Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

The Honoring of Accounts

Author(s): Philip W. Blumstein, Kathryn Groves Carssow, Janet Hall, Barbara Hawkins, Ronald
Hoffman, Ernest Ishem, Caroll Palmer Maurer, Dana Spens, John Taylor and D. Lee Zimmerman

Source: American Sociological Review, Vol. 39, No. 4 (Aug., 1974), pp. 551-566
Published by: American Sociological Association
Stable URL: http://www.jstor.org/stable/2094421
Accessed: 16-12-2015 11:32 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

Sage Publications, Inc. and American Sociological Association are collaborating with JSTOR to digitize, preserve and extend
access to American Sociological Review.

http://www.jstor.org

This content downloaded from 130.194.20.173 on Wed, 16 Dec 2015 11:32:21 UTC
All use subject to JSTOR Terms and Conditions
THE HONORING OF ACCOUNTS*

PHILIP W. BLUMSTEIN
with
KATHRYN GROVES CARSSOW, JANET HALL, BARBARA HAWKINS,
RONALD HOFFMAN, ERNEST ISHEM, CAROLL PALMER MAURER,
DANA SPENS, JOHN TAYLOR andD. LEE ZIMMERMAN

University of Washington

AmericanSociologicalReview1974, Vol. 39 (August): 551-66

An experimentwithina questionnairewasdesignedto isolate factors that would predict the


honoring of accounts. Subjects acted as bystandersand read short vignettesdescribingan
interactioncontainingan offense by one actor, a demandfor an account by the other, and an
account by the former. The context and offense, the type of account, and the status
relationshipbetween demanderand accounter were systematicallyvaried.After readingeach
vignette, subjects rated the offense, the demand, the account, and the accounter on several
dimensions.Factorsfound to affect the predictionof honoringbehaviorwere: the moralworth
of the offender, his penitence, his superior status relative to the demander, and the
offensiveness of the violation. Honorability was predicted by moral worth, the offender's
personal control over the offense, and the appropriatenessof the demand.Differencesin the
predictionof honoringbehaviorand honorabilitywerediscussed.

If the majortaskof sociologyis to discover "changingthe meaning that otherwise might


the foundations of social order, then it is a be given to an act, transformingwhat could be
useful strategy to focus on events that seen as offensive into what can be seen as
threaten or disrupt existing order. A rich acceptable."
source of data is providedby the microscopic Any line of social action has the potential
conflicts that arise among the minutiae of for being interpreted as untoward or offen-
everyday interaction. A simple tabulation of sive. When the initiator of the act has any
such common utterances as, "Excuse me," "I reason to expect such an interpretation,he
did it because .. ." and "What I meant to engages the remedial process. He is likely to
say. . ." makes it apparentthat what we take proffer an account, "a statement . . . to
to be routine, smooth-flowing episodes of explain [the] unanticipated or untoward
interaction are really punctuated by a diverse behavior.. ." (Scott and Lyman, 1968:46).
set of common disturbances. These distur- Such a remedialinterchangeis triggeredby an
bances threaten the workingconsensusunder- offense, opened by an account on part of the
lying the encounter, and momentarilydisrupt offender, and closed either with a gesture of
orderly interaction. Because of the disparity honoring on the part of the offended person
between our intentions and our acts, between (signalling the restoration of interactive
meaningswe hold and those held by others, equilibrium), or else with some form of
between what we want and what others are disengagementprocess following a failure to
willing to give, we are often called upon to honor. The interchange may have many
repair fracturedsociation. We engage in what rounds, where an initial account is rejectedas
Goffman (1971:109) calls remedial work, inadequateand others are proffered,or where
the rejection of an account, in turn, requires
accounting.
*This study was conducted as a project for an
undergraduatehonors seminarat the Universityof The desire to learn the motivation for an
Washington.We wish to thank Lowell Hargensfor offense is only the manifest-and often
his commentson an earlierdraftof this paper. secondary-reason for demandingan account.
551

