Lobbying influence: Meaning, measurement and missing

Executive Summary

All too often, research on the influence of interest organizations in democratic politics produces null findings. What are we to make of these results? In part, the answer may lie in our conception of influence – what it is and what might constitute evidence for it. But even when more complete conceptions of influence are considered in better research designs, null results will still occur. They merit explanation. To address these issues, I will first try to provide a broader conception of influence and its many possible meanings by exploring the older theoretical literature on urban power from the 1950s and 1960s, considering along the way what the different interpretations might tell us about lobbying. And second, I will develop a catalog of null hypotheses and discuss how these bear on interpreting the many null findings in influence research. Finally, I discuss the implications of this analysis for the future of influence research.

Lobbying Influence: Meaning, Measurement and Missing

Along with studies on the mobilization, studies of lobbying influence remain at the heart of research on the role of organized interests in democratic politics. But there is a deep paradox in our attention to influence. Simply put, we all look for it, but rarely find evidence of it. Like the perhaps now successful hunt for the Higgs Boson, we have spent a great deal of time looking for something that is very, very elusive, yet deeply central to our interest in the topic of organized interests. All too often, studies of the influence of organized interests generate only mixed or weak results. Indeed, many who enter the lobbying supermarket to purchase policy are disappointed.

Consider some of the key research findings on lobbying of the last two decades. Mark Smith’s (2000) book, American Business and Political Power, found that when business interests are united in actively supporting a policy proposal the likelihood of Congress acceding to their wishes are markedly diminished. Kollman’s (1998) book, Outside Lobbying, highlighted the public opinion context behind the selection of lobbying tactics, finding that direct lobbying in the face of public opinion opposition has little effect. Page et al’s (1987) analysis of business lobbying via media advocacy found that it more often than not had the opposite of the intended effect. Gray and Lowery and their students found that the number and diversity of interests lobbying state legislatures have only a marginal influence on overall policy liberalism (Gray et al, 2004) or the adoption of health care policies (Gray et al, 2007a; 2007b). Further, the density and diversity of interest systems are far more determined by the size and diversity of legislative agendas than the reverse (Gray et al, 2005). And Gerber’s (1999) analysis of referendum voting in the American states, in The Populist Paradox, found that massive infusions of cash into state referendum campaigns by business interests promoting policies favorable to them almost always fail.

None of these studies of specific issues is unique. The empirical literature as a whole remains extremely mixed in terms of finding evidence that organized interests influence policy. Indeed, three major surveys of the literature – by Richard Smith (1995), Baumgartner and Leech (1998) and Burstein and Linton (2002) – reached remarkably similar conclusions. These were best summarized by Baumgartner and Leech’s (1998, p. 134) statement that ‘the unavoidable conclusion is that PACs and direct lobbying sometimes strongly influence Congressional voting, sometimes have marginal influence and sometimes fail to exert influence’. Even more telling, in the latest major study based on a sample of issues lobbyists were working on, Baumgartner et al (2009) reported in Lobbying and Policy Change that most fail to even get their issue on the policy agenda. Their findings are broadly consistent with the earlier findings of Heinz et al’s (1993) book, The Hollow Core, in which they described the world of lobbying as lacking any of the certainty of a supermarket with its well-defined roles, goals and prices. Rather, the lobbying environment is one governed by uncertainty in goals, means and, especially, the relationships between them.¹

What are we to make of these findings? In part, the answer may lie in our conception of influence – what it is and what might constitute evidence of it. Indeed, it is not obvious given current approaches to studying influence that we can recognize it in all of its complexities, perhaps sometimes mistaking evidence of influence for something very different while ignoring deeper patterns of influence when our surface indicators fail to detect it. More telling, perhaps, we have not been very attentive to what these null findings might mean in a political and policy sense. Indeed, the works cited above typically offer little theoretical guidance about how to interpret their weak findings when they might suggest that there are any number of substantively interesting null hypotheses.

In this article, I address these issues by doing two things. First, I provide a broader conception of influence and its many possible meanings by exploring the older theoretical literature on urban power from the 1950s and 1960s, considering what it might tell us about lobbying. Second, I develop a catalog of null hypotheses and discuss how these bear on interpreting the many null findings in the influence literature. None of these is original; all are implicit and occasionally explicit in the literature. But they are rarely considered together as alternatives. And although several null hypotheses overlap to some degree, they provide somewhat different substantive implications for how we might think about the weak findings in influence studies. Indeed, they may tell us as much or more about the nature of democratic politics than any clear and simple evidence of influence as we usually conceptualize and measure it.

The Nature of Lobbying Influence

The first step toward understanding influence must, of course, be defining what it is. The ‘of course’ element of the prior sentence is, however, all too often simply passed over in many studies of influence. In trying to define influence and how it might be observed, then, it might be useful to look at another older literature – the urban power literature – that struggled mightily with similar questions about defining and observing the related concept of power.² Herbert Simon (1953, p. 501) (also see Polsby, 1963, p. 3; Dahl, 1982, p. 21), in fact, treated these two foundational concepts as essentially synonymous in his seminal work on the nature and character of political power.³

The Meaning and Measurement of Influence

What is influence and how might it be observed? Urban politics scholars sharply divided in trying to answer these questions along the lines first suggested by Herbert Simon’s (1953) ‘Notes on the Observation and Measurement of Political Power’. He distinguished between the exercise of power – when A induces, forces or compels B to do something that B would not otherwise do – and the bases of power or the resources that allow for its exercise. This distinction was followed by most subsequent scholars (Clark, 1968, p. 46), although we will see later that Bachrach and Baratz (1962) later added yet another wrinkle on the nature of power that further divided students of urban politics.

