New Perspectives on the Paradox of Participation

(1)

New Perspectives on the Paradox of Participation

A Theoretical Evaluation of Rational Choice Theory as it Applies to Political Participation

Elin Stendahl Master’s thesis

Department of Government, Uppsala University Supervisor: Per Adman

2020-12-28 19789 words

(2)

i Abstract

Theory is vital for our scientific understanding of the social world. Building, developing, and evaluating theory are therefore central practices within the social sciences. This study performs an evaluation of rational choice theory within the field of political participation.

This is a theoretical framework that has had significant problems reconciling theoretical prediction with empirical findings, causing what is called the paradox of participation. For more than sixty years rational choice theorists have tried to develop new formulations of the theory to avoid this paradox. The purpose of this study is to forward this debate by providing a new perspective on rational choice theory that is purely theoretical. Using the evaluative criteria of falsifiability and leverage, the study finds that a theory using both collective and selective incentives, while also allowing the formulation of the theory to change depending on the form of political participation one wants to explain provides the most promising

approach. However, the evaluation does reveal some issues in connection to selective

incentives. A potential alternative solution to the paradox is therefore briefly discussed, yet a more thorough exploration of this venue is left to future research.

Key words: rational choice theory, paradox of participation, theory evaluation, falsifiability, leverage, parsimony

.

(3)

ii

(Bacharach, 1989). Its aim is to enable prediction and generalisation (Riker, 1995) as well as to organise and clearly communicate what has been discovered through empirical observation (Bacharach, 1998). Gunitsky (2019) says that, like a map, a theory will not be an accurate representation of reality. It will exaggerate certain features and omit other elements of the world. He goes on to conclude that “this simplification is not a limitation of the

cartographer’s skill, but a way of focusing on the salient features of the landscape in order to make the map legible and functional” (2019:711). This, then, is what a theory is meant to do for the researcher: simplify an exceedingly complex social world, in order to make it

understandable.

Since theory is such an integral part of research and the scientific understanding of the social world, it is reasonable to set very high standards for it. This not only means that social scientists must be rigorous when building theory, but also that, as theory develops along with the scientific field, it should be thoroughly and purposefully evaluated to ensure that it keeps living up to these standards. Yet, purely evaluative research endeavours are scarce. Much research within social science performs empirical testing of theory, which is an important part of theory evaluation, but it is arguably not the only necessary part.

This thesis will undertake a purely theoretical evaluation of social science theory. More specifically, the theoretical validity in terms of falsifiability and leverage of rational choice theory as it has been applied to the field of political participation will be investigated. The

(5)

2 primary reason behind this choice lies in the fact that the theoretical predictions produced by rational choice theory severely underestimate the level of political participation in democratic societies. This has been a long-acknowledged circumstance, known as the paradox of

participation, ever since it was first discovered by Downs (1957) in the late 1950’s. Various attempts have been made at altering the theory to avoid the paradox, but for more than sixty years, scholars have been unable to reach consensus on how rational choice theory should best be formulated to avail this issue. Furthermore, several of the attempted solutions have been criticised for either deviating too much from the original assumptions on which rational choice theory is built, or for having been constructed in a way that causes the theory to become tautological.

In other words, there is ample reason to take a closer look at rational choice theory as it applies to political participation. The rest of this introductory chapter will therefore be spent defining political participation with the purpose of setting the outer boundaries of the field to which this thesis applies as well as formulating the research aim and questions which the subsequent study will aim to answer. Part two will examine the paradox of participation in more detail and introduce the original calculus of participation which is the formalisation of rational choice theory. Following this, the third chapter will discuss theory evaluation as a method. The concepts of falsifiability and leverage will be introduced and operationalised to set up the analytical framework necessary to perform this study. Part four provides a more detailed presentation of the various theoretical reformulations of rational choice theory that are evaluated in part five. Finally, the sixth chapter will be a discussion about how rational choice can fit into a greater framework of social science theory when explaining political participation and how this could provide another potential avenue to alleviate the problem of the paradox while the seventh chapter will summarise the main conclusions.

1.1. Defining political participation

In this thesis, a number of attempts at explaining political participation in democratic societies using a rational choice framework will be presented and evaluated. The term

‘political participation’ will thus frequently appear on the coming pages. While none of the theoretical approaches discussed in this thesis attempt to explain political participation in general, but rather specific modes of political participation, there is still reason to take some time to define how this concept is to be understood. Which is the general understanding of

(6)

3 political participation that these theories tap into? By determining the external boundaries of the concept, an external boundary is also set for the study.

During the post-war era, political participation was primarily understood to be synonymous with electoral participation (Ekman & Amnå, 2012). This is apparent in much of the early rational choice accounts within the field and it is therefore no coincidence that the paradox of participation was initially only concerned with voting behaviour. In the early 1970’s, this narrow view on democratic participation was criticised (see e.g., Pateman, 1970). A much broader definition of political participation was presented by Verba and Nie (1972), which opened the field to the study of a whole new range of activities such as protest action and party activity. This definition states that “political participation refers to those activities by private citizens that are more or less directly aimed at influencing the selection of

governmental personnel and/or the actions they take” (1972:2). While this definition is undoubtably more extensive than the previous understanding of political participation it is nonetheless still quite restricted. As pointed out by Ekman and Amnå (2012), this definition focuses solely on behaviour that is aimed at political elites and their actions.

