Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
.
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.
Academy of Management is collaborating with JSTOR to digitize, preserve and extend access to The Academy
of Management Review.
http://www.jstor.org
Theoretical, Information
Processing, and
Attribution
Factors
Situational
Affecting
of
Models
Behavior1
Theory
Organizational
ROBERT G. LORD
JONATHAN E. SMITH
University of Akron
Attributional theories used to explain organizational behavior are overly
restrictive. A ttributional processes may vary with the type of attributional
question addressed and with the level of information processing consistent
with situational or motivational factors. Several attributional principles
are organized into a typology involving type of attributional question and
level of information processing. Boundary conditions affecting the use of
particular principles are identified and the importance of differences
among attributional processes are discussed.
useful to consider several factors that affect these
applications of attributional theories. These restrictive factors can be analyzed using a framework that
focuses on both the type of attribution being made
and the level of information processing compatible
with constraints of the situation and the attributor.
In this paper two dimensions will be used-type of
attribution and level of information processing-to
develop a typology of principles for forming attributions.
Type of Attribution
The type of attribution is important because
several distinct questions have been explored by attribution theory research. In particular, attribution
theory may be applied to understanding causality
for a specific event (Kelley, 1973), to assessing responsibility for a particular outcome (Hamilton,
1980), or to assess the personal qualities of persons
involved in the event being considered (Jones & Davis, 1965).
Attributions concerning causal assessments involve producing explanations for unexpected or
puzzling events. Attributors typically are viewed as
following logical or quasi-scientific processes
through which they assign relative priority to poten-
50
amount of information processing they assume. Indeed, conceptually, one might employ a continuum
that ranges from the assumption that attributors are
highly active, involved information processors to an
assumption that they are passive and limited information processors as one describes differences in attributional theories. Recent research on cognition
and social judgment supports such an approach.
Schneider and Shiffrin (1977) and Shiffrin and
Schneider (1977) argue that a stimulus can evoke
either a controlled cognitive process in which the individual consciously processes information (stimuli)
as he or she makes decisions or an automatic process that does not require conscious monitoring of
stimuli. Evidence from encoding information into
memory also suggests that the process varies in regard to energy or attentional requirements (Hasher
& Zacks, 1979). Encoding may be relatively
automatic and use little energy from the attentional
mechanism, or it may be highly effortful and
elaborate. Finally, this distinction also can be recognized in the area of attribution theory (Feldman,
1981). Kelley (1973), Jones and Davis (1965), and
Weiner (1972) postulate that attributions occur as a
result of individuals seeking explanations for
behavior. They emphasize a rational or effortful
process in which the individual weighs or considers
multiple pieces of information in a conscious attempt to explain situations or understand individuals. However, recent research points out that
sometimes one responds to cues in a "mindless"
manner (Nisbett & Wilson, 1977) or one responds
via an established pattern or script (Abelson, 1976).
In these instances, the attributional process requires
little effort. Moreover, as Lowe and Kassin (1980)
have recently argued, attributions in everyday situations may be guided as much by perceptual factors
as by the cognitive factors emphasized in experimental studies. These two dimensions-type of
attribution and level of information processing-are used to develop the typology of attributional principles presented in Table 1. The principles associated with a high level of information
processing have received widespread attention and,
therefore, will not be discussed further. However,
the attributional principles associated with a low
level of information processing have not received
sufficient consideration.
tial causal agents such as actors, situations, or circumstances. The key issue in assessing causality involves deciding whether the event in question would
have occurred without the presence of a particular
agent (Hamilton, 1980). Thus, necessary factors are
assigned highest causal priority. For example, Kelley's (1973) covariation principles can be viewed as
describing rules for determining whether an event
would have occurred in the absence of particular actors, entities, or circumstances (time) because
events would covary only with necessary factors.
