.,
sults obtained with our short scales can be directly compared anger, outrage, or guilt, depending on the perspective of justice
to results obtained with the original scales. For example, re- sensitivity) and subjective strain felt in response to injustice. As
searchers employing German samples of unknown representa- a further criterion, we chose items that promised to provide the
tiveness might want to compare their results to the normative expected four-factor structure of victim, observer, beneficiary,
data provided by Schmitt et al. (2010) for the original scales. and perpetrator sensitivity. Based on exploratory factor analyses
However, if these researchers use the two-item scales presented of data from Schmitt et al. (2005) and Schmitt et al. (2010), we
here, it has to be established whether their mean levels can in- selected items that loaded clearly on the expected factor with
deed be compared to the mean level of these items when used minimal secondary loadings on other factors. Note that the third
in the context of the original scales. Additionally, researchers indicator (intrusiveness of thoughts) is not represented by the
can conduct different studies concerning one research ques- short scales because the respective items met this latter criterion
tion with either the short scales or the original scales. Direct less optimally than the items of the other two indicators.
comparison of the results is viable only if evidence can be The final set of items can be found in the Appendix. In our
provided that the two items function similarly within different study, items were presented in German. The Appendix presents
scale contexts. This is important because social psychological an English-language equivalent. These items were taken from
research has revealed that various aspects of the measurement the English version of the original justice sensitivity scales
context can affect item characteristics (Knowles & Condon, (Schmitt et al., 2010), which is the result of the translation
2000; Schwarz, 1999). The placement of an item within a scale by a native German speaker fluent in English, and of the inde-
has been shown to determine the item’s correlation with the pendent discussion with two native English speakers, experts in
total score and the variability and mean level of answers to the domain of justice research.
this item (e.g., Knowles, 1988; Knowles, Coker, Scott, Cook, Like the original scales, each set of items representing one
& Neville, 1996; Ostrom, Betz, & Skowronski, 1992; Stein- justice sensitivity perspective was introduced by a short instruc-
berg, 1994). Thus, for the creation of new forms of tests, item tion (see Schmitt et al., 2010), for example, “First, we will
stability has to be established empirically to provide a basis look at situations to the advantage of others and to your own
382 BAUMERT ET AL.
disadvantage” (victim perspective) or, “Finally, we will look at Procedure. German citizens aged 18 and older were ap-
situations in which you treat someone else unfairly, discriminate proached on the streets of different German towns and invited to
against someone, or exploit someone” (perpetrator perspective). participate in a survey on various psychological topics. Among
Response options ranged from 1 (totally disagree) to 6 (totally those who agreed to participate at two times of measurement
agree). that were 6 weeks apart and who provided their contact details
to be contacted again at the second time of measurement, the
Personality: As very short scales were employed in this sample was drawn based on population quotas (see earlier). Re-
study to capture the relevant personality dimensions, high in- sponses at both measurement times were recorded anonymously
ternal consistencies are not expected. As a better estimator of and matched by means of an anonymous personalized code. Par-
reliability, we provide retest correlations (rtt ) across the 6-week ticipants were rewarded 8 Euro for their repeated participation.
interval in our study. Our measures of interest were administered as part of a larger
The Big Five Inventory BFI–10 (Rammstedt & John, 2007) questionnaire either in computer-assisted personal interviews
was used to assess extraversion (α = .63, rtt = .59), agreeable- (CAPIs; 76% of participants at Time 1; 67% of participants
ness (α = .53, rtt = .50), conscientiousness (α = .58, rtt = .59), at Time 2) or in a self-administered paper-and-pencil version.
neuroticism (α = .42, rtt = .49), and openness (α = .30, rtt = .62) CAPIs were conducted by trained interviewers who helped to fill
with two items per dimension. Response options ranged from out the questionnaire on a computer. Self-administered paper-
1 (disagree strongly) to 5 (agree strongly). Despite the brevity and-pencil questionnaires were handed out to the participant,
of their scales and, consequently, low internal consistencies, filled out individually, and returned to the interviewer. The mea-
Downloaded by [University of Connecticut] at 21:16 11 October 2014
Rammstedt and John (2007) obtained satisfactory retest relia- sures were presented in the following order: BFI–10, general
bilities as well as evidence for structural validity and convergent self-efficacy, interpersonal trust, life satisfaction, victim, ob-
validity with peer ratings and other Big Five measures. server, beneficiary, and perpetrator sensitivity.