Another definition, first formulated by Brady and used by Teorell, Torcal and Montero (2007), goes even further than Verba and Nie (1972). In this definition political participation is understood to include any action which is taken by ordinary citizens and that is directed toward influencing some political outcomes (Teorell, Torcal and Montero, 2007). Since political outcomes in the democratic context, especially in economic terms, are not only determined or influenced by political elites, political participation in this sense can also include such behaviour as protest actions aimed at private corporations or boycotting certain products for political reasons.

Studying this broader range of actions as political participation certainly has merit. Intuitively it seems reasonable that actions taken for political reasons, no matter at whom they are

directed, should be understood as political participation. Furthermore, there is no obvious reason to suspect that actions directed towards private actors should be explained completely differently than actions directed towards political elites, given that the motive is to affect political outcomes. However, determining whether this reasoning is in fact empirically correct is too big a question and too far outside the scope of this thesis to be attempted at this point.

(7)

4 Within the rational choice field, research has primarily been concerned with such

participation as is referred to in Verba and Nie’s (1972) definition, that is, actions directed towards political elites (see Leighley, 1995 for an overview of the field). It is also the case that all the theoretical accounts included in the evaluation focus on actions that are directed towards political elites¹. Therefore, it appears most prudent to apply this, slightly more restricted, definition of political participation in this thesis.

1.2. The aim and research questions

Rational choice theory has been said to have a more theoretical basis than many other

theories used within the political participation field, which have been induced from empirical findings (Blais, Young & Lapp, 2000). However, as the original theory has been revised and reformulated to better fit the empirical world, it is important to investigate whether this has had any adverse effects on this theoretical soundness. For this reason, this thesis will approach rational choice from a different angle than most research in the field has done.

While there are some examples of theory evaluation focused on rational choice theory in political participation research, these have mainly been focused on whether the new

formulations of rational choice theory are compatible with the original theory’s underlying assumptions (see e.g., Whiteley, 1995). This evaluation will take another approach. By disentangling the theoretical components from their empirical performance and analysing them in relation to the criteria of falsifiability and leverage, the hope is that this thesis might help advance the debate on how rational choice theory should be used within the field of political participation.

More specifically, the aim is, first, to determine whether the theoretical approaches remain theoretically sound and, second, to conduct a comparison with the purpose of finding which approach provides the most promising solution to the paradox of participation. In pursuit of these goals the evaluation will be centred around the concepts of falsifiability and leverage.

The former will be used as a measure of theoretical soundness, while the latter serves as a means by which to determine which approach is most helpful in furthering our understanding of the decision-making process that leads people either to participate or not. That is, to

1 Bäck, Teorell and Westholm’s (2011) ‘manifestations’ variable partially falls outside this definition of political participation. It establishes whether the individual “has signed a petition, donated money, taken part in a boycott (e.g., of certain products), worn or displayed a campaign badge or sticker, and/or taken part in a demonstration”

(2011:86). While this can contain actions that are directed toward other actors than political elites the deviation from the definition is deemed to be minor enough to be disregarded for the purposes of this thesis.

(8)

5 borrow Gunitsky’s metaphor, which approach provides the best map of political participation.

To this end the following two research questions have been formulated:

1. Are the revised versions of rational choice theory falsifiable?

2. Which of the revised versions of rational choice theory has the highest level of leverage over the problem of explaining what makes individuals participate politically?

2. The Problem of the Paradox and the Original Calculus

The paradox of voting springs from a discrepancy between theoretical prediction and

empirical observation. A basic assumption of rational choice theory is that actors know their goals and can order them transitively² according to preference. Furthermore, actors are assumed to be purposive, that is, they are expected to act strategically with the purpose of achieving their goals and thereby maximising their utility (MacDonald, 2003; Riker, 1995). If individuals act strategically, they can be expected to consider their options carefully,

weighing different potential courses of action against each other and choose whichever option is most likely to promote their goals at the lowest personal cost. Among the first to apply these assumptions to democracy, and more specifically to the act of voting, was Downs (1957). When considering what would induce a rational individual to vote he concluded that

“the reward a man obtains from voting depends upon (1) how much he values living in a democracy, (2) how much he cares which party wins, (3) how close he thinks the election will be, and (4) how many other citizens he thinks will vote” (1957:270). Based on this and other similar works, Riker and Ordeshook (1968) formulated the calculus of voting, which specifies the different aspects of the decision to vote or abstain and their relationship to each other in the manner which is shown in equation (1).

𝑈 = 𝑝 × 𝐵 − 𝐶 (1)

U is the net utility that the individual receives from going to the polls to place their vote. The multiplicative term of the calculus consists of the probability (p) that the individual’s vote decides the election and the benefit (B) that the individual will receive if the preferred candidate wins (compared to another, less preferred, candidate winning). Finally, C signifies

2 If there are three options – A, B, and C – and an individual prefers A over B and B over C, transitive preferences entail that then the individual must also prefer A over C.

(9)

6 the costs to the individual. Voting takes time, not only the time it takes to go to the polls but also the time it takes to reach a decision on who to vote for³. Maybe the individual needs to read about the candidates’ political programmes or watch debates on TV to decided which candidate to vote for. This does not only require time, but also some cognitive effort from the individual, all of which counts into the negative cost term in the calculus. Only when the calculus yields a positive result is the individual expected to vote. That is, voting only occurs when U > 0, or, put differently, when p × B > C.

