ideological and attributional boundaries on public compassion - reactions to individuals and...

7/25/2019 Ideological and Attributional Boundaries on Public Compassion - Reactions to Individuals and Communities Affecte

1/16

Ideological and Attributional BoundariesonPublic Compassion: Reactionsto Individualsand CommunitiesAffected byaNatural Disaster

LindaJ. SkitkaUniversity of Il li noisat Chicago

Thepresent study explored whether i deologically based attr ibu-

ti onsfor whypeopleneed publi cassistance(a) emergeeven in the

context of an external-uncontrollablecauseof need; (b) general -

izeacross different l evels of analysis, for example, across di ffer-

ent forms of assistance, aswell asacrossdi fferent typesof claim-

ants (individuals or groups); and explored (c) the role of

promised reform on will ingness to help those with personal

responsibil i ty for their plight. Ideological and attr ibut ional di f-

ferences in wi ll ingness to help emerged even in the context of a

natu ral disaster. L iberal s tended to suspend the usual conse-

quences of attri but ional anal ysis when making judgments

about humanitari an ai d by helping even thei r responsible; con-

servati ves did not. L iberals and conservatives ali keweremore

supporti ve of aid designed to meet primary than secondary

needs, ai d made cont ingent on future reform, and of help forcommun it ies over indivi duals.

The United Statesis both a democratic and a capitalis-tic society. However, the marriage between capitalism

and democracy is often tense (Dahl, 1989). On one

hand, capitalism focuses on a belief in individualism,

self-reliance, and self-determina tion. Through hard

work and ab ility, people should strive to cultivate market-

able skills an d product s of value to oth ers. In the process,

enorm ous ineq ualities can emerge. Some people ama ss

great power over oth ers (e.g., employers over employ-

ees, land lords over tena nts, creditors over debto rs). On

the other hand, our government was founded on the

democra tic ideal of eq uality of citizenship and its atten-

dant rights. Free speech, the right to vote, and (in some

peoples minds) even a right to a b asic subsistence ar e

seen as basic entitlements of citizenship that n eed not be

earned in the competitive marketplace (Okun, 1975).

These con flicting value orient atio nsind ividua lism an d

se lf -re li ance on one hand , ega l itar ian ism on the

otherlead to d ifferent positions regarding the o bliga-

tion of the collective to help the disadvan tag ed (Dion ne,1991; Mead, 1988), o r th e extent to which people are

committed to the notion of public compassion.

Support for public compa ssion, or for using collective

resources to help the less fort una te memb ers of society,

appears to depend largely on ideologically patterned

attributions for why people are likely to need govern-

ment assistance. Conservatives blame poverty on self-

indulgence and t he lack of mora l stand ard s and intelli-

gence. Liberals see the po or as victims of un just social

practices and structures. These ideological d ifferences

in attributions for poverty predict willingness to expand

social programs. Liberals generally favor, whereas con-

ser vatives oppose, increased spending on social pro-grams (Feather, 1985; Kluegel, 1990; Kluegel & Smith,

1986; Snid erm an &Tetlock, 1986; Williams, 1984).

To bett er ad dress wheth er ideolog ical differen ces in

attributional style transcend scripted attitudinal posi-

tions and instead reflect ideological differences in cogni-

tive and /or af fective style, the present stud y investigated

whether the same ideological and attributional effects

AuthorsNote:This project wassupported bya gran t from the National

Science Fou nd atio n, #SBR-9321150. Than ks to William McC read y,

Ph.D . and the POL staff at Northern Illinois Un iversity for their profes-

sional execution of the survey, and to Jennifer Winquist at the Univer-

sity of Illinois at Chicago for assistance with LISREL and AMOSanalyses. The comments of Jack Dovidio, Sherri Lantinga, Bernard

Weiner, and th ree anon ymous reviewerswere all very helpful at differ-

ent stages of thispro ject, although any remaining flawsa re the result of

causes that are internal and controllable to the author. Correspon-

dence con cerning this article should be ad dressed to Linda J. Skitka,

Ph.D ., Departmen t of Psychology (M/C 285), 1009 BSB, 1007 W. Har -

rison St., U niversity of Illin ois at Ch icago , Ch icago , IL 60607-7137; e-

mail: [email protected].

PSPB,Vol . 25 No. 7, July 1999 793-808 1999 by the Society for Personality and Social Psychology, Inc.

793


2/16

are observed in an atypical need domain and, moreover,

a need domain in which the context pulled more

strongly for extern al-uncontrollable th an internal-

controllable explanations for why people need assis-

tance. Specifically, th e present study reported on ideo-

logical differences in response to victims of a natura l dis-

aster (the 1993 Midwest Flood). Investigating publiccompassion in the context of a natural disaster provides

an excellent opportunity to explore a number of ques-

tions neglected in the social psychological literature,

such as the fo l lowing: (a) Do people st i l l turn to

individual-level explanations for why people need help

even in the context of a natura l disasterthe quintessen-

tial external-uncontro llable cause of need? (b) Are peo-

ple eq ually responsive to disaster victims lon g-term or

secondary needs (e.g., a replacement home) as they are

to disaster victims immediat e sur vival needs (e.g., clean

water, tempora r y shelter)? (c) Are people any more or

less responsive t o comm unity-level req uests tha n

individual-level req uests for disaster aid? a nd (d) Towhat extent d o ideological differences emerge in a con-

text for which there is not a well-rehearsed ideological

script?

PERCEPTIONS OF CAUSALITYIN

THE CO NTEXT OF A NATURAL D ISASTER

Natural d isasters are generally seen a s acts of na ture,

that is, causes of harm that could not be prevented by

anyone, much less those victimized by it (cf. Br un, 1992).

Examining reactions to victims of a natural disaster pro-

vides an excellent test of the boundar y conditions of

attributiona l explanations for ideological d ifferences inwillingness to support public assistance programs.

According t o a ttributional m odels of helping, cognitive

appraisalsfor why people need help should be processed

in a three-stage seq uence. Initial explana tions for why

someone needs assistance (e.g., this familys house was

rendered uninhabitable by the flood) are processed

according t o the following sequence:

1.