Reputations and power bases

To sociologists following the tradition pioneered by Floyd Hunter (1953) in Community Power Structure, assessing power entailed asking informants who held it in a given city. This reputational approach almost always generated an answer to the question. This was not surprising; as the pluralists often noted, the very act of asking who has power assumes that someone does. But the foundation of such reputational specifications of power ultimately rested on what Simon called the bases of power – the resources that were presumed to translate into political influence (Mills, 1956). If you can identify who has the resources to exercise influence then it was commonly assumed that they do, in fact, exercise influence.⁴ It was hardly surprising, then, that when asked who had power, respondents almost always listed those with resources useful in exercising power.

Such attention to the bases of influence and reputational measures of power in the literature on lobbying are more common than we might first assume. While simple reputational measures of the influence of interest organizations are relatively rare, Hrebenar and Thomas built an extensive body of work on the influence of interest organizations in the American states (summarized in Thomas and Hrebenar, 2004) by asking informants who had a reputation for power. Even research focused on decisions – our next approach – often echoes this reputational approach by asking who was influential in their resolution. Again, asking this question makes a prior assumption that someone was influential. But it is even more common in the literature to eschew the intermediary step of asking who was influential and going directly to citing bases of power. Thus, many scholars, such as Schlozman (1984; Schlozman and Tierney 1986), rest their claims of business interest influence largely on the bases of power.

But do resources readily translate into influence? Despite efforts by Schlozman et al (2012) to trace the pathways by which resource bases translate into influence, there remains a very large gap between resource-based, reputational measures of influence and actual decisions. As noted by Baldwin (1971) in his theoretical comparison of power and money in, respectively, political science and economics, the counting, weighing, storing, saving, expending and trading of the two resources is so very different that it is not clear that one can be readily or easily translated into the other.⁵

Pluralists and the exercise of influence

Given these problems, most political scientists followed pluralist Robert Dahl’s (1961; also see: Polsby, 1963; Banfield, 1961) practice in Who Governs? of defining and measuring power via attention to Simon’s other approach: influence exercised on specific decisions. This entailed determining who was for and against specific proposals on the issues, and then counting up the winners and losers so as to assess whether A was able to make B do something that he or she would not otherwise do. A generation of research effort followed in which it was found that influence in urban settings was quite specific to the issue at hand, highly contingent on the context in which a decision was made, and only loosely related to the bases of power that were the focus of scholars following in the footsteps of Hunter. These findings, of course, accord well with more recent neopluralist interpretations of the contingency of lobbying influence (McFarland, 2004; Lowery and Gray, 2004).

It was, of course, not quite so simple as suggested above. As Banfield (1961, p. 8) noted, ‘The workings of influence are hard to see and easy to imagine’. One obvious problem was determining the counterfactual. Did B really not want to do what A preferred? If I am hungry and a colleague asks me to go to lunch, my colleague’s influence on me is only of a very trivial sort. In research on lobbying, we all too rarely specify what would have happened in its absence. To plausibly do so, we must either compare similar decisions in which a similar lobbying tactic or strategy was applied or identify a prior position of the target of influence.⁶ With respect to the first, most studies of lobbying focus on a single decision (Smith, 1995; Baumgartner and Leech, 1998),⁷ which can tell us little about what might have happened in the absence of lobbying.⁸ One of the most important exceptions in this regard is Wright’s (2004) analysis of political action committee campaign contributions on all of the explicit decisions made on tobacco policy in the US Congress, more validly identifying across a similar set of issues when particular contributions mattered.⁹ The alternative approach of identifying a clear prior position is best represented by Yackee’s (2006) work (Yackee and Yackee, 2006) on US regulatory policy and Klüver’s forthcoming book Lobbying in the European Union, which assesses lobbying influence on the European Commission. In both cases, they measured how initial proposals changed in response to the comments or consultations, respectively, by organized interests. In both, the initial proposal provided a strong counterfactual from which to assess lobbying influence. Unfortunately, such counterfactuals are not readily available for most decisions.¹⁰

A second complication that emerged in the urban politics literature concerned the many ways in which influence is exercised. Indeed, Banfield (1961, p. 4) identified multiple pathways:

• (a) influence that rests upon a sense of obligation (authority, respect); (b) influence that depends on the wish of the influencee to gratify the influencer (friendship, benevolence); (c) influence that works by improving the logic or the information of the influencee (rational persuasion); (d) influence that works by changing the influencee’s perception of the behavior alternatives open to him or his evaluation of them, and which does so otherwise than by rational persuasion (for example, selling, suggestion, fraud and deception); and (e) influence that works by changing the behavior alternatives objectively open to the influencee, thus either absolutely precluding him from adopting an alternative unacceptable to the influencer (coercion) or inducing him to select as his preferred (or least objectionable) alternative the one chosen for him by the influencer (positive or negative inducement).