Going back to the four motivations for voting proposed by Downs (1957), it is worth noting that the first point, the value accredited to living in a democracy, was not included in the calculus by Riker and Ordeshook (1968). This omission is not explicitly acknowledged by Riker and Ordeshook, and so their reasons for this choice remain unclear. Nonetheless, their specification of the theory has been seminal within the field and so it will be accepted as the original formulation of rational choice theory in this study. The question of what happened to Downs’ first motivation and what potential consequences this has had for rational choice theory will thus be left unanswered at this time.

Now, considering that the probability that one individual’s vote is decisive in an election is vanishingly small so as to be almost zero, the benefits that the individual would get from the desired political outcome would either have to be enormous, or the costs required would have to be almost non-existent in order for the calculus to yield a positive result. Since this can rarely be expected to be the case, the calculus predicts that almost no-one would in fact turn out to vote. Furthermore, there is the additional issue of free riding which exacerbates the problem. Since political outcomes are generally public goods, rather than private rewards, an individual will enjoy the benefits of their preferred candidate winning even if they did not vote themselves. Given the extremely bad odds that their contribution would in fact make a difference, most people are expected to stay at home and save themselves the costs of participating. Yet, despite this, people nonetheless turn out to vote, and often in large numbers too. Hence, the paradox of voting.

While a significant portion of the research using rational choice theory to explain political participation has focused on the act of voting (see, e.g. Aldrich, 1993; Blais, Young & Lapp, 2000; Ferejohn & Fiorina, 1974; Riker & Ordeshook, 1968), others have extended the use of

3 In some contexts, most notably perhaps in the US, there is also the issue of registering to vote which can take considerable amounts of time and effort.

(10)

7 the calculus to other forms of participation as well. By rephrasing the definitions of the

calculus slightly it can be formulated as a calculus of political participation, rather than as only a calculus if voting. In this more general version, the p-term would signify the likelihood that the individuals’ participation (be it through voting, contacting an elected official or joining a demonstration) is decisive for a specific political outcome. Similarly, B denotes the benefit that the individual receives from the preferred political outcome of the participatory action and C marks the costs incurred by participation. Only when the net utility is positive, that is when the benefits outweigh the costs, will a person choose to participate.

The development to extend the use of the calculus beyond the case of voter turnout has thus also extended the reach of the paradox. Just as the calculus of voting predicts very low turnout, the calculus of participation predicts that most people will not chose to participate.

Hence, it becomes a paradox of participation and it is this conundrum with which this thesis is primarily concerned. By including this broader range of research into the analysis the comparison between different theoretical accounts is made somewhat trickier. However, it also allows the inclusion of such insights as have come from broader participation research.

Leaving these contributions outside the analysis would encroach on the potential scope of the answers that would be possible to give to the research question. Also, it would unnecessarily limit the significance of this research for the ongoing debate on rational choice theory within political participation. How to deal with the issue of comparability between theories that explain different forms of political participation will be further discussed in the final section of the next chapter.

3. Evaluative Criteria as a Method

Unlike most other research methods used within political science, there is little established praxis with regard to theory evaluation. Rather than being a result of conscious decision, this is most likely simply the consequence of the scarcity of this type of work within the political science field. However, where theory evaluation is mentioned, it is usually followed by some manner of list containing criteria by which to judge the soundness and validity of a

theoretical system. Some accounts propose long lists of criteria (see, e.g. Gerring, 2005;

Shoemaker, Tankard, Jr., & Lasorsa, 2004), whereas others only discuss one or a few (see, e.g. Bacharach, 1989; King, Keohane & Verba, 1994). There is certainly merit to both approaches. A wider range of criteria will make for a more encompassing evaluation, covering a larger number of aspects. However, restrictions to scope, time and space usually

(11)

8 mean that such width in the analysis must come at the price of its depth. If one wants to delve more deeply into any particular aspects of the evaluation, this must then put limits on the overall range of the analysis.

When making the decision on which aspects to include in the evaluation, a reasonable guide must arguably be the aim of the research at hand. Hence, this work lays no claim to an ambition of establishing a theory evaluation method praxis. The following arguments are, though stated in general terms, not meant to be generally applicable to all such works as this.

For this thesis, the aim is to determine which approach to solving the paradox of participation is most theoretically sound and most successful in aiding our understanding of the decision- making process that leads people either to participate or not. An initial aspect to be

considered must therefore be whether the theory is falsifiable or not. As will be discussed further below, falsifiability is central to theoretical soundness. Theories that cannot be falsified, should not even be called theories (King, Keohane & Verba, 1994). Hence, this criterion can be considered a threshold that all theoretical approaches must pass before any further analysis can be warranted. There is no point to further analysis if the theory cannot be falsified since then, it fails to be a theory at all.

