Justice and power: when will justice concerns encourage the advantaged to support policies which...

30
European Journal of Social Psychology, Vol. 26, 171-200 (1996) Justice and power: when will justice concerns encourage the advantaged to support policies which redistribute economic resources and the disadvantaged to willingly obey the law? HEATHER J. SMITH and TOM R. TYLER Department of Psychology, University of California, U.S.A. Abstract Two studies are presented which test whether justice can motivate support for government policies and authorities even when such support is not in people’s obvious personal or group interest. In the first study, White San Francisco Bay area residents’ attitudes toward Congressionally-authored affirmative action policies and anti- discrimination laws were investigated. In the second study, African-American San Francisco Bay area residents’feelings of obligation to obey the law were investigated. The results from both studies show a significant relationship between evaluations of social justice and respondents’ political attitudes. More importantly, a significant relationship between relational evaluations of Congress and political attitudes is found in both studies. This relationship suggests how justice can motivate policy and government support even if such support does not yield direct personal or group benefits. Finally, the results from both studies indicate when instrumental and relational concerns will be related to political attitudes. If people identified with their particular advantaged or disadvantaged group, instrumental concerns were more strongly related to their political attitudes, but if people identi3ed with a superordinate category that included both potential outgroup members and relevant superordinate authorities, relational concerns were more strongly related to their political attitudes. Addressee for correspondence: Heather Smith, Department of Psychology, Tolman Hall - 1650, University of California, Berkeley, California, 94720-1650, U.S.A. Support for this research was provided by the American Bar Foundation, Chicago, Illinois and an NIMH postdoctoral fellowship to the first author. We wish to thank Yuen Huo, Jim Lea, Peter Degoey and three anonymous reviewers for their helpful advice. CCC 0046-2772/96/020171-30 01996 by John Wiley & Sons, Ltd. Received 20 March 1994 Accepted 21 November 199.5

Transcript of Justice and power: when will justice concerns encourage the advantaged to support policies which...

European Journal of Social Psychology, Vol. 26, 171-200 (1996)

Justice and power: when will justice concerns encourage the advantaged to

support policies which redistribute economic resources and the

disadvantaged to willingly obey the law?

HEATHER J. SMITH and TOM R. TYLER Department of Psychology, University of California, U.S.A.

Abstract

Two studies are presented which test whether justice can motivate support for government policies and authorities even when such support is not in people’s obvious personal or group interest. In the first study, White San Francisco Bay area residents’ attitudes toward Congressionally-authored affirmative action policies and anti- discrimination laws were investigated. In the second study, African-American San Francisco Bay area residents’feelings of obligation to obey the law were investigated. The results from both studies show a significant relationship between evaluations of social justice and respondents’ political attitudes. More importantly, a significant relationship between relational evaluations of Congress and political attitudes is found in both studies. This relationship suggests how justice can motivate policy and government support even if such support does not yield direct personal or group benefits. Finally, the results from both studies indicate when instrumental and relational concerns will be related to political attitudes. If people identified with their particular advantaged or disadvantaged group, instrumental concerns were more strongly related to their political attitudes, but if people identi3ed with a superordinate category that included both potential outgroup members and relevant superordinate authorities, relational concerns were more strongly related to their political attitudes.

Addressee for correspondence: Heather Smith, Department of Psychology, Tolman Hall - 1650, University of California, Berkeley, California, 94720-1650, U.S.A. Support for this research was provided by the American Bar Foundation, Chicago, Illinois and an NIMH postdoctoral fellowship to the first author. We wish to thank Yuen Huo, Jim Lea, Peter Degoey and three anonymous reviewers for their helpful advice.

CCC 0046-2772/96/020171-30 01996 by John Wiley & Sons, Ltd.

Received 20 March 1994 Accepted 21 November 199.5

172 H. J. Smith and T. R. Tyler

INTRODUCTION

Inequality in the distribution of opportunities and resources is characteristic of most societies. These discrepancies between the advantaged and the disadvantaged can generate anger and resentment; feelings which find expression in collective disorders, like the recent Los Angeles riots. The incredible material and social damage that can be caused by such behaviour underscores the importance for understanding how societal inequities can be resolved before they become a catalyst for destructive attitudes and actions. A solution to this problem requires the consideration of two separate questions. First, what might encourage those advantaged by societal inequities to support political strategies for redistributing economic resources, even when it may cost them some of the advantages they currently enjoy. Second, will those disadvantaged by societal inequities be willing to defer to their feelings of obligation to laws which prohibit destructive actions such as rioting even in the face of experiences of individual or collective injustice.

Two studies are reported. Each study examines the ability of two psychological mechanisms to bridge across differences in economic opportunities and resources. Those two mechanisms are: (1) judgements that policymakers and policies follow principles of relational justice and; (2) identification with the superordinate social category, ‘Americans’, which includes both the advantaged and the disadvantaged, rather than identification with the subordinate social category of one’s own ethnic group. The first study examines the willingness of the advantaged to support pro- active policies designed to redistribute economic opportunities. It is based on interviews with 352 White respondents, who are asked about their support for Congressional policies designed to eliminate economic discrimination and rectify past economic injustices. The second study examines the willingness of the disadvantaged to accept their obligation to obey laws. It is based on interviews with 150 African-American respondents.

STUDY 1: THE ADVANTAGED

Traditional economic models suggest that inequalities in opportunities and resources contribute to a grim cycle of intergroup conflict between the advantaged and disadvantaged. The disadvantaged challenge the status quo, while the advantaged struggle to protect it (e.g. Cook and Hegtvedt, 1986. Homans, 1961; Kabanoff, 1991). An economic view suggests that the effort by the advantaged to defend structural inequality is the inevitable result of people’s self-interests. Not only do the advantaged have the means to protect their advantages, they have the motive. Unless public policies are in people’s self-interest, there is no reason for them to view these policies as fair or to support them.

In contrast to economic self-interest models, social justice research suggests that perceptions and reactions to injustice are not always dictated directly by positions of power (Montada, 1992; Montada and Schneider, 1989). Just as feelings of injustice may motivate participation in collective rebellion and riots (Crawford and Naditch, 1970; Gurr, 1970; Sears and McConahay, 1970), feelings of injustice also may motivate support for constructive and pro-active political attitudes and behaviours

Justice and power 173

(Veilleux and Tougas, 1989). In fact, the recognition of injustice might motivate the advantaged to support policies designed to achieve justice before the disadvantaged pursue the often devastating alternative of collective violence or rebellion, just as, in experiments, it leads people to forego advantages they could obtain (Hoffman and Spitzer, 1982, 1985; Kahneman, Knetsch and Thaler, 1986).

Social justice research suggests two different types of evaluations that may influence people’s reactions to inequities. Distributive justice describes the fairness of outcomes. Feelings of entitlement and injustice are assumed to be reactions to discrepancies in the distributions of outcomes between different people or groups (Deutsch, 1985; Martin, 1986; Moore, 1991). Evaluations of distributive injustice are linked to collective protest, sabotage at work and employee theft (e.g. Greenberg, 1990; Hafer and Olson, 1993; Walker and Mann, 1987). Distributive justice models suggest that people simply determine whether they and others get what they deserve. If this is not the case, they will be more likely to support policies designed to rectify the inequity.

Procedural justice describes the fairness of the methods or the procedures used to determine outcomes (Lind and Tyler, 1988). Evaluations of procedural justice are related to voluntary acceptance of decisions made by authorities, obedience to laws and legal rules, and evaluations of the legitimacy of authorities (e.g. Tyler, 1990; Tyler and Degoey, 1994; Tyler and Lind, 1992; Tyler and Mitchell, 1994). Procedural justice research suggests that if how outcomes are distributed is judged to be unfair, people will support social change, regardless of any outcome differences, but if how outcomes are distributed is perceived as fair, discrepancies in outcomes also may be viewed as justified. In fact, people may use the fairness or unfairness of procedures as an heuristic for determining the justice of the outcome distribution (Folger, 1987; Lind, Kulik, Ambrose and de Vera Park, 1993). For example, people might assume that as long as the current market procedures for allocating economic outcomes are fair, any distributive injustice is based on individual, not group, characteristics. Procedural justice models suggest that if people view the job hiring, firing and promotion in economic markets as governed by non-neutral, discrimatory criteria, they will be more likely to support policies designed to rectify the inequity.

The psychology of justice

The distinction between procedural and distributive justice reflects two different explanations as to why justice is important to people (Tyler, 1994). The traditional distributive justice focus on outcomes reflects an economic explanation as to why justice motivates people. This explanation assumes that people strive to maximize their personal rewards in exchanges with others (Taylor and Moghaddam, 1987). People try to act fairly, and expect others to act fairly toward them because they believe that over time equitable behaviour promotes personal gain (Tyler, 1994). The same model can be extended to explain an interest in procedural justice as ‘informed’ self-interest (Lind and Tyler, 1988). If people believe that the decision-making procedures are fair, they may be willing to sacrifice short-term gains in the interest of long-term gains in outcomes and group harmony (Lind and Tyler, 1988; Thibaut and Walker, 1975).

174 H. J. Smith and T. R. Tyler

Unfortunately, an instrumental account of why justice is important cannot explain why advantaged group members would give up their privileges without a direct increase in their outcomes or other instrumental benefits. Recent research investigations of the influence of process control or ‘voice’ on evaluations of procedural fairness illustrate the limitations of an instrumental model of justice. When evaluating fairness, people appear to want to have the chance to state their case to authorities regardless of whether their statement influences the authorities’ decisions (Lind, Kanfer and Earley, 1990; Lind, Lissak and Conlon, 1983; Tyler, 1989; Tyler and Lind, 1992).

Proposed to explain these non-instrumental effects, Group Value theory (Lind and Tyler, 1988; Tyler, 1989; Tyler and Lind, 1992) suggests that people’s concerns about procedural justice are motivated by an interest in information about their social position and connections to others. Group Value theory assumes that people are concerned about how they are viewed within their important reference groups because such information shapes their images of themselves (Tyler, 1994; Tyler, Degoey and Smith, 1994). While there might be numerous ways to determine how other members of important groups view them, the quality of the interpersonal treatment people receive from relevant group authorities is an especially potent source of information for determining whether they are valued group members. When people evaluate procedures, they are sensitive to what those procedures seem to say about how they are viewed by their group (Tyler and Lind, 1992, p. 32). If group authorities treat people fairly, it communicates respect and value, but if authorities treat people unfairly, it communicates disrespect and social marginality (Tyler et al., 1994).