It seems fair to say that the lobbying literature examines only a limited portion of this menu, with heavy emphasis on the last (inducement or coercion), although more attention recently (Burstein and Linton, 2002; Kersh, 2002) has returned to Bauer et al (1963) and Milbrath’s (1963) emphasis on the third – rational persuasion. But all are possible, including friendship, as seen in the Jack Abramoff lobbying scandal in the United States (Schmidt, 2004). More to the point, such multiple pathways may obscure our ability to observe influence. For example, the strength of the Yackee (2006) and Klüver (forthcoming) approach to measuring influence rests on their assumption that the initial proposals represented sincerely held preferences of the regulatory agencies or the European Commission. However, if prior lobbying operates through rational persuasion then the initial proposals could accord more closely with those of the interest organization, and there would then be nothing to observe using their methodology.

Even if these hurdles are passed, there remain two more serious barriers to assessing influence, both raised by Simon (1953) and neither well addressed in either the urban politics or the lobbying literature. The first is symmetrical feedback. In assessing whether A induces B to do something that he or she would not otherwise do, we are assuming that influence operates in a single direction or that B does not influence A, again through all of the mechanisms outlined by Banfield (1961, p. 4). But this seems unlikely in either urban politics or lobbying. For the latter, access is a scarce resource (Hall and Wayman, 1990; Bond et al, 2000) that – given that the policy agenda is not limited to single issues – can be plausibly used to manipulate the preferences of organized interest with respect to lobbying on any one issue (Peterson, 1992; Loomis, 2007). This means, in combination with the previous concern, that neither the policy positions of A nor those of B may represent unadulterated sincere preferences.

The final problem is perhaps the most severe for studying power or influence: Simon’s (1953, pp. 505–506) rule of anticipated reactions. Neither government officials nor interest organizations take positions in a vacuum. Rather, they plausibly take positions on issues in anticipation of reactions from other actors. Thus, actor A might take Position 1 in lobbying actor B when he or she knows that bargaining will ensue in which a decision to adopt Position 1 – the sincere preference of actor A – results. But such anticipated reactions can, allowing for reciprocal influence, work both ways with actor B – the influencee – taking a hard initial stance in anticipation of the lobbying of A. A lot of what rightfully should be labeled influence probably takes place via such shaping of initial bargaining positions by anticipated reactions, but it is essentially invisible to most research on lobbying. This is most readily seen in the otherwise superb work of Yackee (2006) and Klüver (forthcoming). Their findings that comments/consultations moved regulatory agency and Commission proposals from their initial positions assumes that those initial positions represented sincere preferences that were unaffected by anticipations of what organized interests would say about them.

The problem of anticipated reactions is even more severe in most lobbying situations given the institutional complexity of the policy process.¹¹ While some political systems are more institutionally complex than others, a single actor seldom makes policy via a single decision. This institutional complexity is important in that actors at any one stage of the process might well anticipate reactions at later stages. Thus, anticipated reactions do not just involve A and B, but also C, D, E and sometimes F. To use as an example one of the very best studies now available, Klüver’s forthcoming study of consultation on European Commission proposals does not account for how anticipations on the part of both the Commission and the lobbyists consulting on their proposals about what the European Parliament might do when it receives the final proposal. Either party in the consultation may more readily cede preferred positions in the expectation that the Parliament will recoup their losses later. Alternatively, either party may take a harder line in anticipation that the Parliament will tilt even further from their preferences. In effect, Klüver’s work captures a snapshot – although an important one – of influence that itself might be contaminated by expectations about what will happen later.¹²

The second face of power

If all this were not enough to worry about, we must add an even more serious complication from a third school of thought – the second face of power or democratic elitism school (Bachrach, 1967) – on community power. This approach focused on the pluralists’ exclusive attention to decisions on specific issues, but is equally applicable to the study of the influence of organized interests via attention to lobbying associated with specific legislative or regulatory actions. As Bachrach and Bartz (1962, p. 948) noted:

Of course power is exercised when A participates in the making of decisions that affect B. But power is also exercised when A devotes his energies to creating or reinforcing social and political values and institutional practices that limit the scope of the political process to public consideration of only those issues which are comparatively innocuous to A. To the extent that A succeeds in doing this, B is prevented, for all practical purposes, from bringing to the fore any issues that might in their resolutions be seriously detrimental to A’s set of preferences.

In short, looking at tactics and strategies on specific proposals is simply beside the point if the agenda itself is shaped via an even deeper level of influence, largely rigging the game before it has even begun. Indeed, borrowing from Schattschneider (1960), Bachrach and Baratz note that such a mobilization of bias might be so powerful that little or no effort on the part of those benefiting from the status quo is needed to preclude attention to important issues. There will be no actual decisions to observe. This, then, is the dark matter of influence that is vitally important to the distribution of power in a society, but influence that cannot be readily observed.

