RELATIONAL GOODS AND OVERCOMING BARRIERS TO COLLECTIVE ACTION

Relational goods contribute to understanding why people engage in collective action, notably including political participation, even though, as Olson showed, it would often be more rational for them to free-ride on the activity of others. Relational goods are neither public goods nor private goods but a third type of good. They must be jointly consumed with another person or persons (unlike private goods), but the identity of the other persons matters crucially (unlike the case for public goods). Relational goods can only exist by mutual agreement as part of a relationship with non-arbitrary others, in the context of an interaction. Friendship is a prototypical example. Relational goods can exist along a range of personal contact from direct, where individuals interact face to face, to indirect, where the interaction may be at a distance with a certain type of person. In the indirect case the relational good may frequently take the form of reinforcing a desired social identity. The distinction between “consumption” and “instrumental” goods applies, parallel with other usage. The consumption relational goods are produced independent of any consequences of the action or relationship, while the instrumental ones refl ect consequences, such as from an action that enhances the value of an identity. People who value relational goods may act collectively even if other net benefi ts of action are negative. Larger groups become more, not less, prone to collective activity. Relational goods provide a missing element to understand how the process of mobilization works at the individual level; leaders can infl uence people’s perceptions of what others are doing, of the value of a shared identity, and of the likelihood of success. Some limited empirical evidence is consistent with relational goods playing a role in enhancing collective action.


RELATIONAL GOODS AND OVERCOMING BARRIERS TO COLLECTIVE ACTION
Mancur Olson's The Logic of Collective Action brilliantly used the concepts surrounding public goods to show why collective action is profoundly surprising.He directed attention to the logical inconsistency of assuming that just because a group of people had interests in common they would act to pursue them; in fact, to the contrary, only under special conditions would they take action to pursue such ends.Ever since publication of Olson's seminal work, scholars have worked to reconcile the frequent irrationality of collective action with the inconvenient empirical reality of its widespread existence.Often this work examines the institutions or the preferences that make this behavior appear rational.The logic identified by Olson applies with special force to politics and especially to mass political behavior which, therefore, presents us with a paradox.
Political participation is a form of collective action, and thus from the rational choice perspective it is puzzling that people engage in it.The typical outcome is a public good (such as a government policy or an election outcome).At the same time, each individual's actions (voting, demonstrating, giving money to a campaign, contacting an official) has a small to negligible probability of changing that outcome, even if collectively the actions are powerful.Voting is the most problematic.Why, then, do people take these actions?It is very difficult to argue plausibly that the public good of the outcome provides the reason for action.
Various solutions have been presented to account for the paradoxical observation that people nonetheless do act collectively.This is a subset of the more general problem of explaining cooperation by players in prisoners' dilemma and related social dilemmas (Kaminski 2014 provides an overview of solutions in the literature).A number of authors have specifically addressed the theoretically surprising fact that people do participate in politics instead of free-riding.Many of the answers assume people have some reason beyond making a real difference in achieving the collective political outcome (see Oppenheimer 2012, Bardsley 2005, Blais 2000, Aldrich 1993 for summaries of this literature).They may act from a sense of duty or of fulfilling a social norm, or from the sheer enjoyment of the action, or out of habit, or from an other-regarding sensibility that weighs the needs of the collective over their own self-interest.Some of the theoretical solutions emphasize the role of leaders and leadership in overcoming the tendency to free-ride (Oppenheimer 2012, Frohlich, Oppenheimer, and Young 1971, Uhlaner 1989a).Empirical work on political participation has taken up this point and demonstrated that elites do increase participation, a process labeled recruitment or mobilization (Rosenstone and Hansen 1993, Verba, Schlozman, Brady 1995, Gerber and Green 2000).
While each of the above answers to the puzzle is useful and important, they are conceptually incomplete.The various private goods proposed as reasons for action tend to generate ad hoc explanations.The leadership/recruitment story does provide a systematic link to political forces, but it leaves the process under-specified.In the usual story leaders provide selective incentives and reduce costs, but there are missing pieces in understanding how this process works at the level of the individual citizen.