Furthermore, interpersonal trust was measured with three
items taken from the German Socio Economic Panel (GSOEP- Results
trust: “In general, you can trust people,” “Nowadays, you can’t Bivariate correlations among justice sensitivity scales.
rely on anybody” recoded, and “When dealing with strangers it Table 1 displays means, standard deviations, and correlations
is better to be cautious before trusting them” recoded; α = .59, of the justice sensitivity scales at Time 1 and Time 2. The retest
rtt = .57; Naef & Schupp, 2009). The authors provided evidence correlations show a moderate stability of the scales (between
for the satisfactory retest reliability as well as for the construct .44 and .56).
and criterion validity of their measure. Response options ranged
from 1 (disagree strongly/not trust at all) to 4 (agree strongly/a Simultaneous latent state–trait analysis. A simultaneous
lot of trust). latent state–trait analysis (LSTA; Steyer, Ferring, & Schmitt,
General self-efficacy was measured with 10 items (e.g., “I can 1992; Steyer, Schmitt, & Eid, 1999) of our four justice sensitiv-
always manage to solve difficult problems if I try hard enough”; ity short scales provides an error-free estimator of the scales’ re-
α = .92, rtt = .49; Schwarzer & Jerusalem, 1995) that have liabilities. Moreover, in LSTA, three sources of systematic vari-
been successfully employed in prior research, for example on ance in the manifest variables can be distinguished. First, trait
personal determinants of job stress and burnout among teachers consistency (stability) is defined as the proportion of variance
(Schwarzer & Hallum, 2008). Further studies have provided explained by variance in the stable latent trait. Second, occasion
evidence for its validity (e.g., Scholz, Gutiérrez-Doña, Sud, & specificity is the proportion of variance explained by systematic
Schwarzer, 2002). Response options ranged from 1 (disagree unstable variance at one measurement occasion. Third, method
strongly) to 4 (agree strongly). specificity is defined as the proportion of variance explained by
Internal locus of control and external locus of control were nonequivalent indicators of a construct (Schmitt et al., 2005).
assessed by means of three-item short scales developed specif- The LST model for our four short scales is depicted in Fig-
ically for use in population surveys (Jakoby & Jacob, 1999; ure 1. As explained by Eid, Schneider, and Schwenkmezger
internal: e.g., “I find it best to make decisions myself, rather (1999) and confirmed by simulation studies (Geiser & Lockhart,
than to rely on fate”; α = .76, rtt = .65; external: e.g., “I often
have the feeling that I have little influence over what happens to TABLE 1.—Means, standard deviations, and interscale and retest correlations of
me”; α = .61, rtt = .53).1 The authors reported satisfactory evi- justice sensitivity short scales in Study 1.
dence for the validities of both scales. Response options ranged
from 1 (disagree strongly) to 5 (agree strongly). 1 2 3 4 5 6 7 8
1 JS-Victim (T1)
Life Satisfaction: We employed the German version of the 2 JS-Observer (T1) .45∗∗
3 JS-Beneficiary (T1) .26∗∗ .46∗∗
Satisfaction with Life scale (Diener, Emmons, Larsen, & Griffin, 4 JS-Perpetrator (T1) .22∗∗ .45∗∗ .72∗∗
1985), which consists of five items (e.g., “In most ways my life 5 JS-Victim (T2) .56∗∗ .14∗ .18∗∗ .14∗
.25∗∗ .44∗∗ .30∗∗ .24∗∗ .40∗∗
is close to my ideal”; α = .89, rtt = .66). The authors reported 6
7
JS-Observer (T2)
JS-Beneficiary (T2) .24∗∗ .38∗∗ .54∗∗ .40∗∗ .26∗∗ .47∗∗
high internal consistency of the scale. Pavot and Diener (1993) 8 JS-Perpetrator (T2) .24∗∗ .40∗∗ .40∗∗ .47∗∗ .30∗∗ .54∗∗ .69∗∗
provided evidence for the discriminant and convergent validity M 3.26 3.36 2.15 2.36 3.17 3.07 2.07 2.31
SD 1.51 1.32 1.16 1.30 1.50 1.29 1.08 1.26
of the scale. Response options ranged from 1 (disagree strongly)
to 7 (agree strongly). Note. N = 539 (Time 1). N = 338 (Time 2). JS-Victim, JS-Observer, JS-Beneficiary,
JS-Perpetrator = justice sensitivity from the victim, observer, beneficiary, and perpetrator
perspectives. The mean reported is the arithmetic mean of individual scale means. Response
options ranged from 1 (totally disagree) to 6 (totally agree).
∗
1Translation of the original German scales provided by us. p < .05. ∗ ∗ p < .01.
JUSTICE SENSITIVITY SHORT SCALES 383
JS- JS-
JS-Victim JS-Observer Beneficiary Perpetrator
TABLE 3.—Means and standard deviations of personality and life satisfaction tim(s) of the unfairness, correlated highest. By contrast, both
and correlations with justice sensitivity short scales in Study 1. correlated only lowly with victim sensitivity, reflecting the fact
JS- JS- JS- JS- that outcomes of victims, on the one hand, and beneficiaries or
M SD Victim Observer Beneficiary Perpetrator perpetrators, on the other hand, are negatively interdependent.
