5/27/2018 Articol coercitie
1/28
_______________________________________________________________
_______________________________________________________________
Report Information from ProQuestDecember 05 2013 07:58
_______________________________________________________________
05 December 2013 ProQuest
5/27/2018 Articol coercitie
2/28
able of contents1. Risk and power use: Constraints on the use of coercion in exchange......................................................... 1
Bibliography...................................................................................................................................................... 26
05 December 2013 ii ProQuest
5/27/2018 Articol coercitie
3/28
Document 1 of 1
Risk and power use: Constraints on the use of coercion in exchangeAuthor: Molm, Linda DPublication info:American Sociological Review 62.1 (Feb 1997): 113-133.ProQuest document link
Abstract Abstract): Molm investigates how risk and fear of loss constrain the use of coercive power innonnegotiated social exchange relations. An analysis of the use of strategic power helps explain why exchange
partners in previous research have rarely used coercive power, even when their incentives and capacities to
coerce were high.
Full text: HeadnoteI investigate how risk and fear of loss constrain the use of coercive power in nonnegotiated social exchange
relations. An analysis of the use of strategic power helps explain why exchange partners in previous research
have rarely used coercive power, even when their incentives and capacities to coerce were high. Using power isrisky, and actors fear losses from the potential retaliation of their partners far more than they value the prospect
of increased rewards. The risks of coercive power use are especially great in the context of relations of mutual
exchange because of the high reward dependence of actors who have the strongest incentive to use coercion.
Two experiments show that when risk is reduced, particularly the risk of reward loss, both the use of coercion
and the effects of variations in the structure of coercive power increase.
RISK AND POWER USE: CONSTRAINTS ON THE USE OF COERCION IN EXCHANGE*
For the past 15 years, theory and research on social exchange have concentrated on the effects of structural
power on behavior while paying minimal attention to the decision-making processes of actors. Emerson's
(1972a, 1972b) theory of power
dependence relations provided the impetus for the focus on structural power by proposing that structure
determines power use, regardless of actors' intentions to use power or their awareness of the power structure.
The validity of Emerson's thesis has been supported by numerous studies in a variety of experimental settings,
varying along such important dimensions as the information actors have about the power structure, whether
exchanges are negotiated or nonnegotiated, the manipulation of power, and so forth (see Molm and Cook 1995
for a review). Support for his prediction that an imbalance or inequality in structural power produces a cor
responding inequality in exchange benefits that favors the less dependent actor in an exchange relation is one
of the more robust findings in the social science literature.
Recent experimental research suggests, however, that this conclusion does not hold for coercive power in
social exchange relations. When actors in nonnegotiated exchange relations control both rewards and
punishments for each other, their interaction is influenced primarily by the structure of reward power (Molm
1988, 1990).1 The structure of coercive power (actors' control over negative outcomes for each other) has
almost no effect on exchange behavior. A structural advantage on coercive power does not lead to a more
favorable rate of reward exchange for the advantaged actor, nor does it lead to an increased use of coercion.
Actors rarely punish each other, even when their partners fail to provide rewards on which they are highly
dependent.
I develop and test a theory arguing that risk and fear of loss constrain the use of coercion in social exchange.
Actors who have the strongest incentive to use coercion in exchange relations are disadvantaged on reward
power and thus face greater risk of retaliation from their advantaged partners. This structural disadvantage,combined with an aversion to loss (Kahneman and Tversky 1979), inhibits coercion in most relations of mutual
reward dependence. Two experiments show that when risk is reduced by changing the structure of
dependence, both the use and effectiveness of coercive strategies increase.
05 December 2013 Page 1 of 26 ProQuest
http://search.proquest.com/docview/218821543?accountid=15859http://search.proquest.com/docview/218821543?accountid=158595/27/2018 Articol coercitie
4/28
THEORETICAL BACKGROUND
Social Exchange and Reward Power
Emerson's (1972a, 1972b) theory of social exchange, commonly called "power-dependence theory," predicts
that behavioral patterns of exchange are determined by the structural relations of dependence among actors.
The theory's scope is restricted to four conditions:2 (1) actors who seek to increase outcomes they positively
value and decrease outcomes they negatively value, (2) outcomes that obey a principle of satiation or decliningmarginal utility, (3) relations of mutual dependence in which the rewards obtained from others are contingent on
rewards given in exchange, and (4) recurring exchanges over time between the same actors or sets of actors
(Emerson 1981; Molm and Cook 1995).
The simplest form of social exchange occurs between two actors, A and B, who can be either individual actors
or collective actors whose members behave as a unit. Each actor must be able to provide some reward
(positively valued outcome) to the other (e.g., advice or status); that capacity is the actor's resource in that
relation. Each actor is dependent on the other to the extent that the outcomes valued by one actor are
contingent on exchange with the other. When an actor provides rewards for another, he or she incurs cost--
minimally, opportunity costs (i.e., rewards foregone by not pursuing other alternatives) and sometimes other
costs (e.g., investment costs, the actual loss of a material resource, or costs intrinsic to the behavior itself).
The mutual dependence of actors on each other for rewards provides the structural basis for their exchanges
with one another and for their power over each other. A's power over B is defined as the level of potential cost
that A can impose on B (Emerson 1972b). Four assumptions state the core ideas of power-dependence theory:
Assumption 1. A's power over B derives from and is equal to B's dependence on A.
Assumption 2. B's dependence on A increases with the value of the rewards that A can produce for B and
decreases with B's alternative sources of those rewards.
Assumption 3. Actors initiate exchanges by giving rewards to other actors on whom they are dependent.
Assumption 4. Actors increase exchanges that are relatively more rewarding or less costly, decrease those that
are relatively less rewarding or more costly, and change behaviors when rewards decline or costs increase.
The alternatives that reduce dependence (Assumption 2) typically take the form of other exchange partners who
are potential sources of the same (or equivalent) rewards (e.g., alternative dating partners or alternative sources
of expert advice). When connected by a focal actor, alternative relations form larger exchange networks, which
consist of three or more actors (e.g., B-A-C). If B and C are alternative exchange partners for A, then the B-A
and A-C relations are negatively connected: The more frequently A exchanges with B, the less frequently A
exchanges with C.
Within a relation, each actor's dependence (and power) can vary independently of the other's. Their joint or
relational dependence is described by two dimensions: average power and power imbalance. Average power
(the average of the two actors' dependencies on each other) is a measure of the absolute strength of the actors'power over each other; power imbalance (the difference between the two actors' dependencies on each other)
is a measure of their relative power over each other. Two friends may have high average power over each
other, for example, but low power imbalance. In a power-imbalanced relation, the less dependent and more
powerful
actor has a power advantage in the relation, and the more dependent actor is power disadvantaged.
The structure of power and dependence in a relation affects the behavioral pattern of exchange that develops
over time. When actors engage in recurring exchanges with partners in negatively-connected networks,
Assumptions 1 through 4 imply the following relations between structural power and the frequency and
distribution of exchange:
Theorem I. A's initiations of exchanges with B increase as A's dependence on B increases (and B's power over
A increases).
Theorem 2. The average frequency of exchange in a relation increases as average power and dependence in
05 December 2013 Page 2 of 26 ProQuest
5/27/2018 Articol coercitie
5/28
the relation increases.
Theorem 3. The asymmetry of exchange in a relation increases as power imbalance increases and favors the
more powerful (less dependent) actor. If A has a power advantage, then A's use of power will increase over
time (until restrained by A's own dependence), as evidenced by either (1) increased rewards to A from B (i.e.,
increased costs for B), or (2) decreased rewards to B from A (decreased costs for A).3
In short, in imbalanced relations, exchange rates change in favor of less dependent, more powerful actors(Cook and Emerson 1978; Cook et al. 1983; Markovsky, Willer, and Patton 1988; Molm 1990). Underlying this
structural effect is the same law that governs economic exchange: When demand is high (high value) and
supply is scarce (few alternatives), one pays more for the things one values.