"Relational goods" mitigate these conceptual difficulties and provide an important answer to the puzzle of collective action.As I have proposed elsewhere (Uhlaner 1989b, Uhlaner 2014), when we add "relational goods" to the set of outcomes valued by individuals, collective action becomes more expected under a much wider set of circumstances.They help explain why people act for a jointly shared end instead of free-riding, without requiring that we postulate altruism or a sense of obligation.Moreover, models which include relational goods clarify the process of mobilization by leaders, the function of duty, and the reasons that identification with some group can increase activity (such as racial and ethnic groups in the contemporary United States).Attention to relational goods also leads to the conclusion that an increase in group size can enhance activity.
Before showing how relational goods mitigate the collective action problem, they need to be defined.Understanding their role will require distinguishing among several types.After defining relational goods and these types, this paper moves on to a simple discussion of how they alter individuals' decision to act or not to act.The paper then moves to some empirical examples drawn from identity politics and some speculation about revolution.

DEFINING AND CATEGORIZING RELATIONAL GOODS
Relational goods are neither public goods nor private goods, but a third type.The concept was developed in parallel by Donati (1986), Gui (1987), Uhlaner (1989b) and, with somewhat different content, Nussbaum (1986). 1 The discussion below develops the theory as presented in Uhlaner (1989bUhlaner ( , 2014)).Relational goods can only be obtained by a person in conjunction with others.They are not private goods, as these can be consumed alone (such as a banana, or the psychic reward of feeling as if you have done your duty by voting).Neither are they public goods as these are shared with others but the utility to the individual does not depend upon them being shared (clean air may not be excludable but its value to me does not depend upon you).In contrast, relational goods can only exist by mutual agreement as part of a relationship with non-arbitrary others, in the context of an interaction.They therefore also differ from club goods, where excludability exists but the identity of the others in the club is not relevant.Relational goods, in contrast, exist only when shared with some specific other person or people in some specific set.They require joint action by a person and non-arbitrary others to exist, appear simultaneously in the utility functions of two or more people, and the utility depends upon that joint presence.Friendship is a prototypical relational good, but so is acceptance by others and other aspects of sociability.A beautiful sunset is a public good that can be enjoyed on one's own; the additional enjoyment from sharing it with a friend is a relational good.Relational goods may include "social approval, solidarity, a desire to experience one's history, friendship and its benefits, the desire to be recognized or accepted by others, the desire to maintain an identity, other aspects of sociability ...." (Uhlaner 1989b, p. 255).
The characteristic of a relational good that it must involve a relationship between someone and some specific other or others, not just an anonymous relationship, raises some definitional difficulties and helpful analytic complexities. 2For many relational goods, such as friendship or love, as ordinarily understood, the interaction must be with a specific person.For others, such as Nussbaum's (1986) "civic commitment" or Sugden's (2002Sugden's ( , 2005) "fellow-feeling," the relationship may be with any or many other individuals including ones known only as a "type" of person rather than as a specific named person.In the sunset example from above, or while sitting in a theater, I may even derive a relational good, a pleasure, from sharing the experience with strangers.I do not know their names nor expect to see them again, but I do recognize them as somehow like me (very much in the sense of Sugden's fellowfeeling).While complete anonymity disqualifies a good from being relational (for example, the value of "money" depends upon mutual agreement, but the value is independent of whoever provides it), a very short step away from anonymity may be sufficient to create the conditions for a relational good, yet pure anonymity would take us outside the realm of relational goods.The more impersonal relational goods are important for understanding political participation.
While all relational goods require interaction with identified others, the variation in the range of personal contact leads to conceptualizing relational goods as lying along a spectrum from direct to indirect (Uhlaner 1989b).The distinction between direct and indirect relational goods depends on whether the contact among those interacting is in person, with specific known individuals, (direct) or at a distance, such as with a type of person or some collectivity with which one identifies and 2 This characteristic is closely related to "identity" on Bruni and Zamagni's list (2007, pp. 242-243).considers oneself a member, (indirect).Donati's primary and secondary relational goods (2011, p. 158) capture a similar distinction.These types roughly correspond to interaction with a primary group versus with a secondary group.There are degrees of indirectness, so the distinction suggests a continuum rather than a strict dichotomy.
In the direct case, individuals interact face-to-face but in any event they know each other personally.The relational good here arises from interaction with a specific known other individual, and thus mutual knowledge and mutual observation are both likely.The paradigmatic example is friendship.