Since Congress is responsible for the policies of concern in this study, the actions of Congress can communicate both relational and instrumental information. People can use governmental policies to judge both the favourability of their benefits and their position in American society. These motives enable a comparison between the influence of Congressionally-related relational and instrumental evaluations. If people are motivated by instrumental concerns, they should be especially sensitive to the proposed policy’s impact on them. Given that the current distribution of outcomes is unfair, evaluations of proposed solutions should be based on whether or not the proposed policies will rectify the injustice. In contrast, a group value model of justice emphasizes people’s evaluations of how Congress treats its constituents. If people believe that they are respected and valued group members, they should be more likely to support authorities’ decisions, even if the outcomes are not in their interest. Neutral, trustworthy and respectful authorities communicate status recognition (‘standing’ in the group) to individual group members (Tyler and Lind, 1992).

Group identification and justice

The group-value model suggests how justice can motivate support for public policies that rectify current inequities, even when they do not serve people’s obvious self-interests. This model offers an optimistic alternative to the grim scenario suggested by more economic or instrumental models of justice. However, there may be limits to people’s concerns about relational issues. Evaluations of the neutrality,

Justice andpower 175

trustworthiness and respectfulness of treatment by authorities should only influence justice evaluations if the authorities are included within important social categories or groups. Furthermore, endorsement of social change policies may depend on whether people see the disadvantaged as part of their ‘community of responsibility’ or not (Montada, 1992; Montada and Schneider, 1989). If people view the targets of injustice as outside the scope of their community of responsibility, the fairness and respectfulness of their treatment may be irrelevant (Deutsch, 1985; Optow, 1990; Tyler and Lind, 1988).

Alternatively, if group boundaries are redefined to include both the disadvantaged and the advantaged, recognition of unfair treatment of some category members may encourage support for governmental policies even if they are not in people’s personal self-interest. As demonstrated by social dilemma research, introducing a common group membership means individual group members pay less attention to distinctions between themselves and other group members and give more weight to common group interests as opposed to their personal self-interest (Brewer and Kramer, 1986; Kramer, 1991; Wit, 1989). Similarly, investigations of ingroup bias suggest that if superordinate categories are salient or important, between group bias and discrimination is decreased (Gaertner, Dovidio, Anatasio, Bachman and Rust, 1993). Distinguishing between levels of inclusiveness suggests that when a superordinate category is more important to people, inequities between different groups represent an intragroup situation with implications for collective cooperation and harmony. In contrast, if a particular group is more important to people, inequities between different groups represent an intergroup situation with different groups competing for resources and power (Azzi, 1994; Kramer, 1991).

The different levels of inclusiveness represented by different social categorizations suggests that it is not general relational concerns that are important for understanding the antecedents to justice evaluations but rather the relational concerns that correspond with the social category or group membership that is most important or salient to the evaluator. If people feel more identified with a larger collective category, they may be more sensitive to the relational information communicated by collective policies and collective policymakers, and less concerned with the instrumental consequences of any proposed changes. Furthermore, they should be more likely to include the disadvantaged group within a common superordinate categorization and therefore, be more motivated to rectify between group inequalities. We expect that people who feel close to Americans in general will pay more attention to relational aspects of fairness when evaluating Congressional policies. However, people who feel more positively toward their advantaged group should be motivated to protect and maintain their immediate group’s interests and therefore, they should be relatively uninterested or insensitive to the unjust outcomes or treatment of other groups. We expect that they will pay more attention to instrumental aspects of fairness when evaluating the same policies.

Of course, it is important to recognize that instrumental concerns are defined in this study as benefits and costs to the advantaged group, while relational concerns are defined as an interest in how superordinate category authorities treat its citizens. Identification with one’s racial group may well increase the importance of relational concerns within the group and decrease instrumental concerns at the interpersonal level. However, even though the meaning of instrumental and relational concerns

176 H. J . Smith and T. R. Tyler

may be different at different levels of categorization, we think it is appropriate to distinguish relational and instrumental issues at the collective level when the goal is explaining support for policies designed to redistribute economic opportunities and outcomes at the collective level.

Identification with a superordinate category may influence the relative importance of instrumental and relational concerns to policy endorsement for two reasons. First, identification with a superordinate category rather than a subordinate group may change people’s views of resource interdependencies. For example, people who identify with Americans in general may view their outcomes as interdependent with all the other members of their group, including the disadvantaged (Kramer, 1991; Sherif, Harvey, White, Hood and Sherif, 1961). Favourable outcomes for others are related to favourable outcomes for themselves. However, when people identify with an advantaged group, the outcomes for the disadvantaged may be viewed, if anything, as negatively related to favourable outcomes for themselves. In other words, instrumental concerns about costs and benefits are still important, they simply are interpreted differently depending on one’s level of identification.

In contrast, the Group Value model suggests that identification with superordinate categories encourages people to sacrifice their advantages because they value their relationship to the group, not because they foresee eventual benefits for themselves. In other words, identification may not only shape people’s perceptions of the resource interdependencies between themselves and others, it also may shape their sense of self (Taijfel and Turner, 1986). When people identify with a particular group, they value their positive emotional connection to that group (Tyler and Degoey, 1994; Tyler and Dawes, 1993). Because they value the group, whether the group values them has direct consequences for their feelings of self-worth (Tyler et al., 1994). However, if people do not value a particular group membership, the relational information communicated by group authorities or other group members should be relatively unimportant to their sense of self- worth. In other words, the reason that superordinate categorizations may encourage greater concern with fair and just treatment of all groups included within the larger categorization is the potential information that authorities as representatives of important groups can communicate to individual group members about their value to the group, and through it, their self-worth (Tyler et al., 1994).

The first study addresses four questions. The first question is whether evaluations of how economic opportunities are distributed are as important to policy endorsement as the recognition of unfair differences in the economic outcomes for different groups. The second question is a comparison of instrumental evaluations of Congress with relational evaluations of Congress as predictors of policy endorsement. The third question is to explore whether identification with the superordinate category of Americans in general influences the relative importance of relational and instrumental concerns to support for policies designed to reduce economic inequities between groups. Finally, the relationship between identification, respect and self-esteem will be explored as a test of whether identification influences the relative importance of instrumental and relational concerns to policy evaluations because it changes the sources of information important to people’s self- evaluations.

Justice andpower 177

Method

Respondents

Random digit telephone sampling was used to contact San Francisco Bay area residents between January and June of 1992. The sample was restricted to English- speaking adults who identified themselves ethnically as White. Of those eligible for interviews, a completed interview was obtained from 352 respondents (66 per cent response rate). The mean age of respondents was 44 (standard deviation= 15.55 years). Forty-eight per cent of respondents were high school graduates or less; 36 per cent had some college education; 25 per cent were college graduates; and 19 per cent had some post-college education. Thirteen percent of the respondents had a household income of less than 20000$; 21 per cent had an income of 20001 to 40000$; 22 per cent had an income of 40001 to 60000$; and 42 per cent had an income over 60 001$. Further details about the sampling design and the demographic characteristics of the sample are available from the authors.

Measures

Summary statistics and intercorrelations for all the variables included in this study are presented Table 1.

Policy endorsement A single measure of respondents’ attitudes toward Congressionally-authored solutions to economic justice was created (Cronbach’s alpha = 0.67). Respondents indicated whether they felt that Congress should have the power to make laws about who is fired, hired and promoted in private business and whether a qualified black person should be offered a job before an equally qualified white person. They also indicated their support for Congressional laws designed to prevent job discrimination against members of different racial and ethnic groups, and their support for laws designed to give African-Americans preferential treatment in hiring, firing and promotion in order to compensate for past discrimination. Responses could range from strongly favour (1) to strongly oppose (7). Finally, respondents also reported their overall satisfaction with affirmative action policies and anti-discrimination laws. Scores of 7 indicated basic unhappiness with government efforts, scores of 1 indicated basic happiness with government efforts and scores of 4 indicated feeling neither one way nor the other.

Measures of procedural and distributive justice Relative deprivation research demonstrates that it is evaluations of collective injustice, not individual injustice, that are associated with political attitudes and actions (e.g. Dube and Guimond, 1986; Petigrew, 1985; Walker and Mann, 1987). Therefore, in this study, we focus on evaluations of collective injustice rather than feelings of individual injustice or deprivation.

To measure distributive injustice, two difference scores were created based on respondents’ ratings of wages and the general economic situation for Whites and African-Americans. Evaluations of the general economic situation could range from excellent (1) to poor (7). For pay rates, respondents indicated whether most African- Americans or White Americans received more than they deserved (l), less than they deserved (7) or just about the right amount of pay (4). One difference score was

Tabl

e 1.

In

terc

orre

latio

ns a

nd s

umm

ary

stat

istic

s - St

udy

1

Var

iabl

e M

S.D

. 1

2 3

4 5

6 7

8 9

10

11

12

13

1. P

olic

y en

dors

emen

t 2.

Pro

cedu

ral j

ustic

e 3.

Dis

trib

utiv

e jus

tice

4. R

elat

iona

l

5. I

nstr

umen

tal

6. R

elat

iona

l con

cern

s 7.

Ins

trum

enta

l co

ncer

ns

8. I

dent

ific

atio

n w

ith

Am

eric

ans

9. I

d w

ith U

.S. x

rel.

cnrn

s 10

. Id

with

U.S

. x in

str.

cnrn

s 11

. Id

entif

icat

ion

with

ad

vant

aged

12

. R

espe

ct b

y ot

her

adva

ntag

ed

13.

Id w

ith a

dv. x

re

spec

t 14

. Se

lf-es

teem

eval

uatio

ns

eval

uatio

ns

4.03

1.

37

4.77

4.

34

3.85

3.23

4.