Punishment and Coercion in Social Exchange
Traditionally, social exchange theory and its analysis of power assumed an additional scope condition: that only
rewards or posi
tively valued outcomes were exchanged (Blau 1964; Emerson 1972a, 1972b; Homans [1961] 1974). The costs
of exchange were restricted to opportunity costs, and punitive actions by one actor toward another were
excluded. As Heath (1976) and others (Blalock 1987; Molm 1988) have argued, however, there is no logical
reason for this restriction. Although an exchange "enables both participants to be better off than they would
have been without it . . ., they need not necessarily be better off than they were before" (Heath 1976:19). For
example, a spouse might prefer to spend more time at the office than at home, and might begin working more
hours. But if family members react with anger and disapproval, the pattern is less likely to persist: Cutting back
on work hours becomes preferable to enduring unpleasant family interactions. Consequently, the spouse begins
to spend more time at home, and the disapproval stops. The spouse is not better off than before the coercion
began--remem
ber, more hours at work are preferred--but he or she is better off than persisting with more work hours and
enduring frequent battles at home.
I define coercion as the use of punishment to obtain rewards from another. B coerces A by punishing A's failure
to reward B and withholding punishment when A's rewards to B are forthcoming.4 Thus, in exchanges involving
mutual rewards actors engage in mutually contingent (but not necessarily equal) exchanges of rewards,
whereas in coercive
exchanges one actor provides rewards to another in exchange for the other's withholding of expected
punishment.
An actor's capacity to obtain rewards from another through either reward exchange or coercive exchange is
determined by their power to reward or to punish the other. Structurally, forms of power based on control over
positive outcomes (reward power) or negative outcomes (coercive power) are parallel, and both can be
conceptualized in terms of dependency. When A controls rewarding outcomes for B, B is dependent on A forobtaining the things B finds desirable; when A controls punishing outcomes for B, B is dependent on A for
avoiding the things B finds aversive. This structural comparability between reward power and coercive power
makes it possible to define parallel dimensions of power and power imbalance for the two power bases, and to
extend Assumptions 1 through 4 to coercive power.
Rewards and punishments are defined in relation to an actor's current situation (Van Houten 1983). In general,
actions that increase positive value or decrease negative
value are rewarding; those that decrease positive value or increase negative value are punishing. A move to a
small house might be rewarding to a homeless person, for example, but punishing to a person accustomed to a
manor.
Analytically, the two bases of power share a common foundation that enables them to be transformed from one
into the other. If rewards are regularly given, they can be withheld as punishment; if punishments are regularly
given, they can be withheld as rewards (Bacharach and Lawler 1981; Blau 1964). As we shall see, the capacity
05 December 2013 Page 3 of 26 ProQuest
5/27/2018 Articol coercitie
6/28
of an actor to punish another using either form of power, through either direct punitive action or the withholding
of rewards, is central to this analysis: This capacity provides the means to use either reward power or coercive
power to retaliate against a partner for using coercion.
The Research Program
This study is part of a broader program that compares the effects of reward power and coercive power in social
exchange relations. The aim is to incorporate punishment and coercion into the framework of social exchange theory. The unit of analysis is the dyadic relation, which is embedded in a larger exchange network. In
addition to the four scope conditions described earlier, several other conditions define the boundaries of this
research program:
First, coercive power is studied in the context of relations in which actors have the capacity both to reward and
to punish each other. Such relations are distinct from purely coercive relations (Willer and Anderson 1981),
which do not provide the opportunity for mutual reward exchange. Second, actors' resources are capacities to
perform behaviors that produce valued outcomes for others-- not to provide tangible goods. I assume that the
only costs of performing such behaviors are opportunity costs.5 Third, exchange transactions are reciprocal, not
negotiated. In reciprocal exchanges, actors' contributions to the exchange are separately performed and
nonnegotiated (Blau 1964; Emerson 1981)-- actors initiate exchanges without knowing what they will receive in
return, and the reciprocity of exchange is determined over time.
Fourth, all actors have alternative partners with whom they can exchange; they also have alternative behavioral
choices (i.e., they can reward or punish). Therefore, both the use of reward or coercive power and responses to
power use are voluntary. Fifth, although actors can cease exchanging with a partner, they cannot "leave the
field" of influence (i.e., they cannot avoid another's rewards or punishments); nor can they change the structure
of the network. And sixth, actors have full information about the values governing the relative dependencies in
their own exchange relations, but not in distal relations in the network.
Earlier experiments in the program systematically compared the effects of average power and power imbalance
for reward
power and coercive power under these conditions (Molm 1988, 1989, 1990). Results from these experiments
repeatedly showed that the predicted relations between structural power and behavioral exchange (Theorems 1
through 3) held only for reward power. The structure of coercive power had almost no effect on exchange; in
particular, actors with greater power to inflict losses on their partners obtained no greater benefits from
exchange. Even actors with a coercive power advantage equal to and opposing a partner's reward-power
advantage ended up on the losing end of the exchange.
These findings seemed to confirm the position of the classical exchange theorists, who excluded punishment
and coercion from the scope of their theories. Both Homans ([1961] 1974) and Blau (1964) argued that coercion
is ineffective and produces resistance and retaliation rather than compliance. Other theorists also share thisposition (Boulding 1989; Deutsch 1973; Tedeschi, Schlenker, and Bonoma 1973), suggesting that the power to
punish has different effects than does the power to reward because of the con
sequences of its use (i.e., the effects of punishment on the target of coercion).
Subsequent experiments have failed to support this explanation, however. When computer-simulated actors
were programmed to use strong and consistent strategies of coercion against real subjects (by inflicting
monetary losses whenever subjects failed to reward them), their coercive power was highly effective (Molm
1994). Simulated actors who were disadvantaged on reward power successfully coerced their advantaged
partners to exchange with them most of the time. Other researchers (e.g., Willer 1987) have also found that
coercive power, when used, is effective.
But real subjects do not apply coercion consistently in exchange relations. Subjects in both strong and weak
power positions typically punish their partners on fewer than 5 percent of all opportunities, and these
punishments include both coercive punishment (punishment contingent on the other's failure to reward) and
05 December 2013 Page 4 of 26 ProQuest
5/27/2018 Articol coercitie
7/28
retaliatory punishment (punishment contingent on the other's prior punishment).
In short, coercion is effective when used, but most actors do not use it enough to influ
ence another's behavior. Understanding why coercive power is ineffective, therefore, requires understanding
why actors use it so seldom. In the remainder of this paper, I develop and test a theoretical explanation
proposing that coercion in the context of relations of mutual exchange is risky, and that risk constrains its use.
A THEORY OF COERCION IN SOCIAL EXCHANGEI develop four main arguments. First, unlike reward power, the use of coercive power is not structurally induced
by power advantage but is behaviorally motivated by the receipt of insufficient rewards from an exchange
partner. Second, a disadvantage on reward power provides the incentive to use coercion in exchange relations,
but it also increases the risk of retaliation by the advantaged partner. Third, loss aversion heightens this risk by
increasing the subjective value of losses. Fourth, risk is reduced and coercion increased only when the partner's
exchange frequency is so low that there is little to lose.
Structurally-Induced Power Use versus Strategic Power Use
As power-dependence theorists have long asserted, the use of reward power is structurally induced by a
reward-power advantage, regardless of actors' intent to use power or to influence another's behavior (Cook and
Emerson 1978; Emerson 1972b). Structural power advantage leads to power use (Theorem 3) because
advantaged actors have access to more or better alternative partners (Assumptions 1 and 2); when they pursue
exchanges with these alternative partners (Assumption 3), they inadvertently withhold rewards from their more
dependent partners (i.e., they "use power" over them). In the process, they drive up others' opportunity costs of
obtaining the rewards they control, while lowering their own opportunity costs of obtaining their partners'
rewards. As long as actors respond by following the behavioral principles in Assumption 4--increasing rewarding
exchanges and avoiding costly exchanges--the inequality predicted by Theorem 3 will result.
In short, in power-imbalanced relations, the use of reward power is an unintended byproduct of advantaged
actors' exchanging more frequently with alternative partners. For example, a business that favors one supplier
may not intend to decrease another supplier's price for the same product; a person who spends more time with
one friend may not intend to acquire another friend's favors at lower cost. Nevertheless, these actions will have
such effects if the actors are structurally advantaged in the relations. Their greater access to alternative partners
provides both the capacity and the incentive to use reward power.