Provided that more than one theoretical approach passes the threshold of falsifiability, it is necessary to include some criterion by which to compare different theoretical approaches to be able to say something about which one provides the most promising way forward, beyond the paradox of participation. For this task, there are several potential candidates. Gerring (2005) proposes a list of sixteen criteria for theory evaluation, among which there are many that could be helpful when comparing theories. Yet, as mentioned above, the constraints on the length of this study makes it objectionable to include too many such criteria. Therefore, the analysis will be centred around the concept of leverage, formalised by King, Keohane and Verba (1994) in their highly influential work Designing Social Inquiry. However, the concept of leverage is closely akin to the arguably more established principle of parsimony. King, Keohane and Verba themselves disqualify parsimony as a relevant principle by which to evaluate social science theory, yet they do so on what will be shown to be rather questionable grounds. Therefore, both concepts will be defined below in order to determine what

distinguishes them from each other. As the discussion below aims to show, parsimony will prove to be an integral part of leverage and thus it can play a useful role in terms of making the evaluation regarding leverage clearer and more systematic.

(12)

9 As is the case with any research project, this thesis inevitably comes with its own limitations.

Since a restricted number of criteria will be used for this evaluation, its conclusions will be vulnerable to such criticism as might be derived from the use of a different or more extensive set of criteria. However, this thesis does not aim to pronounce conclusive judgement on which theoretical approach is the ‘right’ one. Instead, this study is primarily an attempt to spark a new sort of discussion among theorists within the field of political participation. For more than sixty years empirical testing has failed to develop rational choice theory to take it beyond the paradox of participation. By approaching the problem from a methodologically different perspective the hope is to bring new insights that might have been obscured by the prevalence of empirical work within the field. Any conclusions from this project should therefore be considered in light of this. It is not in opposition to empirical work, nor does it wish to discourage similar evaluations in future. Rather, the hope is to work in tandem with empirical research on the journey to develop better rational choice approaches and to draw more attention to purely theoretical considerations. Criticism based on such considerations would therefore be welcome.

3.1. Falsifiability

A chief concern in any evaluation of theory must certainly be the question of its falsifiability.

On this issue Stinchcombe writes that “social theorists should prefer to be wrong rather than misunderstood” as “being misunderstood shows sloppy theoretical work” (1968:6). A similar view is expressed by King, Keohane and Verba in claiming that “vastly more is learned from theories that are wrong than from theories that are stated so broadly that they could not be wrong even in principle” (1994:19). The principle of falsifiability thus acknowledges that theories that lay claim to explain everything, simultaneously end up explaining nothing because they must either be so complex or so vaguely formulated as to give no real guidance with regards to understanding, describing, or explaining what has been empirically observed.

The desirability of falsifiability might appear counter intuitive. Seeing as the very point of social science is arguably the search for knowledge and complete understanding of the social world, deliberately constructing theories so that they can be proven to be wrong naturally seems strange. What many philosophers of science agree on, however, is the fact that theories, no matter how many times they are confirmed by empirical findings, can never be conclusively found to be true. According to Popper (1968), this gives rise to an asymmetry between verification and falsification of theory, making the latter much more important than

(13)

10 the former because only by this means can the line be drawn between what can be considered science and that which cannot.

King, Keohane and Verba support a more lenient view by regarding falsification not as a tool to disqualify theory, but rather to determine its limits of applicability. They claim that “the question is less whether, in some general sense, a theory is false or not […] than how much of the world the theory can help us explain” (1994:101). As previously mentioned, theory

building within the social sciences is a matter of abstracting from empirical findings in order to identify and systematically formulate general behavioural patterns. Theorising can thus be described as an act of consciously simplifying complex phenomena. Given this, it must be expected that such empirical findings as do not perfectly fit with social science theory will be frequently forthcoming. If we were, as Popper suggests, to completely reject theories based on such findings, there would be very few social science theories left. Certainly, rational choice theory should long since have been stricken from the records of political science according to this logic. The fact that this practice does not appear to have gained much influence within social science supports instead the post-popperian interpretation of

falsification supported by King, Keohane and Verba (1994). Falsifiability thus becomes an important criterion of evaluation because it can teach us valuable lessons about the bounds of applicability of certain theories.

One further argument for why falsifiability is a specifically important aspect to analyse with regards to rational choice theory and its development over the last sixty years can be found in the following words by King, Keohane and Verba:

“Science proceeds by […] adjustments; but they can be dangerous. If we take them too far we make our theories and hypotheses invulnerable to disconfirmation. The lesson is that we must be very careful in adapting theories to be consistent with new evidence.” (1994:104)

Since the developments of rational choice theory have tended to be precisely of this kind – adjustments aimed at making the theory more consistent with empirical evidence – there is reason to take falsifiability very seriously when considering rational choice theories. Even though the original calculus suffers from the very opposite problem, that does not guarantee that its successors’ attempts do not over-compensate and thus fail to formulate their

theoretical accounts in a way that allows for falsification.

Having thus established the why of falsifiability, it is time now to move on to the how. A theory can be said to be falsifiable if it can be specified which empirical evidence would be

(14)

11 needed to convince us that it is indeed wrong (King, Keohane & Verba, 1994; Popper, 1968).

While this might sound simple, it is in fact an exercise that relies on the existence of several properties to be possible. On this point, accounts vary slightly but some common

characteristics can be identified.