Extraversion 3.50 0.90 –.14∗∗ –.15∗∗ –.14∗∗ –.07
Second, correlations of the justice sensitivity short scales
Agreeableness 2.46 0.81 –.18∗∗ .10∗ .13∗∗ .15∗∗ and the Big Five factors were low to medium, corroborating the
Conscientiousness 3.46 0.71 –.16∗∗ .00 –.06 –.02 notion that justice sensitivity is not independent from, but cannot
Neuroticism 4.07 0.71 .20∗∗ .12∗∗ .12∗∗ .06
Openness 3.36 0.72 –.19∗∗ .04 –.05 .08
be reduced to global factors of personality. Third, our results are
Interpersonal trust 2.95 0.66 –.25∗∗ –.05 –.04 .02 highly similar to those reported by Schmitt et al. (2005) for
General self-efficacy 3.14 0.49 –.24∗∗ .01 –.12∗∗ –.05 the original victim, observer, and beneficiary scales with the
Internal locus of control 2.62 0.73 –.15∗∗ .01 –.06∗ .03
External locus of control 3.76 0.73 .18∗∗ .05 .11∗∗ –.08∗ Big Five factors measured with the BFI and NEO–FFI. Our
Life satisfaction 3.23 0.83 –.23∗∗ –.07∗ –.12∗∗ .06 results are also largely consistent with those found by Schmitt
et al. (2010) for the original victim, observer, beneficiary, and
Note. N = 539. JS-Victim, JS-Observer, JS-Beneficiary, JS-Perpetrator = justice sen-
sitivity from the victim, observer, beneficiary, and perpetrator perspectives. The mean perpetrator scales and the Big Five facets measured with the
reported is the arithmetic mean of individual scale means. Response options for extraver- Revised NEO Personality Inventory (NEO PI–R). Fourth, the
sion, agreeableness, conscientiousness, neuroticism, openness, internal and external locus correlations found with further personality dimensions and life
of control ranging from 1 (disagree strongly) to 5 (agree strongly). Response options for
interpersonal trust and general self-efficacy from 1 (disagree strongly/not trust at all) to satisfaction are consistent with theoretical assumptions and thus
4 (agree strongly/a lot of trust). Response options for life satisfaction from 1 (disagree support the validity of our short scales.
Downloaded by [University of Connecticut] at 21:16 11 October 2014
satisfaction with social life in particular should be reduced. For main assessment, a pretest is conducted to select instruments
observer sensitivity and beneficiary sensitivity, we also found for the introduction of new topics into the SOEP.
negative correlations with life satisfaction. It seems plausible Ages ranged from 16 to 93 years (M = 53.95, SD = 18.63).
that for justice-sensitive persons, perceiving high degrees of in- Regarding marital status, 51.2% of the participants reported
justice in the world has the potential of reducing satisfaction that they were married or in a stable partnership; 24.1% were di-
with life because they experience injustice as extremely aver- vorced, permanently separated from their former partner, or wid-
sive also when they see themselves as beneficiaries or neutral owed; and 23.9% were single. Regarding education, 23.8% in-
observers of injustice. dicated that they had graduated from high school (Abitur), 30%
Interestingly, perpetrator sensitivity was unrelated with life from intermediate secondary school (Realschule), and 34.2%
satisfaction. Perpetrator-sensitive persons are assumed to avoid from secondary school (Hauptschule). The remaining partici-
committing injustices, and can be expected to prevent their own pants indicated that they had no formal education, were still
feelings of guilt and, thus, threats for life satisfaction. Impor- attending school, or had received formal education in a country
tantly, however, specific situational constellations can be as- other than Germany.
sumed where perpetrator-sensitive persons do feel guilty about This sample did not differ from the representative sample
their own wrongdoings so that their life satisfaction suffers dis- recruited by Schmitt et al. (2010; N = 2,510) regarding gen-
proportionally. For example, when faced with unemployment der, χ 2(1) = 0.60, p = .44, d = .03; or marital status, χ 2(2) =
and forced to rely on welfare, perpetrator-sensitive persons can 2.04, p = .36, d = .05. However, the samples differed in mean
be assumed to have lower life satisfaction than less perpetrator- age (M Schmitt = 47.97, SDSchmitt = 17.79), t = –8.96, p < .01,
Downloaded by [University of Connecticut] at 21:16 11 October 2014
sensitive persons because they feel guilty about not contributing d = –.33, and in educational level, χ 2(2) = 30.35, p < .01, d
to the workforce. This situation-specific effect of perpetrator = .19. In Schmitt et al.’s (2010) sample, relatively more partic-
sensitivity among unemployed persons has been recently sup- ipants indicated that they had graduated from secondary school
ported by evidence obtained with the ultrashort scales presented (Hauptschule; 42.5%) and intermediate secondary school (Re-
in this article (Stavrova, Schlösser, & Baumert, in press). In this alschule; 34.5%), and relatively fewer had graduated from high
research, for employed persons, like in this study, perpetrator school (Abitur; 17.5%).
sensitivity and life satisfaction were unrelated.