This content downloaded from 130.194.20.173 on Wed, 16 Dec 2015 11:32:21 UTC
All use subject to JSTOR Terms and Conditions
552 AMERICANSOCIOLOGICAL
REVIEW
An offense requires that an account be 1973:108). A repeated violator is saddled
demanded in order to nullify the negative with a "personattribution,"and is imputed to
implications concerning the offender's regard have impaired moral character; hence, his
for the identity of the offended party accounts are not honored.
(Goffman, 1971:122). Furthermore, the of- In choosing whether to honor an account,
fended party must demand an accounting, the demander acts in many ways like a legal
insofar as he has a stake in the continuous tribunal, integrating all the evidence to
flow of interaction, in order to parry the establish the "true" responsibility for the
offense's stigmatizing implications for the offense. Because normal actors are awarethat
identity of the offender and his orientationto the responsibility attribution process is
the moral order. In focusing on these triggered immediately following an offense,
motivational factors, we are ignoring situa- they are eager to manipulate this process to
tions where the demandermight wish to sever their advantage. Hence the ubiquity of
the relationship, to pick a fight, or to alter accounts claiming such forms of reduced
radically the offender's situational identity. responsibility as external culprits or forces,
Given this restriction,we may assumethat the uncontrollable urges and impulses, un-
demanderhas a positive stake in normalizing foreseeability of the consequences of the
the interaction, and should thereforeabet the deed, lack of intention for such consequences,
offender in issuing the proper ritual expres- etc. (Goffman, 1971:98-9, 110-11; Scott and
sions (Garfinkel, 1956:423). At the same Lyman, 1968:47-9). As Goffman (1971:112)
time, however, he must demand a minimal and Weinstein(1966:396) argue,the offender
level of satisfaction with the account(s) given; will often boldly proclaimwhat he would like
hence, he will honor only accounts meeting the other to interpret as his true motivation,
normativespecifications. in an attempt to dissociate the evaluationof
A successfulaccount is one honored by the the offense from the evaluation of his moral
demander. But honoring ultimately depends character. A successful claim to reduced
on the total meaning held by the offended personal control over one's acts is likely to
party, incorporatingnot only the account, but preserveface and insurehonoring.
also the demander'sstake in restoringorder, An offender may also return to a proper
and his construction of the nature of the moral position by a display of penitence. By
offense. The decision whetheror not to honor showing proper respect for the rule he broke,
involves integratingmany different elements the offender lays claim to the rightto reenter
and assessing the goodness of fit between the moral gracesof the offended party, who,
offense and account. The purpose of this by demanding an account, becomes the
paper is to examine some of the inferential momentary guardianof respectability (Goff-
elements that go into this process, and to use man, 1971:107). Showing penitence, like
them to predictthe likelihood of an account's claimingreducedpersonalresponsibility,splits
being honored. the identity of the offender. He asserts his
An account should be honored when the own guilt for the act and accepts the
moral character of the offender has been momentaryblow to his moralcharacter,while
restored. Goffman (1971:108) emphasizes at the same time he reaffirmshis overriding
that remedial work allows the offender to righteousness (awareness of the rules) and
offer a corrective reading of his behavior, acknowledgesthe offended's right to demand
indicatingthat if he has appearedto have the an account (Gusfield, 1967:179; Sykes and
wrong relationship to the rules, he can be Matza, 1957:666). Since rituals of contrition
expected to return to the right path. Scott are so commonly the final step in restoringan
and Lyman (1968:47), in their discussionof encounter to moral balance, their omission is
the invocation of the account accident, argue likely to be a barrierto honoring,which is the
that honoring occurs precisely because the offended party's only way of announcing
same accident does not occur repeatedly to normalization.
the same person, and is therefore not bound In all of his accounting behavior, the
to that person's social character.As attribu- offender must establish the credibility of his
tion theory informs us, an effect will tend to account by assuring its internal consistency
be attributedto the one of its possible causes and its congruence with all the "facts" (e.g.,
with which it covaries over time (Kelley, his relevantpast conduct, the consequencesof

This content downloaded from 130.194.20.173 on Wed, 16 Dec 2015 11:32:21 UTC
All use subject to JSTOR Terms and Conditions
THEHONORINGOF ACCOUNTS 553
the offense and their foreseeability, the usual 1969). It should be the case that those of
motivation for such an act, etc.). In remedial lower status are more constrained to honor
work, as in all of social interaction, attention accountsby their superiors.
to the credibility of one's communicative Relative status not only affects the
behavioris a ubiquitous burden(Hovlandand constraints to honor an actual account, but
Weiss, 1952; Weiler and Weinstein, 1972; somewhat independently,has implicationsfor
Weinstein, 1966). All things being equal, whether that account deserves honoring. So,
accounts are honored to the extent the for example, while a subordinate might be
demanderfinds them credible. likely to honor a superordinate'saccount, he
In deciding whether or not to honor an might not feel privately that the account was
account, the offended must consider-all adequate. At the same time, an uninvolved
things being equal-if the account were bystander might or might not feel, in the
adequate to the circumstances and the moral abstract,that the account was appropri-
offense. In real encounters, however, actors ate. The relativestatus between offender and
do not have the luxury of "all things being offended should have major influence on
equal." Situationalfactors must enter into the overt honoring behavior, somewhat reduced
honoring decision, including, for example, impact on the offended party's private
characteristicsof the relationshipbetween the evaluation of honorability, and still smaller
parties prior to the offense (part of what impact, in the abstract, on the account's
Scott and Lyman [1968:53] call "back- worthinessof being honored.
ground expectancies"), characteristicsof the A final concern in predicting account
offense itself, and characteristicsof the initial honoring is the nature of the offense that
demandpresentation. instigated the remedialprocess. As Scott and
Violations, demands, and accounts do not Lyman (1968:51) point out, a frequently
occur in a social vacuum, but are "situated potent account is to justify an act by
according to the statuses of the interactants" minimizing its offensiveness. If the offender
(Scott and Lyman, 1968:46). Makingcertain can successfully argue a denial of injury, he
demands with certain gestures, and giving will suffer less deterioration of moral
certain accounts, all have implications for character. "An account is treated as illegit-
both the identity of the speaker and the imate when the gravity of the event exceeds
identity projectedfor the other. The offended that of the account" (Scott and Lyman,
party lays claim to an identity possessingthe 1968:54).1
right to take offense at such an act, and
attempts to altercast the offender into an METHOD
identity where accountability is required. If Data aimed at predictingthe likelihood of
the offender provides an account without account honoring were provided by an
objection, it is tantamount to his accepting experiment within a questionnaire. Subjects
the identity into which he is being cast. Not were treatedas bystanders,whose reactionsto
only are identities establishedand modified in written interaction vignettes provided the
this way, but the remedial process often basic data. The assumptionwas made that the
reflects and reaffirms identities already description of a common violation followed
established. Such identities are often coordi- by a demand and then an account, would
nated through a pattern of relative status. As allow subjects to predict confidently the
Scott and Lyman (1968:57) illustrate, "the ensuing reaction of the demander,to infer the
vulnerabilityof actorsto questions concerning
their conduct varieswith the situation and the
1 Whilethis position seemsreasonable,it mightbe
status of the actors. Where hierarchies of
arguedthat the perceivedoffensivenessof a violation
authority govern the social situation, the is entirelya function of the successof the accountin
institutionalized office may eliminate the offsetting the negative implications of the act.
necessity of an account, or even prevent the Therefore,the seriousnessof the violation,judgedin
question from arising." Studies have shown a vacuumwould have no independenteffect on the
that the same communicativeexigencies will likelihood of honoring;whereas,the seriousnessof
the deed in its accounted context would have a
be met quite differently depending on the strong impact. A somewhat less social construc-
status relationship between the actors (e.g., tionist argumentwould hold that both effects might
Jones et al., 1963; Weinsteinand Beckhouse, exist independently.