The literature on the influence of organized interests addresses these issues only obliquely and incompletely. For example, interest organizations play a crucial role in the policy agendas literature, helping to move new issues onto the agenda (Baumgartner and Jones, 1993, pp. 175–192). However, this is perhaps the most weakly developed portion of the agendas literature. More importantly, when bias is well and fully mobilized so as to render some nascent policy issues beyond the pale, there would be no behavior to observe in the sense that it is studied in the agendas literature. In a sense, this powerful second face of power is as invisible to agenda researchers as it is to pluralists studying the fate of issues once they reach the agenda. Indeed, pluralist attention to the exercise of influence thereon is simply not useful for addressing such deeper issues (McFarland, 2004, pp. 153–155). Perhaps, the closest we come to observing such a mobilization of bias comes in research noting the difference in the policy agendas of the public and those of elites (Schlozman et al, 2012; Kimball et al, 2012). But this only hints at the import of this mode of influence. The public too may be influenced by a mobilization of bias so that they do not recognize the legitimacy of potential grievances and they remain deeply latent.

In sum, interpreting, much less measuring, political power or influence is difficult, difficult in a way that the urban politics literature never really solved. We are unlikely to fully solve these problems either. Much of what is substantively meaningful influence will always remain unobservable. Even that which we can observe using superior research designs that sharply isolate competing determinants of policy outcomes – such as in the research of Wright (2004), Yackee (2006), Yackee and Yackee (2006), and Klüver (forthcoming) highlighted here – are to some degree likely to be less than fully anchored by unambiguous counterfactuals, to remain less than fully specified in terms of identifying all of the possible pathways through which influence is expressed, to be contaminated to one degree or another by symmetrical influence across bargaining parties and by anticipated reactions, and to be largely inattentive to the deep, underlying assumptions of the policy system that determine why some issues are raised in the first place and others not.

An Inventory of Null Hypotheses

Given these problems, it is not surprising that many studies of influence have failed to find evidence of it. But even when we define and measure influence in a more inclusive way, what does it mean when we find null results? In many cases, authors of studies do not specify a null hypothesis, instead offering off-hand reasons – often methodological (Burstein and Linton, 2002, pp. 398–400) – that prevented the analysis from finding evidence of the influence that was surely there. In others, authors offer a more substantive account for the null results. But a full range of plausible explanations for the null findings is rarely considered, some whose implications are quite different in political terms from others. To address these problems, a broader range of null hypotheses is developed below.

The behind the veil hypothesis

This first null hypothesis rests on our earlier discussion of the many problems inherent in measuring influence. Given these difficulties, we might fail to find evidence of influence if it is hidden from us via our attention to an often-limited menu of influence modes on actual decisions. This is clearly consistent with the second face of power view of Bachrach and Baratz (1962). In their view, as well as that of Schattschneider (1960), influence so permeates the system that the playing field is severely canted to begin with. If powerful organizations fail to exercise any further influence on the margins of the political and policy discretion that remains, well, this is merely leaving the last few chips on the table to sooth the feelings of those who have lost. After all, to keep up the appearances of democratic play in a system that is anything but, one must not seem too greedy.

However, this behind the veil hypothesis applies equally well to pluralists’ attention to the exercise of influence on actual decisions. In the earlier discussion of measuring influence, much was made of Simon’s (1953) concerns about anticipated reactions and asymmetrical influence. If we look at how the public position of B changed in response to the public lobbying tactics and strategies of A, the lack of any observed change in B’s behavior may well be due to prior shaping of B’s public position in anticipation of A’s response or the shaping of A’s influence activities in response to B’s preferences. This problem, of course, is compounded in a complex institutional setting where C, D and E might also shape the preferences, practices and expectations of both A and B. And it is further compounded if we fail to account more completely for the multiple ways in which overt influence might be expressed (Burstein and Linton, 2002, p. 400). In short, attention to only observable exercises of influence may miss much, thereby producing null results.

The crowded agenda hypothesis

A second null hypothesis highlights differences in the nature of the failure of an lobby organization to secure its policy objectives. That is, a policy proposal may fail in at least two ways: it may substantively lose to the opposition on the proposal or it may simply not secure space on the policy agenda. While these differences should be evident in the locus of research (whether one is studying the agenda setting or the decision stage of the policy process), their meaning is quite different in the sense of who the opponents are. In the latter, the opponents are those with different substantive preferences on a policy. For the former, however, the opponents are the proponents of all other policy proposals and even exogenous events. Simply put, the policy agenda is limited, and simple attention to issues is a very scarce resource (Baumgartner and Jones, 1993; Keim and Baysinger, 1988, p. 175). Given competition for attention, the efforts of all organizations lobbying on all other issues are responsible for how policy attention is allocated. And exogenous events – wars and economic crises, especially – may fundamentally reshape the agenda so that even an issue of the cusp of passage might be summarily hustled off the playing field. Thus, a failure to secure space on the policy agenda may tell us little about the substance of a policy.

But this competition for scarce agenda space also makes it difficult to assign blame for failure. Because competition at the agenda stage includes all possible policies, their proponents and exogenous circumstances, any precise linkage between a lobbyist’s activities and success may be obscured. Such team efforts generate what Alchian and Demsetz (1972) identified as metering problems or an obfuscating of the relationship between the level and quality of inputs into a production process and the quality and level of outputs. As a result, we may not be able to identify sources of influence failure at the agenda stage of the policy process beyond noting ‘difficult circumstances’.