Taken together, the pattern of results found in Study 1 supports Material. Like in Study 1, for all measures, we aggregated
the construct validity of the presented two-item scales for victim, the items by calculating individual mean values. Items and in-
observer, beneficiary, and perpetrator sensitivity, and provides structions were employed in German and English translations
evidence for the differential relevance of the constructs for life provided here were taken from the original literature.
satisfaction. Our two-item short scales were employed to measure victim
sensitivity (α = .77), observer sensitivity (α = .77), beneficiary
sensitivity (α = .82), and perpetrator sensitivity (α = .85). Re-
STUDY 2: ITEM STABILITY OF THE SHORT AND LONG sponse options ranged from 0 (totally disagree) to 5 (totally
SCALES agree). Note that we accepted a change in response scale from
In Study 2, we further investigated the construct validity of Study 1 to Study 2 to maximize the comparability of the results
our short scales by testing the correlations of the justice sensi- of the short scales with results obtained with the original justice
tivity perspectives with the individual inclination toward social sensitivity scales in the study of Schmitt et al. (2010).
comparison (Gibbons & Buunk, 1999), as well as toward pos- Furthermore, the G-SOEP pretest sample contained measures
itive and negative reciprocity (Perugini, Gallucci, Presaghi, & for social comparison and reciprocity. The individual inclination
Ercolani, 2003). to engage in social comparison was measured with six items (“I
As a second goal of our research, we examined the item sta- always pay a lot of attention to how I do things compared with
bility across scale contexts. Specifically, we wanted to know how others do things,” “I often try to find out what others think
whether results obtained with the two items of our short scales who face similar problems as I face,” “I often compare how I am
could be directly compared to results obtained with these items doing socially (e.g., social skills, popularity) with other people,”
in the context of the original scales. For this purpose, we em- “I always like to know what others in a similar situation would
ployed data for the short scales in a representative German sam- do,” “If I want to learn more about something, I try to find out
ple and compared those data to the representative data obtained what others think about it,” and “I am not the type of person who
with the original scales by Schmitt et al. (2010). Using multi- compares often with others” reverse coded; α = .77). Response
group confirmatory factor analyses (MGCFA), we first tested options ranged from 1 (totally disagree) to 7 (totally agree). This
(a) for measurement invariance across scale contexts, and (b) scale is a German short version of the questionnaire developed
for differences in latent parameters (factor variances, covari- by Gibbons and Buunk (1999). As Schneider and Schupp (2011)
ances, and means). showed, the items represent two correlated facets (comparisons
of abilities and comparisons of opinions, r = .50) and are highly
reliable. Known group comparisons and correlations with other
Method psychological variables supported the validity of the facets. As
Sample. As part of the regular pretest module of the Ger- in our study, we found a single factor to explain item variance
man Socio-Economic Panel (G-SOEP), a representative sample sufficiently (40.48% explained variance), a finding consistent
of the German population (N = 1,030; 55.9% women) was re- with results of Gibbons and Buunk (1999), so we combined
cruited by the certified poll agency TNS Infratest in May and both facets into one scale.
June 2011. The G-SOEP is an annual household panel that has The individual readiness to repay positive interpersonal
been conducted in Germany since 1984. Each year, prior to the behavior reciprocally (positive reciprocity) and to retaliate
386 BAUMERT ET AL.
negative interpersonal behavior (negative reciprocity) were TABLE 4.—Means and standard deviations of justice sensitivity short scales and
measured with three items each (positive reciprocity: “If some- correlations with validation constructs in Study 2.
one does a favor for me, I am ready to return it,” “I am ready to JS- JS- JS- JS-
undergo personal costs to help somebody who helped me be- M SD Victim Observer Beneficiary Perpetrator
fore,” and “I go out of my way to help somebody who has been
kind to me before”; α = .79; negative reciprocity: “If I suffer Social com- 3.47 1.22 .23∗∗ .18∗∗ .22∗∗ .02
a serious wrong, I will take my revenge as soon as possible, no parison
Positive 2.91 1.49 –.10∗ .14∗∗ .01 .24∗∗
matter what the costs,” “If somebody puts me in a difficult po- reciprocity
sition, I will do the same to him/her,” and “If somebody offends Negative 5.95 1.00 .31∗∗ .03 .08∗ –.07∗
me, I will offend him/her back”; α = .78). Response options reciprocity
ranged from 1 (totally disagree) to 7 (totally agree). Items were M 2.93 3.69 2.38 4.23
adopted from the questionnaire developed by Perugini et al. SD 1.52 1.41 1.30 1.61
(2003), who provided evidence for its structural validity and the Note. N = 1,030. JS-Victim, JS-Observer, JS-Beneficiary, JS-Perpetrator = justice
good reliabilities of the scales. Further empirical results support sensitivity from the victim, observer, beneficiary, and perpetrator perspectives. The mean
the validity of the short scales (Caliendo, Fossen, & Kritikos, reported is the arithmetic mean of individual scale means. Response options for the justice
sensitivity scales ranged from 0 (totally disagree) to 5 (totally agree). Response options for
2012). In this sample, in an exploratory principal axis factor social comparison, positive and negative reciprocity ranged from 1 (totally disagree) to 7
analysis with oblique rotation, we replicated the two-factor so- (totally agree).
∗
lution of the original scale with almost perfect simple structure. p < .05. ∗ ∗ p < .01.
Downloaded by [University of Connecticut] at 21:16 11 October 2014
TABLE 5.—Fit indexes for measurement invariance tests and tests of equality of latent parameters in Study 2.