This content downloaded from 130.194.20.173 on Wed, 16 Dec 2015 11:32:21 UTC
All use subject to JSTOR Terms and Conditions
554 AMERICANSOCIOLOGICAL
REVIEW
motivation and dispositions of the accounted, Georgebeing fellow players,(2) Dan being the
and to give a normative evaluation of the captain of the team on which they both play,
behaviorin question. or (3) Dan being a younger student who has
just joined the team. For each of the eighteen
Instruments different situation-by-relationship combina-
Each instrumentwas presentedto subjects tions there were six different types of
as an "impressionformation questionnaire," accounts.3 In the teammate example, one of
containing six vignettes, each on a separate them was:
page, followed by a battery of questions. Each
vignette described two people in interaction: George responds, "I didn't realize there
an offense was committed by one, an account was a practiceyesterday."
demanded by the other, and an account
offered by the first. While the vignettes did not represent a
There were two types of instruments;in systematic sampling of all remedial inter-
one, each vignette was followed by a page of changes, they were designed to maximize the
standardsemantic differential items designed variability along certain contextual dimen-
to obtain the subject's impressions of the sions. One shortcoming of the sampling was
accounted.In the other, the same vignette was the reduced variabilitycreatedby the absence
followed by a series of contextual questions of intentionally inadequate accounts designed
concerningthe act, the demand, the account, to jeopardize the working consensus, e.g.,
and the internal workings of the accounter. "Whocaresabout basketballpractice."
These questions were standardized across
vignettesand providedthe majorvariables. ExperimentalDesign
The factorialcombinationof six situations,
Vignettes six account types, and three levels of
relationship (demander superordinate, de-
Six different basic vignette situations were
mander and accounter equal, demander
created, each having several variable forms.
subordinate),provided 108 unique vignettes.
Hereis a samplevignette:
These were combined in a modified Graeco-
Dan and George have both been playing Latin square design with each subject
first string for three years on a high school receiving six vignettes: one each of the six
varsity basketballteam. Three days before situations, one each of the six account types,
a league game, Dan and George are in the and two each of the three status relationships.
locker room.
Dan asks George, "How come you missed Subjects
the last practice?The team needed you."2 Subjects were students in sections of
introductory sociology or in a course on the
Each of the six vignette types had three family at the Universityof Washingtonduring
sub-types, varying the relative status of the winter quarter 1973. Half the subjects
demander and accounted, e.g., (1) Dan and received semantic differential instruments,
and half received the contextual questions.
2In this example the victim of the offense is a Withineach half there were eighteen different
group to which both demander and accounter booklet types (108 vignettes, with six per
belong. The other basic vignetteswere: a machinist booklet type), which were randomly distrib-
asked why he is not wearing safety goggles (the uted to subjects. Vignettes were randomly
victim is the accounted);a buyer for a department
store asked why she has not returneda borrowed ordered within each booklet. From each half
book (the victim is the demander);a family member of the subjects, 180 valid booklets were
asked why she failed to bringhome a cake for Dad's returned,providingten subjectsper condition.
birthday (the victim is a person outside the
encounter);one person asked by another why he 3The six account types were: appeal to (1) a
allowed his car to pollute (the victim is society in transsituational internal cause (a trait of the
general);and a newly-wedaskedwhy she has not yet accounted), (2) a situational internal cause (a
sent thank-younotes for weddinggifts (a violation temporary state of the accounted), (3) a cause
of etiquette). Note that the setting of the vignette, external to the accounted,(4) a denial of injuryin
the nature of the offense, and the nature of the the offense, (5) a denialof the victim'srightto claim
demandwereall confounded. an offense, and (6) an appealto a higherloyalty.

This content downloaded from 130.194.20.173 on Wed, 16 Dec 2015 11:32:21 UTC
All use subject to JSTOR Terms and Conditions
THEHONORINGOF ACCOUNTS 555
Analysis the subject's need to maintain response
In asking subjects to respond to vignettes, consistency. The measure of Account Ade-
we were not concerned with their assessment quacy (D) used the dimension "adequate,"
of what they would do under such circum- and the measure of Account Credibility (D)
stances, but rather with their impressionsof was the sum of two items, "believable"and
various elements of the accounting episode, "sincere,"which were correlatedr = .740.
and how these would affect their assessment The questions aimed at tapping the
of the likelihood that honoring would be the subject'spersonalmoralassessmentbegan,
outcome. We were not interestedin individual
judgments or in correlations based on the You, as an outside observer, may feel
differently about [the accounter's] answer
experimental subject as unit of analysis. than [the demander] felt. On the following
Rather we were concerned with propertiesof scales indicate how you personally felt
episodes, which were taken to be the mean about [the accounter's] answer.
scores on all variables for a given stimulus
unit. Hence the 108 unique vignettes were The measure of Account Credibility (S) was
units of analysiswith the raw data for subjects again the sum of "believable"and "sincere"
(ten per unit) averagedto give properties of (correlated r = .811). Account Adequacy (S)
vignettes. Correlations and regressionswere was the sum of two items, "adequate"and the
based on these 108 observations.4 responseto the question,
Measures Do you feel [the accounter's] answerwas
Five measures characterizingthe account appropriate?
were generatedfrom the questionnaireitems.
The major dependent measure, the subject's The correlationwas r = .902.
assessment of the likelihood that the de- Two measures characterizedthe demand:
mander would actually honor the account Demand Appropriateness, based on the
(Honoring) was the sum of two items based question,
on the following question:
Do you feel the question [the demander]
After hearing [the accounter's] answer, asked [the accounted] was appropriate?
how do you think [the demander] will
behave? and DemandLegitimacy,measuredby,