09

7.06

-0.1

5

-0.0

4

4.55

2.53

0.26

1.99

1.14

-

0.87

0.2

9::

-

0.88

0.

21**

0.

39**

-

1.19

0.

23**

0.

17

-0.0

5 -

1.05

0.

29**

-0.

08

0.03

0.

18**

-

0.81

0.

33**

0.

38**

0.

17**

0.

92**

0.

19**

-

0.79

0.

23**

-0.

03

0.23

**

0.16

**

0.95

**

0.14

*

1.93

0.

02

0.09

0.

17**

-0.

00

-0.2

0 -0

.15*

*

1.23

-0

.05

-0.0

1 0.

06

-0.0

2 0.

03

0.02

0.98

0.

11*

-0.0

7 0.

04

-0.0

7 -0

.02

-0.0

4

1.68

-0

.02

-0.2

0 -0

.20*

* -0

.01

0.14

**

0.06

1.16

-0

.06

-0.1

7 0.

22

-0.0

3 0.

17*

0.08

1.00

0.

07

0.02

0.

03

-0.0

7 0.

00

0.00

1.06

-0

.07

-0.1

1 0.

12*

0.05

0.

01*

-0.0

5

-

-0.0

5 -

-0.0

4 0.

18**

-

-0.0

6 0.

02

0.21

** -

0.02

-0

.26*

* -0

.02

0.01

-

0.05

-0

.24*

* -0

.02

0.01

0.

26**

-

-0.0

6 0.

04

0.03

0.

05

-0.0

9 0.

11*

-

0.08

-0

.19*

* -0

.08

-0.0

8 0.

13*

0.32

**

-0.1

0

Not

e. N

= 35

2. E

ntrie

s ar

e Pe

arso

n co

rrel

atio

ns. L

ow s

core

s ind

icat

e le

ss p

olic

y en

dors

emen

t, m

ore

proc

edur

al ju

stic

e, m

ore

dist

ribut

ive j

ustic

e, g

reat

er in

stru

men

tal b

enef

its,

mor

e re

spec

tful C

ongr

essi

onal

trea

tmen

t, le

ss id

entif

icat

ion w

ith A

mer

ican

s in

gene

ral,

mor

e id

entif

icat

ion

with

the

adva

ntag

ed, m

ore

resp

ect a

nd h

ighe

r sel

f-es

teem

. *p <

0.05

, **

p<O

.Ol.

Justice and power 179

created based on respondents’ ratings of whether African-Americans and Whites generally receive the pay that they deserve. A second difference score was created based on evaluations of the general economic situation for Whites and African- Americans. The two difference scores were averaged to create an index of outcome discrepancy (Cronbach’s alpha =0.58). The relatively low alpha reflects the fact that most respondents evaluated the relative outcomes for the disadvantaged as very unfair ( M = 4.63, S.D. = 0.72).

Three items were combined to measure evaluations of the fairness of economic opportunities or procedural justice: the first question asked whether White and African-Americans have equal opportunities to succeed in the workplace. The second question asked whether African-Americans have as much chance to succeed as Whites and the last question (reversed scored) asked whether African-Americans still face economic discrimination (Cronbach’s alpha = 0.76). Responses could range from agree strongly (1) to disagree strongly (7).

Measures of relational and instrumental evaluations of Congress As a measure of relational evaluations of Congress, five items asking respondents about Congressional treatment of its constituents were combined (Cronbach’s alpha = 0.72). The first question (reversed scored) asked whether Congressional representatives are dishonest. The second question asked whether Congress is honest in the way it makes decisions. The third question asked whether Congress obtains the information necessary for informed decisions. The fourth question asked whether Congress is concerned about protecting the average citizen’s rights. The fifth question asked whether Congress gives equal consideration to the views of all the different groups in America. Responses could range from agree strongly (1) to disagree strongly (7).

As a measure of instrumental evaluations of Congress, respondents reported whether affirmative action policies or anti-discrimination laws had increased or decreased their job opportunities, whether affirmative action policies or anti- discrimination laws had increased or decreased the job opportunities for most African-Americans. For each item, scores of 1 indicated that job opportunities had decreased, scores of 7 indicated that job opportunities had increased, and scores of 4 indicated that there had been no change in job opportunities. Respondents’ answers were combined to create a measure of whether policies would favour themselves and other Whites (Cronbach’s alpha = 0.69).

Identification with one’s advantaged group and Americans in general Agreement with two items was used to measure respondents’ identification with their advantaged group: the first question asked whether respondents were proud to think of themselves as Whites. The second question asked if when the achievements of Whites were praised, whether it felt like a personal compliment (Cronbach’s alpha = 0.52). Responses could range from agree strongly (1) to disagree strongly (7). As a measure of respondents’ identification with Americans in general, respondents rated Americans as a group on a scale from 0 to 10 with 10 representing extremely favourable or warm feelings and 0 representing extremely unfavourable or cold feelings‘.

‘We recognize that measures of pride or affective feelings toward a particular group membership are just one aspect of identification as first defined by Tajfel and his colleagues (Tajfel, 1981). One goal for future research will be to include more sophisticated measures of identification that include both a cognitive and an evaluative component as well as an effective component of identification.

180 H . J . Smith and T, R . Tyler

Feelings of respect Three items, each asking respondents to imagine that most Whites knew them well, were used to measure respondent’s evaluations of whether most members of their advantaged group respected them (Cronbach’s alpha = 0.64). The first question asked whether most Whites would think that the respondent had accomplished a great deal, the second question asked whether most Whites would approve of the respondent’s lifestyle and the third question asked whether most Whites would respect their values. Responses could range from agree strongly (1) to disagree strongly (7).

Self-esteem Agreement with three items taken from the Rosenberg self-esteem scale (Rosenberg, 1979) was used to measure respondents’ self-esteem. Responses could range from agree strongly (1) to disagree strongly (7). The first item asked respondents whether they felt sure of themselves, the second item, asked respondents whether they felt satisfied with themselves and the third item asked whether respondents felt proud of what they had accomplished (Cronbach’s alpha = 0.80).

Overall perceptions of justice

Before considering the potential influence of either distributive or procedural justice evaluations on attitudes toward Congressionally-authored social change policies, it is important to establish that respondents do indeed view themselves as relatively advantaged and African-Americans as relatively disadvantaged. When respondents’ evaluations of the economic situation for the advantaged are compared with their evaluations of the economic situation for the disadvantaged, they clearly indicate the respondents’ view that the economic situation for the disadvantaged is much less fair and satisfying (4347) = 10.00, p < 0.0001). Moreover, a majority of respondents agreed that the disadvantaged do not have the same opportunities to succeed (52.9 per cent), do not have equal chances to succeed (55.7 per cent) and still face racial discrimination (74.9 per cent). The fact that a significant number of respondents evaluated current economic opportunities and outcomes as unfair for the disadvantaged provides evidence that evaluations of justice can be considered separately from social position or power.

Procedural justice, distributive justice and policy evaluations

Multiple regression analyses were used to test whether evaluations of procedural justice, distributive justice or both were related to policy endorsement2. When both procedural and distributive justice judgements were entered simultaneously, a significant amount of the variance in policy endorsement was explained (adjusted

’Initial regression analyses that included only demographic variables explained 6 per cent of the variance in policy endorsement. Female respondents were significantly more likely than male respondents to support affirmative action policies and anti-discrimination laws. No other demographic variables were significant.

Justice andpower 181

R-squared = 8.9 per justice appears to be p < 0.01, versus beta

cent, p < 0.001). Furthermore, the weight given to procedural greater than the weight given to distributive justice (beta = 0.24, = 0.12, p < 0.05).

Instrumental and relational evaluations of Congress

The significant contribution of procedural justice evaluations to policy endorsement demonstrates that it is worthwhile to determine whether people’s interest in procedural justice reflects instrumental or relational concerns. The relative importance of procedural justice to policy attitudes may reflect an interest in the relational implications of policies and policymakers’ behaviour and not just potential instrumental benefits. Multiple regression analyses were used to test whether instrumental or relational evaluations of Congress, or both were related to policy endorsement. When both instrumental and relational evaluations of Congress were entered simultaneously, a significant amount of the variance in policy endorsement was explained (adjusted R-squared = 11.3 per cent, p < 0.001). Although both types of evaluations were significantly related to policy endorsement, the weight given to relational judgements appears to be greater than the weight given to instrumental judgements (beta=0.26, p<O.OOl, versus beta=0.18, p <0.001).

Superordinate categories and policy endorsement.

The previous regression analyses document the importance of both relational evaluations and procedural justice for policy endorsement. However, there may be limitations to the influence of procedural justice and relational evaluations. We hypothesize that relational evaluations and procedural justice are most likely to promote an interest in justice when respondents include both the targets of the injustice and the authorities proposing any changes within valued groups or social categories. If respondents exclude the disadvantaged and national policymakers from the groups they value, then instrumental evaluations and distributive justice should be more important. As an initial test of this idea, respondents’ feelings toward Americans in general were used to measure identification with Americans in general; a superordinate category that includes the advantaged, the disadvantaged and Congress as a symbolic authority.

Confirmatory factor analysis Recent research (Huo, Smith, Tyler and Lind, 1994; Tyler, 1994) shows that evaluations of distributive justice are influenced by instrumental concerns while evaluations of procedural justice are influenced by relational concerns. Therefore, evaluations of procedural justice and relational evaluations of Congress were combined to create a single index of relational concerns (Cronbach’s alpha = 0.66) and evaluations of distributive justice and instrumental evaluations of Congress were combined to create a single index of instrumental concerns (Cronbach’s alpha = 0.64). Confirmatory factor analysis was conducted to test whether relational judgements; judgements about procedural justice and the neutrality, honesty and trustworthiness of Congress, could be distinguished from instrumental judgements; judgements about distributive justice and the effectiveness of Congressional policies. Covariance matrices were computed

182 H. J . Smith and T. R. Tyler

using listwise deletion of missing data and analysed with the maximum likelihood method of estimation in LISREL-8 (Joreskog and Sorborn, 1993)3. An examination of the goodness of fit indices for the two-factor solution supports a distinction between relational and instrumental judgements. The two-factor solution yielded a non- significant chi square ( ~ ~ ( 3 6 ) = 23.00, as.), and values of the Goodness of Fit Index of 0.99 and the Adjusted Goodness of Fit index of 0.98. Both values are above 0.90, a convention generally reflecting a good fit to the data (Pedhazur and Schmelkin, 1991).