RMSEA
Model Value 90% CI SRMR CFI χ2 df χ 2 (df )
Equal latent variances .068 [.061, .074] .029 .977 362.30 40 1.79(4)
+ equal latent covariances .070 [.064, .076] .042 .972 439.03 46 76.73(6)∗∗∗
+ equal latent means .093 [.087, .099] .067 .946 813.25 50 374.22(4)∗∗∗
Note. RMSEA = root mean square error of approximation; SRMR = standardized root mean square residual; CFI = Comparative Fit Index; JS-V, JS-O, JS-B, JS-P = justice
sensitivity from the victim, observer, beneficiary, and perpetrator perspectives.
∗∗∗
p < .001.
of Table 5). In a first model, we restricted latent means of only Second, perpetrator sensitivity, and to a lesser degree observer
the factor victim sensitivity to be equal, whereas for the other sensitivity, were positively related to the inclination to pay back
factors, latent means were allowed to differ across scale con- favors one received, thus, reflecting the prosocial tendency in-
texts. In a second model, we restricted the latent means of victim volved in these justice sensitivity perspectives (Schmitt et al.,
and observer sensitivity to be equal across scale contexts. This 2005). As victim sensitivity is assumed to reflect a rather self-
second model did not have an acceptable fit and had a decrease oriented justice concern (Gollwitzer et al., 2005), the negative
in CFI > .01 compared to the preceding, less restrictive model. correlation with positive reciprocity is consistent with expec-
Therefore, we tested a third model in which we restricted the tations. Victim-sensitive persons most probably perceive favors
latent means of victim sensitivity and beneficiary sensitivity to done to them as legitimate compensations of prior disadvantages
be equal, whereas the latent means of observer and perpetrator that they suffered. Therefore, they should not be motivated to
sensitivity were allowed to differ across scale contexts. As this invest their own resources to return a favor. Interestingly, ben-
model was acceptable, we finally tested a fourth model with eficiary sensitivity was uncorrelated with positive reciprocity.
equal latent means across scale contexts for victim, beneficiary, Prior research using the original scales has demonstrated the
and perpetrator sensitivity. This model yielded acceptable fit in- prosocial motivation of beneficiary-sensitive people (Lotz et al.,
dexes and a decrease in CFI < .01 compared to the preceding, 2011). However, from a beneficiary perspective, a favor done to
less restrictive model. These results indicate that latent factor oneself is perceived as an unjust own advantage that potentially
means are unaffected by scale contexts for victim sensitivity, disadvantages a third person. Thus, beneficiary-sensitive per-
beneficiary sensitivity, and perpetrator sensitivity, but not for sons should be highly motivated to compensate disadvantaged
observer sensitivity. others, particularly if the disadvantage is causally linked to their
own advantage, rather than reciprocally pay back favors that
Discussion they receive.
As a first goal of Study 2, we aimed at providing further ev- Third, the correlation of victim sensitivity with the tendency
idence for the construct validity of our two-item scales for jus- to negatively reciprocate reflects victim-sensitive persons’ incli-
tice sensitivity by investigating correlations with theoretically nation to get even and punish a perpetrator as long as this does
related constructs. Importantly, the distinct patterns of correla- not involve own costs (Gollwitzer et al., 2005). Observer- and
tions for the justice sensitivity perspectives with the individual beneficiary-sensitive persons do not share this rather antisocial
inclination toward social comparison (Gibbons & Buunk, 1999; tendency to the same degree. Moreover, perpetrator-sensitive
Schneider & Schupp, 2011), positive and negative reciprocity persons are rather reluctant to take revenge assumedly because
(Perugini et al., 2003) are clearly in line with theoretical as- this could constitute an own violation of justice standards.
sumptions. Finally, the correlations revealed in this study are low to
First, the constructs of victim and beneficiary sensitivity in- moderate. We argue that the size of the correlations consistently
volve the readiness to compare one’s own outcome or treatment reflects the degree of conceptual overlap between the justice
with that of others. In a related vein, observer sensitivity in- sensitivity perspective and the validation constructs. Each cor-
volves the readiness to draw social comparisons among other relate captures an aspect of the justice sensitivity perspectives.
people. Perpetrator sensitivity, by contrast, entails the readiness However, the validation constructs are clearly different from the
to anticipate or detect one’s own transgressions, thus reflecting justice sensitivity perspectives in that they are not specifically
a reliance on internalized justice standards rather than on com- related to the strength of emotional reactions to injustice and
parison with the behavior of others. Consistent with this notion, they do not differentiate between perspectives toward injustice.
in distinction from the other sensitivities, perpetrator sensitivity Regarding the second goal of this research, we empirically
was unrelated to social comparison. tested whether comparisons of results across scale contexts
388 BAUMERT ET AL.
could be meaningfully conducted for our two justice sensitivity psychological processes through which objective constellations
items. This is crucial because the placement of items within a of rewards or characteristics of decision-making procedures can
scale has been shown to affect item characteristics (Knowles & lead to psychological strain and health impairment.