The two dimensions, "favorably" and "ac- Do you feel that [the demander] had the
ceptingly," were correlated r = .868. (All right to ask that question of [the
items were seven-pointpolar opposite scales.) accounted]?
The four other account measuresfit into a
two-by-two classificationscheme: (1) whether These two measuresseemed to tap somewhat
the adequacy or credibility of the account different content; their correlation was r =
were being measured, and (2) whether the .618.
subject were to role-take the demander's The Offensiveness of the Violation was
perspective or to give his or her own moral measuredby askingthe subject to
perspective.The demanderoriented questions
began, ... indicate [his] feelings about the
seriousness of the behavior . . . being
objected to.
After hearing [the accounter's] answer,
[the demander] will behave in a certain The six offenses
were presented for separate
way. But how will he reallyfeel about [the
accounter's] answer? responses(e.g., "Georgemissingthe basketball
practice").RelativeStatus was the sum of two
This wording was purposely used to minimize items:
In general in their relationship, is [the
4The use of mean data increasesreliabilityby demander] or [the accounted] more likely
reducingerrorvariance(Shaycoft, 1962). to give adviceand makesuggestions?

This content downloaded from 130.194.20.173 on Wed, 16 Dec 2015 11:32:21 UTC
All use subject to JSTOR Terms and Conditions
556 REVIEW
AMERICANSOCIOLOGICAL
and The inference measures were expected to be
interrelated, but no causal model was
.. . more likely to need help or guidance designated. Among the dependent measures
from the other? (the accounts block) the expectations were
straightforward:Honoringbehaviorshould be
The items correlated r = .756, with a high a joint function of both the perceived
score given for the demanderbeing superordi- credibility of the account and its judged
nate. adequacy, with adequacy also being a partial
Three measures tapping the subjects' function of credibility. The five accounts
inferences about the causes and motivations measureswere not expected to be uniformly
behind the account were Incorrigibility, affected by other measures. It was argued
Personal Causation, and Penitence. Incorrigi- above that RelativeStatus wouldX have greatest
bility was based on the question, impact on Honoring, less impact on the
demander'sevaluation of adequacy, and least
How likely do you think it is that in a impact on the subject's evaluation.The effect
similarsituation [the accounted] would do of status on credibility was an open question:
something like [the offense] again? high status people are often given greater
trust; yet since they are seldom called to
PersonalCausationwas based on the question, account, any demand might signal their true
untrustworthiness. The model within the
In your opinion, was the true reason [the situation block reflects a simple temporal
accounter committed the offense] within sequence in the bystander's interpretation
his control, or was it out of his hands? process.
Penitence was measuredby, RESULTS
Before evaluating the predictionsfrom the
How bad do you think [the accounted] felt
about [the offense] ? model in Figure 1, let us consider the
variation produced by the experimental
manipulations. Table 1 shows the variance
Finally, the MoralWorth of the accounter
explained in each of the measures by the
was measuredby the sum of four items from
manipulation of situation, account, and the
the semantic differential questionnaire: "un-
interaction of these two.5 It is encouragingto
trustworthy/trustworthy," "worthless/valu-
observe that the two manipulationsaccounted
able," "irresponsible/responsible,"and "sin- for such a substantialbulk of the variation.
cere/insincere" (the coefficient alpha was
Among the five account measures,the nature
.820).
of the offense (the situation manipulation)
had no significant independent effect, while
Model the account manipulationexplained substan-
Predictedrelationshipsamongthe measures tial variation, both alone and in interaction
were generated from the earlier theoretical with situation. Moral Worth and the three
discussion,andhavebeen illustratedin Figure1. inference measureswere significantlyaffected
The four blocks in the figurerepresentclasses by both manipulations as well as their
of measures,whose causal sequencingreflects
the inferential steps by which people process 'Variation explained by the situation and
information about an encounter, beginning account manipulationswere the R2's derived from
with backgroundexpectanciesand an offense, two dummy variable analyses, each with five
proceeding to a demand, and ending with an independent variables.Because these two sets of
variableswere orthogonal,there was no redundant
account. Briefly, it was expected that explained variance. Variation explained by the
Incorrigibility,PersonalCausation,and Offen- interactionof the two variableswas the incrementin
siveness would have negative effects on Moral R2 providedby the twenty five dummy crossprod-
Worth, Credibility, Adequacy, and Honoring, ucts, over and above that explained by the ten
while the effect of Penitence would be dummy variables alone. For all measures but
RelativeStatus,the relationshipmanipulationand its
positive. interactionsfailed to explain significantportionsof
Within the three blocks with more than variance.Hencetheircontributionsarenot presented
one measure, some predictions were made. in the table.

This content downloaded from 130.194.20.173 on Wed, 16 Dec 2015 11:32:21 UTC
All use subject to JSTOR Terms and Conditions
THEHONORINGOF ACCOUNTS 557

Measures
Situational

Measures
Inference MoralWorth

LIIIIL LIE

AccountsMeasures

- RelativeStatus 7 - Penitence
2- Offensiveness 8 - MoralWorth
3- DemandAppropriateness 9 - Credibility(D)
4- DemandLegitimacy 10 - Credibility(S)
5- Incorrigibility II - Adequacy(D)
6- PersonalCausation 12 - Adequacy(S)
13 - Honoring
Figure1. ModelIncorporatingMeasuredVariables

This content downloaded from 130.194.20.173 on Wed, 16 Dec 2015 11:32:21 UTC
All use subject to JSTOR Terms and Conditions
558 AMERICANSOCIOLOGICAL
REVIEW
Table 1. Variance Explained (R2 ) in Measured Variables by Manipulated
Variables and Their Interaction

Manipulated Variable
Measured Variable Sum
Situation Account Interaction

Honoring .070 .457** .167t .694**


Adequacy (D) .027 .448** .173 .647**

Adequacy (S) .023 .606** .185** .814**


Credibility (D) .077 .278** .307** .663**

Credibility (S) .042 .441** .262** *745**

Moral Worth 130* .499** .194** .824**


Incorrigibility .125* .365** .193* .684**
Personal Causation .181** .428** .219** .829**
Penitence .162** .280** .205* .647**

Relative Status .080 .093 t .077 .250

Offensiveness .611** .030 .085 .726**

Demand Appropriateness .462** .051 .150 .633**

Demand Legitimacy .435** .033 .159 .627**

t .05 ' p < .10.