To test whether identification with Americans influenced the relative importance of relational and instrumental concerns to policy endorsement, two interaction terms were included in a simultaneous multiple regression analysis. The first term tested whether identification with Americans moderated the relationship between relational concerns and policy endorsement. The second term tested whether identification with Americans moderated the relationship between instrumental concerns and policy endorsement. If identification with Americans moderates the relationship between justice concerns and policy endorsement, the regression coefficient for the respective interaction terms should be significant. As shown in Table 2, both interaction terms were significant.

A median split of the distribution was used to classify respondents as identifying more or less closely with Americans in general (median = 7.00). As expected, relational concerns appear more closely related to policy endorsement for respondents who identified more closely with Americans in general ( r = 0.42, p < 0.001) than for respondents who identified less closely with Americans ( r = 0.29, p < 0.01). Further- more, instrumental concerns appear more closely related to policy endorsement for respondents who identified less closely with Americans (r=0.33, p < 0.01) than for respondents who identified more closely with Americans in general ( I = 0.06, n.s.).

Although the results from this regression analysis are suggestive, they do not allow us to distinguish between identification with the United States as a superordinate category and identification with one’s particular advantaged group. If some respondents interpreted Americans in general as a reference to their advantaged group, the fact that they feel close to Americans may not indicate that they have expanded their social category to include the disadvantaged. The fact that identification with Americans in general and identification with the advantaged group are positively related (r(352) = 0.26, p < 0.01, see Table 1) suggests that this may be true. Therefore, one way to more clearly understand the relationship between group identification and justice evaluations is to compare subgroups of respondents who did distinguish between identification with their advantaged group and identification with Americans in general.

A second median split (median = 5.00) was used to create a group of respondents who identified relatively more or less with the advantaged group. Respondents then could be classified into one of four groups: negative feelings toward both Americans in general and their advantaged group, pride in their advantaged group but negative feelings toward Americans in general, positive feelings toward both Americans in general and their own advantaged group and finally, positive feelings toward

’The average distribution of respondents’ beliefs about job opportunities were highly skewed (ratio of skewness to standard error of skewness of more than 141). Inverse transformations failed to lead to any more than a minimal movement towards normality. Therefore, these variables were analysed in their original format. However, in order to meet the assumptions required for a confirmatory factor analysis, the average response across all six items, rather than the responses to individual variables were used.

Justice and power 183

Table 2. Study 1

The relationship between identification, justice evaluations and policy endorsement-

Policy Endorsement (n = 343) Regression without Regression with interaction terms interaction terms

Identification with Americans (A) Relational evaluations (B) Instrumental evaluations (C) A x B A x C

Adjusted R2

0.07 0.09 0.32*** 0.33*** 0.19*** 0.20*** - -0.10* - 0.15**

0.14*** 0.17***

Note. Unless otherwise noted, the entries are standardized regression coefficients for an equation in which all variables are entered simultaneously. *p < 0.05, **p < 0.01, ***p < 0.001.

Americans in general and relatively less pride in their advantaged group. The three most theoretically interesting groups are those respondents who reported relatively greater warmth toward Americans in general but relatively less pride in their advantaged group (more identification with Americans in general), those respondents who reported both warmth towards Americans in general and pride in their advantaged group (equal identification with Americans in general and the advantaged) and those respondents who reported relatively less warmth toward Americans in general and relatively greater pride in their advantaged group (more identification with the advantaged). Respondents who reported not feeling close to either their advantaged group or to Americans in general were excluded from the analysis (n = 92).

As shown in Table 3, level of relative identification does change the relative importance of instrumental and relational concerns to policy endorsement. For respondents who identified more with Americans in general, the weight given to relational concerns appears to be greater than the weight given to instrumental concerns (beta = 0.36, p < 0.001, versus beta = 0.30, p < 0.01). Similarly, relational concerns were more closely related to policy endorsement than instrumental concerns for respondents who reported relatively equal identification with both Americans in general and the advantaged (beta = 0.26, p < 0.001 versus beta = 0.20, p < 0.10). In contrast, for respondents who identified more with their advantaged group, the weight given to instrumental concerns appears to be greater than the weight given to relational concerns (beta = 0.35, p < 0.01, versus beta = 0.28, p < 0.05).

Relationship between self-esteem and feelings of respect

We hypothesize that the reason identification with particular groups changes the relative influence of relational and instrumental concerns on policy endorsement is that identification determines what sources of information are important to self- evaluations. To test this idea, we can compare the relationship between feelings of respect from members of the advantaged group and feelings of personal self-esteem for those respondents who identify with the advantaged and for those respondents who do not. We hypothesize that closer identification with a particular group makes

184 H . J . Smith and T. R. Tyler

Table 3. Relative identification, justice evaluations and policy endorsement -Study 1

Policy endorsement

More indentification with Americans" Relational concerns Instrumental concerns

Adjusted R2

0.35*** - 0.36**** - 0.28** 0.30***

0.11*** 0.07** 0.19****

Equal identification with Americans and the advantagedb Relational concerns 0.28*** - 0.26* * * Instrumental concerns - 0.23** 0.20*

Adjusted R2 0.07*** 0.04** 0.10***

More identification with the advantaged" Relational concerns Instrumental concerns

Adjusted R2

0.39*** - 0.28** - 0.44**** 0.35***

0.14*** 0.18**** 0.24****

Note. Unless otherwise noted, the entries are standardized regression coefficients for an equation in which all variables are entered simultaneously. *p<O.lO, **p<O.O5, ***p<O.Ol, ****p<O.OOOl aRespondents reported greater identification with Americans in general and less identification with the advantaged (n = 63). bRespondents reported greater identification with the advantaged and greater identification with Americans in general (n = 1 13). "Respondents reported greater identification with the advantaged and less identification with Americans in general (n = 75).

how respectfully and fairly other group members treat people more personally relevant. Therefore, feeling respected by most advantaged group members should be more closely related to self-esteem for people who identify more closely with the advantaged than for people who do not identify as closely with the advantaged.

Identification with the advantaged and feelings of respect were entered as an interaction term in a simultaneous multiple regression analysis. If identification with the advantaged moderates the relationship between feelings of respect and self- esteem, the regression coefficient for the interaction term should be significant. The first regression equation included identification with the advantaged and feelings of respect and accounted for a significant amount of variance in feelings of self-worth (adjusted R-squared = 10 per cent, p < 0.001). However, feelings of respect appeared to be more important than identification with the advantaged group (beta = 0.3 1 , p < 0.001, versus beta= 0.05, n.s.). A second regression equation included a significant interaction term (beta= -0.13, p < 0.01) and significantly increased the variance explained for self-esteem (change in R-squared = 1.6 per cent, F( 1,344) = 6.34, p < 0.05). Feelings of general respect from one's advantaged group were more closely related to feelings of self-esteem for respondents who identified with the advantaged (r(187) =0.46, p <0.001) than for respondents who did not (r( 154) = 0.21, p < 0.01). Furthermore, a comparison between two subgroups; (1) those who identified more with the advantaged than with Americans in general and (2) those who identified more with Americans in general than with the advantaged also support the predicted pattern. Self-esteem was more closely related to within- group respect for respondents who identified more closely with the advantaged compared to Americans in general (r(76) = 0.42, p < 0.001) than for respondents who identified more closely with Americans in general compared to the advantaged (r(64) = 0.27, p < 0.05).

Justice and power 185

Discussion

These results offer optimistic evidence that a desire for justice is related to support among the advantaged for pro-active policies designed to redress inequalities. First, it is clear that evaluations of justice are not a direct consequence of social power or position. Most advantaged respondents evaluated the economic opportunities and outcomes available for the disadvantaged as unfair. More importantly, evaluations of both distributive and procedural justice were significantly related to policy endorsement. Second, the influence of relational evaluations of Congress on policy support suggests why justice can be important even when it does yield clear personal or group benefits. Agreement that Congress treats people and makes decisions fairly and honestly was significantly related to policy endorsement even after considering people’s beliefs that particular policies would increase the jobs available for themselves or their group. These results suggest that when people evaluate policies, they consider more than just their potential instrumental rewards or costs. People also consider how respectfully the group authorities responsible for public policies treat group members (see also Tyler, 1989, 1990; Tyler and Degoey, 1994).

While the strong influence of procedural justice and relational evaluations on policy endorsement supports a relational model of justice (Tyler, 1994; Tyler and Lind, 1992), distributive justice and instrumental evaluations also influenced policy endorsement. The fact that both relational and instrumental concerns contribute to policy endorsement demonstrates the value in determining when each might be more important. Political policies are much more likely to be successful if those for whom the policy is not clearly beneficial are more interested in relational issues than in instrumental issues. The results from this study show how identification can influence the relative importance of relational and instrumental concerns to policy endorsement. For respondents who identified more closely with Americans in general, a superordinate category, relational concerns were more closely related to policy endorsement than were instrumental concerns. In contrast, for respondents who identified more closely with their advantaged group, instrumental concerns were more closely related to policy endorsement than were relational concerns. The influence of identification suggests one method for facilitating support for policies that require people to share their resources and opportunities.

Finally, the relationship between self-esteem, respect and identification with the advantaged group suggests evidence for the Group Value model argument that identification shapes people’s sense of themselves as well as their views of resource interdependencies. As predicted, there was a significantly closer relationship between feelings of self-worth and feelings of respect by members of the advantaged group for respondents who identified more closely with their advantaged group than for respondents who identified with Americans in general. If identification only changed people’s views of resource interdependencies, whether or not they valued the source of respectful treatment should not influence the relationship between self-esteem and feeling respected so clearly.

Together, these results provide important evidence for the argument that if people identify more closely with a superordinate category that includes both themselves and relevant policymakers, they will be less interested in distributive justice and the instrumental benefits of a particular policy and more interested in procedural justice and relational evaluations of the policymakers. In contrast to more traditional

186 H . J . Smith and T. R . Tyler

economic models, these results show when and why people might be willing to support policies out of a sense of fairness rather than out of personal or group self- interest.