2.

3.

To the extent that ideological differences reflect base-

line propensities to make interna l-controllable versus

external or uncontrollable attributions,liberals and con-

servatives will have different affective reactions toward

claimants and vary in their willingness to provide public

assistan ce. Co nsiderable research supports th e close

connections b etween cognitive a ppraisal, affective

arousal, and intentions to help (Amato, 1986; Batson,

Duncan, Ackerman, Buckley, & Birch, 1981; Meyer &

Mulher in, 1980; Reisenzein , 1986; Weiner, 1986, 1995;

Weiner, P err y, & Magnu sson , 1988). Similarly, oth erresearch is consistent with ideological differences in

attributions for why people need help and subsequent

affective reactions and decisions about whether to assist

different claimants (Skitka, McMurray, & Burroughs,

1991; Skitka & Tetlock, 1992, 1993a; Zucker & Weiner,

1993).

Because a flood ( and particularly a flood of th e mag-

nitude of the Midwest Flood) is seen as a catastrophic

an d exter na l-unco ntrollable event, at first pass it would

seem that such a disaster would tr igger unq ualified sym-

pathy and compassion and a subsequent desire to

respond to the n eeds of those caught by it.1 Consistent

with this notion, it is considered to be nearly political sui-cide for legislators to vote against providing Federal

Emergency Man agement Agency (FEMA) assistance

(Berenson, 1994; Goodwin &Smith, 1995). Therefore,

the doma in of disaster assistance providesn early an ideal

backdrop to examine the extent to which obser ved ideo-

logical differences in willingn ess to suppor t public assis-

tance reflect truly d ifferent cognitive-affective orienta-

tions toward th inking ab out the needy, or instead reflect

well-rehearsed ideological scripts that both direct a nd

justify peoples position o n p ublic a ssistan ce.

Examination of reactions to disaster victims also pro-

vides an o pportunity to explore an interesting juxtaposi-

tion of necessar y versus sufficient causes for why peopleneed help. The flood wasthe primary cause of whyclaim-

ants needed assistance: If th e flood h ad not occurred,

there would be no need for clean water or food , tempo-

rar y shelter, or a replacement hom e. Lack of insurance,

although a contributing factor to why many of the vic-

tims of the 1993Midwest Flood n eeded help, was not th e

root cause of claimants need for assistance.

The prediction tha t people will respond with unq uali-

fied compa ssion to victims of a na tura l disaster, however,

presumes that at least in this context, people will not

adhere to their typical tendency to ignore situational

information in favor of focusing on personal causes of

behavior. One of the most pervasive findings in social

psychology has been th e tenden cy of people to turn to

dispositional rather than situational explanations for

others behavior (e.g., Ross, 1977). The tendency to

make the fundamental attribution error occurs even

when people are made explicitly aware of situational

constraints on their own aswell as targets behavior (e.g.,

Gilbert &Jones, 1986). In short, there issome suggestion

tha t even in the face of a blata nt extern al-unco ntrollable

794 PERSONALITYAND SOCIAL PSYCH OLO GY BU LLETIN


3/16

explanation for why people need help, perceivers may

nonetheless focuso n featureso f persons when deciding

whether to help (cf. Gilbert, 1989; Quattrone, 1982).

It is hypothesized here that perceivers will evaluate

whether th ere are any personal ca uses for why disaster

victims require assistance, and to the extent that per-

sonal causes are revealed (e.g., Mr. Carlson did no t haveflood insuran ce), these reasons will becom e focal in per-

ceivers a ttrib ution s for why victims need assistan ce. To

the extent that perceivers come to conclusions that a

given targets need for assistance is interna l and control-

lable, they will be less willing to support giving that victim

disaster assistance. In add ition, to the extent tha t ideo-

logical differen ces reflect a consistent cog nitive style (cf.

Wan ke &Wyer, 1996) rath er tha n a situa tiona lly scripted

response, conser vatives should be more likely tha n liber-

als to invoke personal attributions for why people need

assistan ce, with subsequen t implications on their willing-

ness to h elp.

PRIMARYVERSUS SECONDARYNEEDS

Studying helping and allocation decision making in

the context of a na tural disaster also allows for examina-

tion of the boundaries of public compassion. In most

studies of helping behavior, the dependent variable is

whether help is provided. Shifting focus to distributions

of public assistance, the question is usually not whether

help is provided to a single need y oth er, but expa nd s to

become a decision a bout who receives help, how much

help, and what kind of help to provide. To date, there has

been very little investigation of what kinds of help peo-

ple think are appropriate to provide when there is apotential ran ge of responses to a problem.

Disaster victims need a variety of resources, ranging

from clean water and temporar y shelter to the replace-

ment of a home. Although attributiona l theories of help-

ing are a gnostic with respect to differential predictions

as a function of type of aid, some research hints that

there a re ind eed con straints on peoples willingness to

respond to different kinds of requests as a function of

whether the resources being a sked for a ddr ess primary

or second ar y needs. For example, Bickman a nd Kamzan

(1973) foun d tha t people were con sidera bly more reluc-

tant to help someone requesting money in a grocery

store to buy a tube of cookie dough (a relative luxuryitem) tha n milk.

In the spirit of Maslows need hierarchy (Inglehart,

1977), public compassion may be primed m ost power-

fully in resource domains that are critical for physical

sur vival, such a s food, health, or shelter. For exa mple,

although job training augments the likelihood that

those who are unemployed can satisfy their primary

needs, providing people with jobs or ad ditional training

may be perceived as a relative luxur y given th e num ber

of jobs available at th e low end of the income d istribu-

tion. Dignity, status, and upward mobility may be per-

ceived to be extras, not to be doled out even by liberals to

people who h ave brough t ill-fort une on them selves. But

when resources are linked directly to survival, are at least

some perceivers willing to expan d the moral commun ity

to include even those who placed themselves at risk?Similarly, Okun (1975) argues that liberals and conser-

vativesd iffer in what they consider to be rights, or entitle-

ments, of citizenship. Liberals are more likely than con-

servatives to see a minimum subsistence a s a right (like

the right to vote or to police protection) th an someth ing

that has to be ear ned com petitively in the ma rketplace.