The other guy did it hypothesis

Metering problems, however, do not apply only to the agenda stage of the policy process. We know that lobbying success rests to some considerable degree on matching appropriate strategy and tactics to the circumstances in which a policy is considered (Kollman, 1998; Beyers, 2008; Eising, 2007; Marshall, 2010; Binderkrantz and Kroyer, 2012). A null result in an attempt to exercise influence may be due, therefore, to a failure to adopt an appropriate set of strategies and tactics. Determining this might be easy if we focus on issues where only one organization lobbies against another.

This is rarely the case, however, on the more substantively important issues we typically study.¹³ Debates over the issues studied by Heinz et al (1993), for example, were conducted by rich constellations of interest organizations, all of which were trying to shape the policy outcome, not all using the same tactics and strategies. If blame rests on the failure to adopt the best approach to exercising influence, it may be difficult to identify which organization used the ‘wrong’ strategy. A given organization may have gotten it right, but its influence efforts might have been swamped by those on the same side who got it wrong. The metering problem does not disappear when we consider lobbying coalitions (Hojnacki, 1997; Hula, 1999; Gray and Lowery, 1998). Even with some coordination among their members, it may not be clear which organizations among them failed to pull their own weight in executing the agreed upon strategy and tactics. All those on the winning side can claim credit equally, as do all those on the losing side share blame, although failures might generate considerable finger pointing among former allies.

The two-sided hypothesis

While policy issues might have multiple dimensions and might be framed in any number of ways, by the time it comes to a final decision contests typically devolve to only two sides (Heinz et al, 1993; Baumgartner et al, 2009; Mahoney, 2008). If there is a resolution, one side must win, more or less. This implies that the other side must lose, more or less. Indeed, contrary to the central claim made at the beginning of this analysis about the frequency of null findings, the two-sided character of policy decisions suggests that influence failures and successes must be balanced in an accounting identity. This means that when we account for null results on one side of the issue we are at the same time explaining lobbying success on the other side. But all too often, studies with null results make little effort to look at what the lobbyists on the other side did right. Why did their lobbying tactics and strategies work?

Importantly, however, even if debates typically have only two sides, this does not mean that lobbyists were working actively on both. Indeed, in what should not have been a surprising finding, Baumgartner et al (2009) found that government officials themselves often constituted one side with organized interests arrayed on the other. However, more than lobbyists and government officials influence public policy, and their influence may be more or less direct. As Dahl (1961, p. 59) noted:

…among all of the persons who influence a decision, some do so more directly than others in the sense that they are closer to the stage where concrete alternatives are initiated or vetoed in an explicit and immediate way. Indirect influence might be very great but comparatively difficult to observe and weigh. Yet, to ignore indirect influence in the analysis of the distribution of influence would be to exclude what might well prove to be a highly significant process of control in a pluralist democracy.

Perhaps, the most important of these indirect influences is public opinion. In a restatement of pluralist theory, Denzau and Munger (1986) explained why legislators represent even unorganized interests. Simply put, elected officials depend on their votes. But whether the winning side includes many or few organized interests, and whether the other influences on policy are more direct via the actions of government officials or as indirect as public opinion, perhaps the most plausible explanation of why an interest organization failed to exercise influence is because some other constellation of interests did.

The status quo hypothesis

There is always a status quo, and it is often very strong (Baumgartner et al, 2009, p. 242). This means that even when there is no obvious opposition, as in the two-sided hypothesis, victory in the sense of imposing influence is not assured. This raises its own ambiguities about claimed influence. If the status quo is strong and one lobbies in support of it then success will be signaled by a non-event. Indeed, a lack of any policy change constitutes evidence of success even for those seeking to preserve the status quo who do nothing to defend it. It is obviously difficult to link a person’s effort – known or unknown to the effort, but especially the latter – to the absence of an event.¹⁴

This problem of the power of the status quo is also problematic for lobbying because it means that successes are likely to be lumpy and rare. As Keim and Baysinger (1988, p. 175) have noted in regard to the latter, ‘some political issues are resolved in discreet, all or nothing, decisions’. Further, simply getting on the policy agenda (Baumgartner et al, 2009) is difficult, much less altering the status quo. However, the combination of lumpiness and rarity becomes even more problematic when we note that the payoffs from altering the status quo may be very large indeed.¹⁵ Still, this combination of lumpiness, rarity, uncertainty and large payoffs might help us to understand why organizations continue to lobby even in the face of infrequent returns. That is, given these attributes, lobbying operates as a partial or variable reinforcement schedule (Ferster and Skinner, 1957). It has long been noted that under such schedules rats will continuously press a lever in hopes of a reward of a food pellet and gamblers will stay at slot machines long after it might seem reasonable to do so (National Research Council, 1999, p. 39). These characteristics of some policy goods might, therefore, feed into cognitive biases that make lobbying an addictive behavior, with the lobbyist as the enabling pusher.