* p < .05.

** p < .01.

interaction; but again, the account had the to the account'scorrectivereading.


greatest independent effect. These data lend
systematic support to the assertion that RelationshipsAmong DependentMeasures
people do not react so much to what we do, Before evaluating the overall regression
but ratherto the interpretationwe providefor model, let us considerthe relationshipsamong
our acts. the measureswithin the accounts block (see
The situational measuresshould obviously Figure 2). Given this submodel, Credibility
have only been subject to situational varia- had no independent effect on Honoring,
tion, and such was exactly the case. The although it did have a sizable impact on
findings for Offensiveness(61 per cent of the Adequacy, which in turn affected Honoring.
variance explained by the situation, while The substantial zero-order relationships be-
only 3 per cent explained by the account and tween the Credibilitymeasuresand Honoring
8.5 per cent by their interaction) reflect the (r = .590 for the demanderand r = .626 for
notion that an act may be judged in the moral the subject) were totally vitiated by the
abstract even after an account has been intervention of Adequacy. Whateverimpor-
invoked. Offensiveness,then, was not simply a tance the credibility of accounts has for
construction based on the subject's response ultimate honoring, it is only in conjunction

This content downloaded from 130.194.20.173 on Wed, 16 Dec 2015 11:32:21 UTC
All use subject to JSTOR Terms and Conditions
THEHONORINGOF ACCOUNTS 559

.836**

Credibility (D) Credibility (S)

.371 *.839*

.379 -.045

Adequacy (D) Adequacy(S)


.080 -.068
-2= 516 * R2 =-.643 **

.430** .413**

Honoring

| 2 =.664 * ** p <.01.

Figure2. Honoringas a Functionof Credibilityand Adequacy.

with other judgments that enter into the consider-along with credibility-when evalu-
assessment of adequacy. It is not surprising atingthe overalladequacyof the account.
that the subject's feeling of credibility alone
affected his judgement of adequacy, but it is 6The substantialzero-orderrelationshipbetween
curious that his imputation of the demander's the two adequacymeasures(r = .818) was too large
assessment of adequacy was colored by both to be accounted for solely by their shared
credibility perspectives. It is as though the dependence on the credibility measures.To assess
how much of this relationshipoccurredbecauseall
subject felt he had a private insight into the measureswere based on a singleset of vignettes,and
account's credibility which could be trans- how much was independent of these stimuli, the
lated into some factor the demander would situation and account manipulations (dummied)

This content downloaded from 130.194.20.173 on Wed, 16 Dec 2015 11:32:21 UTC
All use subject to JSTOR Terms and Conditions
560 REVIEW
AMERICANSOCIOLOGICAL

.250

RelativeStatus Offensiveness

-.003 _

Incorrigibility PersonalCausation Penitence

R-=.020 R2= 114 R2=03

.094 -.136 -.397** .202* .

MoralWorth t of < p< IC


* p < .05.
R2= .185** ** p < .01.

Figure 3. Moral Worth as a Function of Relative Status, Offensiveness and Inference Measures

Relationships Among Independent Measures inference measures or on Moral Worth.7


The effects on the inference measuresand Situational factors had little impact on the
on Moral Worth of the situational factors inference measures,the only exception being
appear in Figure 3. The demand measures the tendency for accounts following more
were not included in the figure because offensive violations to be seen as within the
neither had a significanteffect on any of the offender's personal responsibility and as not
engendering penitence. The offensiveness of
were partialledout. The resultingpartialcorrelation the deed had an independent effect on the
between the two adequacy measures,removingthe accounter's ascribed MoralWorth,but not in
effects of both the manipulationsand the credibility the expected direction:the more offensive the
measures,was only reduced to r = .578, suggesting
that there were other factors common to the two
measuresof adequacythan those we haveisolated.A
similar analysis was performed on the credibility 7This might be attributableto the redundancyof
measures,to determineif their zero-ordercorrelation the demand measures. To assess this possibility,
(r = .836) was produced fully by the vignette multiple-partialcorrelationswere computed, using
variations.The resultingpartial correlationwas r = the demand measuresas independentvariablesand
.679, again suggesting unspecified sources of the appropriatemeasuresas controls. Therewereno
association. significantcoefficients.