With indirect relational goods, face-to-face contact is absent, but a person thinks of himself or herself as part of some group.The outcome of action and hence the value of the relational good depends upon the actions of both self and others.In the intermediate case, persons may interact with some other specific members whose actions they can observe, but they know many other members only by type and have only indirect contact.Farther away from direct contact, a person may think of him or herself as part of some group in which he or she values acceptance.Perhaps the paradigmatic example of such goods is social identity, the feeling of belonging to a particular group in society and being accepted by the other members of that group.Thus the other people in the group may not be precisely identifiable, they may not be known by name, but they are known as people who share a world view or an ideology or an ancestral homeland or a neighborhood.Indirect relational goods arise when the individual values maintaining the connection with this group.Solidary goods, or sociability, fall into this middle ground, where a person feels connected to some specified subset of people even without really knowing who they are.
The spectrum from direct to indirect relational goods is related to the conceptual problem of defining a "group" or collectivity engendering identity.Groups with which people identify vary along a continuum in size, in the ease with which members can recognize each other or reliably scrutinize each other, in their cohesion, in the strength of identification people feel with the group, and in the extent to which others are explicitly excluded.Many of these differences are related to how individuals interact with each other and thus whether the relational goods are direct or indirect.At one end, specific individuals know each other and interact directly.In groups farther along the spectrum, members interact with some other members whose actions they can observe, but they know many other members only as people characterized in a certain way and have only indirect contact.At the far end of the continuum membership depends only upon a subjective sense of identification self-enforced by adherence to perceived norms.Beyond that, relational goods disappear as they do in transactions such as money-based market exchanges where the interaction can be with any interchangeable "other."Bimber et al. (2012, p. 89) discuss "interaction" of organization members occurring along a similar continuum, from "personal interaction" involving "repeated, intentional interaction with known others over time," to "impersonal interaction" which "involves no personal, direct contact with known others" but in a situation with "shared affiliation and interest."As Bimber et al. (2012) note, this continuum is similar to the strong tie, weak tie distinction made by Granovetter (1973Granovetter ( , 1983)).At the impersonal or weak end both of these conceptions differ from relational goods as for the latter to exist the person must value the identity.
The direct relational goods are in the easier of the two types to understand.It seems that (almost) everyone values friendship, and there are probably sound evolutionary reasons for this preference.The indirect relational goods are likely less universal and in any case more variable.While there is a general propensity towards "groupness," as shown by the work of Tajfel (1982, Tajfel et al. 1971, Tajfel and Turner 1979) and Turner (1975Turner ( , 1982) ) and others working on social identity, individuals differ.Individuals also differ in how broad a definition they accept for their group, with "all of humanity" or "all creation" an extreme claimed by some, but credibly only by very few.Relational goods closer to the anonymity pole depend more upon the person's subjective sense of self-identification; the farther from anonymity, the more it depends upon objective inclusion by others.
In addition to the "direct" versus "indirect" dimension, the standard distinction between "instrumental" and "consumption" goods applies to relational goods as well, with the caveat that some actions can yield both consumption and instrumental relational goods.Consumption relational goods come from performing some action or being in some relationship without any further consequence needed to produce them.I watch a sunset with a friend, I derive a consumption relational goods.I join a friend at a protest against killing whales; I enjoy that we acted together and I enhance my sense of myself as an environmentalist.The consumption relational goods are the more straightforward type -they include all the identity and relationship benefits, feeling part of a group, feeling one has done one's duty, feeling liked.In sum, the consumption indirect relational goods involve enhancing one's own sense of identity.
The instrumental relational goods are more complex; they refer to an enhancement of identity as a result of outcomes from some actions.The protest may have some policy consequence, such as stimulating tighter regulation of whaling or increasing the salience of the issue.The policy outcome is available to all members, whether or not they participated, and thus faces the usual problems of public goods.The additional instrumental relational good arises "if action by one's group bolsters the group's political identity" (Uhlaner 1989b p. 257) thereby increasing its power and influence.Crucially, the person only gets his or her share of that enhanced identity by having taken part in the action, thereby strengthening the claim to membership in the group.These instrumental goods are relational because the value to the individual depends both upon his or her actions and those of specific other people.The group's power and prestige depend upon your action and your own actions are necessary as an "entry ticket" to claim the identity.The actions of others enhance the odds of success.
Instrumental relational goods can occur anywhere along the direct to indirect spectrum, although they are more likely and more important with the larger groups.In the face-to-face context, consider the example of people who live on a few blocks in a city who would like recognition as a distinct neighborhood, perhaps marked by respect from city hall, possibly leading to better services or the recognition of shared needs.The probability of success depends upon members acting to attain this collective benefit, but the residents who refuse to take part, especially those on the fringes, can find themselves not really considered part of the neighborhood.