First, there is the issue of operationalisation and measurability. For King, Keohane and Verba this is mostly an issue of validity – actually measuring what one claims to measure. Since social science theories frequently include abstract concepts that cannot be directly observed, this can often be complicated. To this, Bacharach (1989) also adds the issues of reliability and variability. The latter refers to the fact that variables must be defined precisely enough for it to be possible to measure some variation. This condition is most commonly mentioned in reference to the dependent variable in the theory (Gerring, 2005; King, Keohane & Verba, 1994), yet for causal theory it must apply equally to independent variables. Any circumstance that is constant over space and time cannot contribute in any meaningful way in an empirical analysis. In this study it is the theoretical possibility to achieve operationalisation,

measurability and variability that is of primary concern. Even though practical issues can cause significant problems for empirical testing, which in turn can be problematic in terms of falsifiability, such issues lie outside the scope of this evaluation. Rather, the aim is to

establish whether the theories under evaluation are theoretically falsifiable.

Second, theories must avoid formulations that are tautological. This means that an explanation must not be stated in such a way as to make the outcome an automatic consequence (Bacharach, 1989). Claiming that an individual will vote if they go to the polling station, fill in the ballot and put it in the box would thus be a tautological statement.

What should also be clear from this example is the fact that tautological statements make falsification impossible. A tautology will always be true but, more importantly, it will also always be nonsensical and thus completely irrelevant as part of a theory.

Another way of phrasing the issue is proposed by Gerring (2005). Among the sixteen criteria he puts up for causal theory is that of differentiation. He writes that “a cause must be

differentiable from the effect it purports to explain” (2005:174). Hence, defining the dependent variable in terms of some or all of the independent variables would violate this criterion and inevitably lead to a tautological argument. This then further stresses the importance of proper and well thought out operationalisations as discussed above. Again, King, Keohane and Verba turn the attention to abstract concepts as they observe that:

(15)

12

“Explanations involving concepts such as culture or national interest or utility or motivation are suspect unless we can measure the concept independently of the dependent variable that we are explaining. When such terms are used in explanations, it is too easy to use them in ways that are tautological or have no differentiating, observable implications. An act of an individual or a nation may be explained as resulting from a desire to maximize utility, to fulfil intentions, or to achieve the national interest. But evidence that the act maximized utility or fulfilled intentions or achieved the national interest is the fact that the actor or the nation engaged in it.” (1994:110)

When evaluating rational choice explanations of political participation, it will thus be important to investigate whether this kind of circular reasoning can be found. Of course, one might be working with these kinds of assumptions present when developing theory, yet it is important that the product does not reflect this fact. That is, a rational choice theorist who makes assumptions about individuals’ behaviour being purposive and utility maximising will also be working from the assumption that the empirical reality one can observe can be

explained by such behaviour. The formulation of theory thus becomes an issue of filling in the blanks between what the theorist assumes about peoples’ behaviour, and the actual, observed behaviour. The risk of falling into the trap of tautology is therefore apparent if not, however, unavoidable.

Beside these two aspects of ensuring falsifiability King, Keohane and Verba (1994) and Gerring (2005) also point to the importance of the internal consistency of the theory. A theory which can be used to formulate opposing hypotheses about an outcome will be very hard to falsify. However, the remedy for this is to construct formal models (King, Keohane & Verba, 1994). Formalising the theory in mathematical or graphical terms helps ensuring that the arguments included in the theory do not oppose each other and this is what Riker and Ordeshook did all the way back in 1968. Following their example, this has also frequently been the practice since then. It should therefore be possible to feel reasonably comfortable as to the internal consistency of rational choice theory from the outset of this analysis. By formalising the hypothesised relationship between participation and costs as a negative mathematical sign in the calculus, it becomes impossible to use the theory to hypothesise that increased costs will increase participation. The calculus leaves no room for such opposing hypothesis to be drawn. When considering the issue of falsifiability, the analysis will thus primarily be concerned with operationalisation and measurability, as well as ensuring non- tautology of the theoretical statements.

(16)

13 3.2. Leverage as parsimony and breadth

Another concept commonly used when evaluating theory is that of parsimony. At first glance, this principle can be understood as a call for simplicity (Shoemaker, Tankard, Jr. & Lasorsa, 2004) or, more figuratively, as a razor that cuts away those pieces of a theory that are not strictly necessary to explain the phenomenon at hand (Sober, 1981). Its merits thus appear quite intuitive. It seems that there can be no controversy in the claim that a theory should not include such parts that are not necessary for the explanation or description at hand. To allow it would not only complicate our understanding of the world unnecessarily but would also make the theories more difficult to use since more data would be needed for empirical testing.

Hence, parsimony seems to be a sound principle to employ when developing and evaluating social science theory. However, at closer inspection the concept becomes a bit more muddled.

“Unfortunately, the word has been used in so many ways in casual conversation and scholarly writings that the principle has become obscured” says King, Keohane and Verba (1994:20) about parsimony. In contrast to the brief definition of the principle of parsimony given above, they interpret it not merely as a principle for the judging of theories, but rather as an

assumption about the nature of the social world. The preference for simple theories according to this interpretation comes out of the conviction that they are most likely closest to the truth because the social world itself is in truth simple. Unless it can be positively ascertained that this assumption about the social world holds, King, Keohane and Verba argue that the principle of parsimony must be used with great care. Seeing as there is ample reason to believe the social world is not indeed simple, as opposed to some parts of the natural world, parsimony might in fact prove not to be a very prudent principle to apply in the present context.