Con servatives are more likely to endorse the hard form

of self-reliance, such as th ose who do no t work, shall not

eat. Conservatives may therefore be more likely than

liberals to withhold even immediate humanitarian aid

from those who failed to protect themselves against

peril. However, liberals are unlikely to see the fulfillment

of seconda ry n eeds as basic entitlements of citizenship.Liberals and conservatives should therefore be equa lly

reluctant to provide the personally responsible with aid

designed to meet seconda r y, rath er tha n primary, needs.

WHATI F THE PERSO NALLY

RESPONSIBLE REFORM?

The tension between helping the needy without

rewarding people who fail to protect themselves against

risks recurs when considering the possibility of second

chan ces. Reactions to reform can b e based on compa s-

sion , fear of creat ing a slippery slope, or sustained pun i-

tiveness (see a lso Skitka & Tetlock, 1993b, stud y 3). Ifpeople who violated an implicit social con tract to beh ave

responsibly truly refor m, th en some people ma y believe

the repenters deserve to be rewarded an d reintegrated

into the moral community. The underlying logic is that

other free riders might witness this reward con tingency

and be inspired to change their behavior as well. Alterna-

tively, people may continue to be reluctant to help

repentant free riders to the same degree as non free rid-

ers, because they may believe that allowing second

chan ces will undermine m otivation to d o well the first

time around why not third or fourt h chan ces?

Some research has found tha t people were less angr y

and punitive toward personally responsible claimants ifthere were clear indications that the claimants had

refo rm ed ( Schwa rzer &Weiner, 1991; Skitka &Tetlo ck,

1993b). Int erestin gly, con servat iveswere even more will-

ing to help a target who had reform ed than on e who had

never sinned at all (Skitka &Tetlock, 1993b) . H owever,

it is less clear whether the promise of making aid con tin-

gent o n future precaution s will have a similar effect of

returning the personally responsible to the moral

community.

Skitka / PU BLIC COMPASSION 795


4/16

GRO UP S VERSUS INDIVIDUAL

CL AIMS FOR ASSISTANCE

Applying theories of helping to th e doma in of public

compa ssion also raises the question of whether the deci-

sion to help an individual (the usual focus of helping

research) is similar to the decision to help a group of

needy others. Public policy makers rarely consider indi-vidual ca se studies when voting on legislation to inter-

vene in emerg encies like th e Midwest Flood . To wh at

extent are decisions to help groupsdifferent or similar to

decisions to help ind ividua ls?

Some research indicates that people process social

informa tion differently for individual and group targets.

People tend to have poorer recall of group than individ -

ual behaviors, and slower recognition of statements asso-

ciated with groups than individual targ ets (McConn ell,

Sherman, & Hamilton, 1994; Srull, Lichtenstein, &

Rothbart, 1985). O ther research supports the notion

that groups are seen as ha ving less of a real social exis-

tence than individuals, and this in turn affects how peo-

ple process inform ation a bout th em (McConn ell et al.,

1994).

In addition to research that reveals that people

process inform ation d ifferently about groups than ind i-

viduals, other research suggests th at individuals are

evaluated more favorably when judged alon e than when

the same individual is judged a s a member of a n a ggre-

gate or group (Sears, 1983). However, when group

descriptions are provided rather than descriptions of

individuals in a group con text, people were less likely to

make negative inferences about groups than ind ividuals

(C oovert &Reeder, 1990). Oth er research has ind icatedthat empathizing with an individual group member

(e.g., someone with AIDS or who is homeless) subse-

quently leads to more positive reactions toward the

entire groupthat is, AIDS victims and the homeless

overall( Batson et al., 1997), but onlyif the individual tar-

get is not personally responsible for his or her plight.

Too little research h as been do ne to make stron g pre-

dictions about how people will view individual versus

group claimsfor assistance in the context of a natur al dis-

aster. However, based on the evidence collected to date,

people may process information about groups (in this

case, communities affected by the flood) differently

than individuals, which in turn may lead to differentattributiona l and helping thresholds.

In summ ar y, the present study extend ed investigation

of attribution al explanations for ideological differences

in willingness to support public assistance. It was pre-

dicted that liberals and conservatives differ in their

thresholds for ascribing personal responsibility for why

people need public assistance, and that these differences

would be revealed even in a need dom ain for which liber-

als and conservatives have not been handed an easy

script. The tend ency to ascribe varying levelsof responsi-

bility was assessed by tapping general policy preferences

in the disaster aid domain (e.g., degree of support for

federal d isaster aid versus the belief t hat it is citizens

responsibility to protect themselves against natural haz-

ards), as well as reactions to specific claimants (either

individuals or communities).Liberalswere not predicted to be bleeding hearts who

ignore attributiona l inform ation an d provide help of all

types to all comers. Rather, it is proposed that liberals

have come to view basic subsistence a s a righ t of all com-

munity members, and they will therefore suspend the

nor mal co nsequences of th e a ttribution-affect-action

sequence when considering allocations of basic humani-

tarian aid. When aid addresses needs outside of basic

subsistence, the liberal bleeding heart is predicted to

form a scab . Liberals are pred icted to be unwilling to use

the public purse to return flood victims to their prior

stand ard of living, an d will be especially unlikely to pro -

vide this kind of assistance to those who did not takesteps to p rotect them selves again st flood losses.

In contrast, because conservatives do not see basic

subsistence as a fund ament al right or entitlement, they

will not suspend attributional analysis, even when allo-

cating ba sic human itarian aid. C onservatives were pre-

dicted to be willing to provide huma nitarian aid to claim-

ants who came to harm despite taking some personal

precautions, but should be less likely than liberals to

expand t he mora l community to provide even huma ni-

tarian aid to those who recklessly placed themselves at

risk. Of add itional interest was extending investigation

of conservatives willingness to embrace reformare

promises of future reform sufficient for con servatives tosoften their stan ce toward the irresponsible? Fina lly, the

present study also investigated th e explorato r y q uestion

of h ow people respond to requests for assistance from

not only individuals but also groups.