A structural advantage on coercive power, in contrast, provides the capacity to impose costs on a partner, but
no incentive to use that power. An imbalance in coercive power induces the disadvantaged actor to give
rewards to the advantaged partner to prevent punishment (Assumption 3, extended to coercive power), but the
imbalance does not induce the advantaged actor to use power (i.e., to punish the disadvantaged actor for failure
to provide rewards). If coercive power is not used, any effects of the potential to punish are unlikely to bemaintained.
Not all power use is structurally induced, however. Actors also can use power strategically by selectively giving
or withholding rewards and punishments contingent on a partner's prior behavior. Thibaut and Kelley (1959)
assumed the strategic use of power when they discussed the conversion of fate control (structural dependence)
to behavior control (actual influence over another's behavior), and strategic power use has been the focus of
researchers studying the sanctioning of free riders and defectors from cooperation (e.g., Axelrod 1984, 1986;
Yamagishi 1995). When actors use power strategically, they create contingencies that produce predictable
consequences for their partners' behaviors, rather than simply respond to the consequences of their own
behaviors.
Such contingencies are particularly useful in the absence of joint decision-making and negotiated agreements.
In ongoing relations, these contingencies create what Axelrod (1984) called "the shadow of the future"-- the
likelihood that what actors do at any particular time will influence how other actors will respond to them later.
05 December 2013 Page 5 of 26 ProQuest
5/27/2018 Articol coercitie
8/28
This form of in
fluence is possible only when actors exchange repeatedly with one another (the fourth scope condition).
The distinction between structurally-induced and strategically-performed power use has important implications
for predicting when actors will use coercion. Developing the significance of this distinction requires extending
Emerson's model of the actor as learner to comprise a model of the actor as both learner and teacher. Actors
who use power strategically are responding to others' behaviors, but they are attempting to influence theirbehaviors as well.6
Assumptions about Strategic Actors
To analyze strategic power use, I draw on two theories of strategic action and decisionmaking under conditions
of risk and uncertainty: Axelrod's work (1984, 1986) on the evolution of cooperation, and Kahneman and
Tversky's (1979) prospect theory. These two theories and the empirical work on which they are based suggest
four additional assumptions for deriving predictions about the strategic use of power:
Assumption 5. Actors initiate strategic power use by imposing costs on other actors who fail to provide rewards
in exchange.
Assumption 6. Strategies are dynamic and adaptive (Axelrod 1984); once initiated, they are shaped by the
consequences produced by partners' responses.
Assumption 7. The potential consequences of strategic power use include both gain (from compliance) and loss
(from retaliation); the probabilities of both are unknown but they are greater than zero.
Assumption 8. Actors evaluate exchange outcomes (both anticipated and experienced) by a value function that
has three
main characteristics: referent dependence, diminishing sensitivity, and loss aversion (Kahneman and Tversky
1979):
(a) Referent dependence. Outcomes count as gains (or rewards) if they improve an actor's current outcome
level (the status quo); they count as losses (or punishments) if they worsen it.7
(b) Diminishing sensitivity. The marginal value of gains and losses decreases with their distance from the
reference point (i.e., the status quo).
(c) Loss aversion. The negative subjective value of a loss is greater than the positive subjective value of an
equivalent gain.
Assumption 5 is parallel to Assumption 3, recognizing that strategic power is not based on the expectation of
reciprocity that underlies Assumption 3 or on the inadvertent use of structurally-induced reward power, but on
an application of the behavioral principles in Assumption 4. Rather than giving rewards to obtain benefits in
return, actors impose costs (by administering punishment or withholding rewards) on partners who fail to pro
vide them with sufficient benefits. Once initiated, strategies--like other patterns of exchange behavior--should
increase or decrease in frequency according to their consequences; thus, Assumption 6 is parallel toAssumption 4.
Assumption 7 makes explicit the risk and uncertainty involved in the strategic use of power. The exchange
partner may respond with either compliance (increased rewards) or retaliation (decreased rewards or increased
punishment). Complying with coercion produces the immediate benefits of decreased punishment, but in a
continuing relation retaliating against coercion can reduce the likelihood of coercion in the future.
Assumption 8 specifies the value function, taken from Kahneman and Tversky's (1979) prospect theory, which
describes how actors evaluate gains and losses (both actual and
anticipated)8 under conditions of risk and uncertainty. The first two properties of this value function are already
implied, respectively, by contemporary definitions of reward and punishment and by the second scope condition
(declining marginal utility) of exchange theory. The third property, loss aversion, is new and important, and its
validity is supported by numerous studies (Fishburne and Kochenberger 1979; Gray and Tallman 1987;
Hershey and Shoemaker 1980; Kahneman and Tversky 1979, 1984; Molm 1991). Researchers typically have
05 December 2013 Page 6 of 26 ProQuest
5/27/2018 Articol coercitie
9/28
reported coefficients of loss aversion in the range of 2 to 2.5, indicating that the slope of the value function for
losses is roughly twice the slope for gains over the same range (Kahneman, Knetsch, and Thaler 1991; Tversky
and Kahneman 1991). In contrast, most exchange theories implicitly assume that rewards and costs carry equal
weight. The assumption that they do not has important implications for understanding the use and effects of
coercive power.
The Status Quo BiasWhen actions can produce either gains or losses and the probabilities of both outcomes are either unknown or
equal, loss aversion produces a phenomenon known as the status quo bias (Kahneman and Tversky 1984),
applied here to strategic power use:
Theorem 4. When the potential losses from retaliation are roughly equal to the potential gains from compliance,
strategic power use is unlikely. Actors will accept the status quo of a relation rather than risk the consequences
of using power to change it.
Empirically, choices favoring the status quo tend to dominate decision-making under uncertainty (Samuelson
and Zeckhauser 1988). As long as actors stand to lose as much as they might gain and both outcomes are
uncertain (Assumption 7), the subjective overweighting of losses relative to gains (Assumption 8c) will constrain
the strategic use of either reward power or coercive power in social exchange relations. Thus, loss averse
actors will opt for the certainty of the status quo, accepting their current position in a relation rather than risking
the use of power to improve their outcomes. As potential gains increase and potential losses decrease, the
status quo bias should decrease and the use of coercion should increase. Potential gains and losses will vary
with the structure of power and dependence.
Dependence, Risk, and Coercion
Actors' dependencies affect the probability that they will initiate coercion, the probability that their partners will
respond with compliance or retaliation, and the magnitude of gains and losses attached to these alternative
outcomes. Actors' relative dependencies-- their positions of power advantage or disadvantage--influence both
the probability of coercion and the probability of retaliation against coercion.
Theorem 5. The probability that B will initiate a coercive-power strategy to influence A increases as B's reward-
power disadvantage in the relation increases.
This theorem follows directly from Theorem 3 and Assumption 5. Reward-power disadvantage provides the
motivation to use coercion to improve one's outcomes from exchange. A structural imbalance in reward power
reduces the advantaged partner's reward frequency, which in turn motivates the disadvantaged actor to try to
change that behavior and increase rewards from the partner. In contrast, actors who are advantaged on reward
power have little need to use coercion to get what they want.
The notion that a disadvantage on reward power produces the incentive to use coercion appears contrary to
traditional assumptions that it is the "strong" who coerce the "weak." Legitimated and institutionalized forms ofcoercion are, of course, typically used by powerful actors such as the state. But results from both experimental
and nonexperimental studies suggest that noninstitutionalized forms of coercion are more common among
actors who are disadvantaged on reward power and who lack other means of influencing those on whom they
depend for rewards (e.g., low-level managers, low-status spouses, minority political factions, and children)
(Kanter 1977; Molm 1990; Patterson 1982; Raush et al. 1974; Stets 1995).
Power advantage or disadvantage not only affects the probability that B will use coercion against A, it also
affects the probability that A will retaliate. Power-advantaged actors gain more by retaliating and lose more by
complying than do power-disadvantaged actors. Consequently, the probability that a partner will retaliate should
increase with that partner's power advantage:
Theorem 6. The probability that A retaliates rather than complies in response to B's use of coercion against A
increases with A's power advantage in the relation.