However, what King, Keohane and Verba (1994) fail to acknowledge is the fact that the many different ways in which the principle of parsimony have been used does not necessarily demonstrate conceptual confusion but rather points to the presence of more than one possible interpretation of the term. Gunitsky (2019) argues for three distinct interpretations of

parsimony: an aesthetic interpretation, an ontological interpretation, and an epistemological interpretation. These versions of parsimony come with their own assumptions and limitations, making them differently suitable in different contexts (Gunitsky, 2019). The form of

parsimony that King, Keohane and Verba (1994) reject is the ontological interpretation of parsimony. Gunitsky indeed agrees that its assumption about the simplicity of the world is far

(17)

14 better suited for natural sciences, and especially physics, than it is for social sciences.

Equally, the aesthetic interpretation, preferring simplicity for the sake of beauty and elegance, is deemed by Gunitsky to have no justifiable place in social sciences. However, Gunitsky claims that the epistemological interpretation of parsimony should not be as readily slighted.

According to the epistemological interpretation, parsimony is preferrable for two reasons:

firstly because “simplifying assumptions are necessary for useful generalizations” (Gunitsky, 2019:708), and secondly because simple theories are more easily falsified. This second justification for parsimony, however, limits its use by making it only a means to an end, namely that of falsifiability. Rather than having any intrinsic value, it is rendered no more than a tool. As such, it could not qualify as a distinct criterion for theory evaluation but would be relegated to the sphere of theory development as a strategy to ensure falsifiability. On the other hand, there is certainly merit to the first justification of the epistemological

interpretation of parsimony. As put by Gunitsky: “[…] explanations require generalization, and generalization requires abstraction. A theory abstracts from the world in order to make sense of it, not out of a conviction that abstraction is more elegant or more descriptively accurate” (2019: 711). Thus, contrary to the ontological interpretation of parsimony, the epistemological version does not assume the world is simple but rather “assumes the world is complex and seeks to mitigate the consequences of the complexity” (Gunitsky, 2019: 708).

At this point it can therefore be concluded that by employing the epistemological

interpretation of parsimony it can be rendered relevant as a criterion of evaluation suitable for the purpose of this thesis. Parsimony thus gives preference to simple theories before complex ones because they better allow for generalisation. Only by moving away from the details can the bigger picture be seen. While the fact that simpler theories are also more easily falsified cannot be considered a valid argument for using the principle of parsimony as an evaluative criterion, it can be viewed as an added bonus. In practice, what this interpretation of

parsimony means is that when faced with two theories that are used to explain or describe the same phenomenon, the simplest one is judged better. Since they both have the same

observable implications, the added detail of the more complex model must only muddle the bigger picture. By virtue of managing to explain the same phenomenon with fewer variables, the simpler theory proves the added variables in the more complex theory unnecessary and therefore, those details should be shaved away.

(18)

15 Going back to King, Keohane and Verba (1994), their disqualification of parsimony as a relevant principle for evaluating theory leads them to propose another concept instead;

namely that of leverage. Even though the discussion above shows that parsimony should not be disregarded, there is still reason to have a closer look at this idea of leverage. King, Keohane and Verba describe leverage as the principle of “explaining as much as possible with as little as possible” (1994:29). In accordance with the principle of parsimony, this concepts also holds that a theory which uses only a few explanatory variables to accurately explain a complex problem is preferable to one that uses a larger number of variables to achieve the same level of explanation. However, contrary to the principle of parsimony, leverage can be used more broadly, comparing a wider range of theories, because the theories under evaluation must not necessarily have the exact same observable implications. On the contrary, maximising a theory’s leverage can often be an issue of showing that it applies to a wider range of phenomena than those it was originally constructed to describe or explain (King, Keohane & Verba, 1994).

In terms of formal modelling of theories, the difference between parsimony and leverage can thus be explained as follows. Imagine there are three potential versions of the rational choice theory proposed by different researchers. Theories A and B seek only to explain party

activism and they do so equally well, theory C however not only explains party activism, but lays claim to explain the broader phenomenon of political activity. Theory A has the fewest independent variables, while theories B and C have the exact same independent variables.

Now, using the principle of parsimony, only theories A and B can be compared and the one with the fewest independent variables, theory A, is the simplest and thus the preferred theory.

However, if leverage is employed instead, all three theories can be compared. In this case, the judgement becomes less straightforward. Theory A still is the simplest, it fulfils the “as little as possible” part of the leverage concept the best. On the other hand, theory C performs better in terms of explaining “as much as possible”. In this case it will thus be a question of whether the larger scope of theory C is worth the increase in complexity.

Something else that can be seen by this example is that a theory can be highly parsimonious and yet have relatively little leverage as is the case with the first model. By only using the principle of parsimony to evaluate theory, this aspect would be overlooked. In fact, many theories would not even be comparable if the analysis were restricted to the use of parsimony as the only evaluative criterion because they explain slightly different phenomena.

(19)

16 Strictly speaking, the concept of leverage can be used in the same way as the principle of parsimony when applied to two or more theories with the same observable implications.

Parsimony can thus be said to be included within the concept of leverage. In addition to this, leverage also incorporates what Gerring (2005) calls breadth. He writes that“a proposition informing us about many events is, by virtue of this fact, more useful than a proposition applicable to only a few” (2005:173). Leverage can thus be considered as a combination of these two, and in Gerring’s view, distinct principles of parsimony and breadth.