METHOD

SUBJECTS

A sample of 1,015 adult members (representing a

72% response rate) of random-digit d ialed (RDD)

households in the continental Un ited States was con-

tacted between October 15, 1993 and November 15,1993. Most area s of the Mississippi an d Oh io River valleys

were still flooded at this time, or were flooded a second

time. Sampling was based on a multistage cluster design,

structured so tha t each a dult in the U nited States living

in a household with a telephone ha d an eq ual chance of

being selected. This nation al sample was segmented into

six distinct replicatesthat became the source samplesfor

the six versions of the survey (six different orders of

stimulus materials were used to control for possible



5/16

order effects). Therefore, each version of the question-

naire was administered to a complete na tional sample.

The sam ple con sisted of 564 females an d 451 males,

who ranged in age from 18 to 92 (M= 41,SD= 15.87). Of

the respondents, 70% lived in a single family home,

20.5% in an apartment or condo, 5.7% in a mobile

hom e, and 3.7%had some oth er form of residen ce. Thesample was 8.6% African American , 3.3%Na tive Ameri-

can, 3.3% Latino or Hispanic, 0.6% Asian, and 82.2%

Caucasian. Of the sample, 7%had been somehow per-

sonally affected by the Midwest Flood and 18%person-

ally knew someone who had been af fected by the flood.

Explorator y ana lysis indicated that people who were per-

sona lly affected by the flood were less supportive of pro-

viding an y type of a id to flood victims than were those

who had not been personally affected by the flood.

Hypothesis testing controlling for self-interest did not

change the observed pattern of results, so these respon-

dents were retained in all analyses.

PROCEDURE

Professional interviewers from the Public Opinion

Laboratory(POL) facilityat Northern IllinoisUniversity

conducted interviews. The PO L operates a 30-station

computer assisted telephone interviewing (CATI) sys-

tem, using two parallel Novell networks and an Elec-

t ro n i c Qu e s t io n n a i re ( E QTM ) so f tware syste m .

Although a survey approach was used, the survey

included an experimental design tapping respondent

reactions to claimants that varied as a function of type

(community or individual) and responsibility informa-

tion (whether targets took any predisaster preventive

measures, such as purchased flood insurance or builtflood walls or levees).

STIM ULUSMATERIALS

Respondents were asked to give their reactions to

three different communities devastated by the 1993Mid-

west Flood. Although given h ypothetical nam es, th e

descriptions were representative of real communities

affected by the flood. The communities varied in the

extent tha t they had taken protective measures against

f lood damage: Plan Ahead Place , a town that had

invested a share of its tax dollars to build a flood wall or

levee to protect it from up to 50 feet of water ; Tour Town ,

a town that elected not to build a flood wall or leveebecause it would r uin the view an d th erefore nega tively

affect to urism, its major economic ba se; an d Ba il-O ut

by the River, a town tha t did not build a flood wall or

levee because they voted aga inst increasing local ta xes

to build it.

In addition to the communities, respondents were

also asked to respond to three d ifferent individuals

whose homes or b usinesses had been d estroyed by the

flood, a lso representa tive of rea l people affected by the

flood: Mr. Adams, an individual who had always main-

tained flood insurance; Mr. Bell, an individual who pur-

chased flood insurance only after it became obvious that

his property would be flooded; and Mr. Carlson, an indi-

vidual who did not buy flood insurance but who counted

on federal disaster assistance should a flood occur.

A prima rily within-subject design was chosen becausetheories about allocation decisions involve h ow people

cope with competing claims for resources. It follows that

participants need to have information on the entire pool

of ap plicant s. Most real-world a llocation decisions take

on within-subject forms; for example, facultyh iring com-

mittees have access to all applicant files, medical ethics

committeesha ve the filesof all patients waiting for organ

transplants, an d Congress has a n otion of the number

an d type of people who apply for welfare aid when decid-

ing their budget. Community and individual descrip-

tions were presented in counterbalanced order, crossed

with three different ord ers of claiman ts within commu-

nities and individuals to guard against any potentialorder effects.

MEASURES

In addition to providing some basic demographic

information, respondentsanswered questionsabout dis-

aster a ssistance po licies genera lly (see Table 1), in add i-

tion to questions about specific targets that ta pped per-

ceived responsibility for needing assistan ce, blam e for

needing assistance, measures of positive affect (sympa-

thy), nega tive affect ( anger) , an d five measures of will-

ingness to help: support of immediate humanitarian

assistance, money targeted to rebuild homes and busi-

nesses in the same location, bail-out gran ts that wouldprovide victims with the full replacement value of their

home, funds to relocate people out of the affected area,

and whether the provision of any federal aid (of any

kind) should be contingent on a commitment to take

precautionary measures in the future. In addition,

respond ents were asked ho w unfair it would be if a given

claimant did n ot receive federal aid, and th e extent that

they felt each target deserved federal assistance. All

questions were on 7-point scale respon se forma ts.

POLITI CAL ORIENTATION

Several measures of political orientation were also

included in the sur vey. Self-repor ts of both par ty iden tifi-cation and political orientation were assessed using

measures tha t asked ideological orientation (from very

liberal to very conservative), strength of ideological ori-

entation , and for respond ents who were at the midpoint,

more focused questions such as more like a liberal or a

conservative. Party identification was assessed using a

similar format. In addition to these measures, five items

were included from established attitudina l and personal-

itymeasuresof political ideology: how important respon-



6/16

dents believed it was to strengthen law and ord er, to pre-

ser ve respect for autho rity, to mainta in respect for the

U nited States as a world power, to improve politeness in

daily behavior, and to follow Gods will (see also Snider-

man &Tetlock, 1986).