Attributions concerning responsibility assessments involve a somewhat different issue-whether
sanctions should be administered to a particular actor (Hamilton, 1980). Although responsibility generally is assigned to causally predominant sources,
responsibility assessments may be mitigated by factors such as foreseeability, intention, and justification (Fincham & Jaspars, 1980; Heider, 1958).
Moreover, responsibility sometimes may be assigned to sources of low priority in causal explanations. Here the key issue is whether an actor could
have produced a different outcome by behaving differently (Hamilton, 1980). Thus, even if most leaders would behave similarly in a given situation,
leading to similarly bad outcomes and yielding causal assessments to entities and circumstances, each
would be held responsible if by behaving otherwise
she/he could have produced different outcomes. It
is to be noted that leadership positions often may
involve responsibility for outcomes even though
leaders may not be causally important (Pfeffer,
1977).
Attribution of personal qualities emphasizes a
third issue-determining the qualities of individuals
that generalize across situations and time. Thus, the
key question involves determining how to characterize a particular social stimulus. Although personal qualities may be inferred from behaviors or
their outcomes (Jones & Davis, 1965), such generalizations also may be based on characteristics such as
race, sex, cultural background, or appearance.
Thus, implicit theories and social stereotypes also
are related to attributions of personal qualities.
Tablel
Principles Used by Attribution Theories
Type of Attribution
Level of
of
Level
Information
Processing
Causality
High
Covariation
Configuration
Low
Salience
Temporal proximity
Spatial proximity
Responsibility
Intent
Control
Justification
Formal role
Personal
Characteristics
Correspondence
Prototype fit
Critical feature fit
Context
tional performance often is attributed to immediately prior management decisions rather than
to events in the more distant past (i.e., the development of inferior products).
In short, it is argued that causality often may be
assessed by either a passively (salience) or heuristically (proximity) directed search for the first sufficient cause for an outcome. In other words, once a
sufficient cause has been identified, search for or
evaluation of other possible causes should be terminated. Because salient or proximal causes (such
as people) tend to be considered first and often are
plausible explanations, they should predominate as
explanations of causality. Such processes should
place limited demands on the human information
processing system. The present authors feel that
these may be better descriptions of humans' typical
causal assessments than the quasiscientific models-associated with a high level of information
processing in Table 1.
Causality
The amount of information processing needed to
make causal assessments can be minimized in several ways. The foremost is that people generally
may be satisfied with the first "satisfactory" causal
explanation that comes to mind (Taylor & Fiske,
1978). Thus, if there were some mechanism by
which more plausible explanations would be considered first, very few (and frequently only one)
plausible explanations would be considered in most
situations. Taylor and Fiske argue that, for the
most part, attention is such a mechanism and that
it, in turn, is directed to a large extent by salience.
Accordingly, people attend to and "see causality
in" salient sources. To Taylor and Fiske, such
salience arises either from distinctive qualities of the
social stimulus or from the centrality of a particular
cause in an attributor's memory schema.
Their point seems well taken, but there may be
factors other than salience that direct attention to
plausible causes. One may be a tendency to look for
causes that are physically proximal to an effect; this
would be particularly important when visual search
for causes predominate. For example, equipment
breakdowns often are attributed to the operator
rather than defective materials or poor maintenance. Another factor, temporal proximity or
closeness in time, would seem to be particularly important when searching memory for possible
causes. Because common sense suggests that causes
must precede results, one may merely search backward from the event in question to the first plausible cause when making causal attributions. If, as
Hasher and Zacks (1979) argue, temporal information is automatically encoded or if temporal order
can be inferred from well learned scripts (Schank &
Ableson, 1977), such a limited search should be
quite feasible and may not be consciously directed
or monitored. For example, a change in organiza-
Responsibility
Although it is comforting to believe that responsibility assessments are guided by logical principles
such as assessment of intent, control, or justification (and in legal proceedings they are), it is suspected that in many situations (or for many people)
cognitively simpler, albeit less logical, principles
predominate. For example, perceivers may assign
responsibility for role related outcomes to role occupants (Hamilton, 1980), assuming that these individuals, through different actions, could have
produced different outcomes. In administering
sanctions, mitigating circumstances may not even
be considered. Thus, leaders may be replaced when
their subordinates perform poorly, or employees
may be sanctioned for failure to attain goals
without extensive analysis on the part of the sanctioner. Such sanctions may be administered even
52
BoundaryConditions
Personal Characteristics
Sometimes trait or personal characteristic assessments may reflect quasi-logical inferences from
behavior as suggested by Jones and Davis (1965).