This content downloaded from 130.194.20.173 on Wed, 16 Dec 2015 11:32:21 UTC
All use subject to JSTOR Terms and Conditions
THEHONORINGOF ACCOUNTS 561
act, the more uprightits perpetrator.8Of the .250*
inference measures, only Personal Causation
had an independent effect on MoralWorth:if
the offender was seen as responsible for the
act, his moralcharacterwas blemished.9
The inference measureswere significantly RelativeStatus Offensiveness
interrelated.The zero-ordercorrelationof r=
.488 between Incorrigibility and Personal
Causation was predictable from attribution
theory (Kelley, 1973): any deed perceivedas
likely to occur with relativefrequencyshould .118 .174k 457** .247**
also be perceived as emanatingcausally from
the actor. A negative association between
Incorrigibilityand Penitence (r = - .528) was Demand Demand
not surprising, considering that a major Appropriateness Legitimacy
element of contrition is the promiseof future R2 =.250** R2 = .113**
moralconduct. It would have been difficult to
predict a relationship between Personal
Causationand Penitence(r = - .598): one can t .05 <p < .10.
imagine different accounts which would * p<.O5.
produce either a negative or a positive ** pc.OI.
association.'0
Relationships among the situational mea- Figure4. DemandAppropriateness andLegitimacyas
Functionsof RelativeStatusandOffensiveness.
sures appearin Figure 4. Since there were no
firm theoretical grounds for predicting a
causal link between Relative Status and
Offensiveness,the relationshipis presentedas
a significant correlation coefficient: the violations were judged more offensive when
the perpetratorwas of lower relative status.
The cause of this finding was not intuitively Since DemandAppropriatenesswas somewhat
obvious, so rather than attempt to construct an more situational, and Demand Legitimacy
explanation, it was assumedthat it reflected some based more on the identities of the parties,
unique patternof vignette elements.Any interpreta- the regressioncoefficients in the figureare not
tion should be made in light of the significant
interventionof PersonalCausation,since when this startling. Relative Status had no effect on
factor was deleted from the analysis,the Offensive- Appropriateness, but did tend to affect
ness effect on MoralWorthwas not significant(b* = Legitimacy (p = .065). The higher status of
.126). the demandergave him the right to make a
9 Zero-ordercorrelationsbetween inferencemea-
demand. Offensiveness, on the other hand,
sures and Moral Worth were: Incorrigibility,r =
-.129 (n.s.);PersonalCausationr = -.367 (p < .01); affected both demand measures. The more
Penitence,r = .248 (p < .01). offensive a violation, the more likely its
1 0These zero-orderassociationsobviouslycould ensuing demandwould be appropriateand the
not be attributed to the impact of the situational demanderwould be allowed (or required)to
factors,so a regressionwas againperformed,adding make it. "'
the manipulationsto test whether the common
vignettebase would explainthe correlations.Two of
the observedcorrelationswere washed out by this Predictionof Accounts Measures
analysis, while that between Incorrigibilityand
Penitence was only reduced to the partial of r = In order to observe the relationships
- .355 (p < .01). Failureto removethis association between each of the independent measures
suggests that the validity of the ac.-ounter's and the accounts measures,zero-ordercorrela-
expressed penitence was inferred, at leas! in part, tions are presented in Table 2. All of the
from the incorrigibilityimplied by the account,
ratherthan from the content of the accountdirectly. independent measureswere significantlyasso-
Becauseof the durabilityof this associationit was
momentarilyappropriateto treat the measuresas I IRegression analysesincorporatingthe manipu-
though indicatorsof the same variable.A multiple- lated variables failed to wash out the zero-order
partial correlation to assess their joint effect on correlations between Offensiveness and Relative
MoralWorthproduceda non-significantr2 = .0 1. Statusandbetweenthe demandmeasures.

This content downloaded from 130.194.20.173 on Wed, 16 Dec 2015 11:32:21 UTC
All use subject to JSTOR Terms and Conditions
562 AMERICANSOCIOLOGICAL
REVIEW
Table 2. Correlations between Dependent Measures and Independent Measures

Dependent Measure
Independent
Measure Honoring Adequacy Credibility
(D) (S) (D) (S)

Moral Worth .492** .503** .571** .506** .585**

Incorrigibility -.272** -.345** -.430** -.255** -.325**

Personal Causation -.480** -.581** -.701** -.371** -.538**

Penitence .387** .428** .553** .378** .489**

Relative Status -.302** -.241* -.224* -.047 -.112

Offensiveness -.353** -.317** -.343** -.002 -.117

Demand Appropriateness -.217* -.229* -.251** .033 -.144

Demand Legitimacy -.223* -.108 -.154 .031 -.102

* p < .05.

** p < .01.

ciated with the major dependent measure, that implied properregretwere more likely to
Honoring. As expected, an account was be believed.
judged more likely to be honored if the Table 1 showed the considerableimpact of
accounter's moral worth was not impaired,if the situation and account manipulationson
he was unlikely to repeat the offense, if he reactions to the vignettes. It was appropriate
was not totally responsible for the violation, to ask how much of this variancecreated by
and if he was properly repentant. Honoring the manipulationscould be explainedwith the
was more likely when the violation was less measured independent variables. While the
offensive, when the initial demand was data in Table 2 could easily be translatedinto
inappropriate or illegitimate, and when the measuresof the explained proportion of the
violatorwas of higherrelativestatus. total variationin the dependent measures,we
The patterns for both the demander's might instead want to look at the explained
attributed assessment, and the subject's proportion of the variance produced by the
personal assessment, of whether the account manipulations.These quantities are presented
should be honored were parallel to those for under the heading ratio in Table 3.1 2 The
honoring behavior,except that the legitimacy ratios show that the four independent
of the demand did not reach significance. measures had a strong impact on the
Different patterns emerged among the Credi- dependent measures.That so much variation
bility findings. Not surprisingly,there was an could be explained by so few measuresin this
intimate relationshipbetween Credibilityand areaof researchis encouraging.
the moral character of the accounted, and Table 4 presentsregressiondata to evaluate
again, not surprisingly,there was no relation- the model in Figure 1. Looking first at the
ship between Credibilityand factors external prediction of Honoring, we find independent
to the account (situational measures). There effects for MoralWorth,Penitence, Offensive-
were, however, interesting relationships be-
tween Credibility and the inferences drawn 1 2 Only the inference measuresand MoralWorth
from the accounts. Accounts that reducedthe are presentedin the Table,since these were the only
probabilityof the offense's repetition, or that independentmeasuresexpected or observedto vary
denied full responsibility for the offense, or with the accountmanipulation.