More often, we would expect to see instrumental relational goods arise in groups where the interaction is more indirect and the membership more subjective.Thus a woman might contribute to the National Organization for Women to solidify her identification as a feminist and at the same time bolster the recognition of N.O.W. and feminist concerns.For another example, when the pro-Russian separatists in the eastern Ukraine acquired some territory, they enhanced their collective identity as separatists (rather than just people with guns).Thus the instrumental relational goods enhance the value of the identity.And an action can produce both types of relational goods.The participant in the Stonewall Riots (June 27-29, 1969, New York City) gets both an instrumental relational good from the increased visibility of the LGBT rights movement and also gets the consumption relational good of being closer to and more accepted by the people with whom he took action.Probably most instrumental goods do have some consumption component, but this component decreases the more impersonal the actions, the more one moves along the continuum from direct to the far end of indirect.

THEORY: HOW RELATIONAL GOODS SOLVE COLLECTIVE ACTION PROBLEMS
An interesting tradeoff arises between size and scrutiny when relational goods motivate political action.Since direct goods arise when people interact face-toface, they will tend to apply to small groups of people.Powerful though they may be as motivators, with the substantial opportunities for scrutiny in these small groups, the result will still be actions by only a few people.Olson (1965) noted that large groups will face relatively high costs when attempting to organize for collective DECYZJE no 24/2015 DOI: 10.7206/DEC.1733-0092.66 action while small groups will face relatively low costs.In the absence of selective incentives, the incentive for group action diminishes as the group size increases.For political change on a state or national level, one needs numbers, and that will almost surely involve indirect relational goods from identification with a more expansive collectivity -environmentalists, promoters of democracy, upholders of the nation.When the benefits include relational goods, an individual's benefits may increase as more people are involved, so that larger groups are no less likely, and perhaps even more likely, than small ones to act in their common interest.This is in addition to any effect from a larger group being more likely than a smaller one to have a core which spurs action (Oppenheimer 2012).The larger group may, often, provide a larger relational benefit.
When relational goods are included in the utility functions of people deciding whether or not to act, the collective action problem becomes transformed.The relational goods broaden the conditions under which people will act to contribute to a collective good, notably including participating in politics.The process differs by type of relational good.The direct consumption relational goods transform many situations into an assurance game.The indirect and especially the investment relational goods provide opportunities for leaders to increase action.In addition, they can transform the collective action problem to make contributing to the collective good more likely, not less, when the group is larger.Game theory provides the framework for analyzing direct relational goods, since by assumption people are interacting directly with each other.Decision analysis provides the framework for analysis of the indirect goods, since the relevant groups are too large for strategic interaction.
Although the focus thus far has been on the contribution relational goods make to producing collective action, they can also reduce action.If the people with whom one relates instead consider action to have negative value or affirmatively value inaction, then the effect will be the opposite; relational goods will reduce political participation.Thus, (some) people must evaluate action nonnegatively for relational goods to increase participation.If acceptance by someone else or by a group entails refraining from activity, then adding relational goods to the utility function could make someone less likely to participate.Some groups have indeed had norms against certain types of activity.Prior to the latter part of the twentieth century, fundamentalist Christian groups in the United States encouraged their adherents to refrain from politics, and they in fact had voter turnout rates well below those of socio-economically comparable members of the population. 3Friends may encourage each other to avoid contributing to public goods.We thus need also to take account of the circumstances under which DECYZJE no 24/2015 DOI: 10.7206/DEC.1733-0092.66 relational goods can discourage action.Second, for relational goods to lead to more participation, the relational good must be tied to the participatory action, either inherently (received in the course of taking the action) or consequently (increased as a result of taking the action).Clearly, many relational goods have nothing to do with political action, such as the earlier example of watching a sunset with a friend.But just as clearly, they are not infrequently connected to political action.