A principal components ana lysis of self-reported

political orientation, party identification, and the per-sonality/attitudina l items yielded a single, interna lly

consistent solution. Compon ent scores were calculated

using the regression method. High scores reflected

greater conservatism, and low scores on this measure

reflected grea ter liberalism. Respondents scoring below

the 40th percentile (n= 394) were labeled liberals and

respondents scoring above the 60th percentile were

labeled conservatives (n= 412), leaving a total sample

size ofN= 806 for analyses that included political orien-

tation as a variable.2

VALIDATION OFTHE POLITICAL

ORIENTATION MEASURE

Seven ad ditional q uestions on th e sur vey assessed tra -

ditional policy positions on which liberals and conserva-

tives generally differ. If the political orientation mea sure

was valid, significant differences among high and low

scorers should emerge on each of these items. Analysis

supported the validity of the measure. Conservatives

were significantly more angered by poor people who

spend their money on fancy clothes or big cars rather

than spending th e money to feed and clothe their fami-

lies,t(804) = 7.72,p< .001; by someone who collects

welfare because he or she ist oo lazy to get a job,t(804) =

8.70,p< .001; by giving blacks and other minorities

special ad van tages in jobs and schools,t(804) = 13.89,

p< .001; and by government officials interfering and

tr ying to tell us what we can an d cann ot do with our own

lives,t(804) = 5.65,p


7/16

General Att itudesAbout

Federal Di saster Assistance

As can be seen in Tab le 1, the vast prop ort ion o f th e

respond ents strongly believed tha t it was the responsibil-

ity of individual comm unities located in flood plains to

build their own flood walls and levees( 73.3%agreed), in

add ition to it being the responsibility of individuals wholive in flood plains to protect their investment by buying

flood insurance (91.2%). A smaller percentage of the

sample felt that it was the appro priate role of the govern-

ment to build floodwalls and levees (56.6%) or to pro-

vide inexpensive flood insurance to people who live in

flood plains (56.6%). The sample was mixed, however, as

to whether people who did not buy flood insurance were

just as deserving of federal assistance in a natura l disaster

(46.0%) as those who did (44.8%).

As can also be seen in Tab le 1, respon dents var ied in

their support for disaster policies as a function of politi-

cal orientation. Liberals were more in favor of providingfederal disaster aid in all for ms than conservatives and

less likely to feel that it was the individuals or communi-

tyssole responsibilityto cope with the consequences of a

natural disaster such as the 1993 Midwest Flood (see

Table 1 for ad ditional deta il; all differences were statisti-

cally significant atp< .01).

Reactions to Individual

Victimsof Natural Disaster

Ana lysis of reaction s to ind ividua l victims of the Mid -

west Flood generally supported the following predic-

tions: (a) even in th e context o f a n atural d isaster, per-

ceivers still sought out and used information aboutpersonal responsibility when judging how deserving

needy others were of public assistance, an d oth er results

supported the links between political orientation and

judgments of being deserving; (b) liberals were more

likely than conservatives to support providing public

assistance to flood victims, and were specifically more

likely to provide hum anitar ian a id to the irresponsible

than were conservatives; (c) n either liberals nor con ser-

vatives, however, were enthusiastic about providing for

flood victims second ar y needs; and ( d) a id contingent

on future reform was viewed positively by both liberals

and conservatives.

JUDGMENTSOFRESPONSIBIL ITY

The first set of analyses examined whether respon-

dents in fact a ttributed var ying d egrees of personal

responsibility to individual claimants. Consistent with

hypotheses, perceivers were willing to ascribe personal

responsibility and blame fo r n eeding a ssistance to indi-

viduals, despite the external-uncontro llable context in

which their need arose.

Two questions assessed judgments of responsibility

for need ing assistance: The extent to which respond ents

believed that a given individual should have taken

greater precautions against f lood damage, and the

extent to which a given individual was to blam e for his

predicament. These ratings were correlated within indi-

viduals ( rs ranging from .47 to .61, allps < .001), andtherefore were collapsed into a single item for analysis.

Ana lysis of responsibility judgm ents as a fun ction of a 3

(individual target) and 2 (political orientation of the

respond ent) mixed design ana lysis of var iance (ANOVA)

yielded significant main effects for the individual,F(2,

1598) = 974.19,p< .001, 2 = .65; and political orienta-

tion,F(1, 799) = 25.71,p< .007, 2 = .01.

Mr. Adam s, who always maintained flood insurance,

was seen as the most blameless for hispred icament (M=

2.06), in contra st to Mr. Bell, who bo ught insurance a t

the last minute (M = 4.87), or Mr. Carlson, who never

bought flood insurance but instead counted on federal

disaster assistance if the worst happened (M = 6.07).Conservatives rated individuals to be more responsible

(M= 4.60) than did liberals (M= 4.24).3

Con sistent with the hypotheses, participants were sen-

sitive to individual levels of responsibility for needing

assistance, even in the context of a na tura l disaster as the

prima facie cause of need. In ad dition to being an inter-

esting result in itself, th ese ana lyses also represent

man ipulation checks on the extent to which the individ-

u a l t a r g e t d e s c r i p t i o n s m a n i p u l a t e d p e r s o n a l

responsibility.

DESERVINGNESSOFDISASTER A ID:

REACTIONSTO INDIVIDUALS

It was hypothesized tha t pa rticipants would evaluate

the extent to which individuals revealed any personal

causes for needed a ssistance, an d to the extent tha t per-

sonal causes were revealed, tha t these (rath er tha n th e

situationa l context o f the disaster itself) would b ecome

focal in judgments of the extent to which claimants

deserved a ssistan ce.

Resultssupported the notion that participants indeed

relied on personal informa tion to determ ine if one was

deserving, even in the context of a strong situational

alternative such asa natural disaster. In addition, the pre-

dicted ideological differences in reactions to claimants

also emerged. H owever, the an alysis revealed that politi-cal orientation had more of a direct than mediated effect

on judgment s of being deserving.

Two questions tapped the extent to which respon-

dents viewed claimants as deserving of d isaster aid: t he

extent to which each individual was seen as deserving of

federal disaster assistance, and how unfair it would be if

each individual received no assistance. These ratings

correlated a t least atr= .57,p< .001 within ea ch ind ivid-



8/16

ual, and were therefore collapsed together as a single rat-

ing of deservingness.