However, it is suggested that less conscious and less
demanding means of forming impressions of others
may predominate. In many situations, attention
and information processing capacity is directed at
task, situational, or self-related factors while interaction takes place, thereby precluding elaborate
or consciously controlled assessments of personal
qualities. Nevertheless, trait assessments still seem
to be formed in these situations, suggesting more
automatic processes.
How might such processing proceed? On the simplest level, people might be classified by the fit of
their behavior, appearance, dress, and so on with
category prototypes (Cantor & Mischel, 1979;
Lord, Foti, & Phillips, 1982). Alternatively, there
may be critical features that map social stimuli into
contrasting sets, as with sex. Also, people may use
context as a means of assessing traits (people in
libraries are intellectual). Such principles would
equate the perceptual organization that makes stimuli meaningful with trait ascriptions. That is, trait
ascriptions would result from processes by which
people recognize stimuli, not from logical analysis.
On a somewhat more sophisticated level, subjects
may generalize from ascribed traits to other traits.
Thus, they may base inferences on implicit theories
or stereotypes that are thought to be generally applicable. Again, such inferences may not be conscious, may not use much information processing
capacity, and often may not be logical (or correct).
A more cognitively demanding means to assess
personal qualities would incorporate information
on intent, reliability (consistency), or personal versus situational determinism. At this level, one
would expect conscious awareness of an attributional process and a need to devote a substantial
portion of one's controllable processing capacity to
the attributional task.
53
Ordering of Attributional
and Behavioral Processes
Up to this point, the analysis has avoided the
question of how attributions tie into behavior or attitude change. In most typical attributional models,
attributional assessments precede behavior. For example, the Green and Mitchell (1979) model depicts
leaders' causal attributions as preceding their
behavior toward subordinates. This is fully consistent with the logical, cognitive emphasis found in
the attributional models associated with a high level
of information processing in Table 1. However,
54
cesses, but more controlled processes may predominate under the infrequent situations in which sanctions are considered. The consequences of this example for researchers concerned with superior attributional processes are straightforward. That is,
one must not only identify attributional principles,
but must also specify the type of questions and
situational factors that determine which principles
are likely to be used.
A second implication for scientific inquiry concerns the chameleon-like nature of human attributors. It is maintained that the type of information
processes used to form attributions varies with the
motivation of attributors, the form and amount of
information presented them, the available processing capacity to devote to attributional questions,
and the primes or cues inherent in particular situations. Such factors, alone or in combination, may
easily determine whether controlled or more
automatic processes are used in forming attributions. Thus, the troublesome problem of demand
characteristics (Orne, 1969) seems particularly germane to studies of attributional processes. As Lowe
and Kassin (1980) point out, experimenters interested in testing more cognitive (controlled)
models of attributional processes can easily induce
such processes by their choice of methodologies.
A related problem concerns the obtrusiveness and
limitations of data collection strategies. Wong and
Weiner (1981) point out that most data on attributional processes are collected via self-report procedures. Such procedures may prime attributional
processes, leading subjects to think in terms of attributional questions they otherwise would not address. Moreover, such measures may be biased in
that the more automatic processes may be less accessible to subjects (Nisbett & Wilson, 1977; Wilson
et al., 1981) and, therefore, may not be adequately
measured by self-report techniques. There is a clear
need for development of more varied methodologies for measuring attributions. Such methodologies should be designed to tap both perceptual
and more cognitive processes and should not be
based solely on subjects' retrospective insight into
their own information processing.