Theorem 6 applies to both bases of power: A partner who is advantaged on either reward power or coercive
05 December 2013 Page 7 of 26 ProQuest
5/27/2018 Articol coercitie
10/28
power is more likely to retaliate. A partner who controls rewarding outcomes can retaliate by withholding those
rewards; a partner who controls punishing outcomes can retaliate by administering punishment. A partner who
controls both can do either.9
Together, Theorems 5 and 6 imply the same result as Theorem 4: a bias against the use of coercive power, and
in favor of the status quo. A coercive-power advantage reduces the risk of retaliation against the coercer
(Theorem 6) but provides no incentive to use coercion. A reward-power disadvantage provides the incentive touse coercion (Theorem 5) but also increases the risk that the partner will retaliate by decreasing rewards rather
9 The relation between coercive power and risk of retaliation also underlies theories of deterrence, which
propose that the risk of retaliation keeps actors from using punishment when they have equal power to punish
each other (Morgan 1977; Schelling 1960). Lawler's (1986, 1992) extension of deterrence theory, however,
suggests the deterrence effect begins to erode when power is unequal.
than increasing them (Theorem 6). For example, the family that coerced the workaholic spouse to spend more
time at home might, instead, have provoked the workaholic to spend even less time.
This, then, is the dilemma that constrains the use of coercive power in exchange relations: Actors who have the
incentive to use coercive strategies are often too dependent on their partners to risk doing so. As Blalock (1987)
observed, "... substantial degrees of dependence will lead to both a high level of motivation to employ negative
sanctions to the other party and high potential costs of such actions." He suggested that reward-power
disadvantage increases the risk of retaliation, in general, but also noted, "Where exchanges of positively valued
goods accompany the punitive actions, the more dependent party is especially vulnerable to a cutoff of the
exchange relation altogether" (p. 13).
This conflict between incentive and risk exists only for coercive power. For reward power, the condition that
increases the incen
tive to use power--reward-power advantage--simultaneously reduces the risk of the partner's retaliation.
Consequently, whether the use of reward power is strategic or structurally induced, it should increase with
reward-power advantage. For this reason, researchers studying reward-based networks have rarely analyzed
how risk affects actors' decisions to use power (for an exception, see Kollock 1994).
To the extent that actors anticipate retaliation, the risk of reward loss can prevent them from initiating coercion.
If they do initiate it, their reward-power disadvantage increases the chance of retaliation, regardless of their
relative coercive power. Most actors initially do retaliate against a partner's punishment (Lawler and Bacharach
1987; Molm 1990), even though such responses are often shortlived (Molm 1994).
For actors who are uncertain how their partners will respond to coercion and who adapt their strategies to their
partners' behaviors over time (Assumption 6), initial retaliation is likely to suppress further coercion. Like all
behavioral contingencies, coercive strategies take time to work, and to be effective they must be applied
consistently over repeated interactions. But in theface of immediate retaliation, loss-averse actors are unlikely to persist with strategies whose long-term
consequences are uncertain. Thus, retaliation is not an irrational response. If swift retaliation can suppress
another actor's attempts to coerce, the costs of both continued coercion and compliance can be avoided.
Reducing risks and increasing incentives.
While actors' relative positions of power affect the probabilities that one actor will initiate coercion and the other
will either comply or retaliate, their absolute dependencies (i.e., A's dependence on B and B's dependence on
A) affect how much the coercive actor stands to gain or lose from the partner's compliance or retaliation.
Analyzing these dependencies reveals that under some conditions the actual cost of retaliation can be low
enough to encourage coercion, even when the probability of retaliation is high.
Assume that B's disadvantage on reward power provides the incentive for B to coerce A. B's reward
dependence on A determines the range of potential outcomes that A can produce for B by either increasing or
decreasing the frequency of A's rewarding. This range represents the rewards at stake for B.
05 December 2013 Page 8 of 26 ProQuest
5/27/2018 Articol coercitie
11/28
If B's reward dependence is high, changes in A's behavior (for better or worse) can affect B's rewards
substantially. If B's reward dependence is lower, increases or decreases in A's rewarding matter less.
Within this range, the actual rewards A provides for B constitute B's current outcome level. Holding B's
dependence on A constant, according to Theorem 1 this outcome level should increase with A's dependence on
B (i.e., the greater A's dependence on B, the more frequently A should reward B). By Assumption 8a, increases
in this outcome level count as potential gains for B, and decreases are potential losses. Consequently, potentialgains increase and potential losses decrease as A's dependence on B (and the frequency of A's rewarding)
decrease. When the frequency of A's rewarding becomes low enough that potential gains outweigh potential
losses (taking loss aversion into account), coercion should increase:
Theorem 7. The positive effect of rewardpower disadvantage on both the incentive to use coercive power
(Theorem 5)
and the risk of retaliation (Theorem 6) will constrain B's use of coercion against A unless the proportion, p, of
opportunities on which A rewards B, multiplied by the coefficient for loss aversion, c, is less than I - p: pc
5/27/2018 Articol coercitie
12/28
used in previous experiments in this program: Each actor could exchange with two of the three other actors in
the network; thus, each actor had two alternative partners (see Figure 1). Subjects could act toward only one
partner on each exchange opportunity.
Each actor in the network could perform behaviors to produce monetary gains (reward power) or monetary
losses (coercive power) for each partner. Thus, subjects were mutually dependent on one another for their
earnings in the experiment; they gained or lost money solely as a result of the other subjects' behaviors. Thisbehavioral capacity to affect
another's earnings--not money per se--was the exchange resource in the experiment. In contrast to economic
exchange, money was not transferred from one actor to another-- adding to another actor's earnings did not
reduce a subject's own earnings, and subtracting from another actor's earnings did not increase the subject's
own earnings. The only cost of initiating an exchange was the "opportunity cost" of not exchanging with the
alternative partner. To assure that money was valued, subjects were recruited on the basis of their desire to
earn money and instructed that their only concern should be maximizing their own earnings. Other potentially
valued outcomes were removed from the experimental setting (i.e., subjects were seated in isolated rooms and
interacted with each other only through computers).11
An "earnings schedule" displayed at the top of each subject's computer screen informed subjects of the
amounts of money (in points, one point equalled one cent) that they and their two partners could gain or lose
from each other on each exchange opportunity. (These point values are displayed in Figure 1.) The points that
each actor could either add or subtract were fixed by the experimental condition; subjects could vary how much
value they added to or subtracted from a partner's earnings only by varying the frequencies of their rewarding or
punishing behaviors.
Subjects engaged in reciprocal (nonnegotiated) exchanges for 250 exchange opportunities. On each
opportunity, all actors simultaneously chose (1) which of their two potential partners they would act toward, and
(2) whether to add to or subtract from the earnings of that partner. In effect, these two choices gave each
subject four behavioral options. Because each side of the exchange was individually performed, subjects could
gain (or lose) money from one, both, or neither of their partners on each opportunity. The potential for obtaining
rewards from both partners at the same time in reciprocal exchanges makes an asymmetrical exchange of
rewards beneficial to the actor who exchanges less often.
05 December 2013 Page 10 of 26 ProQuest
5/27/2018 Articol coercitie
13/28
After all subjects made their choices on each opportunity, they were reminded on the screen of their own
choices and informed of their partners' choices (they were told that each partner added n points to their
earnings, subtracted n points from their earnings, or did not act toward them). Their own point totals changed to
reflect any gains or losses; their partners' total points were not shown.
The Manipulation of RiskThe experiment manipulated the risk of power use within a basic network structure in which actors B and D
were highly dependent on A and C respectively for rewards, reward power was imbalanced in the A-B and C-D
relations in favor of A and C, and coercive power was imbalanced in these relations in favor of B and D (see
Figure 1). That is, in all four conditions, B and D could potentially gain more from A and C (8 points per
opportunity) than from each other, but A and C
could gain more from each other (6 points) than from B and D. B and D's greater dependence on A and C for
rewards gave A and C a reward power advantage over B and D. The opposite was true of coercive power: B
and D could inflict greater losses on A and C (a loss of 8 points) than A and C could inflict on each other, but
they could potentially lose more from each other (6 points) than from A and C.This basic structure provided both the incentive (reward-power disadvantage) and the capacity (coercive-power
advantage) for B and D to use coercion to obtain increased rewards from A and C. Within this structure, I
manipulated the risks that B and D incurred if they did so, (1) from their partners' retaliatory punishment, and (2)
from their partners' reduction of the rewards on which B and D are highly dependent.