From this discussion it should thus have become clear that leverage is a more encompassing concept than that of parsimony. In the following theory evaluation, finding the theoretical approach which has the highest leverage while at the same time satisfying the criterion of falsifiability will therefore be the goal. However, rather than using leverage directly in the analysis, dividing the concept into its composite parts of parsimony and breadth and discussing them separately will make the evaluation clearer and more systematic. By using these principles for the analysis instead of only talking of leverage, any potential confusion regarding what aspect of leverage is referred to can be avoided. The two-by-two table in figure 1 shows how breadth and parsimony vary to achieve different levels of leverage.

Breadth

Few observable implications Many observable implications

Parsimony Low

complexity

Medium leverage Little breadth/high parsimony

High leverage

Much breadth/high parsimony High

complexity

Low leverage Little breadth/low parsimony

Medium leverage Much breadth/low parsimony

Figure 1. How parsimony and breadth interact to provide various levels of leverage.

3.3. A note on comparability

In the introduction the external boundaries of the concept of political participation were set by clarifying more precisely how the concept is defined for this thesis. Another important aspect to consider is whether there are any significant internal boundaries within the concept of political participation. In other words, can it be expected to matter for the evaluation which specific form of participation a certain theoretical model seeks to explain? Answering this question is crucial for the comparability of different theoretical accounts and thus how the forthcoming evaluation can be executed.

(20)

17 The argument has been made that different forms of participation should be expected to warrant different explanations (see e.g., Bäck, Teorell & Westholm, 2011). Again, it is important to consider the aim of this study. The relevant concern is the theoretical validity of rational choice theory, not its empirical validity. Using the criteria of falsifiability and

leverage, the purpose is to say something about the theoretical soundness of various rational choice accounts. Naturally, another, equally important step is to investigate whether such theoretically sound accounts have any empirical use, but that step is not included in the present study. Such investigations of the theory’s empirical validity make up the bulk of research within the field. In fact, most of the articles from which the theoretical accounts under evaluation have been taken, are studies of this kind. This means that for all the rational choice theories presented and evaluated below, the empirical validity has already been studied and confirmed.

Therefore, the issue of empirical validity can be set aside for the present, allowing this thesis to focus solely on the falsifiability of the various components of rational choice theories and their respective level of leverage. An example of what this could mean is that, for the present, it is irrelevant whether it makes any empirical sense to try to explain individual voting

behaviour using collective rationality. What matters is whether such a theoretical argument is falsifiable and if it has more leverage over the problem than other theories explaining

political participation. If it is found to be falsifiable and able to improve our theoretical understanding in terms of leverage, it will be tested empirically. Should it then become apparent that it lacks empirical validity, that is, if the theory should be falsified, collective rationality would be rejected as an incentive that can explain voting. This also gives a

glimpse of why it is important to make sure that theories are in fact falsifiable. If they are not, the empirical investigation might conclude, on false grounds, that collective rationality does explain individual voting behaviour.

Now, this is a very crude example of the difference between studying the theoretical and the empirical validity of theory. Hopefully, though, it is successful in illuminating why it will be possible to evaluate the different theoretical accounts jointly, despite the fact that they explain different forms of participation. Since the various forms of participation with which the theoretical accounts are concerned all fit within the definition of political participation, evaluating the falsifiability of the theoretical components in relation to political participation in general will allow for conclusions to be drawn as to their theoretical validity, even if their empirical validity in relation to specific forms of participation might remain undetermined.

(21)

18 4. Presentation of the Theoretical Approaches Under Evaluation

Before proceeding to conduct the evaluation, it is necessary to give an overview of the material which is to be analysed. By thus presenting the various theoretical approaches found in the literature on political participation the following analysis will be afforded more

transparency, much like one would be expected to descriptively present the data used in an empirical analysis to make any subsequent causal analysis more transparent. The presentation of the relevant theoretical approaches will be followed by a brief discussion of the potential limitations caused by this selection.

In identifying relevant rational choice theoretic accounts to include in this evaluation, a number of selection criteria have been applied. First, the theories must aim to explain

political participation as defined by Verba and Nie (1972). Second, they must be built on the underlying assumption that actors are utility maximisers, and third, they must explicitly and exclusively belong to the rational choice theory tradition. These restrictions entail leaving some heavily cited works outside the analysis. Ferejohn and Fiorina’s (1974) study building on a minmax regret framework or Verba, Schlozman and Brady’s (1995) Civic Voluntarism Model which includes rational choice theory within a bigger theoretical framework are examples of such theoretical approaches that do not fit the criteria set for this evaluation.

Forth, and finally, the account must bring some new element or argument to the calculus of participation. What this means is that the theorists must propose either a reformulation of the calculus or a new interpretation of some or all of its elements. Having thus narrowed down the relevant population of theories, the selection has aimed to include those approaches which are most frequently cited within the field.

4.1. Theoretical Approaches

One of the first attempts at reconciling the paradox of participation was made by Riker and Ordeshook (1968). They took notice of the fact the original calculus of voting appeared to distinguish between different types of effects on the expected utility of voting; those that were dependent on the individual’s contribution to the outcome, and those that were not. In the calculus this is represented by the fact that B is multiplied by p, indicating a dependency, while C is not and must thus be independent of the individual’s contribution. Given that the effects on the expected utility can be divided into these two categories, Riker and Ordeshook argued that a further subdivision should also be possible, namely into positive and negative

(22)

19 effects. Thus, C is a negative independent effect and B is a positive dependent effect. Beside these there should also be a positive independent effect and a negative dependent effect according to Riker and Ordeshook. In the end, they choose to exclude the positive dependent effects from their formulation of the theory arguing that the relationship between these effects and the expected utility gained from voting would be too complex to warrant their inclusion.