The black hole hypothesis

The failure to exercise influence must also be considered over time. That is, whereas many organizations lobby once and then leave the policy process, many others are repeat players (Anderson et al, 2004; Berkhout and Lowery, 2011). For repeat players, a successful exercise of influence at Time 1 could influence their success at Time 2. This is because effective use of influence at Time 1 will shift the status quo. This means that an even further exercise of influence is more difficult at Time 2 for the winning side at Time 1 (Baumgartner et al, 2009). If this were not true, a small imbalance of influence at Time 1 would lead to a runaway reaction where victories are multiplied to the point of extremity. Much as a gravitational black hole is collapsed to a singularity from which not even light can emerge, a small initial imbalance will lead to one set of interests dominating the policy process completely.

This, of course, is inconsistent with what we know from theories of macro-politics (Wlezien, 1995; Erikson et al, 2002) as well as, more generally, from problems of declining marginal returns in many economic and social activities. With regard to macro-politics, much like a thermostat, the policy preferences of the public and the actions of a range of public officials tend to move counter-cyclically so that more conservative policies in one period will shift preferences and policies in the other direction. Further, the preferences and, perhaps more importantly, the efforts of even successful organizations are presumably as subject to the logic of diminishing returns as are the consumers of any good or service. More to the point, in accounting for failures to exercise influence, we need to consider where the status quo is with respect to the broader set of preferences of an interest organization and how earlier efforts to exercise influence fared.

The degree of difficulty hypothesis

This hypothesis provides less an explanation of failure as it does an alternative way of thinking about it. It is also related to the status quo argument in the sense that it suggests that not all status quos are equal. Because policy debates typically devolve to two sides when proposals are decided, we tend to think of influence as a dichotomous variable. Of course, the threshold of having a proposal pass or fail is a meaningful one. But like Olympic diving, in which the quality of a dive is weighted by its degree of difficulty, some lobbying objectives are more challenging than others. And there are also meaningful degrees of failure that may well bear on future efforts to alter the status quo. As Clark (1968, p. 334) noted in the context of organized interests lobbying in urban settings:

A partisan group in a community may start with little chance of an alternative being accepted. By waging a vigorous fight they may reach a point where acceptance or rejection is touch and go. Ultimately, of course, the measure will either pass or fail, but we should not judge this group to have had influence only if it passes. The move from an almost certain failure to a near miss is a mark of their influence. Similarly, a victory by a partisan group in a situation in which they would have won doing nothing is no indication of influence.

This is important because it might matter greatly next time whether an initial failure was by a large or small margin. The point is simply that we should consider more fully both the initial prospects of success and how close to success a lobbying effort came.

It is not always about winning hypothesis

Interest organizations must balance achieving their policy goals with other critical needs, such as mobilization and maintenance. This means that lobbying is often really about something other than influence. Quite simply, interest organizations are first and foremost organizations. Their overriding priority, then, will be surviving in a world of scarce resources, and their actions – like those of all organizations (Pfeffer and Salancik, 1978) – will be primarily focused on securing the resources they need to survive (Grote and Lang, 2003; Lowery, 2007). A failure to exercise influence on a specific issues may, then, represent a victory on a mobilization and/or maintenance task or even influence on another issue entirely. Indeed, if the several stages of the influence production process are linked in complex ways, as research now suggests (Lowery and Gray, 2004), then it is plausible that lobbying is often about some other purpose than simply winning.

Many examples can be cited. When the National Organization for Women lobbied on the highly salient issue of President Bush’s judicial nominations, it almost certainly realized that it had little chance of winning. But maintaining a fierce profile on court nominees proved to be a very effective tool in retaining members and securing new ones. When Christian Right organizations lobbied the US Congress in support of President Bush’s proposal to privatize Social Security, it was likely less due to any deep antipathy to the New Deal program than in exchange for more energetic support by the Bush Administration of proposals to limit abortion and restrict gay rights (Peterson, 1992). And when Common Cause abruptly switched its agenda from government reform to opposing the MX missile, it likely did so not because its leaders expected to win or because its members cared deeply about the MX, but because a few patrons who supplied the lion’s share of the organization’s funds did (Rothenberg, 1992). In short, losing on a specific question on the table may, in fact, be irrelevant if the goal of lobbying was something else entirely.

The selection bias hypothesis

This hypothesis is less about specific lobbying studies than about why the literature as a whole finds so little evidence of influence. Perhaps, we are looking in the wrong place, looking for lost automobile keys under a nearby streetlight rather than near where they were lost simply because the light is better under the streetlight. This account has two distinct elements. First, as Baumgartner and Leech (1998, p. 176) noted from their survey of articles of the literature, ‘the modal type of interest group study … over the postwar period is a cross-sectional comparison of a few groups working on a single issue at one point in time. Such a research approach seems a perfect strategy for producing unexplained variation between studies’. ‘It is a recipe’, they note, ‘for the creation of a contradictory and noncumulative literature’. In effect, the research designs of these studies defined away critical elements of context that determine when influence is effective. While their criticism has led to renewed reliance on large-N studies, small-N studies have not disappeared and remain problematic for studying influence due to their inattention to context.