This content downloaded from 130.194.20.173 on Wed, 16 Dec 2015 11:32:21 UTC
All use subject to JSTOR Terms and Conditions
THE HONORINGOF ACCOUNTS 563
Table 3. Variance Explained in Dependent Measures by Selected Independent
Measures, and Ratio of that Variance to the Variance Explained by
the Intersection of the Respective Independent Measures and the
Manipulated Variables

Dependent Measurea

Independent Measure Honoring Adequ cy Credi'bility


(D) (S) (D) (S)

R2 ratio RZ o R2 r atio R2 ratio R2 ratio

Moral Worth .242 .337 .253 .377 .326 .393 .256 .385 .342 .458

Inference Measures .246 .299 .349 .491 .521 .586 .176 .162 .333 .367

Incorrigibility .074 .107 .119 .181 .185 .223 .065 .045 .106 .367

Personal Causation .230 .291 .337 .475 .491 .554 .137 .155 .289 .329

Penitence .150 .201 .183 .275 .306 .364 .143 .167 .239 .299

Moral Worth and


Inference Measures .359 .454 .446 .634 .634 .721 .331 .395 .501 .592

R2 in all cases significant beyond the .01 level.

Table 4. Standardized Partial Regression Coefficients for Regression of


Accounts Measures on Moral Worth, Inference Measures, and
Situational Measures

Dependent Measur..
Independent

(D) (S) (D) (S)

Moral Worth .384** .362** .384** .427** .438**

Incorrigibility -.043 -.071 -.092 -.045 -.045

Personal Causation -.085 -.301** -.358** -.094 -.228*

Penitence .194* .091 .16 .196t .218*

Relative Status -.174* -.108 -.075 -.014 -.021

Offensiveness -.253** -.139 -.130t .021 .036

Demand Appropriateness .030 -.175t - .162* .023 -.132

Demand Legitimacy -.069 .154 .097 .079 .061

R2 .466** .513** .689** .340'" .510**

t <
.05 < p .10.

* p < ..05.

** p < .01.

This content downloaded from 130.194.20.173 on Wed, 16 Dec 2015 11:32:21 UTC
All use subject to JSTOR Terms and Conditions
564 REVIEW
AMERICANSOCIOLOGICAL
ness, and Relative Status. Honoring was a and one near significant effect (p =..079 for
joint function of the righteousness of the Credibility D); whereas, Personal Causation
accounter,his display of ritualrepentance,the affected the subject's trust, but not the
unimportance of the transgression,and the demander's.1 4 There was no support for
relative superior status of the accounter. either the common finding that high status is
Qualities of the demand were irrelevant to accompanied by greater trust, nor for the
honoring, as were the likelihood that the opposite contention that higher status per-
offense would be repeated and the personal sons, by virtue of having committed an
responsibilityof the offender. offense, are more likely to be deemed
Turning now to a comparision of the untrustworthy.Note that for none of the five
adequacy measureswith Honoring,recallthat dependent measures was there a significant
the only expectation generated was that the effect for Incorrigibility or for Demand
effect of Relative Status would decline as one Legitimacy.'5
moved from Honoring to Adequacy (D), then
to Adequacy (S). Such was the case, although DISCUSSION
there seemed to be no sizable difference
The purpose of this study was to
between the two adequacy measures. Status
investigate the factors contributing to the
was seen by subjects as relevant to overt
restoration of smooth flowing interaction
honoring behavior, but not to either the
demander's nor the subject's view of honor- following an accountable violation. The
research question was framed in terms of a
ability. Moral Worth had quite the same
effect on all three measures, while Personal bystander's prediction of whether or not an
Causation had a strong impact on the account would be honored by the offended
party. The most important predictor of
adequacy measuresalone. This result suggests
that accounts were deemed more appropriate honoring was too obvious to be discussed
directly-the nature of the account itself. The
when they reduced personal responsibility,
content of the accounts provided in the
but this judgment did not carryover into the
vignettes-both the general account types and
demander'sovert response.
the accounts specific to the situated viola-
Penitence affected the subject'sevaluation
tions-explained a great deal of the variation
but not the demander'simputed judgment. in honoring, and in other related measures.
Offensiveness had less impact on adequacy This was particularlystrikingwhen compared
than on honoring (p = .064 in the subject's
case and not significant in the demander's).
question. In all cases the regression coefficients
The natureof the offense seemed to affect the obtained produced a reduction in slope significant
actual response to the violation, but not the beyond the .05 level, suggestingthat the observed
internal reaction to it. In comparison, the relationships between Moral Worth and the
characterof the demand affected the internal accounts measureswere producedby the account,
judgment of adequacy (p = .079 for the rather than by the offender's preaccount moral
stature.
demander), while it was irrelevantto honor- 1 4In order to consider again the impact of the
ing. redundancyof Incorrigibilityand Penitence, multi-
MoralWorthhad even a greaterimpact on ple-partialswere calculated,treatingthe five account
the Credibilitymeasuresthan on Honoringor measuresas dependent variables.The only signif-
icant findings reflected the observed significant
Adequacy. 3 Penitence had one significant regressioneffects for Penitence.
I I Perhaps Incorrigibility failed to produce
significanteffects becauseof its redundancywith the
1 3There remains a question of interpretation other inference measures.To assessthis
possibility,
concerningall the significantMoralWortheffects in regression analyses were performed eliminating
the Table: Are people who are perceived to be either PersonalCausation,Penitence,or both. When
generally more righteous also more likely to have one was omitted, the other invariably became
their accounts honored, or do accounts affect significant;but Incorrigibilitydid not. Whenboth
righteousness,and hence affect honoring?The data were omitted, Incorrigibilitybecame significantfor
in Table 1 indicated that Moral Worth was more each of the five account measures.It can be argued
strongly affected by the account than by the that Incorrigibilitywas the least rich of the three
situation.By addingthe accountmanipulationto the substantivelyoverlappingvariables,and hence was
regressionanalysesreportedin Table4 and observing eclipsed by the others. Further multiple-partial
any reduction in the Moral Worth effects, it was correlations showed no joint effects of the two
possible to gain an imperfect purchase on this demandmeasureson any of the dependentmeasures.