I manipulated the risk of retaliatory punishment by varying A and C's capacity to punish B and D. This risk is
present and moderate in the networks in Figures la and
1b, in which A and C have moderate coercive power over B and D (A and C control 4/10 of B and D's potential
losses), and absent in the networks in Figures lc and ld, in which A and C's coercive power over B and D, and
hence the risk of their retaliatory punishment, is reduced to zero. By Theorem 8, eliminating A and C's power to
punish B and D should increase B and D's use of coercion.
As the theoretical analysis indicated, the reward-power disadvantage that provides B and D with the incentive to
coerce rewards from A and C can also constrain coercion. To reduce potential losses and increase potential
05 December 2013 Page 11 of 26 ProQuest
5/27/2018 Articol coercitie
14/28
gains, I reduced A and C's reward dependence on B and D (which, in turn, should reduce the frequency of their
respective rewards to B and D). In the networks in Figures la and lc, A and C's reward dependence is moderate
(B and D control 4/10 of A and C's potential gains); in Figures lb and ld, their reward dependence is reduced to
zero (B and D have no reward power over A and C). When A and C's reward dependence is zero, the frequency
of their rewards to B and D should be minimal, and B and D's potential losses from a reduction in the frequency
of those rewards should be lower than their potential gains. Consequently, by Theorem 7, B and D's use ofcoercion should increase.
When actors had no capacity to reward or punish another actor, they could still choose the option of adding or
subtracting zero points. The instructions informed subjects that choosing to add or subtract zero points was
equivalent to not acting toward that partner and would be displayed as such on the other's screen. Thus, such
choices carried no symbolic value and gave subjects a "no action" option; coercion was always voluntary.
Previous experiments in this program studied networks like the one in Figure la, in which both sources of risk
are present. In this structure the low use of coercion has been particularly puzzling because actors B and D
have both the incentive and the capacity to coerce, and their risk of punitive retaliation is reduced (but not
eliminated) by B and D's punishment advantage. Because reward-advantaged actors typically reward their
disadvantaged partners about 35 percent of the time in this and similar networks (Molm 1990), however, the
potential loss from A and C's withholding of rewards, weighted by loss aversion, is higher than the potential gain
from an increase in A and C's reward frequency. This is true even if we generously assume that compliance is
as likely as retaliation and ignore the additional loss that might result from A and C's retaliatory punishment.
Dependent Variable
The dependent variable is the average frequency with which B and D use their coercive power. I divided the
frequency of each actor's punishing behavior toward the focal partner (A or C) by the number of exchange
opportunities (250), thus standardizing the measure to vary from 0 to 1. I then averaged the standardized
frequencies of B's punishment of A and D's punishment of C to create a single measure of their use of
punishment.
Hypotheses
The experiment tests two hypotheses:
H^sub 1^: B and D will punish A and C more frequently when the risk of reward loss is low than when it is
moderate.
H^sub 2^: B and D will punish A and C more frequently when the risk of retaliatory punishment is zero than
when it is moderate.
Results
05 December 2013 Page 12 of 26 ProQuest
5/27/2018 Articol coercitie
15/28
Table 1 shows the mean use of punishment in the four experimental conditions. A twoway analysis of variance
on these means found a strong and significant effect of the risk of reward loss (F^sub 1,36 = 19.77, p
5/27/2018 Articol coercitie
16/28
reward frequency to zero, but it did re
duce it enough (from .39 to .23, based on means during the first 100 trials) that B and D's potential losses from
a decrease in A and C's rewarding were well below B and D's potential gains from an increase in their
rewarding, even with losses weighted for loss aversion.
While removing the risk of retaliatory punishment had little effect by itself, it enhanced the effects of reducing the
risk of reward loss over time. Reducing the risks of both retaliatory punishment and reward loss produced thehighest frequency of punishment, increased (rather than decreased) punishment over time, and increased the
reward frequency of the target of coercion over time. No other condition produced all of these effects. Note,
however, that reducing both sources of risk--by removing A and C's coercive power, and B and D's reward
power-- created a structure of pure coercion (Willer and Anderson 1981 ) in which neither mutual reward
exchange nor mutual punishment was possible. In the structures that are the focus of this research program
(i.e., networks in which actors can engage in the mutual exchange of both rewards and punishments), risk
constrains the use of coercion.
The lack of an overall effect of the risk of punitive retaliation is surprising in some respects, but supports results
of earlier experiments in this program. Molm, Quist, and
Wiseley (1994) found that the structure of only reward power, not coercive power, affected subjects' tendencies
to retaliate against a programmed partner's use of coercion. Reward-power-advantaged actors were more likely
to retaliate, using both reward reduction and retaliatory punishment, regardless of their relative coercive power.
The interaction between the two sources of risk and trial block suggests that reward loss and retaliatory
punishment may affect the use of coercion at different times. Because insufficient rewards from a partner
motivate coercion in the first place, the prospect of changes in those rewards, for better or worse, may be the
only salient factor in the initial decision to use coercion. Once coercion is initiated, its actual consequences,
including punitive retaliation, determine its subsequent use.
EXPERIMENT 2:
05 December 2013 Page 14 of 26 ProQuest
5/27/2018 Articol coercitie
17/28
THE EFFECTS OF COERCIVE POWER UNDER LOW RISK
Experiment 1 showed that reducing risk, particularly the risk of reward loss relative to reward gain, increases the
use of coercion. It could not examine, however, how increased use of coercion affects the impact of structural
variations in coercive power. Apart from the manipulation of risk, the absolute or relative strength of coercive
power remained stable. Experiment 2 was designed to address that question.
Experiment 2 tested how a target of power use responds to varying levels of reward power and coercive power
when the risks of using both bases of power are equivalent for the user. With risk reduced and punishment
increased, the effects of coercive power should also increase. Actors with relatively greater coercive power
should be more successful than relatively weaker actors at increasing their partners' reward frequency.
Design
Because reducing the risk of reward loss leaves actors with only coercive power over their partners, this
experiment cannot compare the effects of reward power and coercive power in networks in which all actors
have both. Instead, I compared the effects of the two bases of power when each is an actor's sole power source
by placing an actor with only coercive power (B) and an actor with only reward power (C) in direct competition
with each other for exchange with a third actor, A, who has only reward power (Figure 3). (The fourth actor in
the network, D, provides an alternative for B and C, but D offers only minimal rewards and thus D's behavior is
not of interest.)
In all conditions, both B and C are highly and equally dependent on A for rewards; both actors have a strong
(and equal) incentive to secure A's rewards through strategic power use. And, in all conditions, both sources of
risk for B's use of coercion are equal to those in the lowest risk condition of Experiment I (i.e., Figure ld). A has
no power to punish and no reward dependence on B. Thus, both the incentive to use power and the risks of
power use are equal for both reward power and coercive power, and both bases of power must be used
strategically.
The completely randomized design compares three conditions that vary B's power
over A relative to C's power over A: (1) B with lower power than C (low coercive, high reward power, Figure 3a)(2) B and C with equal power (Figure 3b), and (3) B with greater power than C (high coercive, low reward
power, Figure 3c). Subjects exchanged for 200 opportunities.
The main dependent variable is the relative frequency with which A rewards B, computed by dividing the
frequency of A's rewards to B by the total frequency with which A rewards both B and C.
Hypotheses
Initially, A should prefer to exchange with C, the actor with the power to reward A, in all conditions. But over
time, if B punishes A when A fails to reward B, A's rewards to B should increase in proportion to B's relative
power over A. The initial effects of the base of power that B and C control should disappear; only their relative
power should matter.
H1: The relative frequency with which A rewards B rather than C will vary directly with B's relative power over A.
H2: Support for Hypothesis 1 will increase over time.
05 December 2013 Page 15 of 26 ProQuest
5/27/2018 Articol coercitie
18/28
H3: By the last 50 opportunities, the frequency of A's rewards to B when B's coercive power is high and C's
reward power is low will equal the frequency of A's rewards to C when C's reward power is high and B's
coercive power is low.