Second, most influence studies – whether small- or large-N – focus on substantive large and controversial issues on which many organizations are engaged. In the latter, for example, it is common to sample on the basis of media attention to the issue.¹⁶ But these are precisely the issues on which influence might be least expected because public opinion is likely especially constraining (Smith, 2000). In fact, the vast majority of policy proposals attract very little lobbying (Baumgartner and Leech, 2001). Although the stakes involved in these issues may be narrower and less controversial, they are often, nonetheless, very important to those concerned. It could well be that lobbying is far more successful on these kinds of proposals.¹⁷ But we will not know this for sure unless we stop confusing studying lobbying influence with studying lobbying influence on controversial issues.

The milker bill hypothesis

In an extreme form of the reciprocal influence argument discussed earlier, this hypothesis switches the direction of influence, with politicians extorting campaign funds from interest organizations and the lobbyists who represent them by introducing bills that compel them to pay protection money (Mueller and Murrell, 1986; Coughlin et al, 1990; Shugart and Tollison, 1986). This leads to a radically different view of what failure means. Organized interests are viewed – to mix metaphors in perhaps an appalling way – as dairy cows providing the mother’s milk of politics. Politicians introduce bills or encourage agencies to propose new regulations to expropriate rents. The resulting rush to the capitol constitutes an auction that ‘provides valuable information whether regulator action or inaction will be more lucrative to politicians themselves; it helps to identify the likely payers and to set the amounts of compensation to be paid’ (McChesney, 1997, p. 36). From this perspective, a failure of ‘advocacy’ indicates that an organization has paid sufficient bribes to politicians to keep an issue off the agenda. The ‘successful’ opponent would be an accidental beneficiary of a game that is really being played between others.

Although this explanation is unlikely to appeal to most political scientists, it has not yet been clearly rejected in favor of another, more widely accepted model of interest influence, if one that also disagrees with the standard pluralist demand model.¹⁸ In this other model, also arising in economics, interests are seen as so powerful that they fully determine the issues on the policy agenda and how they are dealt with.¹⁹ In effect, these two economics-based models of influence deny any possibility of reciprocal influence, but assign causal agency to a different partner in the bargaining dyad. To date, empirical analysis has not been able to distinguish these two models.²⁰ Thus, the milker bill hypothesis remains a theoretically plausible explanation for some null results with failures indicating the offer of a sufficient bribe on the part of the other side of the issue.

The unprincipled agent hypothesis

This explanation too redraws the usual lines of influence found in the literature by inserting the lobbyist as an independent – and powerful – actor between officials and the organizations he or she represents. To a considerable degree, the literature on lobbying the lobbyist is a superconducting transmitter linking issues of concern to clients to responsible officials with no loss of energy and with perfect fidelity. In an important sense, then, lobbyists have become invisible in the lobbying literature. They are merely servants who advise clients and execute their orders (Kersh, 2000). But agency problems may easily arise in lobbying due to adverse selection such that the lobbyist agent may poorly seek influence on behalf of clients (Stephenson and Jackson, 2009).

Of the several conflicts of this type that have been noted by Lowery and Marchetti (2012), two are important here. The first concerns contract lobbyists with multiple clients. The interests of two clients might conflict at one point in time (Rosenthal, 2001, p. 103). In choosing to represent either client with greater fidelity, he or she will fail the other. Indeed, as in the case of Jack Abramoff and his Indian casino clients (Stone, 2006), the lobbyist may represent neither client’s interest, playing them off against each other to provide benefits to the lobbyist. But such conflict might also arise over time if a contract lobbyist will remain in contact with government officials long after the client of the day has left town.²¹

The second potential conflict giving rise to agency problems is even worse and bears directly on the null hypothesis in influence studies. The same incentives that might lead any type of lobbyist to suggest more services than required (Lowery and Marchetti, 2012) might well lead him or her to eschew ultimate victory for a client. As Rosenthal (2001, p. 102) quoted one lobbyist, ‘You should never kill a bill so bad [sic] that [the opposition] can’t come back next year, or you’ll be out of business’. Such incentives might be especially powerful for in-house lobbyists and representatives of membership organizations. Quite simply, contract lobbyists have more than one client. Lobbyists with but one client might be well inclined to leave some doubt as to the final resolution of an issue or, at a minimum, search for a new one (Sills, 1957) in order to preserve their jobs.

The pluralist hypothesis

Despite the many alternatives reviewed until now, probably the default explanation for null results on the influence of organized interests remains pluralist theory as developed by Robert Dahl (1961) and, especially, David Truman (1951). In their view, explaining why organizations lobby was not problematic. Like-minded individuals naturally come together in response to disturbances in the policy environment. The population of organized interests formed in response to policy disturbances was assumed to validly reflect the distribution of salient interests in society where few are without access to any means of pursuing influence (Dahl 1967, p. 130). But lobbying by these organizations was then viewed as highly constrained in a manner that precludes effective influence in the sense that it is typically used in the literature. Of the several types of ‘influence tactics’ noted by (Banfield, 1961, p. 4) and discussed earlier, emphasis is placed on the more benign – largely providing technical information to elected officials so as to sharpen their linkage of preferences to policies (Bauer et al, 1963; Milbrath, 1963) rather than exercising strong-arm tactics in the sense of making A doing something he or she would really, really prefer not to do. The reason is rooted in the electoral incentives (Mayhew, 1974). Given election-induced attention to constituents, only information that better links elected officials to their voters can sway officials (Burstein and Linton, 2002). The resulting policy outcomes will thus reflect the will of the public, if perhaps weighted by issue salience (Truman, 1951; Dahl, 1961, 1967; Key, 1964; Denzau and Munger, 1986).