BTUDY 2 THE DISADVANTAGED

The purpose of the second study is to test the same four questions explored in the first study, but from the perspective of the disadvantaged rather than the advantaged. Since the disadvantaged lack resources and opportunities, their willingness to give them up is not an adequate test of justice motivations and identification. Instead, the focus of this study is on feelings of obligation to obey social rules. This focus is important because decreasing inequalities can awaken feelings of entitlement among the disadvantaged (Davies, 1962; Gurr, 1970. Sears and McConahay, 1970). Traditional relative deprivation models propose that as the resources and opportunities available for the disadvantaged increase, so will their expectations (Davies, 1962; Pettigrew, 1985). If the available resources do not keep pace with their rising expectations, the disadvantaged will be extremely motivated to participate in collective violence and rebellion. Furthermore, an economic view suggests that as the disadvantaged gain greater power, they will use their power in self-interested ways. In addition to their motive to gain resources and opportunities, they will increasingly have the means. In such a situation, the concerns of the advantaged about how their interests will be defended as they lose resources appear well-founded and resistance to resource redistribution seems appropriate. However, if the disadvantaged are willing to defer to their feelings of obligation to obey laws, the advantaged can have greater confidence that they can redistribute resources without creating risks to themselves. Redistribution can occur within a framework of law and government which protects the rights and interests of all groups.

The first goal of the second study is to determine whether evaluations of procedural and distributive justice are related to feelings of obligation and legitimacy. The second goal is to compare instrumental evaluations of Congress with relational evaluations of Congress as predictors of obligation and legitimacy. The third question is to explore whether identification with the superordinate category of Americans in general influences the relative importance of relational and instrumental concerns to feelings of obligation to obey the law. To test whether identification influences the relative importance of instrumental and relational concerns because it changes how people think of themselves and their relationships to other category members, the influence of identification on self-esteem also will be explored.

Method

Respondents

Random digit telephone sampling was used to contact San Francisco Bay area residents between January and June of 1992. The sample was restricted to English-

Justice and power 187

speaking adults who identified themselves ethnically as African-American and produced 45 interviews. In order to obtain an adequate African-American sample, telephone numbers in calling areas with relatively large African-American populations (Oakland and Richmond) were oversampled, yielding an additional 105 respondents. The second oversample included significantly more female respondents than the first sample, but there were no significant differences between the two samples for income level, education level, politics or any of the variables included in the study.

The mean age of respondents was 41 (standard deviation= 15.52 years). Forty-five per cent were male, 55 per cent were female. Thirty-five per cent of respondents were high school graduates; and 10 per cent had some post-college education. Twenty- eight per cent of the respondents had a household income of less than 20 OOO$; 30 per cent had an income of 20001 to 40000$; 29 per cent had an income 40001 to 60000$; and 13 per cent had an income over 60001$. Further details about the sampling design and the demographic characteristics of the sample are available from the authors.

Measures

Summary statistics and intercorrelations for all the variables included in this study are presented in Table 4.

Legitmacy and obligation A single measure of respondents’ feelings of obligation and legitimacy was created (Cronbach’s alpha = 0.66). Respondents rated how positively they felt toward Congress and the United States Supreme Court, and whether they respected Congress as a government institution on a scale from 0 to 10 with 10 representing extremely favourable or warm feelings and 0 representing extremely unfavourable or cold feelings. The other four items measured respondents’ feelings of obligation to obey federal authorities. The first question asked whether decisions made by the federal government should be accepted even when people disagree. The second question asked whether people should obey the laws made by Congress even if they do not believe the law is right. The third question (reversed scored) asked whether there are situations in which it is all right for people to disobey the government and the last question (reversed scored) asked whether the respondents could think of situations in which they would stop supporting government policies. Responses could range from agree strongly (1) to disagree strongly (7). Because the answers for these last two items were highly skewed, a square root transformation was used to normalize the distributions before they were combined with the other five questions.

Measures of procedural and distributive justice As in Study 1, two difference scores based on respondents’ ratings of the pay rates and economic situations for Whites and African-Americans were combined to measure distributive justice (Cronbach’s alpha = 0.54). Almost all the respondents viewed the disadvantaged as having fewer deserved outcomes than the advantaged ( M = 5.32, S.D. = 0.86). The relatively low alpha reflects this consensus. The measure of procedural justice also included the same three items used in the first study (Cronbach’s alpha = 0.81). Because the answers for these items were all highly skewed, a square root

Tabl

e 4.

Var

iabl

e M

S.

D.

1 2

3 4

5 6

7 8

9 10

11

12

13

Inte

rcor

rela

tions

and

sum

mar

y st

atis

tics - St

udy

2

1. O

blig

atio

n an

d 3.

26

0.88

-

legi

timac

y 2.

Pro

cedu

ral j

ustic

e 3.

47

0.60

0.

21*

-

3.

Dis

trib

utiv

e jus

tice

5.32

0.

86

0.16

* 0.

35**

-

4.

Rel

atio

nal

4.02

1.

04

0.59

**

0.22

**

0.14

-

5. I

nstr

umen

tal

4.54

1.

31

0.30

**

0.11

0.

14

0.49

** -

6. R

elat

iona

l con

cern

s 3.

92

0.89

0.

59**

0.

33**

0.

17*

0.91

**

0.46

** -

7. I

nstr

umen

tal

4.86

0.

91

0.33

**

0.23

**

0.50

**

0.47

**

0.93

**

0.48

** -

8. Id

entif

icat

ion

with

6.

55

2.27

-0

.38*

* -0

.16

-0.2

0*

-0.3

5**

-0.2

8**

-0.3

6**

-0.3

3**

-

9. I

d w

ith U

.S.x

re1.

-0

.36

1.02

0.

12

0.23

-0

.01

0.10

-0

.06

0.13

-0

.06

-0.0

1 -

10. I

d w

ith U

.S. x

re

l. -0

.32

1.10

-0

.06

0.08

-0

.14

-0.0

7 -0

.09

-0.0

5 -0

.13

0.03

0.

48**

-

11.

Iden

tific

atio

n w

ith

1.81

1.

17

-0.0

3 -0

.20*

-0

.23*

* -0

.08

-0.0

3 -0

.10

-0.1

0 0.

05

-0.0

2 0.

09

-

12.

Res

pect

by

othe

r 2.

18

1.43

0.

13

-0.0

2 -0

.07

-0.0

1 0.

02

-0.0

1 0.

00

0.00

0.

09

-0.0

4 0.

34**

-

13.

Id w

ith d

isad

v x

0.34

1.

69

-0.1

7 -0

.07

-0.0

9 0.

21**

-0.

16

-0.2

1**

-0.1

5 0.

15

0.10

0.

02

0.47

**

0.39

**

14.

Self-

este

em

1.70

1.

04

-0.1

4 -0

.02

0.09

0.

04

0.07

-0

.03

0.03

0.

18*

0.05

0.

44**

0.

06

-0.0

2

eval

uatio

ns

eval

uatio

ns

conc

erns

Am

eric

ans

cnrn

s

cnrn

s

disa

dvan

tage

d

disa

dvan

tage

d

resp

ect

Not

e. N

= 15

0. E

ntrie

s ar

e Pe

arso

n co

rrel

atio

ns. L

ow s

core

s ind

icat

e le

ss p

olic

y en

dors

emen

t, m

ore

prec

edur

al justice, m

ore

dist

ribut

ive

just

ice,

gre

ater

inst

rum

enta

l be

nefit

s, m

ore

resp

ectfu

l Con

gres

sion

al tr

eatm

ent,

less i

dent

ifica

tion w

ith A

mer

ican

s in

gene

ral, m

ore i

dent

ifica

tion w

ith th

e dis

adva

ntag

ed, m

ore

resp

ect a

nd h

ighe

r sel

f- es

teem

. *p<

0.05

, **p

< 0.

01, *

**p<

O.O

Ol.

a &

Justice andpower 189

transformation was used to normalize the distributions before they were combined into a single index.

Measures of relational and instrumental evaluations of Congress As in Study 1 , five items asking respondents how they believed Congress treated its constituents were combined to measure relational evaluations of Congress (Cronbach’s alpha = 0.63). The first question (reversed scored) asked whether Congressional representatives are dishonest. The second question asked whether Congress is honest in the way it makes decisions. The third question asked whether Congress obtains the information necessary for informed decisions. The fifth question asked whether Congress gives equal consideration to the views of all the different groups in America. Responses could range from agree strongly (1) to disagree strongly (7).

Three items were combined to measure instrumental evaluations of Congress (Cronbach’s alpha = 0.61). The first question asked whether Congressional representatives try to be fair to the people in their district and not just to special interest groups. The second question asked whether the decisions made by Congress generally favour people like the respondent. The third question asked whether the views of average citizens influence the decisions made by Congress. Responses could range from agree strongly (1) to disagree strongly (7).

Identification with one’s dhadvantagedgroup and Americans in general AS in Study 1, agreement with two items was used to measure respondents’ identification with their disadvantagedgroup (Cronbach’s alpha = 0.35). Respondents’ ratings of Americans as a group were used to measure respondent’s identification with Americans in general.

Feelings of respect and self-esteem The same three items used in Study 1 were used to measure respondent’s evaluations of whether most members of their disadvantaged group respected them (Cronbach’s alpha = 0.77). The same three items from the Rosenberg self-esteem scale used in Study I were used to measure respondents’ feelings of self-esteem (Cronbach’s alpha = 0.75).

Results

Overall perceptions of justice

As might be expected, if respondents’ evaluations of the economic situation for the disadvantaged are compared with their evaluations of the economic situation for the advantaged, they view the situation for the disadvantaged as significantly less fair and satisfactory (t(l50)= 11.56, p < 0.0001). A majority of respondents also agreed that the disadvantaged do not have the same opportunities to succeed (63. 3 per cent), do not have equal chances to succeed (71.3 per cent) and will face racial discrimination (94.0 per cent). Finally, a significant proportion of respondents reported that it was okay to disobey the government (39.2 per cent) and that there were situations in which they would stop supporting government policies (58.0 per cent). The fact that a large number of respondents did not feel obligated to obey federal authorities shows that support and deference to government and law is far from automatic for the disadvantaged.