The following notation will be used.We have individual decision makers 1, 2, … n.Each evaluates the option of acting or not acting -for example, voting or abstainingin terms of the usual consumption goods (such as fulfilling a general sense of citizen duty) and instrumental benefits (such as the utility from the outcome produced by one candidate winning rather than the other).The marginal expected benefit of action instead of abstention for person i is constant and given by Ba i .This includes all the non-relational benefits (including consumption terms and appropriately discounted instrumental terms) and subtracts the costs (such as time and money).The relational goods will be notated as R i for person i; if they vary according to action or inaction they will be notated as Ra i or Rn i .We assume that both Ra i and Rn i are nonnegative (0 implies each player is indifferent to the other person) and that Ra i is obtained when both individuals act and Rn i when both do not act.We restrict the model to the behavior of people who do value the relational good of being like each other or are at worst indifferent.Thus Ra i  0 and Rn i  0. However, the value of the relational good may differ if they act (for example if they join each other at a protest) than if they abstain (for example if they share the knowledge that they both stayed away from a protest).We make the further mildly restrictive normalization assumption, for ease of exposition, that we can choose units for the terms so that the absence of relational goods and the non-relational outcome of no action are both zero points, allowing us to drop the "a" subscript for the B term.In this model, the total payoff to each player is an additive function of the net non-relational benefits and the relational benefits.
First consider the case of direct consumption relational goods, with two individuals.For simplicity we start with the symmetric case, where both players have the same payoffs from each action, and thus we can drop the "i" subscript for now.The table below summarizes the payoffs to the players of acting or not acting.The B terms could easily be positive as even if costs exceed the individual's utility from the discounted public good benefits, the B terms also include the nondiscounted consumption benefits.This game has a dominant solution under the following conditions."Act" dominates if B > -Ra and B > Rn; by the nonnegativity assumption, the first condition necessarily holds if the second one is met."Don't Act" dominates if B < -Ra and B < Rn; now the second condition necessarily holds if the first one is met.Thus inaction dominates if B + Ra < 0, that is if the costs are so high that the net nonrelational benefits of action are less than the relational benefit of joint action.Action dominates if the net nonrelational benefit of action outweighs the relational benefit of inaction.
There is no dominant solution to this symmetric game if -Ra < B < Rn.These players are in a coordination game.If B is 0, that is if the nonrelational net benefit of action equals its cost, then they are better off with inaction if Rn > Ra and with action if Ra > Rn.Positive values of B will make action preferable even with lower values of Ra while negative values of B will make inaction preferable even with higher values of Ra.Since by assumption the players are in direct contact when there are direct relational goods, the conditions for coordination will be present, but whether they end up acting or abstaining will depend upon these values.
In asymmetric cases, adding relational goods to the payoffs leads to more complex conclusions.We reintroduce the "i" subscript in Table 2 to allow for each player to place different values on the relational goods and to have different net nonrelational benefits from action.The game has the structure of Battle of the Sexes (Luce and Raiffa, 1957;Rapoport 1966chapter 8, Shubik 1984) if the B i are zero, Ra 1 > Rn 1 and Ra 2 < Rn 2 .Both players prefer making the same choice as the other to making different choices, but they disagree as to whether action or inaction (or, in the Battle of the Sexes story, boxing or opera) is preferable.Since these are direct relational goods the players are in contact with each other and thus we assume they can communicate.Then they can not only coordinate with each other but can also negotiate and offer side payments from their own payoffs to induce the other player to take a preferred action.In general the person who places a higher value on the relational good will be the one to provide a larger side payment while the one who cares less about similarity is in a stronger bargaining position.Unlike in a simple assurance game (Oppenheimer 2012) where players are equally invested in the outcome, in this case the person who cares more about the other, that is the one for whom Ra and Rn are larger, is more likely to shift.
Consider the extreme asymmetric case where Player 2 places no value on either relational good, Ra 2 = Rn 2 = 0, and B 2 is negative.For player 2, "Don't act" becomes dominant and, acting independently, player 2 always prefers inaction.If player 1 has information about the payoffs for player 2, she can achieve Rn by also not acting.But if B 1 plus Ra 1 is larger than Rn 1 by more than |B 2 |, B 1 + Ra 1 > Rn 1 + |B 2 |, player 1 can induce player 2 to act by offering a sidepayment greater than |B 2 |.Conversely, if Ba 2 is positive for player 2 but B 1 + Ra 1 + B 2 < Rn 1 for player 1, player 1 will offer a sidepayment for inaction.Whether the players end up both acting or both not acting will depend upon both the relative values of the Ra and Rn terms and upon how these payoffs compare with the magnitude of the B i terms.Note that under either assumption about B 2 the key to player 2's power and negotiating position is that player 1 places more value on the relational goods.Of course, explaining inactivity is easy.Relational goods suggest more joint collective action than one might otherwise observe when some people's nonrelational preferences would lead them to be active, and they bring others along.This is more likely when people place more value on being alike through shared action than on being alike in shared inactivity.Thus, there is some role for elite mobilization, by enhancing coordination and side payments or by manipulating perceptions of the other participants' values.As the number of people increases, any structure that facilitates communication and negotiation can push the outcome towards activity (or inactivity).Also, if direct communication diminishes, for example as the number of people increases, then the perception of the payoffs of other players becomes even more open to manipulation by other actors, such as leaders.However, as long as direct relational goods involve individuals who are in direct contact, the room for manipulation is constrained.