Condon, 2000; Schwarz, 1999), including item means (Knowles Importantly, trait consistencies and occasion specificities ob-
et al., 1996). First, measurement invariance was established tained in the LSTA in Study 1 suggest that justice sensitivity
(scalar invariance), thus indicating that differences in ratings as assessed with our short scales might be less stable and more
across scale contexts are not due to differences in response bi- sensitive to change than with the original scales. At the same
ases, but rather actual differences in the latent construct. Second, time, estimated reliabilities are high and only slightly lower
we tested for differences in latent variances, covariances, and compared to the original scales (Schmitt et al., 2005). It seems
factor means. Our analyses are informative for future research possible that the short scales are more sensitive to factors that
that might aim to compare results obtained with different scale generate true change because these factors might at least partly
versions. For such comparisons it is necessary to know about be indicator-specific. If this is the case, occasion specificity will
potential effects that the scale versions might have. We did not decrease with the number of indicators that are combined in a
find differences in these parameters across two representative scale. Thus, increased occasion specificity of the short justice
samples that responded to our sets of two justice sensitivity sensitivity scales might be due to employing only two of the
items either in the context of the short scales or in the context three indicators used in the original scales (and dropping intru-
of the original scales. This is important because it shows that siveness of thoughts). Future studies could test this speculation
the internal structure of the construct assessed by our two-item by systematically varying the number of indicators realized in
Downloaded by [University of Connecticut] at 21:16 11 October 2014
scales does not change across scale contexts. the scales and comparing occasion specificities. If the greater
Additionally, partial invariance of latent factor means could sensitivity to true change of the short scales is corroborated in
be established. Thus, except for observer sensitivity, latent fac- further research, they will be useful for research on the dynam-
tor means can be considered to be unaffected by scale versions. ics of justice sensitivity across time and situations. For example,
This means that samples that responded to the two justice sen- decreases in victim sensitivity could be tested as a potential out-
sitivity items in different scales (short vs. original) can be com- come of therapeutic or preventive interventions in the context
pared regarding victim, beneficiary, or perpetrator sensitivity, of burnout where victim sensitivity represents a substantial per-
knowing that potential differences in factor variances, covari- sonal risk factor (Hessler et al., 2012). The short scales should
ances, and latent means are not attributable to effects of the be handled with caution, however, when the researcher’s main
scale contexts. However, with regard to mean levels in observer interest lies in stable effects of justice sensitivity, specifically
sensitivity, our results show that differences between samples across time intervals longer than the 6-week interval chosen in
cannot unambiguously be attributed to sample characteristics or our Study 1.
potential differences in treatment, but that these differences are When testing the generalizability of previous findings on jus-
confounded with the effects of the scale context. tice sensitivity obtained with the original scales, the question
arises as to whether results from the short scales can be directly
compared to the former. Researchers have called for caution
GENERAL DISCUSSION regarding the comparability of test forms (Knowles & Con-
In this research, we created ultrashort scales for the eco- don, 2000). By means of MGCFAs, in Study 2, we tested item
nomic assessment of justice sensitivity from a victim, observer, stability across scale contexts to provide an empirical answer
beneficiary, and perpetrator perspective. In Study 1, we pro- to this question. Our results show that latent factor variances
vided evidence in an LSTA for the factorial validity and the and covariances obtained from our two-item sets can indeed
good reliabilities of these scales. Moreover, factor correlations be considered to be unaffected by scale contexts. For victim,
as well as correlations with the Big Five personality factors, beneficiary, and perpetrator sensitivity, in addition, latent factor
with interpersonal trust, self-efficacy, and locus of control as means are comparable across scale contexts. For example, mean
further personality dimensions and with life satisfaction sup- results obtained in samples of unknown representativeness can
port the validity of the short scales. In Study 2, these findings be compared with the normative data provided by Schmitt et al.
are complemented by consistent patterns of correlations among (2010) for the respective two items. By contrast, for observer
our justice sensitivity scales and individual inclinations toward sensitivity, latent mean differences will be confounded with dif-
social comparison and positive and negative reciprocity. In con- ferences caused by the scale contexts. Thus, it seems advisable
clusion, we provide a very useful instrument for the assessment for future research to provide normative data for the short scales.
of justice sensitivity in large-scale surveys as well as in clinical
screenings. This will allow future research to test the general-
izability of previous findings regarding individual differences Limitations
in the perception of injustice and the strength of reactions to As a limitation of this research, it has to be noted that in Study
unfairness. Moreover, our instrument opens up opportunities to 2, the representative samples differed with regard to some de-
gain more insight into the role that subjective injustice plays mographic characteristics. Thus, it cannot be excluded that the
in the explanation of psychological well-being and physical differences in latent factor means of observer sensitivity between
health. For example, justice sensitivity has been shown to be these samples were caused by demographic differences rather
systematically linked to biases in information processing, such than by item contexts as we suggested earlier. Furthermore, re-
as attention, interpretation, and memory for unjust information sponse scales of our justice sensitivity items changed from Study
(e.g., Baumert, Gollwitzer, Staubach, & Schmitt, 2011). Thus, 1 (1–6) to Study 2 (0–5). This is important to keep in mind when
the assessment of justice sensitivity in studies on health and comparing scale means between studies (Tables 1 and 4). Po-
well-being promises a more fine-grained understanding of the tentially, differences in scale means between Studies 1 and 2 are
JUSTICE SENSITIVITY SHORT SCALES 389
due to inconsistent response options. Note that we used response Gollwitzer, M., Schmitt, M., Schalke, R., Maes, J., & Baer, A. (2005). Asym-
options from 0 (totally disagree) to 5 (totally agree) in Study metrical effects of justice sensitivity perspectives on prosocial and antisocial
2 to obtain maximal comparability with the study of Schmitt behavior. Social Justice Research, 18, 183–201.
et al. (2010) employing the original justice sensitivity scales. Greenberg, J. (2004). Stress fairness to fare no stress: Managing workplace stress
by promoting organizational justice. Organizational Dynamics, 33, 352–365.