190 H . J . Smith and T. R. Tyler

Procedural justice, distributive justice and obligation to obey

Multiple regression analyses were used to test whether evaluations of procedural justice, distributive justice or both are related to feelings of obligation4. When both procedural and distributive justice judgements were entered simultaneously, a significant amount of the variance in feelings of obligation was explained (adjusted R-squared = 4.0 per cent, p < 0.05). However, only procedurnal justice judgements were significantly related to feelings of obligation (beta =0.17, p < 0.05, versus beta = 0.1 1, as.). However, a regression equation that included just distributive justice judgement explained a significant amount of variance (adjusted R- squared = 2.1 per cent, p < 0.05), as did a regression equation that included just procedural justice evaluations (adjusted R-squared = 3.6 per cent, p < 0.05).

Instrumental and relational evaluations of Congress

The significant contributions of procedural justice to feelings of obligation suggests the potential importance of a relational model of justice. Therefore, multiple regression analyses were used to test whether relational evaluations, instrumental evaluations or both were related to feelings of obligation to obey the law. When both instrumental and relational evaluations of Congress were entered simultaneously, a significant amount of the variance in feelings of obligation was explained (adjusted R-squared = 33.6 per cent, p < 0.001). However, only relational evaluations were significantly related to feelings of obligation (beta = 0.58, p < 0,001, versus beta = 0.02, n.s.). A separate regression equation that included just instrumental evaluations explained a significant amount of variance (adjusted R-squared = 8.5 per cent, p < 0.001), as did a regression equation that included just relational evaluations (adjusted R-squared = 34 per cent, p < 0.001).

Superordinate categories and feelings of obligation

The contribution of both distributive and procedural justice, instrumental and relational evaluations of Congress to feelings of obligation suggests that it is worthwhile to explore when either type of justice or Congressional evaluation might be more important to government support. The results from the first study suggest that if respondents identify with Americans in general, they will be less influenced by instrumental concerns and more sensitive to relational issues when they evaluate their feelings of obligation to obey the law.

Confirmatory factor analysis As in the first study, evaluations of procedural justice and relational evaluations of Congress were combined to create a single index of relational concerns (Cronbach's alpha = 0.70) and evaluations of distributive justice and instrumental evaluations of Congress were combined to create a single index of instrumental concerns (Cronbach's alpha = 0.5 1). Confirmatory factor analysis was conducted to test whether relational judgements; judgements about procedural

41nitial regression analyses that included only demographic variables explained 1 per cent of the variance in federal government support. None of the demographic variables were significant.

Justice andpower 191

justice and the neutrality, honesty and trustworthiness of Congress, could be distinguished from instrumental judgements; judgements about distributive justice and the effectiveness of Congressional polices. Covariance matrices were computed using listwise deletion of missing data and analysed with the maximum likelihood method of estimation LISREL-8 (Joreskog and Sorbom, 1993). An examination of the goodness of fit indices for the two-factor solution supports a distinction between relational and instrumental judgements. The two-factor solution yielded a non- significant chi square ( ~ ~ ( 4 5 ) = 30.43, n.s.), and values on the Goodness of Fit Index of 0.96 and the Adjusted Goodness of Fit index of 0.95. Both values are above 0.90, a convention generally reflecting a good fit to the data (Pedhazur and Schmelkin, 1991).

To test whether identification with Americans influenced the relative importance of relational and instrumental concerns to feelings of obligation, two interaction terms were created. The first term tested whether identification with Americans moderated the relationship between relational concerns and feelings of obligation. The second term tested whether identification with Americans moderated the relationship between instrumental concerns and feelings of obligation.

As shown in Table 5 , the regression coefficients for both interaction terms were significant. As in the first study, a median split of the distribution was used to classify respondents as identifying more or less closely with Americans in general (median = 7.00). Feelings of obligation were not influenced by instrumental concerns for respondents who identified with Americans (r(80) = 0.06, n.s.) but instrumental concerns were significantly related to feelings of obligation for respondents who did not identify with Americans (r(68) = 0.33, p < 0.01). In contrast, feelings of obligation were more closely related to relational concerns for respondents who identified with Americans (r(80) = 0.42, p < 0.01) then for respondents who did not identify with Americans (r(68) = 0.29, p < 0.01).

To further explore the relationship between identification and instrumental or relational concerns, a second median split was used to classify respondents into the same four groups used in the subgroup analysis described in Study 1 (median (identification with disadvantaged) = 1 .OO). As in Study 1, the focus will be on the three most theoretically interesting groups; (1) those respondents who reported

Table 5. Study 2

The relationship between identification, justice evaluations and obligation to obey-

Obligation to obey (n = 147) Regression without Regression with interaction terms interaction terms

Relational evaluations (B) 0.51*** 0.53 ** *

Instrumental evaluations (C) 0.02 0.03

A x C - 0.53

Identification with Americans (A) -0.19** -0.18**

A x B - -0.53

Adjusted RZ 0.37*** 0.38***

Note. Unless otherwise noted, the entries are standardized regression coefficients for an equation in which all variables are entered simultaneously. *p <0.05, **p <0.01, ***p < 0.001.

192 H. J. Smith and T. R. Tyler

Table 6. Relative identification, justice evaluations and obligation to obey -Study 2

Obligation to obey

More identification with Americansa Relational concerns Instrumental concerns

Adjusted R2 Equal identification with Americans and the disadvantagedb

Relational concerns Instrumental concerns

Adjusted R2 More identification with the disadvantagedC

Relational concerns Instrumental concerns

Adjusted R2

0.62*** - 0.65*** - 0.01 0.12

0.37*** 0.00 0.37***

0.53*** - 0.55**

0.26*** 0.01 0.24** 0.17 -0.05 -

0.27 0.38* - - 0.36* 0.23

0.12* 0.10* 0.13* ~ ~~

Note. Unless otherwise noted, the entries are standardized regression coefficients for an equation in which all variables are entered simultaneously. *p (0.05, **p < 0.01, ***p<O.OOl. aRespondents reported greater identification with Americans in general and less identification with the disadvantaged (n = 35). bRespondents reported greater identification with the disadvantaged and greater identification with Americans in general (n = 43). ‘Respondents reported greater identification with the disadvantaged and less identification with Americans in general (n = 34).

relatively greater warmth toward Americans in general but relatively less pride in their disadvantaged group (more identification with Americans in general), (2) those respondents who reported both warmth towards Americans in general and pride in their disadvantaged group (equal identification with Americans in general and the disadvantaged) and (3) those respondents who reported relatively less warmth toward Americans in general and relatively greater pride in their disadvantaged group (more identification with the disadvantaged). Respondents who reported not feeling close to either their disadvantaged group or to Americans in general were excluded from the analysis (n = 35).

As shown in Table 6 , relational concerns were significantly related to feelings of obligation for those respondents who identified more with Americans in general while instrumental concerns were not important (beta = 0.65, p < 0.001 versus beta = 0.12, n.s.). Similarly, relational concerns were more closely related to policy endorsement than instrumental concerns for respondents who reported relatively equal identification with both Americans in general and the disadvantaged (beta = 0.55, p < 0.001 versus beta = 0.05 n.s.). However, for respondents who identified more strongly with their disadvantaged group, both instrumental concerns and relational concerns appear equally important (beta = 0.27, n.s. versus beta = 0.23, n.s.). Although the regression coefficients were not significant when both instrumental and relational concerns were entered simultaneously, a separate regression equation with just relational concerns included explained a significant amount of the variance (adjusted R-squared=12, p<O.O5) as did a regression equation with just instrumental concerns (adjusted R-squared = 10 per cent, p < 0.05).

Justice and power 193

Relationship between self-esteem and feelings of respect

We hypothesize that if people do not value a particular group membership, the relational information communicated by group authorities or other group members should be relatively unimportant to their sense of self-worth. As an indirect test of this argument, we can compare the relative importance of feeling respected by other members of one’s disadvantaged group to feelings of self-worth for respondents who identify relatively more closely with their disadvantaged group and respondents who identify relatively less strongly. The first regression equation included both identification with the disadvantaged group and respect and accounted for a significant amount of variance in feelings of self-worth (adjusted R-squared = 19 per cent, p < 0.001). However, feelings of respect appeared to be more important than identification with the disadvantaged group (beta = 0.47, p c 0.001, versus beta = - 0.10, n.s.). A second regression equation including an interaction term between respect and identification with the disadvantaged significantly increased the variance explained for self-esteem (change in R-squared = 3.6 per cent, F(3, 145) = 6.38, p < 0.05). Feelings of general respect from the disadvantaged group were more closely related to feelings of self-esteem for respondents who identified with the disadvantaged (r(70) = 0.56, p < 0.001) than for respondents who did not (r(80) = 0.32, n.s.). Furthermore, a comparison between two subgroups, (1) those who identified more with the disadvantaged group than with Americans in general and (2) those who identified more with Americans in general than with their disadvantaged group also support the predicted pattern. Self-esteem was significantly more closely related to respect for respondents who identified more with the disadvantaged compared to Americans in general (r(35) = 0.66, p < 0.001) than for respondents who identified more with Americans compared to the disadvantaged (r(36) = 0.25, n.s.).

Discussion

The results from this study show how justice motivations and social categorizations can influence reactions to economic inequalities for the disadvantaged as well as the advantaged. First, procedural justice evaluations rather than distributive justice evaluations were significantly related to feelings of obligation to obey the law. This pattern of results replicates previous research that shows procedural justice is a stronger influence on support for social rules and authorities than is distributive justice (Lind and Tyler, 1988). On the other hand, distributive justice may be a more important influence on pro-active attitudes and behaviour such as supporting the redistribution of resources as in the first study than on reactive attitudes and behaviour such as obeying the law. Similarly, relational evaluations of Congress rather than instrumental evaluations of Congress proved to be closely related to feelings of obligation to obey the law. Even though respondents clearly recognized their disadvantages, their feelings of obligation were linked more closely to respectful treatment by Congress than to instrumental issues.