Participants viewed claimants who took different lev-

els of precautions as differentially deserving of federal

disaster a ssistan ce,F(2, 1600) = 324.37,p


9/16

equally supportive of providing humanitarian aid to

individuals who took precautions against flooding, but

conservatives would be less supportive than liberals of

providing even humanitarian aid to those who had not

taken f lood precautions. Results supported both

hypotheses.

A within -subject ANOVA that compa red support forhumanitarian aid, rebuilding, buyout, or relocation of

flood victims indicated significant differences in partici-

pants willingness to support different kinds of aid,F(3,

2,976) = 443.02,p< .0001, 2 = .31. As can be seen in

Table 2, participants were m uch more supportive o f

humanitarian aid than aid that was designed to address

more second ar y needs.

Also as predicted, liberals and conservatives were

equally supportive of providing Mr. Ada ms with hum an i-

tarian assistance, F(1, 804) = 2.94, ns, 2


10/16

kinds of h elp revealed th at respondent s supported pro-

viding huma nitarian aid and assistance designed to

rebuild more to communities than to individuals, but

were equally unenth usiastic about providing individuals

and communities aid to b uy out their homes or to relo-

cate to a different area . A more deta iled d escription of

these results is reported below.JUDGMENTSOFRESPONSIBIL ITY

Consistent with reactions to individuals, respondents

attributed varying levels of responsibility for needing

assistance to the three different communities, Plan

Ahead Place, Tour Town, and Bail-Out by th e River.

Respond ents reported their perceptions of whether the

communities could have done more to prevent being

damaged by the flood, the communities blameworthi-

ness, an d whether each community was responsible for

needing government assistance. Because these items

were significantly intercorrelated within each commu-

nity (rs ranged from .26 to .49, allps < .001), they werecollapsed in to a single scale of responsibility.

An analysiso f the 3(co mmunity) by 2 (political orien-

tat ion) mixed-design ANOVA with th e depen den t vari-

ab le of responsibility yielded a significant ma in effect for

community,F(2, 1856) = 88.78,p< .001, 2 = .14. Plan

Ahead Place was seen as the least responsible of the three

communities(M= 2.06), with Tour-Town (M= 4.78) and

Ba il-Ou t b y the River (M = 4.78) as equally and more

responsible (the comparison between the latter two

means y ie lded anF < 1) . No o ther e f fec ts were

significant.

These results indicated tha t communities tha t did not

take precautions for protecting themselves against a

flood by buildin g either a flood wall or levee were seen as

equa llya nd more responsible for incurring dama ge due

to the flood than th ose communities tha t did, regardless

of why they failed to take precautions (i.e., to avoid rais-

ing taxes or because of a concern abo ut disturbing their

tour ist-based eco nom y). Because Tour Town a nd Ba il-

Ou t by the River did n ot d iffer in perceived responsibil-

ity, they were collapsed toget her in subsequent analyses.4

It sho uld be noted tha t the effect size for the ma nipu-

lation of r esponsibility was considerably smaller a t t he

community level of analysis (2 = .14) than what was

observed at the individual level of analysis(

2

= .65), sug-gesting that it may be more difficult to make attributions

of responsibility with respect to communities than it is

for individuals.

DESERVINGNESSOFDISASTER AID:

REACTIONSTO COMMUNITI ES

Because so little previous work had been done exam-

ining how people think about collectives relative to ind i-

vidua ls, especially in help setting s, it was diff icult to gen-

erate strong hypotheses regarding whether people

would similarly turn to community chara cteristics (such

aswhether theyvoted to build a flood wall) when making

judgments regarding being deserving of disaster aid.

Results indicated tha t at least in this context, people did

base their judgmen ts on the extent to which communi-ties took prior precaution s against flooding.

Two questions tapped the extent to which respon-

dents viewed comm unities as deser ving of disaster aid:

the extent to which each community was seen as deserv-

ing of federal disaster assistance, and h ow unfair it would

be if each comm unity received n o a ssistan ce. These rat-

ings correlated at least atr= .51, p


11/16

AID DESIGNED TO MEET

PRIMARY VERSUSSECONDARY NEEDS

The next set of analyses addressed whether people

were (a) less enthusiastic about pro viding for comm uni-

tys secondary relative to their primary needs, and (b)

whether ideological differences emerged in willingness

to provide humanitarian assistance to those communi-

tiestha t did not take precautions against flood dam age.

The hypothesis that participants would be more sup-

portive of humanitarian aid tha n aid designed to return

people to their predisaster standard of living was sup-

port ed. A within -subjects ANOVA that comp ared sup-

port for human itarian aid, aid to help affected commu-

nities rebuild their towns, aid to provide communities

with buyouts, or aid designed to relocate the commun ity

to a safer location yielded a significant effect for type of

help on support,F(3, 2310) = 511.14,p< .0001, 2 = .40

(see Tab le 2). Altho ugh Tukeys tests indicated tha t a ll

pair-wise compa risons were significant, there was clearlythe greatest amount of support for immediate humani-

tar ian assistance, an d to some extent to rebuild th e com-

munities. Participants were considerably less eager to

provide aid to provide buyouts or to re locate the

community.

Also para lleling the find ings with individua ls, liberals

(M= 6.13) and conservatives (M= 6.09) were equally

supportive of providing humanitarian aid to Plan Ahead

Place,F(1, 802)


12/16

included target as a factor ( i.e., the community/individ-

ual distinction) are reported below.

RESPONSIBIL IT Y, AFFECT,

AND DESERVINGNESS

Analysis of responsibility judgments (the average of

the extent to which respond ents blamed the target, and

their belief that th e target should ha ve taken greater pre-

cautions) as a function of a 2( target: community or indi-

vidual) by 2 (poli t ical orientat ion) mixed-design

ANOVAind icated tha t commun ities(M= 4.29) and in di-

viduals(M= 4.32) were held equally responsible for their

plight,F(1, 770)


13/16

Previous research has hinted that there may be differ-

ences in the way that liberals and conservatives react to,

or differentiate between, primar y and secondary needs.