In addition, a replication of the LSTA in Study 1 could further
Greenberg, J. (2006). Losing sleep over organizational injustice: Attenuating
corroborate the high reliabilities of the short scales and test our insomniac reactions to underpayment inequity with supervisory training in
speculations on potential causes of the increased occasion speci- interactional justice. Journal of Applied Psychology, 91, 58–69.
ficities. Finally, particularly for cross-cultural comparisons, the Greenberg, J. (2010). Organizational injustice as an occupational health risk.
English translation provided in the Appendix could be validated The Academy of Management Annals, 4, 205–243.
by an independent back-translation that has not been undertaken Hessler, C., Pretsch, J., Hillert, A., & Schmitt, M. (2012). Effects of justice
so far. sensitivity and effort-reward-imbalance on the mental health of teachers.
Manuscript in preparation.
Conclusion Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance
structure analysis: Conventional criteria versus new alternatives. Structural
Despite these limitations, this research provides important ev-
Equation Modeling, 6, 1–55.
idence for the high reliability and validity of the ultrashort scales Jakoby, N., & Jacob, R. (1999). Messung von internen und externen Kon-
for justice sensitivity. We are confident that the availability of trollüberzeugungen [Measuring internal and external locus of control].
these very short scales will trigger much social justice research, ZUMA-Nachrichten, 45, 61–71.
thus making use of the great potential of large-scale surveys as Knowles, E. S. (1988). Item context effects on personality scales: Measuring
Downloaded by [University of Connecticut] at 21:16 11 October 2014
well as clinical screenings. changes in the measure. Journal of Personality and Social Psychology, 55,
312–320.
ACKNOWLEDGMENTS Knowles, E. S., Coker, M. C., Scott, R. A., Cook, D. A., & Neville, J. W. (1996).
We thank Nadine Knab for literature searches, Kerstin Dorsch Measurement induced improvement in anxiety: Mean shifts with repeated
assessment. Journal of Personality and Social Psychology, 71, 352–363.
for help in preparing the manuscript, and Jane Zagorski for
Knowles, E. S., & Condon, C. A. (2000). Does the rose still smell as sweet?
helpful comments on this article. Item variability across test forms and revisions. Psychological Assessment,
12, 245–252.
REFERENCES Lotz, S., Baumert, A., Schlösser, T., Gresser, F., & Fetchenhauer, D. (2011).
Baumert, A., Gollwitzer, M., Staubach, M., & Schmitt, M. (2011). Justice sen- Individual differences in third-party interventions: How justice sensitivity
sitivity and the processing of justice-related information. European Journal shapes altruistic punishment. Negotiation and Conflict Management, 4, 297–
of Personality, 25, 386–397. 313.
Beauducel, A., & Wittmann, W. W. (2005). Simulation study on fit indexes Lovas, L., & Wolt, R. (2002). Sensitivity to injustice in the context of some
in CFA based on data with slightly distorted simple structure. Structural personality traits. Studia Psychologica, 44, 125–131.
Equation Modeling, 12, 41–75. Mikula, G., Petri, B., & Tanzer, N. (1990). What people regard as unjust: Types
Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. and structures of everyday experiences of injustice. European Journal of
Sociological Methods & Research, 21, 230–258. Social Psychology, 20, 133–149.
Byrne, B. M., Shavelson, R. J., & Muthén, B. (1989). Testing for the equivalence Milfont, T. L., & Fischer, R. (2010). Testing measurement invariance across
of factor covariance and mean structures: The issue of partial measurement groups: Applications in cross-cultural research. International Journal of Psy-
invariance. Psychological Bulletin, 105, 456–466. chological Research, 3, 111–121.
Caliendo, M., Fossen, F., & Kritikos, A. (2012). Trust, positive reciprocity, Mohiyeddini, C., & Schmitt, M. (1997). Sensitivity to befallen injustice and
and negative reciprocity: Do these traits impact entrepreneurial dynamics? reactions to unfair treatment in a laboratory situation. Social Justice Research,
Journal of Economic Psychology, 33, 394–409. 10, 333–352.
Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit indexes for Muthén, L. K., & Muthén, B. O. (1998–2010). Mplus user’s guide (5th ed.).
testing measurement invariance. Structural Equation Modeling, 9, 233–255. Los Angeles, CA: Author.
Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The Satisfaction Naef, M., & Schupp, J. (2009). Measuring trust: Experiments and surveys in
with Life scale. Journal of Personality Assessment, 49, 71–75. contrast and combination (SOEP Papers on Multidisciplinary Panel Data
Eid, M., Schneider, C., & Schwenkmezger, P. (1999). Do you feel better or Research No. 167). Berlin, Germany: DIW Berlin.
worse? The validity of perceived deviations of mood states from mood traits. Ostrom, T. M., Betz, A. L., & Skowronski, J. J. (1992). Cognitive representation
European Journal of Personality, 13, 283–306. of bipolar survey items. In N. Schwarz & S. Sudman (Eds.), Context effects in
Elovainio, M., Kivimäki, M., Vahtera, J., Keltikangas-Järvinen, L., & Virtanen, social and psychological research (pp. 297–311). New York, NY: Springer-
M. (2003). Sleeping problems and health behaviors as mediators between Verlag.
organizational justice and health. Health Psychology, 22, 287–293. Pavot, W., & Diener, E. (1993). Review of the Satisfaction with Life Scale.
Fergus, T. A., Valentiner, D. P., McGrath, P. B., Gier-Lonsway, S. L., & Kim, Psychological Assessment, 5, 164–172.
H.-S. (2012). Short forms of the Social Interaction Anxiety Scale and the Perugini, M., Gallucci, M., Presaghi, F., & Ercolani, A. P. (2003). The personal
Social Phobia Scale. Journal of Personality Assessment, 94, 310–320. norm of reciprocity. European Journal of Personality, 17, 251–283.
Geiser, C., Crayen, C., & Enders, C. K. (2012). Advanced multivariate data anal- Rammstedt, B., & John, O. P. (2007). Measuring personality in one minute
ysis with Mplus. Wiesbaden, Germany: VS-Verlag für Sozialwissenschaften. or less: A 10-item short version of the Big Five Inventory in English and
Geiser, C., & Lockhart, G. (2012). A comparison of four approaches to account German. Journal of Research in Personality, 41, 203–212.
for method effects in latent state–trait analyses. Psychological Methods, 17, Rothmund, T., Gollwitzer, M., & Klimmt, C. (2011). Of virtual victims and
255–283. victimized virtues: Differential effects of experienced aggression in video
Gibbons, F. X., & Buunk, B. P. (1999). Individual differences in social com- games on social cooperation. Personality and Social Psychology Bulletin, 37,
parison: Development of a scale of social comparison orientation. Journal of 107–119.
Personality and Social Psychology, 76, 129–142. Rousseau, V., Salek, S., Aubé, C., & Morin, E. M. (2009). Distributive jus-
Gollwitzer, M., Rothmund, T., Pfeiffer, A., & Ensenbach, C. (2009). Why and tice, procedural justice, and psychological distress: The moderating effect
when justice sensitivity leads to pro- and antisocial behavior. Journal of of coworker support and work autonomy. Journal of Occupational Health
Research in Personality, 43, 999–1005. Psychology, 14, 305–317.
390 BAUMERT ET AL.
Schmitt, M., Baumert, A., Gollwitzer, M., & Maes, J. (2010). The Justice Sensi- Vandenberg, R. J. (2002). Toward a further understanding of and improvement in
tivity Inventory: Factorial validity, location in the personality facet space, de- measurement invariance methods and procedures. Organizational Research
mographic pattern, and normative data. Social Justice Research, 23, 211–238. Methods, 5, 139–158.
Schmitt, M., & Dörfel, M. (1999). Effects of justice sensitivity and procedural
injustice in the workplace on job satisfaction and psychosomatic well-being.
European Journal of Social Psychology, 29, 443–453. APPENDIX
Schmitt, M., Gollwitzer, M., Maes, J., & Arbach, D. (2005). Justice sensitiv-
ity: Assessment and location in the personality space. European Journal of
Justice sensitivity short scales in English (and German).
Psychological Assessment, 21, 202–211.
Schmitt, M. J., Neumann, R., & Montada, L. (1995). Dispositional sensitivity No. in
to befallen injustice. Social Justice Research, 8, 385–407. Original Item
Schneider, S., & Schuup, J. (2011). The social comparison scale (Data Docu- Perspective Scale Wording
mentation No. 55). Berlin, Germany: DIW.
Scholz, U., Gutiérrez-Doña, B., Sud, S., & Schwarzer, R. (2002). Is general Victim 6 It makes me angry when others are
self-efficacy a universal construct? Psychometric findings from 25 countries. undeservingly better off than me.
European Journal of Psychological Assessment, 18, 242–251. (Es ärgert mich, wenn es anderen unverdient
besser geht als mir.)
Schwarz, N. (1999). Self-reports: How the questions shape the answers. Amer-
7 It worries me when I have to work hard for
ican Psychologist, 54, 93–105. things that come easily to others.
Schwarzer, R., & Hallum, S. (2008). Perceived teacher self-efficacy as a pre- (Es macht mir zu schaffen, wenn ich mich für
dictor of job stress and burnout: Mediation analyses. Applied Psychology: An
Downloaded by [University of Connecticut] at 21:16 11 October 2014