Finally, identification shaped the relative importance of instrumental and relational concerns to feelings of obligation for the disadvantaged in the same way that identification shaped the relative importance of instrumental and relational

194 H. J. Smith and T. R . Tyler

concerns to policy endorsement for the advantaged. Instrumental evaluations of Congressional policies and policymakers were more closely related to feelings of obligation for respondents who identified more closely with their disadvantaged group, while relational evaluations of Congressional policies and policymakers were more closely related to feelings of obligation for respondents who identified more closely with Americans in general. The significantly closer relationship between self- esteem and feelings of respect from members of the disadvantaged among those respondents who identified more with the disadvantaged compared to those respondents who identified more with Americans further supports the argument that identification does not just shape people’s self-interests, it also shapes their self- images.

The influence of identification on the relative importance of relational and instrumental concerns suggests that just as identification with a superordinate category can facilitate the sharing of resources by the advantaged, it can also encourage support and deference to federal authorities by the disadvantaged. The consistent pattern of responses across both advantaged and disadvantaged respondents suggests that how people define the relationship between themselves, possible outgroup members and superordinate authorities appears to be a much more potent influence on policy endorsement and feelings of obligation to obey the law than their objective position in the larger society.

GENERAL DISCUSSION

The results from both studies suggest that justice is a motive that can be independent of personal or group interest and power. Although evaluations of distributive justice and instrumental evaluations of Congress were significantly related to political attitudes in both studies, procedural justice and relational evaluations of authorities proved to be as strong an influence on political attitudes. The significant relationship between relational evaluations on Congress, in particular, and political attitudes demonstrates that issues of fairness are not always the direct consequences of self- interest and instrumental concerns, but can be important for relational reasons as well (Tyler, 1994).

During the 1970s, a number of social psychological studies focused on issues of resource scarcity under the general rubric of the study of social dilemmas. Those studies demonstrated that people are more likely to act in the interests of their community when they identify with that community. Such effects were variously interpreted as reflecting either the operation of instrumental judgements, in particular, changes in judgements about the likelihood that others in the community would reciprocate, or ‘pure’ identity effects linked to emotional attachments to the community. Recently, Tyler and Degoey (1994) tested these models in the context of a social dilemma and demonstrated that ‘pure’ identity effects do occur.

While the type of interdependence issue represented by social dilemmas presents groups with difficulties in social coordination, group efforts to solve such problems are aided by individual’s awareness that it is in their own, as well as in the group’s, interests for resources to be conserved. If they are not, both the individual and the

Justice andpower 195

group suffer. Once there is no water, both the advantaged and the disadvantaged will die. The focus of our concerns is with a more difficult type of interdependence issue, which is becoming increasingly important in the 1990s. That problem is confict among socio-economic, age and ethnic groups over access to shared societal resources. This problem poses a particularly difficult challenge because the self- interest of the advantaged does not necessarily lie in the redistribution of resources. Further, these conflicts are facilitated by the salience of subordinate identifications along ethnic, gender, age and economic lines, identifications which enable people to more easily dismiss the superordinate identity they all share as members of a common society.

If, as we argue, an interest in maintaining positive relationships to important groups can promote an interest in justice that does not depend on self-interest and instrumental rewards, it becomes especially important to determine when relational concerns might outweigh instrumental concerns. Across both studies, the relative importance of instrumental and relational concerns to political attitudes was related to whether people valued their particular advantaged or disadvantaged group or the superordinate category more positively. Instrumental concerns proved to be important to political attitudes if people identified with their particular advantaged or disadvantaged group, but if people identified with a superordinate category that included both potential outgroup members and the relevant authorities, relational concerns were more closely related to political attitudes. The fact that identification influenced the relative importance of relational and instrumental concerns in the same way for both the advantaged and disadvantaged suggests how valued superordinate categorizations can transcend local group interests regardless of their initial social position.

However, because these data are correlational, any causal inferences should be made cautiously, if at all. An important goal for future research will be experimental tests of the relationship between identification, justice motivations and collectively oriented behaviours. While all the causal relationships that are implied by our argument have not been tested experimentally within the same study, there is evidence that supports individual causal connections. First, experimental research shows that when a common group membership is made salient, people will behave in that group’s interests. In intergroup contexts, when the salience of a disadvantaged group membership is made salient, people are more likely to notice, resent and protest distributive inequities between the advantaged and the disadvantaged (Kawakami and Dion, 1993; LaLonde and Silverman, 1994; Smith, Spears and Oyen, 1994). In intragroup contexts, when a common group or category is made salient, cooperative responses to social dilemmas are increased while conflict and discrimination between groups included in the larger categorization are decreased (Brewer and Kramer, 1986; Gaertner et al., 1993). Given the strong influence of group salience on subsequent behaviour and attitudes documented in this experimental research, it seems extremely plausible to assume that identification with particular groups or categories moderates the relationship between justice motivations and political attitudes. Second, both experimental and longitudinal research demonstrate that evaluations of procedural and distributive justice influence support for both authorities and specific policies (Tyler and Lind, 1992). Therefore, proposing that issues of justice shape political attitudes rather than the reverse also appears a safe and reasonable assumption.

196 H . J . Smith and T. R . Tyler

It is also important to recognize that this is an investigation of attitudes rather than behaviour. We do not mean to imply that people’s abstract policy judgements can be translated directly into particular types of behaviour. One question for future research is to determine whether identification and justice concerns are related to actual behaviour in the same way that we show they are related to judgements of policy. However, there is a long history of research that shows political attitudes, do indeed, predict political behaviour, suggesting that this is a plausible extension (Tyler, 1989).

The promise of superordinate categories

The suggestion that superordinate categorizations can promote group-oriented attitudes and behaviours is not new. Both resource and identification models of intergroup relations have proposed superordinate goals or categories as a method for reducing intergroup conflict and promoting cooperation (Sherif et al., 1961; Brewer and Kramer, 1986; Gaertner et al., 1993; Turner, Hogg, Oakes, Reicher and Wetherell, 1987). However, most previous research, from the Robber’s Cave studies to social dilemma experiments, has focused on instrumental issues; how superordinate categories change the way in which people evaluate resource interdependencies. In contrast, this research suggests the importance of relational issues; how superordinate categories can change the way people view themselves and their relationship to both category authorities and other groups now included within the categorization.

This distinction between relational and instrumental evaluations suggests why differences in identification influences policy and government support. When unfairly treated groups and relevant authorities are included within important categorizations, relational concerns become more important than instrumental concerns. How fairly and respecfully all category members are treated can carry implications about one’s personal self-worth, while protecting or gaining personal and group benefits may become less critical. However, identification with one’s particular ethnic group may exclude authorities (and the potential beneficiaries of the policy) from valued groups. Therefore, the relational implications of fair and respectful treatment by superordinate authorities and other outgroup members should be viewed as less personally relevant and group-or self-interest should drive evaluations of political policies and authorities. Evidence for this second possibility is suggested not only by our results but also by other investigations of attitudes toward affirmative action. For example, White respondents who reported feeling ‘politically alienated’ or overlooked by the federal government (perhaps viewing the federal government as part of an outgroup) held significantly more negative attitudes toward affirmative action than White respondents who did not (Kluegel and Smith, 1986).

One might argue that identification with superordinate categories facilitates pro- social attitudes because instrumental concerns are now redefined to represent costs and benefits at the collective level rather than at the group or personal level. While this interpretation is entirely compatible with our approach, it misses the important insights a focus on relational issues offers. First, limiting oneself to an instrumental understanding of group-oriented attitudes and behaviours makes it difficult to

Justice and power 197

explain why relational aspects of justice, regardless of identification level, are related to policy endorsement or feelings of obligation to obey the law. Second, it is difficult to explain from an instrumental perspective why identification with a group shapes the relative importance of feeling respected by the other group members to feelings of self-worth. The relational emphasis on self-evaluations suggests that members of a superordinate category might choose to focus on their particular group membership within the category to a greater extent. If people feel that superordinate authorities do not respect them, then to protect or enhance their positive self-image they might emphasize or substitute other groups as a source of information about personal status and worth (Tyler et al., 1994). While this is a question for future research, experimental research shows that the self-esteem of people who are the targets of prejudice and discrimination is much Iess likely to suffer if they attribute their treatment to their group membership and not to a personal quality (Crocker and Major, 1989; Dion, 1986).

Conclusion

Traditionally advantaged Americans have encouraged the disadvantaged to improve their situations by ‘creating new wealth’, a solution to social inequities which has the benefit of requiring little or no sacrifice on the part of the advantaged. However, there is no longer an underdeveloped American frontier and American standards of living are no longer rising, so creating new wealth is becoming an unrealistic avenue to economic improvement. In fact, many observers predict a future era of minimal economic growth, and perhaps even economic shrinkage, as the United States moves into an era of global economic competition. These changing social realities require increased attention to the possibilities for the ‘redistribution’ of economic wealth as a solution to problems of inequity.

Self-interest models predict that the advantaged will resist redistribution, using their power to defend their advantages. One example of such an effort in California is the passage of proposition 13, a political referendum designed to preserve existing economic advantages by limiting the power of government to tax economic assets. This initiative suggests one possible scenario for the social issues which will define the United States for the rest of the 20th century: a series of self-interested actions by powerful political constituencies such as the elderly, the well-off and others. Conversely, the same self-interest models suggest that the disadvantaged will use their own growing political power to gain control for themselves, and will then pass their own set of self-interested policies. As Richard Zeiger, the editor of the California Journal says ‘Eventually the have-nots are going to win and seize control of the political mechanisms. I would not want to be an old white person in California 30 years from now’ (Zeiger in Reeves, 1994).

This paper suggests the possibility of a less pessimistic vision of the future based on the demonstration that both the advantaged and the disadvantaged are motivated to act in ways consistent with their judgements about: (1) what is fair and (2) what aids the members of social groups with which they identify. The findings suggest that the advantaged will support redistributive policies designed to aid the disadvantaged if they judge policies and policymakers to be relationally fair. Further, they will be concerned about the welfare of disadvantaged others if they include the

198 H. J . Smith and T. R. Tyler

disadvantaged as members of their self-defined social groups. Coversely, the disadvantaged will respect the rule of law and the policies of government, if they regard those policies as consistent with principles of relational fairness. Further, they will be concerned about the welfare of all the members of groups with which they identify -including the advantaged.