One potential approach would be to investigate
attributions by analyzing protocols generated by
subjects while performing the behavior in question.
As Ericsson and Simon (1980) note, to be accurate
such verbal reports should be concurrent rather
than retrospective and should describe ongoing processes rather than judgments requiring inferences or
generalizations. Alternatively, one could use openended questions to probe attributions rather than
the more common ratings of several experimenter
provided causal sources (e.g., ability). Ratings of
explicit causal sources are troublesome because
items may prime the causal source being rated and
thereby may affect ratings. One recent study (Kroeck, 1982) compared open-ended responses that
were content analyzed by a procedure based on Elig
and Frieze (1979) to ratings of a list of plausible
causal sources (e.g., effort, ability, luck). Kroeck
found frequent mention of perceptually based information (e.g., appearance) when using an open
format measure of attributions. Moreover, such information was more predictive of rating on some
dimensions (leadership potential and promotability)
than was a structured format in which subjects
rated experimenter provided causes.
A final theoretical concern stems from the potential for what can be labeled hybrid attributional
processes. Such processes would involve combinations of the principles presented in Table 1. Hybrid
processes would seem particularly likely when attributional questions are addressed over a period of
time in which the informational or motivational
perspective of the attributor may change or when
attributional processes are related to other
behavioral or attitudinal processes. One example of
a hybrid process may be when information that is
encoded, stored, and retrieved by automatic processes is evaluated by more controlled processes at
some later point in time. For example, workers may
be categorized as either high or low performer types
over many interactions and with little systematic or
conscious processing by superiors. Later on, superiors may consciously assess subordinates ability, a
process that on the surface resembles a controlled
process. However, as Lingle and Ostrom (1979)
have demonstrated, this latter judgment may be
derived from prior judgments. Futhermore, prior
judgments may lead to selective recall of information consistent with already formed opinions
(Hastie, 1981). The reverse process also may occur.
Once a quality is attributed to an individual by
some conscious, controlled process, this label may
lead to subsequent automatic encoding, storage,
and retrieval of information consistent with such a
label. For example, if good performance is ex56
Applied Implications
The typology of attributional principles presented in Table 1 also has several implications with
more direct practical relevance. One concerns the
likely effectiveness of attempts to train or alter attributional processes through instructions. It is
believed that training would be effective only for
the more active, consciously controlled processes.
Training automatic processes should be more difficult because they are overlearned, they may not be
controllable by subjects, and subjects may be unaware of or may even deny the use of automatic
processes. In support of this contention, it is
noteworthy that Schneider and Shiffrin (1977) did
train automatic processes affecting the input of information into short term memory, but such training took 1,500 trials with each subject.
The present authors are pessimistic concerning
the ability to train automatic processing easily or to
alter it through instructions to subjects, but they do
think that there are many means to produce shifts
from automatic to more controlled processing. Motivational factors, information processing parameters, or primes and situational cues may all be used
to induce more controlled or effortful attributional
processes.
An additional consequence of extensive use of
more automatic attributional processes warrants
discussion. This concerns the possibility that skillful
individuals may be able to manipulate or control
the attributions others form. Snyder (1979) argues
References
Cronshaw, S. F., & Lord, R. G. Perceputal versus cognitive
determinants of causal attributions and leadership. Paper
presented at the meeting of the Midwest Academy of Management, Columbus, Ohio, 1982.
Abelson, R. P. Script processing in attitude formation and decision making. In J. S. Carrol & J. W. Payne (Eds.), Cognition
and social behavior. Hillsdale, N. J.: Erlbaum, 1976.
Berscheid, E., Graziano, W. G., Monson, T. C., & Dermer, M.
Outcome dependency: Attention, attribution, and attraction.
Journal of Personality and Social Psychology, 1976, 34,
978-989.