As Figure 4b shows, in the first trial block the frequency with which B punished A did not vary by condition. But
in successive trial blocks, the three conditions diverge as actors with low coercive power punish more frequently
in response to declining rewards from A, while actors with high coercive power decrease their punishment asA's rewards increase. A comparison of Figures 4a and 4b shows that the timing of these increases and
decreases in B's punishment of A corresponds closely with changes in the relative frequency of A's rewards to
B. Thus, in the high-coercive-power condition, when A's rewarding increases sharply in the third trial block, B's
punishment also declines markedly. And, in the low-coercive-power condition, the decrease in A's rewarding in
the second trial block prompts an increase in B's punishment. When increased use of punishment fails to
compensate for B's weak coercive power, however, B's punishment again declines, although it remains
substantially higher in the low-coercive-power condition than in the high-coercive-power condition.
05 December 2013 Page 16 of 26 ProQuest
5/27/2018 Articol coercitie
19/28
The contingency of punishment. These results show, as predicted, that coercive power is more effective when
used more frequently. But frequency per se is a poor measure of coercion. To use punishment to coerce re
wards, actors must make their punishment contingent on the partner's failure to reward.
05 December 2013 Page 17 of 26 ProQuest
5/27/2018 Articol coercitie
20/28
To examine whether the effectiveness of B's coercive power did result from an increase in contingent
punishment (as the patterns in Figure 4 suggest), rather than random, or noncontingent, punishment, I
regressed the relative frequency with which A rewarded B in the last 50-trial block on measures of the frequency
and the contingency with which B punished A during the entire
200 trials. The effects of the frequency and contingency of B's punishment were estimated after controlling for
the effects of the experimental manipulation of coercive power and reward power (represented by two
dichotomous variables). I computed the contingency of B's punishment on A's nonrewarding by the difference
between two conditional probabilities:
the conditional probability of punishment from B (P^sub B) given no rewards from A ( R^sub A^) on the previous
opportunity, minus the conditional probability of punishment from B (P^sub B^) given prior rewards from A
(R^sub A^). Overall, the mean contingency of B's punishing A was .33, and this mean did not vary by power
condition.
05 December 2013 Page 18 of 26 ProQuest
5/27/2018 Articol coercitie
21/28
The results, summarized in Table 2, show the strong effects of coercive and contingent punishment. Not only
does the contingency of B's punishment have an independent effect on the distribution of A's rewards, but the
contingency of B's punishment is the strongest predictor of the relative frequency of rewards from A. The
greater the contingency of B's punishment on A's nonrewarding, the more likely A was to reward B rather than
C. With contingency controlled, the frequency of punishment has a negative effect on A's rewarding, as would
be expected.
SUMMARY AND CONCLUSIONS
I have examined how risk of retaliation and fear of loss constrain the strategic use of coercive power in social
exchange relations. Theories of choice in situations involving risk and uncertainty, particularly Axelrod's (1984,
1986) evolutionary theory and Kahneman and Tversky's (1979) prospect theory,
help explain why actors rarely use coercive power, even when their incentives and capacities to coerce are
high. Power use is risky (i.e., it may produce either gains or losses), and actors fear loss far more than they
value gain. Consequently, actors in imbalanced exchange relations will accept insufficient and unequal rewards
rather than risk the consequences of using their power, and they will reduce their use of power strategies that
produce immediate retaliation, sometimes at the cost of long-term gains.
In general, a structural advantage on either reward power or coercive power reduces risk by reducing losses
from retaliation by partners holding that same base of power. Thus, actors with greater coercive power face less
risk of loss from another's punitive retaliation, and actors with greater reward power face less risk from the
other's reward withholding. But because the incentive to use coercion comes not from the power to coerce but
from dependence on another for rewards, coercion is still risky, even for actors who are advantaged on coercive
power. Reward dependence creates the desire to make the partner give more, but it also creates the fear that pressuring for more will lead to losing what the partner is currently
giving. This conflict between potential gain and potential loss constrains coercion until the partner's reward
05 December 2013 Page 19 of 26 ProQuest
5/27/2018 Articol coercitie
22/28
frequency is low enough that potential losses from the partner's reward withholding, even when weighted by
loss aversion, are lower than potential gains.
The results of both the theoretical analysis and Experiments 1 and 2 suggest that actors are more likely to use
coercion in purely coercive structures, or at least in relations in which one actor's reward dependence is too low
for that actor to benefit from exchange. In purely coercive structures, there is no mutual reward exchange to risk
losing. These forms of coercion, while highly visible (master and slave, mugger and victim, torturer andprisoner), are relatively rare. My interest throughout this research program has been, instead, in the use and
effects of coercive power in relations of social exchange in which both actors are dependent on each other for
rewards. This analysis suggests that if the average reward power of the actors is high enough to sustain a
mutually beneficial exchange relation (albeit an unequal one),
then the risk of losing a partner's rewards, however insufficient and unequal those rewards might be, is likely to
be great enough to keep coercion relatively low and, consequently, relatively ineffective.
There are two interesting ironies to this analysis. First, the same principle that should make coercive power
highly potent when it is used--the preference for minimizing loss over maximizing gain--instead actually
constrains its use. Loss aversion can prevent an actor from punishing a negligent business partner, for
example, even though the partner's same aversion to loss should increase the effectiveness of punishment. As
previous experiments have shown, coercive power applied contingently and consistently is highly effective in
increasing rewards from a partner, even when the partner's coercive power is equal (Molm 1994).
A second irony is that the structural conditions that encourage the use of coercion by
reducing risk should decrease its effectiveness. Actors are more likely to sanction an exchange partner who
rarely rewards them than one who rewards them more frequently, but their chances of successfully coercing
such an uninterested partner are also less. In general, the lower a partner's reward dependence (and the more
infrequent the partner's rewarding), the more likely and the less successful coercion becomes. Coercion can be
effective even under these conditions, as Experiment 2 demonstrated, but the coercer must rely solely on the
threat of punishment to maintain the other's rewarding rather than use coercion as a tool to establish a rate of
reward exchange that is more favorable to the coercer. Punishment is far more effective when combined with
rewards (e.g., punishing low levels of exchange but rewarding high levels) and when used in a structure in
which actors' joint dependence, or average power, is high enough that both can benefit from the relation (Molm
1994; Van Houten 1983).
While this analysis implies that coercion will be used only sporadically in relations of
mutual reward dependence, there are some conditions under which its use, even in these relations, should
increase. First, coercion should be more likely when either norms or lack of alternatives constrain a partner from
withdrawing from the relation, thus reducing the risk of reward loss. (The partner's fre
quency of rewarding could still decline, of course.) In long-term or committed relations, various combinations ofsocial norms, contractual obligations, and restricted alternatives tend to keep people in relations such as
marriages, friendships, and business or political partnerships. The reduced risk of losing a valued exchange
relation in a contractual relationship such as marriage may be one reason why researchers have found that
punitive and coercive behavior is more common among family members than among strangers or casual
acquaintances (Birchler, Weiss, and Vincent 1975; Halverson and Waldrop 1970; Winter, Ferreira, and Bowers
1973).
Second, a decline in a partner's reward dependence and reward frequency can prompt coercive tactics. When
mutual reward dependence is relatively high, as it must be when relations are first established, coercion is
unlikely. But over time, actors' dependencies on one another can decline, and they often decline unequally. As
a result, it is not un
common to see coercive tactics employed by workers who are about to be laid off, individuals whose marriage
partners have ceased to care for them, or political factions that have lost their clout with mainstream parties.
05 December 2013 Page 20 of 26 ProQuest
5/27/2018 Articol coercitie
23/28
When a decline in mutual dependence and an increase in power imbalance are combined with barriers to
dissolving relations, coercion should be particularly likely to increase.
Third, although I have argued that coercion is most beneficial for those actors disadvantaged on reward power,
there may be some conditions under which advantaged actors, who have less to lose from the partner's reward
withdrawal, can also benefit from coercion and will be more likely to use it. One condition is when an actor has
much stronger coercive power than reward power; a second is when reward exchange entails substantial costs, over and above opportunity costs. Under these
conditions, coercion may be more effective and less costly than reward exchange, even asymmetrical
exchange. Both conditions exist, for example, in the case of slavery: The coercive power of slaveholders
typically exceeds their power in the economic marketplace; and paying workers involves real costs. When these
conditions exist, actors are likely to replace reward exchange with coerced exchange (as we see in the example
of slavery), or at least to reduce
their rewarding below the level that would be necessary to maintain the partner's rewards without coercion.