Indirect relational goods involve interaction at a distance where others are not aware of a player's choices and do not have specific strategic contact, and thus these are better modeled via a decision analysis rather than via a game.The simple decision situation for instrumental indirect relational goods is summarized in Table 3. Recall that by definition the person must act in order to partake in the group's identity.For consistency with common decision theory notation, we explicitly use "c" to represent the cost of acting.R is the value of the instrumental relational good from achieving recognition.We assume all the other nonrelational costs and benefits of action, appropriately discounted, are captured in B. The probability that enough other people in the group act for the group to gain visibility is q.

Table 3
Others in the group -treated as "nature" Enough others act for recognition Too few act for recognition Probability q Probability 1 -q Person 1 Act R -c + B -c + B Don't Act 0 0 Then the individual will act as long as qR + B -c > 0. Note that q here is quite different from "p" in the paradox of voting.It is not the probability of being pivotal; it is instead the probability of the group's success, and thus may indeed be nontrivial.If B is small, or even negative, and if c is large, then a larger qR will be needed to produce action.
Decisions to act based on these indirect instrumental goods can easily be affected by the actions of leaders, and thus these instrumental relational goods readily serve as a basis for mobilization.Leaders are especially important in affecting the provision, and therefore the effects, of instrumental relational goods.Leaders can shift whether instrumental relational goods lead to action by shifting people's perceptions of whether or not other people will act.They can influence whether one believes that acting is expected in order to be perceived as a member in good standing of the group.This precondition might be more persuasive where political lines are defined communally, so that group membership entails multiple benefits and exclusion carries multiple costs.In addition, leaders can shift people's perceptions of the probability of an action being successful, the q term.They may suggest that fewer participants are needed for success and that many will be forthcoming.The model thus implies the opposite of free-riding, in that persuading people that "everyone else will do it" increases action instead of reducing it.Leaders can also mobilize by affecting perception of "R," the instrumental good obtained by acting, that is the value which the group derives from its collective recognition.These considerations suggest that R might be especially high where the lines of political cleavage are still being contested and groups are vying for political legitimacy.
Indirect consumption relational goods also present multiple opportunities for mobilization by leaders.The situation is somewhat more complex than with indirect instrumental relational goods, because consumption relational goods can be obtained by behaving "like" the other through inaction, an option our model excluded in the instrumental case.The decision maker again starts by looking at her own nonrelational benefits and costs, appropriately discounted.The relational goods component of the decision adds three major components to her estimates: first, the probability that the nonrelational benefits and costs of relevant others will incline them to act, or not; second, her estimate to the value of the relational consumption goods to these others; and, third, her own valuation of relational consumption goods.The more the decision maker values being like others, and the lower her estimate of the probability that they care about being like herself, the more her choice will lean towards the expected choice of others, and that in turn depends upon her estimate of the probability that their nonrelational benefits exceed their costs (or the reverse).The decision maker could be swayed towards action by a leader who convincingly conveyed that "everyone else" had a large net benefit from action.More generally, since so much of this decision depends upon the individual's perception of other people's costs and benefits, any information provided by another party can affect the decision.
As Olson (1965) noted, and many others have repeated since, groups in which people can scrutinize each other's actions can more readily enforce contributions to a collective good.Partly for this reason, and partly because of some additional less plausible impediments4 , Olson (1965) said that large groups will face relatively high costs when attempting to organize for collective action while small groups will face relatively low costs.Olson's assertion that larger groups will find it more difficult to provide collective goods has attracted substantial criticism.(See, for example, Sandler (1992, p. 10) for a discussion of the conceptual problems in this claim and for a review of the literature.)Nonetheless, when relational goods motivate political action, scrutiny matters and thus larger groups might indeed be at a disadvantage.But an interesting tradeoff between size and scrutiny arises leading to the opposite conclusion.When the benefits include relational goods, an individual's benefits may increase as more people are involved, so that under certain conditions larger groups are no less likely, and perhaps even more likely, than small ones to act in their common interest.Since most political change requires participation by large groups, this result has powerful implications.