Binning, J. F., & Lord, R. G. Boundary conditions for performance cue effects on group process ratings: Familiarity versus
type of feedback. Organizational Behavior and Human Performance, 1980, 26, 115-130.
58
Green, S. G., & Liden, R. C. Contextual and attributional influences on control decisions. Journal of Applied Psychology,
1980, 65, 453-458.
McAllister, D. W., Mitchell, T. R., & Beach, L. R. The contingency model for the selection of decision strategies: An empirical test of the significance, accountability, and reversibility. Organizational Behavior and Human Performance, 1979,
24, 228-244.
Hamilton, V. L. Intuitive psychologist or intuitive lawyer? Alternative models of the attribution process. Journal of Personality and Social Psychology, 1980, 39, 767-772.
Mitchell, T. R., & Wood, R. E. Supervisor's response to subordinate poor performance: A test of an attributional model.
Organizational Behavior and Human Performance, 1980, 25,
123-138.
Jones, E. E., & Davis, K. E. From acts to dispositions: The attribution process in person perception. In L. Berkowitz (Ed.),
Advances in experimental social psychology (Vol. 2). New
York: Academic Press, 1965, 219-266.
Orne, M. Demand characteristics and the concept of quasicontrols. In R. Rosenthal and R. Rosnow (Eds.), Artifacts in
behavioral research. New York: Academic Press, 1969,
143-179.
Kipnis, D., Schmidt, S., Price, K., & Stitt, C. Why do I like thee:
Is it your performance or my orders. Journal of Applied
Psychology, 1981, 66, 324-328.
Phillips, J. S., & Lord, R. G. Determinants of intrinsic motivation: Locus of control and competence information as components of Deci's cognitive evaluation theory. Journal of Applied Psychology, 1980, 65, 211-218.
Knowlton, W. A., & Mitchell, T. R. Effects of causal attributions on a supervisor's evaluation of subordinate performace.
Journal of Applied Psychology, 1980, 65, 459-466.
Phillips, J. S., & Lord, R. G. Causal attributions and perceptions of leadership. Organizational Behavior and Human Performance, 1981, 28, 143-163.
Kroeck, K. G. Supervisory responses to subordinate goal attainment. Doctoral dissertation, The University of Akron, 1982.
Ross, M., & Sicoly, F. Egocentric biases in availability and attribution. Journal of Personality and Social Psychology, 1979,
37, 322-336.
Lingle, J. H., & Ostrom, T. M. Retrieval selectivity in memorybased judgments. Journal of Personality and Social
Psychology, 1979, 37, 180-194.
Schank, R., & Ableson, R. P. Scripts, plans, goals, and understanding: An inquiry into human knowledge structure.
Hillsdale, N. J.: Erlbaum, 1977.
Schneider, W., & Shiffrin, R. M. Controlled and automatic human information processing: I. Detection, search, and attention. Psychology Review, 1977, 84, 1-66.
59
Shiffrin, R. M., & Schneider, W. Controlled and automatic human information processing: II. Perceptual learning,
automatic attending, and a general theory. Psychological
Review, 1977, 84, 127-190.
Wilson, T. D., Hull, J. G., & Johnson, J. Awareness and selfperception: Verbal reports of internal states. Journal of Personality and Social Psychology, 1981, 40, 53-71.
Witkin, H. A., & Goodenough, D. R. Field dependence and interpersonal behavior. Psychological Bulletin, 1977, 84,
661-689.
Wong, P. T. P., & Weiner, B. When people ask "why" questions, and the heuristics of attributional search. Journal of
Personality and Social Psychology, 1981, 40, 650-663.
Staw, B. M., & Ross, J. Commitment in an experimenting society: A study of the attribution of leadership from administrative scenarios. Journal of Applied Psychology, 1980,
65, 249-260.
Wood, R. E., & Mitchell, T. R. Manager behavior in social context: The impact of impression management on attributions
and disciplinary actions. Organizational Behavior and Human
Performance, 1981, 28, 356-378.
60