Thus, in some cases, the use of reward power may depend on the structure of coercive power, just as the use
of coercive power depends on the structure of reward power.
These considerations suggest several directions for further research, such as how coercion is affected by
barriers to leaving relations, changes in reward dependencies, or variations in the costs of exchange. Studies of
such effects should benefit from the analysis of risk introduced in this study. As I have shown, the use of
coercive power cannot be predicted from the structure of coercive power alone. Incorporating coercion within
the framework of social exchange theory requires considering the risk of power use, the dynamic effects of
changing reward levels in relations, and interactions between reward power and coercive power.
Footnote* Direct correspondence to Linda D. Molm, Department of Sociology, University of Arizona, Tucson, AZ 85721
([email protected]). This work was supported by two grants from the National Science Foundation (SES-
8921431 and SES-9210399). I thank Phillip Wiseley, Theron Quist, David Richmond, Angelina Quesada, Anne
Lane, Lisa O'Laughlin, and Lawrence Ducchesi for their assistance in conducting the research, and the ASR
Editors and anonymous reviewers for their helpful comments.
Footnote1 Studies of bargaining relations also report noncomparable effects of reward power and coercive power on
concession behavior (Bacharach and Lawler 1981; Lawler and Bacharach 1987).
Footnote2 The scope conditions of a theory describe the set of conditions under which the theory is assumed to hold;
power-dependence theory is restricted to social interaction that meets these four scope conditions.
Footnote3 Theorists have recently developed several formal models based on power-dependence or alternative theories
that predict exact distributions of exchange benefits in imbalanced relations under certain conditions of
negotiated exchange (see Social Networks, vol. 14, June 1992).
Footnote4 Some theorists include other elements in their definitions of coercion. Many political theorists restrict coercion
to the threat or use of severe deprivation, physical violence, or loss of life (Bachrach and Baratz 1963; Bierstedt
1950; Dahl 1957). Some also specify that coercion must entail restriction of freedom (Bierstedt 1950; Etzioni
1968). And some define coercion only as threatened sanctions, arguing that if the threat is carried out, coercion
has failed (Bachrach and Baratz 1963; Willer and Markovsky 1993). Because my aim is to compare structures
of reward and punishment power that differ only on whether the sanction is positive or negative, 1 exclude these
additional stipulations. Threat is an essential element in "one-shot" coercive transactions, but not in repeated
exchanges between the same actors. Instead, the contingency of punishment on nonreward gradually induces
05 December 2013 Page 21 of 26 ProQuest
5/27/2018 Articol coercitie
24/28
compliance over time.
Footnote5 Unlike an exchange of material goods (e.g., money for food), performing a behavior that produces valued
outcomes for another (e.g., commenting on a colleague's paper or visiting a sick friend) involves no actual
transfer of resources, does not deplete the actor's supply of resources, and primarily entails opportunity costs.
To simplify the analysis, I further assume that enacting such behaviors involves only opportunity costs (othercosts might include fatigue, effort, investment costs, etc.).
Footnote6 While coercive strategies are typically purposive attempts to increase the partner's rewarding, they are
sometimes shaped without awareness. Some punitive behaviors, for example, are initially elicited by emotions
and then maintained by the rewards they produce. This seems to explain the rapid development of coercive
strategies employed by young children, who quickly learn to cry, whine, or throw tantrums to increase attention
from adults (Patterson 1982).
Footnote7 This principle is shared, more generally, by all theories that posit some process of value adaptation. In
particular, see Thibaut and Kelley's (1959) analysis of the comparison level (CL) of exchange relations.
Footnote8 Although the value function was originally drawn to account for patterns of decision-making about future
events, Tversky and Kahneman (1991) suggest it is equally valid for describing reactions to past experience. My
research on subjects' affective responses to their exchange partners' behavior supports this view. I measured
affect by a scale comprised of five semantic-differential items. Averaged across different power positions, I
found that the negative affect produced by one unit of punishment from an exchange partner was 2.3 times
stronger than the positive affect produced by an equivalent unit of reward (Molm 1991).
Footnote10 Gender effects were not of theoretical interest in this study; gender was included as a block variable solely
for control purposes. Preliminary data analyses found no effects of gender, and gender is omitted from the
analyses reported here.
Footnote11 The exchange of "money for money" is, of course, descriptively unlike exchanges in natural
Footnotesettings, whether these involve exchanges of similar benefits (e.g., friendship for friendship) or different benefits
(e.g., advice for approval). Because exchange theory assumes that the domain of value does not affect
theoretical predictions (i.e., actors behave in similar ways to obtain the outcomes they value, whether those
outcomes are money or approval or something else), what isFootnoteimportant for testing the theory is assuring that the benefits used in the exchange are valued by the subjects.
Exchange researchers typically meet this assumption by using money as the exchange benefit in experiments
and recruiting subjects on the basis of their desire to earn money. In this study, the use of money also makes it
possible to create parallel forms of reward and punishment by operationalizing these concepts as gains and
losses of the same valued outcome. Earnings in the experiment averaged $15 for a two-hour session. Subjects
were guaranteed a minimum of $6 and could earn a maximum of $25 in some positions of some networks.
ReferencesREFERENCES
ReferencesAxelrod, Robert. 1984. The Evolution of Cooperation. New York: Basic Books. ______. 1986. "An Evolutionary
Approach to Norms." American Political Science Review 80:1095-111. Bachrach, Peter and Morton S. Baratz.
05 December 2013 Page 22 of 26 ProQuest
5/27/2018 Articol coercitie
25/28
1963. "Decisions and Nondecisions: An Analytical Framework." American Political Science Review 57:641-51.
Bachrach, Samuel B. and Edward J. Lawler. 1981. Bargaining: Power, Tactics, and Outcomes. San Francisco,
CA: Jossey-Bass. Bierstedt, Robert. 1950. "An Analysis of Social Power." American Sociological Review 15:
730-38. Birchler, Gary, Robert L. Weiss, and John P. Vincent. 1975. "Multimethod Analysis of Social
Reinforcement Exchange between Maritally Distressed and Nondistressed Spouse and Stranger Dyads."
Journal of Personality and Social Psychology 31:349-60. Blalock, Hubert M. 1987. "A Power Analysis ofReferencesConflict Processes." Pp. 1-40 in Advances in Group Processes, vol. 4, edited by E. J. Lawler and B. Markovsky.
Greenwich, CT: JAI Press. Blau, Peter M. 1964. Exchange and Power in Social Life. New York: Wiley. Boulding,
Kenneth E. 1989. Three Faces of Power. Newbury Park, NJ: Sage. Cook, Karen S. and Richard M. Emerson.
1978. "Power, Equity and Commitment in Exchange Networks." American Sociological Review 43:721-39.
Cook, Karen S., Richard M. Emerson, Mary R. Gillmore, and Toshio Yamagishi. 1983. "The Distribution of
Power in Exchange Networks: Theory and Experimental Results." American Journal of Sociology 89:275-305.
Dahl, Robert A. 1957. "The Concept of Power." Behavioral Science 2:201-5. Deutsch, Morton. 1973. The
Resolution of Conflict. New Haven, CT: Yale University Press. Emerson, Richard M. 1972a. "Exchange Theory,
Part I: A Psychological Basis for Social Exchange." Pp. 38-57 in Sociological Theories in Progress, vol. 2,
edited by J. Berger, M.
ReferencesZelditch, Jr., and B. Anderson. Boston, MA: Houghton-Mifflin. ______. 1972b. "Exchange Theory, Part II:
Exchange Relations and Networks." Pp. 58-87 in Sociological Theories in Progress, vol. 2, edited by J. Berger,
M. Zelditch, Jr., and B. Anderson. Boston, MA: Houghton-Mifflin. ______. 1981. "Social Exchange Theory." Pp.
3065 in Social Psychology: Sociological Perspectives, edited by M. Rosenberg and R. H. Turner. New York:
Basic Books. Etzioni, Amitai. 1968. The Active Society: A Theory of Societal and Political Processes. New York:
The Free Press. Fishburne, Peter C. and Gary A. Kochenberger. 1979. "Two-piece von Neumann-Morgenstern
Utility Functions." Decision Sciences 10:50318. Gray, Louis N. and Irving Tallman. 1987. "Theories of Choice:
Contingent Reward and Punishment Applications." Social Psychology Quarterly 50:16-23. Halverson, Charles
F., Jr. and Mary R. Waldrop. 1970. "Maternal Behavior toward Own and Other Preschool Children: The Problem
of 'Ownness."' Child Development 41:839-45. Heath, Anthony F. 1976. Rational Choice and Social Exchange: A
Critique of Exchange Theory. Cambridge, England: Cambridge University Press. Hershey, John C. and Paul J.