This content downloaded from 130.194.20.173 on Wed, 16 Dec 2015 11:32:21 UTC
All use subject to JSTOR Terms and Conditions
THEHONORINGOF ACCOUNTS 565
to the insignificant amount of variation to bear, how bad was the misdeed, and what
explained by the nature of the violation and ritual gestures were mobilized to atone, all
the situation in which it occurred.It cannot seemed irrelevant. In deciding actually to
be overstated that people respond to our honor an account, its honorabilityseems only
symbolic restructuringof our deeds, much to have been a part of what was necessary.
more than to the deeds themselves. Honoringwas a blend of honorability,proper
It was reassuringto observethat within the ritualdisplay and situationalconstraints.
variation produced by the manipulationsof A general problem concerning researchin
account and situation, a small number of this area has become evident as a result of this
measures could be summoned to provide a initial study: There is a criticalneed for more
high level of prediction. The post-account sophisticated taxonomies for arrangingsitua-
moral character of the offender, plus the tions and interaction contexts. In attempting
combination of inferences drawn from the to createvignettes, our majordifficulty was in
account had uniformly strong impact on the classifying situated violations and account
honoring of the account, on the account's types. In the latter case, the classification
adequacy,and on its credibility. scheme used by Scott and Lyman(1968) was
Five dependent measureswere included in helpful; but ultimately it was only partly
order to determine the internal dimensions adequate for mapping the universe of
leading immediately to a decision to honor or accounts observedin the real world.
not to honor an account. Not surprisingly,the A major shortcomingof the present study
relationships among honoring behavior, the was the use of written vignettes as stimulus
account's adequacy, and its credibility were materials.While this researchstrategyis often
quite intimate. Subjects expected the of- objectionable, the results obtained seem
fended person'sbehavioralresponseto parallel informative enough, while they await further
closely his internaldispositionsconcerningthe validation. An important next step for
account, and they further expected that the removing some of the limitations of this
latter would also parallel their own moral technique might be to create videotape
evaluations.Honoringlargelydependedon the recordings of actual account behavior which
adequacy of the account, what we might call could vary in quite the same ways as the
its honorability. written vignettes.
While honorability was a major facet of Another weaknessof the presentstudy was
honoring, these two were by no means its failureto integrateadequatelythe remedial
measures of the same variable, as was amply process into other themes in the current
demonstratedby the differentialeffects of the literature on interaction and interpersonal
independent measures on them. While the control. It was suggested above that the
moral character of the accounter strongly remedialprocess normalizesthe encounterby
affected both honoring and honorability, reestablishing the orderly arrangement of
none of the other causalfactors affected both social identities among participants.As Scott
measures. Honoring was influenced by the and Lyman (1968:59) argue, every remedial
ritual repentance displayedin the account, by interchange is a "manifestation of the
the offensivenessof the violation, and by the underlying negotiation of identities." In an
relativesuperiorstatus of the accounted.None attempt to make explicit the relationship
of these factors played a significant role in between remedial work and identity negotia-
honorability. Instead, this internal response tions, we are beginninga study to considerthe
was a product of the mitigation of personal effects of varying the degree of consensus
responsibility for the deed and the appropri- about situationalidentities on the interaction
ateness of the demand. Honorability hinged processes that follow an accountable viola-
partly on a considerationof motivation(what tion. With this design, we hope to be able to
caused the accounter to do it?), partlyon the predict more firmly the maintenance or
credibility of the account, and partly on the destructionof social order.
demand(if the demand should not have been
posed in the first place, how could it deserve
to be honored?). Beyond these factors, REFERENCES
honorability was determined in a situational Garfinkel,H.
vacuum:,what status constraints might come 1956 "Conditions of successful degradation

This content downloaded from 130.194.20.173 on Wed, 16 Dec 2015 11:32:21 UTC
All use subject to JSTOR Terms and Conditions
566 REVIEW
AMERICANSOCIOLOGICAL
ceremonies."AmericanJournalof Sociol- Shaycoft,M. F.
ogy 61 (March):420-4. 1962 "The statistical characteristicsof school
Goffman,Erving means."Chapter5 in J. C. Flanagan,J. T.
1971 Relations in Public: Microstudiesof the Dailey, M. F. Shaycoft, D. B. Orr,and I.
PublicOrder.New York: BasicBooks. Goldberg, Project Talent: Studies of the
Gusfield,J. R. AmericanHighSchool. Pittsburgh:Univer-
1967 "Moralpassage: the symbolic process in sity of PittsburghProjectTalentOffice.
public designations of deviance." Social Sykes,G. M. and D. Matza
Problems15 (Fall):175-88. 1957 "Techniquesof neutralization:a theory of
Hovland,C. I. andW.Weiss delinquency." American Sociological Re-
1952 "The influence of source credibility on view 22 (December):664-70.
communication effectiveness." Public Weiler,J. and E. Weinstein
OpinionQuarterly15 (Winter):635-50. 1972 "Honesty, fabrication,and the enhance-
Jones, E. E., K. J. Gergenand R. G. Jones ment of credibility." Sociometry 35
1963 "Tacticsof ingratiationamongleadersand (June):316-31.
subordinates in a status hierarchy." Weinstein,E. A.
PsychologicalMonographs77 (whole num- 1966 "Toward a theory of interpersonaltac-
ber 566). tics." Pp. 394-8 in Carl W. Backmanand
Kelley,H. H. Paul F. Secord (eds.), Problemsin Social
1973 "The processes of causal attribution." Psychology.New York: McGraw-Hill.
American Psychologist 28 (Febru- Weinstein,E. A. and L. S. Beckhouse
ary):107-28. 1969 "Audience and personality factors in
Scott, M. B. and S. M. Lyman presentationof self." SociologicalQuarter-
1968 "Accounts." American Sociological Re- ly 10 (Fall):527-37.
view 33 (February):46-62.

This content downloaded from 130.194.20.173 on Wed, 16 Dec 2015 11:32:21 UTC
All use subject to JSTOR Terms and Conditions

You might also like