By considering multiple levels of a ssistan ce, the present

study determined that liberals public compassion was

not constrained by attributional analysis when aid

addressed primary relative to nonprimary needs (e.g.,clean water, food, a nd immediate shelter vs. a buyout).

Liberals supported providing those who did not take

flood precautions with humanitarian aid to the same

extent as those who ha d, whereas conservatives were less

supportive of providing even immediate h uman itarian

aid to tho se who had not ta ken actions to protect them-

selves aga inst risk. Libera ls an d conservatives alike, ho w-

ever, were unenthusiastic about using federal disaster

assistance to provide assistance beyond immediate

human itarian aid, especially for those who had not taken

flood precautions in th e first place. Although not

uncommo n forms o f d isaster assistance, willingness to

provide flood victims with money to either rebuild, buyout, or relocate was markedly less enth usiastic tha n will-

ingn essto provide flood victims with assistan ce designed

to meet prima r y sur vival needs. Finally, liberals and con-

servatives alike were also more willing to suppor t provid-

ing federal assistance if that aid was mad e contingent on

recipients taking future precautions against f lood

damage.

Taken t ogether, these results seem to support the

notion that liberals are n ot mind lessly ignoring attribu-

tional informat ion; instead, they are making considered

judgments about when to (and when not to) use that

informa tion in making a judgmen t about whether a tar-

get is deser ving. When a llocatin g resources designed tomeet primar y needs, such as human itarian aid to people

affected by a na tural disaster, liberals appear to suspend

the usual consequences of attribution-affect-action

sequence and help even those who recklessly placed

them selves at h igh risk.

The resultspo int to a potentially complex interplay of

cognitive and motivational concerns that direct peoples

willingness to help th e d isadvanta ged. Specifically, the

finding t hat liberals seem to suspend the typical conse-

quen ces of the a ttribution-affect-action sequence und er

some circumstances (e.g., when allocating huma nitarian

aid) seems to be a par ticularly ripe domain for further

research. Similar to recent ideas about prejudice devel-

oped by Devine and her colleagues (Devine, 1989;

Devine, Monteith , Zuwerick, &Elliot, 1991), it could be

that even though liberals make internal-controllable

attributions for why some of the claimants needed assis-

tance, t hese attributions were inconsistent with a set of

internalized values that dictated that they should help

everyone meet their primary needs. Conflict between

liberals principled commitment to egalitarianism and

the conclusions of attributional analysis may be espe-

cially strong when deciding whether to provide claim-

ants with humanitarian assistance, presuming tha t liber-

als do see humanitarian aid as more of a right than a

good to be distributed.

Research indicates that l iberals f ind trade-offs

between lives and money to be especially painful andawkward , and th ey will go out of their way to avoid these

kinds of choices; however, conservativesd o not (Skitka &

Tetlock, 1993b). Therefore, it seems reasonable to

hypoth esize tha t whenever possible, libera ls are likely to

resolve the aroused conflict between not helping the

irresponsible and egalita rian values tied to saving lives by

suppressing th e conclusions of causal ana lysis, an d help-

ing even the irrespon sible if resource availab ilityperm its

it. Just as low-prejudiced peo ple experience m ore co m-

punction about an y negative reactions they have toward

ethnic minorities or homo sexuals, liberalsm ay feel com-

punction about their negative reactions to the person-

ally responsible when considering their basic survivalneeds; reacting negatively may threaten liberals self-

image as being open-minded and egalitarian.

In contrast, conservatives values are shaped much

more by a com mitmen t to self-reliance and a work ethic

than by a commitment to egalitarianism. Conservatives

are not likely to experience any particular value conflict

or compunction about withho lding assistance from t he

persona lly responsibledo ing so is con sistent with their

core values. According to this analysis, conservatives

would have little value conflict in this cont ext, and th ere-

fore little motivation to suppress acting on the con clu-

sions of attributional analysis.

These ideas are a lso consistent with Judd and Kros-nicks (1989) mod el of political memor y organization,

which suggests tha t specific issue positions are organ ized

in memor y aroun d multiple crowning postures or values

(e.g., equa lity, freedo m) . A specific policy position (e.g.,

flood assistance) is likely to be evaluated against these

crownin g postures, which are likely to be organized ver y

differently for liberals and conservatives. In a related

vein, Tetlocks value pluralism mod el (1986; Tetlock,

Peterson, & Lerner, 1996) suggests that liberals hold

more conflicting values and are more likely to recognize

an d be sensitive to value conflict. Altho ugh o ne implica-

tion of thisis that liberalsma y have more difficulty main-

taining a perfectly constrained or consistent political

belief system, it also suggests that liberals may be more

likely to recognize conflicts between attributional ana ly-

sis and crowning values and , if they have sufficient cogni-

tive resources to bring to bear to the prob lem, to be able

to suppress the effect of the for mer to a llow action as a

function of the latter (cf. Gilbert, 1989; Gilbert, Pelham,

& Krull, 1988). This interpretation suggests that the

on e-step att ributio na l process implicit in Weiners a ttri-



14/16

butio na l mod el of helping (1986, 1995) may need to b e

revised to reflect a stage theory similar to Gilberts

(1989) mod el of how people make tra it inferences. Peo-

ple may spontaneously make responsibility judgments,

which then are only corrected for later in the judg-

men tal sequence if the perceiver has sufficient cognitive

capacity or motivation to do so.In addition to being a fruitful area to explore in its

own right, the above analysis also points to some pro-

vocative predictions about what might h appen if a lloca-

tors work under conditions of scarcity, followed by a

period of abundance. My previous research has found

that under scarcity, liberals and conservatives alike deny

lifesaving resources (e.g., organs for transplant, azido-

thymid ine [AZT] treatm ent) to the personally responsi-

ble. Under no scarcity, liberals tend to help everyone,

but conservatives tend to still withhold help from those

responsible for their predicament (Skitka & Tetlock,

1992).