Note that the relational good argument is not simply equivalent to a threshold model (Granovetter 1978).In a threshold model, people become more likely to act as others around them join some action, presumably because they feel safer, or perceive the action as more normatively appropriate, or for some other unspecified reason that increases with numbers.Each person has some different number of other participants that is "enough" for him or her to join in.Unlike the relational good argument, however, all that matters is the number of other participants.With relational goods, however, the identity of the other participants is critically important.A handful of other participants might be enough to induce action if they are the right people, while a crowded square of people with whom one does not identify would not suffice.

EVIDENCE
Some empirical observations support the claim that relational goods exist and can affect the provision of public goods through collective action, although the evidence is indirect.Various studies use the concept "social pressure" to explain behavior, but in many of these cases the phenomenon is better understood in terms of persons valuing and pursuing relational goods.The field experiments which show that shaming and publicity increase voter turnout provide one piece of evidence.The relationship between social media use and participation in Arab Spring protests provides another piece.Examination of the role of attachment to the community and perceived norms of behavior in influencing participation in an ethnic enclave provides another piece.And finally, although the logical link is longer, some support comes from the positive correlation between subjective political representation and political participation.In all of these cases, other interpretations of the phenomena are possible and may also, simultaneously, be valid.However, relational goods provide a parsimonious explanation of these varied observations, and thus the evidence supports their plausibility and conceptual usefulness.
Voter turnout presents not just a puzzle for collective action theory but a topic of widespread interest to politicians and among political scientists.Get out the vote (GOTV) efforts try various approaches to increase activity, such as phone calls, different types of paper and electronic mail, or personal visits.Gerber and Green (2000) devised a field experiment methodology to evaluate the effectiveness of different techniques.A flood of studies has followed examining different mobilization techniques among different populations in different contexts.One consistent overall finding is that techniques are more effective the more closely they simulate a personal request (Green and Gerber 2008;Green, McGrath, and Aronow 2013).Personal recruitment by a known person more generally increases the chances of acting (Verba, Schlozman, Brady 1995).The literature does not explain why personal requests are more effective.I argue that requests are more effective when they evoke relational goods, and relational goods are more likely to be relevant when people interact directly with someone they know or with whom they identify.
For a better test of whether relational goods are operative, one would like to be able to categorize the personal contacts by whether or not the individual cares about the contactor and his or her good opinion.Short of that, a bit more clarity comes by considering the subset of field experiments which involve publicizing a person's actions to presumably relevant others.(See Uhlaner 2014 for a summary of these results.)While these experiments clearly suggest that relational goods operate as hypothesized, they fall short of a full test as there is not explicit measure of whether or not an individual cares about the opinion of others.The relational goods discussion assumes the effects are much stronger when the targeted individual cares about (and thus presumably acquires relational goods by interacting with) the others who are informed than when they do not.
One study did indeed measure this key aspect.Le (2013Le ( , 2014) ) examined political participation in a politically active ethnic enclave.She asked people whether they felt part of the group but also asked two additional questions important for assessing the role of relational goods.These questions were whether the person cared about being thought of as a member of the group and whether the person thought that the group had a norm of political participation.She asked the latter, as well as whether or not they participated, for several different types of both electoral and nonelectoral activities.Le found that indeed participation was higher among people who felt like part of the group, thought the group identity was important, and also thought that identity carried with it a norm of participation.The effect was weakest for actions done in private and strongest for those done visibly to others.Relational goods thus provide a parsimonious theoretical basis for her results.
Increasing scholarly attention to the effects of the digital sphere on participation has also produced some indirect empirical support for the importance of relational goods in pushing payoffs towards action.Many authors have argued that social media and related digital technologies will increase collective action by greatly decreasing the costs of information and the costs of organizing.Some also point out that these reduce the costs of action and argue about what it takes for a digital action to count as political participation.This has been formulated as the question: "Is clicking the "like" button political participation?" (van Deth 2014, Theocarcis 2015).While these points surely have merit, they do not encompass the full import of the digital space.If they did, then mere access to the digital sphere would be enough to increase participation.But that is not the case.In a study of activity within three organizations 5 , focused on the role of new digital technologies, Bimber et al. (2012) found that persons who generally had extensive access to and use of digital media were no more likely (or only trivially more likely) than other members to participate in the organization.Instead, participation depended more upon feeling personally connected to the organization, identifying with the organization's goals, finding social value in involvement with the organization.In other words, people who received relational goods from their organization activities were the ones more likely to participate.