H. Schoemaker. 1980. "Risk Taking and Problem Context in the Domain of Losses: An Expected Utility
Analysis." Journal of Risk and Insurance 47:111-32. Homans, George C. [1961] 1974. Social Behavior: Its
Elementary Forms. New York:
ReferencesHarcourt Brace and World. Kahneman, Daniel and Amos Tversky. 1979. "Prospect Theory: An Analysis ofDecision Under Risk." Econometrica 47:263-91. ______. 1984. "Choices, Values, and Frames." American
Psychologist 39:341-50. Kahneman, Daniel, Jack L. Knetsch, and Richard H. Thaler. 1991. "The Endowment
Effect, Loss Aversion, and Status Quo Bias." Journal of Economic Perspectives 5:193-206. Kanter, Rosabeth
Moss. 1977. Men and Women of the Corporation. New York: Basic Books. Kollock, Peter. 1994. "The
Emergence of Exchange Structures: An Experimental Study of Uncertainty, Commitment, and Trust." American
Journal of Sociology 100:313-45. Lawler, Edward J. 1986. "Bilateral Deterrence and Conflict Spiral: A
Theoretical Analysis." Pp. 107-30 in Advances in Group Processes, vol. 3, edited by E. J. Lawler. Greenwich,
CT: JAI Press. ______. 1992. "Power Processes in Bargaining." The Sociological Quarterly 33:17-34. Lawler,
Edward J. and Samuel B. Bacharach. 1987. "Comparison of Dependence and Punitive Forms of Power." Social
Forces 66:44662.
ReferencesMarkovsky, Barry, David Willer, and Travis Patton. 1988. "Power Relations in Exchange Networks." American
05 December 2013 Page 23 of 26 ProQuest
5/27/2018 Articol coercitie
26/28
Sociological Review 53: 220-36. Molm, Linda D. 1988. "The Structure and Use of Power: A Comparison of
Reward and Punishment Power." Social Psychology Quarterly 51: 108-22. ______.1989. "An Experimental
Analysis of Imbalance in Punishment Power." Social Forces 68:178-203. ______.1990. "Structure, Action and
Outcomes: The Dynamics of Power in Exchange Relations." American Sociological Review 55:42747.
______.1991. "Affect and Social Exchange: Satisfaction in Power-Dependence Relations." American
Sociological Review 56:475-93. ______. 1994. "Is Punishment Effective? Coercive Strategies in SocialExchange." Social Psychology Quarterly 57:75-94. Molm, Linda D. and Karen S. Cook. 1995. "Social Exchange
and Exchange Networks." Pp. 20935 in Sociological Perspectives on Social Psychology, edited by K. S. Cook,
G. A. Fine, and J. S. House. Boston, MA: Allyn and Bacon. Molm, Linda D., Theron M. Quist, and Phillip A.
Wiseley. 1994. "Imbalanced Structures, Unfair Strategies: Power and Justice in Social Ex
Referenceschange." American Sociological Review 59:98121. Morgan, Patrick M. 1977. Deterrence: A Conceptual
Analysis. Beverly Hills, CA: Sage. Patterson, Gerald R. 1982. Coercive Family Process. Eugene, OR: Castalia
Publishing. Raush, Harold L., William A. Barry, Richard K. Hertel, and Mary Ann Swain. 1974. Communication,
Conflict, and Marriage. San Francisco, CA: Jossey-Bass. Samuelson, William and Richard Zeckhauser. 1988.
"Status Quo Bias in Decision Making." Journal of Risk and Uncertainty 1:7-59. Schelling, Thomas C. 1960. The
Strategy of Conflict. New York: Oxford University Press. Stets, Jan E. 1995. "Identity and Status in Interaction."
Presented at the annual meetings of the American Sociological Association, 1995, Washington, DC. Tedeschi,
James T., Barry R. Schlenker, and Thomas V. Bonoma. 1973. Conflict, Power, and Games. Hawthorne, NY:
Aldine. Thibaut, John W. and Harold H. Kelley. 1959. The Social Psychology of Groups. New York: Wiley.
Tversky, Amos and Daniel Kahneman. 1991. "Loss Aversion in Riskless Choice: A Reference-Dependent
Model." The Quarterly Journal of Economics 106:1039-61.
ReferencesVan Houten, Ron. 1983. "Punishment: From the Animal Laboratory to the Applied Setting." Pp. 13-44 in The
Effects of Punishment on Human Behavior, edited by S. Axelrod and J. Apsche. New York: Academic Press.
Willer, David. 1987. Theory and the Experimental Investigation of Social Structures. New York: Gordon and
Breach Science Publishers. Willer, David and Bo Anderson, eds. 1981. Networks, Exchange and Coercion: The
Elementary Theory and Its Applications. New York: Elsevier. Willer, David and Barry Markovsky. 1993.
"Elementary Theory: Its Development and Research Program." Pp. 323-63 in Theoretical Research Programs:
Studies in the Growth of Theory, edited by J. Berger and M. Zelditch Jr. Stanford, CA: Stanford University
Press. Winter, William D., Antonio J. Ferreira, and Norman Bowers. 1973. "Decision-Making in Married and
Unrelated Couples." Family Process 12:83-94. Yamagishi, Toshio. 1995. "Social Dilemmas." Pp. 311-35 in
Sociological Perspectives on Social Psychology, edited by K. S. Cook, G. A. Fine, and J. S. House. Boston, MA:
Allyn and Bacon.AuthorAffiliationLinda D. Molm is Professor of Sociology at the University of Arizona. Her primary research interest is the
experimental analysis of theories of social exchange and power. Her recent book, Coercive Power in Social
Exchange (Cambridge University Press, 1997), summarizes her decadelong study of coercive power. She is
now engaged in a project comparing reciprocal and negotiated forms of exchange. She is co-editor (with Lynn
Smith-Lovin) of Social Psychology Quarterly.
Subject: Risk; Sociology; Power;Publication title:American Sociological Review
Volume: 62
Issue: 1
05 December 2013 Page 24 of 26 ProQuest
5/27/2018 Articol coercitie
27/28
Pages: 113-133
Number of pages: 21
Publication year: 1997
Publication date: Feb 1997
Year: 1997Publisher:American Sociological Association
Place of publication:Albany
Country of publication: United States
Publication subject: Psychology, Sociology, Population Studies, Social Sciences: Comprehensive Works
ISSN: 00031224
CODEN:ASREAL
Source type: Scholarly JournalsLanguage of publication: English
Document type: Feature
Accession number: 03161444
ProQuest document ID: 218821543
Document URL: http://search.proquest.com/docview/218821543?accountid=15859
Copyright: Copyright American Sociological Association Feb 1997
Last updated: 2012-07-24Database: ProQuest Central
05 December 2013 Page 25 of 26 ProQuest
http://search.proquest.com/docview/218821543?accountid=15859http://search.proquest.com/docview/218821543?accountid=158595/27/2018 Articol coercitie
28/28
BibliographyCitation style: APA 6th - American Psychological Association, 6th Edition
Molm, L. D. (1997). Risk and power use: Constraints on the use of coercion in exchange. American Sociological
Review, 62(1), 113-133. Retrieved from http://search.proquest.com/docview/218821543?accountid=15859
_______________________________________________________________Contact ProQuestCopyright2013 ProQuest LLC. All rights reserved. - Terms and Conditions
05 December 2013 Page 26 of 26 ProQuest
http://www.proquest.com/go/contactsupporthttp://www.proquest.com/go/contactsupporthttp://search.proquest.com/info/termsAndConditionshttp://search.proquest.com/info/termsAndConditionshttp://www.proquest.com/go/contactsupport