Importantly, pluralism emphasizes the very weaknesses of the organized interests formed in response to policy disturbances. Had organized interests been truly influential, there should have been no policy disturbance to begin with in the absence of a natural disaster or some unexpected event. Rather, government activity itself often constitutes a powerful disturbance to which organized interests respond. Indeed, studies of this ‘demand’ function of lobbying in the United States and the European Union have often found that greater government attention to policy issues powerfully draws organized interests into lobbying (Leech et al, 2005; Gray et al, 2005; Messer et al, 2011). However, this attraction does not alarm pluralists. As Salisbury (1990, p. 228) has suggested, ‘the descent on Washington of so many hundreds of associations, institutions and their agents does not mean that these private interests have acquired greater sway or even a more articulate voice in the shaping of national policy. In many ways, the opposite is true’. In his view, we should not expected to find that organized interests are especially influential, even if sometimes useful in adding specificity to policy and better linking voters to politicians. Instead, they are weak, and our weak evidence of influence simply reflects this impotence.

Conclusion

Much of our research in lobbying influence generates null findings (Baumgartner and Leech, 1998). As Burstein and Linton (2002, p. 398) summarized their results²² from a meta-analysis of studies on policy influence:

Virtually all sociologists and political scientists publishing in the top journals hypothesize that … interest groups … influence public policy, and it is safe to assume that they generally expect the impact to be substantial. This hypothesis is not as well supported by the data as we might expect. The impact of political organizations is significantly different from zero, by conventional statistical tests, only about half the time, and important in policy terms (as assessed by the authors) in just over a fifth.

Indeed, this conclusion almost certainly underestimates the number of failures given the usual bias against publishing null findings. This inconsistency in our expectations and findings on such a central issue in research on organized interests merits explanation.

To begin to do so, I first outlined a broader conception of influence by exploring the older literature on urban power from the late 1950s and 1960s. The implication of this discussion is that we need better specified models and stronger research designs. Better specifications are needed to account for reciprocal influence between lobbyists and officials, the multiple pathways of influence, and anticipated responses to these within both bargaining dyads and across multiple institutions. Stronger research designs are also needed if we are to provide strong counterfactuals from which to assess evidence of influence. However, perhaps, the strongest lesson to draw from this portion of the analysis is one of modesty. Influence – in all its forms – is very complex and hard to observe. At best, our research designs capture only brief snapshots – however important – of influence in democratic systems. Moreover, and perhaps more troubling, there is some inherent tension between the use of strong research designs and the scope of our claims about influence. Better, stronger designs in terms of establishing strong counterfactuals and isolating specific modes of influence by their nature will focus our snapshots even more tightly than will weaker designs, thereby restricting the scope of our inferences about influence.

The second part of the essay developed a catalog of null hypotheses implicit in the literature and discussed how these bear on interpreting its null findings. In a narrow, practical sense, these should be useful to scholars for the simple reason that publishing null result is sometimes difficult absent a strong rival explanation. This is unfortunate given the importance of null findings to the traditional Popperian falsificationist model of science (Popper, 1935). From this epistemological perspective, null findings – irrespective of the habits of editors – are especially important given that they are sufficient on their own to demolish the substantive hypothesis of influence we are examining. Indeed, what one scholar has called ‘the dirty little secret’ of the lobbying literature – that lobbying is often not very effective – will, of necessity, remain a dirty little secret so long as null results remain unpublished or plausibly underrepresented in the literature.²³

Still, the falsificationist perspective is not the only game ongoing in our field’s epistemological village. Perhaps, a more realistic model of social science is provided by Ime Lakatos (1970). In his view, no amount of null findings can falsify an hypothesis without offering an alternative explanation that accounts for what the other can explain, can explain the null findings, and can at the same time tell us something new and interesting. From this epistemological point of view, the typical reticence of editors to publish null finding – absent a viable and interesting null hypothesis – is appropriate.

It is not enough, however, to merely cite the null hypotheses discussed here after the fact. They need, instead, to be developed as rival accounts that can stand on their own.²⁴ This requires sorting through them before testing influence models and developing test implications that might help us assess their import. The last is certainly necessary because they differ markedly in terms of how and where they apply to the policy process and in their implications for democratic polities. But the null explanations examined here certainly offer a wealth of rivals to the hypothesis that organized interests influence the selection of public policies. In addition, crucially, while they certainly do not agree with each other in sometimes small and other times very large ways, each offers insights about the nature of democratic politics that are interesting. For this reason, they should be seriously considered as we develop our substantive hypotheses, deduce test implications, develop our research designs and select our measures. And doing so may help us with all those Lakatosian editors when it comes time to publish.