5
The American Legion, MoveOn, and the AARP (American Association of Retired Persons).MoveOn began as a solely digital organization but has branched into the physical world.The American Legion began with a chapter structure which still exists.AARP was a well-established classic interest group.DECYZJE no 24/2015 DOI: 10.7206/DEC.1733-0092.66 A second test may be even more telling.Many commenters have credited social media for the Arab Spring protests of 2010-2011 but have, at best, offered an information story to explain why the digital media mattered.In that case, again, mere access to social media should distinguish participants from non-participants.However, Arab Barometer data collected a few months after the protests show that social media use, taken alone, has only a weak relationship to whether or not a person protested, and even less after accounting for age (Uhlaner and Niayesh 2015).Persons whose friends participated are indeed more likely than other people to protest.But persons who both use social media and whose friends protested were substantially more likely to protest themselves than people for whom only one of those conditions was true.
Finally, there is some very indirect evidence.If relational goods are an important part of the toolkit used by leaders to mobilize action, then we would expect that mobilization will be more effective among persons whose connection to a leader is such that relational goods are available.I suggest that persons who feel that there is someone or some group that looks out for their interests are more likely to be in such a relationship.If this argument is correct, then we should see that, taking account of everything else, these people participate more.The 1989 Latino National Political Survey asked whether people thought someone looked out for their interests and also included extensive questions on political participation.The survey also had items on the other variables which are associated with participation.Controlling for those, indeed those people who felt connected to a leader were consistently more likely to be participatory, especially in nonelectoral activities, namely ones where they tend to be working more directly with other people (Uhlaner, 2002).The positive relationship between descriptive representation (for example based on race, ethnicity, and gender) and voter turnout (Uhlaner andScola 2015, Uhlaner andLe 2015) is also consistent with a role for relational goods in increasing activity.

IMPLICATIONS/CONCLUSION
Olson's Logic of Collective Action (1965) vividly provided the essential message that collective action could not be taken for granted.People could not be expected to invest work in a collective endeavor just because they had interests in common, even in circumstances where joint efforts could clearly achieve their ends.Unless selective incentives were on offer, individuals could be expected to free-ride, except in some DECYZJE no 24/2015 DOI: 10.7206/DEC.1733-0092.66 limited circumstances (such as a small group).This argument by Olson provided a useful corrective to assumptions that sociologists and political scientists were making about political behavior.The first wave of survey researchers had been shocked to discover that people were generally uninformed.Early social movement theorists assumed that the joint interests of workers would be enough for them to join a protest.However, even in the 1965 book, Olson himself recognized that some collective action does take place and devoted much his argument to discussing conditions under which it might appear.Subsequent scholars have taken up this task of reconciling the logic of free-riding with the empirical fact of collective behavior.
This essay has argued that relational goods increase the pathways to collective action or, in some cases, inaction.Recognizing that people value relational goods and considering them in people's payoffs provides additional understanding of the processes that can encourage collective action.Notably, relational goods provide a tool for looking inside the black box of leader mobilization of activity.When people value being like others whom they value, leaders can work to change perceptions of the valued group and of its actions.Paradoxically, in the case of direct relational goods among people in close contact, the person who values interaction the least can drive the collective outcome.If the person who values the relational good least does value political action, for example, then he or she can induce activity among friends who are indifferent to politics but who want to maintain the friendship.
The explosion of social media and digital interaction suggests an area for future work.This sphere multiplies the opportunities for people to connect with each other horizontally without hierarchical leadership structures.It also expands the opportunities to connect with other people and to identify with groups of them.To the extent that the digital sphere simply enhances communication among people who are already connected, it adds technology but probably does not change content.And some of the interactions among people who do not otherwise communicate may be too thin to affect non-digital behavior.However, some aspects of the digital sphere may indeed lead to the creation and propagation of additional forms of relational goods and corresponding additional opportunities for affecting collective action.
The key theoretical point added by relational goods is that behavior occurs in a social context, and that social context can alter the behavior of individual decision makers.Ignoring this context leaves us with a too-truncated view of behavior.Relational goods as a construct provide a systematic way to incorporate this social context.