If liberals do feel compunction about not allocatingaid d esigned to add ress primar y needs equa lly to a ll, to

reduce the awkwardness of these conflicted feelings,

they may need to genera te justifications that a ccentuate

the deserving of those chosen and the undeserving of

those rejected under scarcity. These justifications may

come to be functionally autonomous of the resource

constraints that originally motivated them. Claimants

rejected under scarcity could come to be seen as less wor-

thy o f assistance, even after scarcity constraints are

removed, as allocators attempt to make their attitudes

consistent with their a ction s (Festinger, 1957). B ecause

only the undeserving are going without assistance, scar-

city could easily become self-perpetuating (cf. Ross &Ellard , 1986).

Although the results of the present study should be

viewed with some caut ion because they relied on reac-

tions to hypothetical claimants, the present study pro-

vides a good test of t he genera lizability of attributional

models of helping and allocation preferences. The d ata

were collected durin g the height of the flood crisis in the

Midwest, when the devastation of homes, farms, and

communities were common fodder for the evening

news. Although th e targets presented in th e survey were

no t real claim an ts, they were highly vivid on es given th e

context in which the survey was conducted ( cf. Amato ,

1986). Moreover, by using an experimental design

within a national survey, the present study was able to

maximize internal validity while gaining a great d eal of

external validity. The use of a representative sample is

especially important, as theories of allocation prefer-

ences move away from an exclusive focus on the alloca-

tion of wages and rewards to explore allocation prefer-

e n c e s i n so c i o p o li t ic a l c o n t e x t s. Al lo c a t i o n o f

sociopolitical resources, such as welfare or federal emer-

gency assistance, may n ot have the same psychological

meaning when judged by an interested as compared to a

disinterested respond ent. Relying exclusively on college

students for hypothesis testing in these domains is rely-

ing on peo ple who have yet to becom e invested in main -

taining the current system (i.e., they are less likely to

have pa id income taxes). Therefore, an additionalimportan t contribution of the present study is that it not

only extended our understanding of the attributional

and ideological bound aries on public compassion, but it

also tested these hypotheses using a representative

sample.

Social psychological research has made enormous

progress in und erstand ing when individuals will be will-

ing to extend help to another needy individual (for

reviews, see Cla rk, 1991; Schroed er, Pen ner, Do vidio, &

Piliavin, 1995). The present stud y cont ributes to what we

know about prosocial beha vior beyond this by initiating

investigation of public compassionthat is, willingness

to comm it collective resources to assist the less fort una temembers of the community. Although there are some

strong para llels between individual-level helping beha v-

ior an d public compassion, there a re also some impor-

tant distinctions. For exam ple, the need contexts being

add ressed in the public sphere are generally of a much

larger scale, and involve longer term commitments, than

do individual-level inter ventions. In add ition, th rough

the use of public resources, people can provide assis-

tance for problems tha t they individually (or locally)

could not as effectively resolve. Willingness to support

using tax dollars to help disaster victims, the poor, or

other targeted constituencies involves costs, but not the

same level of persona l risk that in dividually inter veningin an emergency may involve. In add ition, public deci-

sions to intervene also involve tra de-offs tha t may n ot

typically cha racterize individual-level decisions to in ter -

vene in an emerg ency. For example, spendin gxamount

of dollars on d isaster aid mean s that there is that much

less money left to spend on other public programs, such

as education, defense, or infrastructure. H ow do people

believe that we sho uld dea l with th ese compet ing claims

on the pub lic purse? When are people prepared to say

that we should increase taxes to help more people at the

margins, and when do they draw the line and say enough

is enough ?

Public compassion as a research topic can draw on

and inform not only theories of prosocial behavior, but

also theo ries of distributive justice an d allocat ion beha v-

ior. Rawls (1971) defined distributive justice as the set of

principles that people use to decide how to distribute

both the benefitsa nd burdensof socialcooperation. Pre-

sumably one benefit of social cooperation is that com-

munities can provide a safety net for their citizens. In

add ition to further exploration of the psychology of pub-



15/16

lic compassion and how people think we should distrib-

ute the b enefits of social cooperation, further research

should a lso begin to explore how people think the bur-

dens of social cooperation (e.g., taxes, nuclear waste

dumps, and wartime conscription) should be shared and

distributed.

NOTES

1. The 1993 Midwest Floo d was associat ed with th e large st amoun tof precipitation, river levels, area o f flood ing, an d econ omic losses inU.S. recorded history (Interagency Flood Plain Management ReviewCo mmitt ee [IFMRC], 1994; Natio nal Weath er Service, 1994).

2. This strategy was used for identifying political groups for the fol-lowing reasons:( a) the principal compon ents analysisr evealed a singlefactor solution with the measuresused, so it was reasonable to treat lib-eralism/conservatism a s a single bipolar d imension, at least with th issample;( b) using self-identification alone has been judged as problem-atic, because people in d ifferent areas of th e countr y may view them-selveso nly in termso f their neighbo rs (e.g., someon e who callshimselfor herself a conservative in Berkeley, California m ay not loo k very simi-lar attitudinally to someone who calls himself or herself a conservativein Tuscaloosa, Alabama ), so self-iden tificatio n was weighted with per-

sonality and attitudina l scale items to mo re accurately gauge relativedegrees of liberalism an d con servatism, a method judged as superiorto self-identification alone (e.g., Snider man & Tetlock, 1986; Ston e,1983); (c) a com parison of different iden tification strategies indicatedthat the present method (dropping the middle 20%and using scoresd e r ive d f rom a p r inc ip a l c omp one n t s a na ly s i s o f b o t h s e l f -identification items, as well as more subtle personality/attitudinalitems) yielded the best discrimination on the validation ch eck items;and finally, (d) including a moderate group did no t yield any inform a-tive resultso r chan ge the resultso bserved with a dichoto mous strategy.

3. U nless explicitly stated other wise, all pa ir-wise comparisonsbetween any means described in the resultssection weresignificant at p

ideological and attributional boundaries on public compassion - reactions to individuals and...

Documents