REFERENCES

Azzi, A. (1994). From competitive interests, perceived injustice and identity needs to collective action: Psychological mechanisms in ethnic nationalism. In B. Kapferer (Ed.), Nationalism, ethnicity and violence. Oxford: Oxford University Press.

Brewer, M. B., & Kramer, R. M. (1986). Choice behavior in social dilemmas: Effects of social identity, group size and decision framing. Journal of Personality and Social Psychology, 50: 543-549.

Cook, K. S., Lk Hegtvedt, K. A. (1986). Justice and Power: An exchange analysis. In H. W. Bierhoff, R. L. Cohen and J. Greenberg (Eds), Justice in social relations. New York: Plenum.

Crawford, T. J., & Naditch, M. (1970). Relative deprivation, powerlessness and militancy: the psychology of protest. Psychiatry, 33: 208-233.

Crocker, J. and Major, B. (1989). Social stigma and self-esteem: the self protective properties of stigma. Psychological Review, 9 6 608-630.

Davies, J. C. (1962). Toward a theory of revolution. American Sociological Review, 27: 5-19. Deutsch, M. (1985). Distributive justice: A social-psychological persepctive. New Haven: Yale

University Press. Dion, K. L. (1986). Responses to perceived discrimination and relative deprivation. In J. M.

Olson, C. P. Herman and M. P. Zanna (Eds), Relative deprivation and social comparisons: the Ontario symposium. Hillsdale: Lawrence Erlbaum Associates.

Dube, L., & Guimond, S. (1986). Relative deprivation and social protest: the personal group issue. In J. Olson, C. P. Herman and M. Zanna (Eds), Relative deprivation and social comparison. Hillsdale, New Jersey: Lawrence Erlbaum Associates.

Folger, R. (1 987). Reformulating the preconditions of resentments: a referent cognitions model. In J. Master and W. Smith (Eds), Social comparison, social justice and relative deprivation. Hillsdale, New Jersey: Lawrence Erlbaum Associates.

Gaertner, S. L., Dovidio, J. F., Anastasio, P. A., Bachman, B. A,, & Rust, M. C. (1993). The common ingroup identity model: recategorization and the reduction of intergroup bias. In W. Stroebe and M. Hewstone (Eds), The european review of social psychology, (Vol.4

Greenberg, J. (1990). Employee theft as a reaction to underpayment inequity: The hidden cost

Gurr, T. R. (1970). Why men rebel. Princeton: Princeton University Press. Hafer, C. L., & Olson, J. M. (1993). Beliefs in a just world, discontent and assertive actions by

working women. Personality and Social Psychology Bulletin, 19: 30-38. Hoffman, E., & Spitzer, M. L. (1982). The Coase theorem: Some experimental tests. Journal of

Law and Economics, 25: 73-98. Hoffman, E., & Spitzer, M. L. (1985). Entitlements, rights and fairness: An experimental

examination of subjects’ concepts of distributive justice. Journal of Legal Studies, 14:

Homans, G. C. (1961). Social behavior: its elementary forms. New York Harcourt, Brace, Jovanovich.

Huo, Y. J. Smith, H. J., Tyler, T. R., & Lind, E. A. (1994). The effects of relative social identification on justice concerns: Is assimilation necessary to maintain cohesion amid diversity? Unpublished manuscript, University of California, Berkeley.

Joreskog, K. G., & Sorbom, D. (1993). LZSREL 8. Chicago, I L Scientific Software International.

pp. 1-26).

of pay cuts. Journal of Applied Psychology, 75: 561-568.

259-297.

Justice and power 199

Kabanoff, B. (1991). Equity, equality, power and conflict. Academy of Mangement Review, 16:

Kahneman, D., Knetsch, J. L., & Thaler, R. (1986). Fairness and the assumptions of

Kawakami, K., & Dion, K. L. (1993). The impact of salient self-identities on

416-441.

economics. Journal of Business, 59 (suppl.): 285-300.

relative deprivation and action intensions. European- Journal of Social Psychology, 23: 52 5540.

Kluegel, J. R., & Smith, E. R. (1986). Beliefs about inequality: Americans’ views of what is and what ought to be. New York: Adegruyter.

Kramer, R. M. (1991). Intergroup relations and organizational dilemmas: The role of categorization processes. Research in Organizational Behavior, 13: 191-228.

LaLonde, R. N., & Silverman, R. A. (1994). Behavioral preferences in response to social injustice: The effects of group permeability and social identity salience. Journal of Personality and Social Psychology, 68: 78-85.

Lind, E. A., Kulik, C. T., Ambrose, M., & de Vera Park, M. V. (1993). Individual and corporate dispute resolution: using precedural fairness as a decision heuristic. Administrative Science Quaterly, 38: 224-25 1.

Lind, E. A., Kanfer, R., & Earley, P. C. (1990). Voice, control and procedural justice: instrumental and noinstrumental concerns in fairness judgments. Journal of Personality and Social Psychology, 59: 952-959.

Lind, E. A., & Tyler, T. R. (1988). The social psychology of procedural justice. New York: Plenum Press.

Martin, J. (1986). The tolerance of injustice. In J. Olson, C. P. Herman and M. Zanna (Eds), Relative deprivation and social comparison: the Ontario symposium. Hillsdale: Lawrence Erlbaum Associates.

Montada, L. (1992). Attribution of responsibility for losses and perceived injustice. In L. Montada, S. Filipp and M. Lerner (Eds), Life crises and experiences of loss in adulthood. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.

Montada, L., & Schneider, A. (1989). Justice and emotional reactions to the disadvantaged. Social Justice Research, 3: 313-344.

Moore, D. (1991). Entitlement and justice evaluations: who should get more and why? Social Psychology Quarterly, 54: 208-223.

Optow, S. (1990). Moral exclusion and injustice: and introduction. Journal of Social Issues, 46:

Pedhazur, E., & Schmelkin, L. P. (1991). Measurement, design and analysis: an integrated

Pettigrew, T. F. (1985). Three issues in ethnicity: boundaries, deprivations and perceptions. In

Reeves, R. (1994, January 23). The tax revolt that ruined California. The Sun Francisco

Rosenberg, M. (1979). Conceiving the self. New York: Basic Books. Sears, D., & McConahay, J. (1970). Comparison levels and the Watts riots. Journal of Social

Issues, 26: 121-140. Sherif, M., Harvey, 0. J., White, B. J. Hood, W. R., & Sherif, C. W. (1961). Intergroup conflict

and cooperation: The robbers cave experiment. Norman: The University of Oklahoma Book Exchange.

Smith, H. J., Spears, R., & Oyen, M. (1994). The influence of personal deprivation and salience of group membership on justice evaluations. Journal of Experimental Social

Tajfel, H . (1981). Human groups and social categories: studies in socialpsychology. Cambridge:

Tajfel, H., & Turner, J. (1986). The social identity theory of intergroup behavior. In S.

Taylor, D. M., & Moghaddam, F. M . (1987). Theories of intergroup relations: International

Thibaut, J., & Walker, L. (1975). Procedural justice: a psychological analysis. Hillsdale, N . J.:

1-29.

approach. Hillsdale: New Jersey: Lawrence Erlbaum Associates.

J. M. Yinger and S. J. Cutler (Eds), Major social issues. New York: Free Press.

Chronicle pp. 5, 8, 9, 10.

Psychology, 30: 277-299.

Cambridge University Press.

Worchel, (Ed.), Psychology of intergroup relations. Chicago: Nelson Hall.

social psychological perspectives, New York: Praeger Publishers.

Erlbaum.

200 H. J . Smith and T. R . Tyler

Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S., & Wetherell, M. S. (1987). Rediscovering

Tyler, T. R. (1989). The psychology of procedural justice: a test of the group-value model.

Tyler, T. R. (1990). Why people obey the law. New Haven: Yale University Press. Tyler, T. R. (1993). Governing amid diversity: Can fair decision-making procedures bridge

across competing public interests? Unpublished manuscript, University of California, Berkeley.

Tyler, T. R. (1994). Psychological models of the justice motive: The antecedents of distributive justice and procedural justice. Journal of Personality and Social Psychology, 67: 850-863.

Tyler, T. R., & Dawes, R. (1993). Fairness in groups: Comparing the self-interest and social identity perspectives. In B. Mellers and J. Baron (Eds), Psychologicalperspectives on justice: theory and application. Cambridge: Cambridge University Press.

Tyler, T. R., & Degoey, P. (1994). Collective restraint in a social dilemma situation: The influence of procedural justice and community identification. Accepted pending suitable revision, Journal of Personality and Social Psychology.

Tyler, T. R., Degoey, P., & Smith, H. J. (1994). Understanding why the justice of group procedures matters: A test of the psychological dynamics of the group-value model. Unpublished manuscript, University of California, Berkeley.

Tyler, T. R., & Lind, E. A. (1988). Intrinsic versus community-based justice models: when does group membership matter? Journal of Social Issues, 46: 83-94.

Tyler, T. R., & Lind, E. A. (1992). A relational model of authority in groups. In M. Zanna (Ed.), Advances in Experimental Social Psychology, 25: 115-191.

Tyler, T. R., & Mitchell, G. (1994). Legitimacy and the empowerment of discretionary legal authority: The United States Supreme Court and abortion rights. Duke Law Journal, 43:

Veilleux, F., & Tougas, F. (1989). Male acceptance of affirmative action programs for women: the results of altruistic or egoistical motives? International Journal of Psychology, 24: 485496.

Walker, I., & Mann, L. (1987). Unemployment, relative deprivation and social protest. Personality and Social Psychology Bulletin, 13: 275-283.

Wit, A. P. (1989). Group efficiency and fairness in social dilemmas: an experimental gaming approach. Doctoral dissertation, Groningen, The Netherlands: University of Groningen.

the social group: a self-categorisation theory. Oxford: Blackwell.

Journal of Personality and Social Psychology, 57: 830-838.

703-81 5.