Hence, the theoretical approach that they propose takes the shape of the following calculus:

𝑈 = 𝑝 × 𝐵 − 𝐶 + 𝐷 (2)

The addition of the D-term represents the inclusion in the theory of positive effects derived from voting that are independent of the “individual’s contribution to the occurrence of a favourable outcome” (Riker & Ordeshook, 1968:27). Riker and Ordeshook argued that among these positive effects were five satisfactions which arise from (i) complying to the

‘ethic of voting’, (ii) affirming one’s allegiance to the political system or (iii) a certain

political party, (iv) deciding and going to the polls⁴, and (v) affirming one’s political efficacy.

A common feature of several of these satisfactions is that they tap into what might be called a sense of civic duty (Ferejohn & Fiorina, 1974).

Moving beyond the issue of voting, Muller and Opp (1986) develop the rational choice theory aiming to explain participation in rebellious collective action. What they propose is that when considering collective forms of political actions, individuals may consider group influence and political efficacy in their calculations. They point out that “what is individually rational is collectively irrational” (1986:473) since inactivity among those who expect to benefit from successful collective action, diminish the likelihood of such an event taking place. This consideration motivates Muller and Opp to formulate the calculus as follows:

𝑈 = 𝐵(𝑝_𝑔+ 𝑝_𝑖) − 𝐶 + 𝐷 (3)

In this formulation, pg denotes the probability that the group’s contribution is decisive for the outcome. Thus, even though the individual efficacy might still be negligible, action could be motivated by a belief in the efficacy of the group as a whole.

In a highly methodologically driven article, Finkel and Muller (1998) propose a rational choice theory of collective protest action which they call the ‘collective interest model’.

4 “These items are usually regarded as costs, but for those who enjoy the act of informing themselves for the decision, who get social satisfactions out of going to the polling booth, etc., these supposed costs are actually benefits.” (Riker & Ordeshook, 1968:28)

(23)

20 According to this model individuals’ participation in protest activity depend on (1) their level of discontent with current provisions of public goods, (2) their belief that collective action can be successful, and (3) their subjective political efficacy. Hence, this model shares some characteristics with the theoretical approach proposed by Muller and Opp (1986) seeing as they both allows for a collective element in the probability of action success. In addition to the three elements of the collective interest model presented above, Finkel and Muller also include past participation in their formal model. The reasons for this are purely

methodological. The motivation is that social and psychological incentives can be subject to attitude rationalisation and reciprocal causal relationships if they are not measured in the temporally correct order (i.e. before the act of participation). Therefore, past participation should not be considered part of the explanation behind the decision to participate or not.

Rather, it is a methodological tool which enables a more precise identification of those theoretically motivated aspects of the explanation which might otherwise become muddled by statistical biases.

Finkel and Muller (1998) do not present a formal equation for the collective interest model.

How it should be specified is thus not precisely clear. Inferring the specification from the theoretical discussion and the empirical test of the model that is performed in the article, however, results in equation (4). In addition to public goods benefits and collective and individual efficacy mentioned as parts of the collective interest model above, this

specification also includes the perceived necessity of group unity (NGU). This term is closely connected to the individual’s belief in their own influence by capturing the prevalence of collective rationality. This is the belief that the group is stronger than each individual, and that it is important for everyone to do their part in order for the group to succeed (Finkel &

Muller, 1998).

𝑈 = 𝐵 (𝑝_𝑖 + (𝑁𝐺𝑈 × 𝑝_𝑔)) − 𝐶 (4)

Finally, Bäck, Teorell and Westholm (2011) propose a novel approach to tackling the paradox. They argue that different forms of participation could possibly be connected to different forms of motivations and incentives. Rather than specifying one calculus to explain all forms of participation, they propose using a range of different models. Hence, voting (v), party activity (pa), contacting elected officials (c), and manifestions (m) are considered to be connected to their own, specific calculus. While no precise formulations of these calculi are presented by Bäck, Teorell and Westholm, their hypotheses regarding the relationships

(24)

21 between these four forms of participation and a range of different incentives are presented.

Based on these hypotheses, the following four equations can be specified (see table 1 for further explanation of the notions).

𝑈_𝑣 = 𝑝 × 𝐵 + 𝐷_𝑠𝑛+ 𝐷_𝑝𝑖− 𝐷_𝑟− 𝐶 (5)

𝑈_𝑝𝑎 = 𝑝 × 𝐵 + 𝐷_𝑝𝑖+ 𝐷_𝑟+ 𝐷_𝑒− 𝐶 (6)

𝑈_𝑐 = 𝑝 × 𝐵 − 𝐶 (7)

𝑈_𝑚 = 𝑝 × 𝐵 + 𝐷_𝑒 − 𝐶 (8)

The first difference between the models is not directly observable in the equations. This concerns the multiplicative term which Bäck, Teorell and Westholm calls collective

incentives. What they argue is that “collective incentives are less important for the two modes of participation that are directly linked to the electoral system (voting and party activity) than for the two remaining modes (contacting and manifestions)” (2011:79). The reason for this is

“the strong correspondence between the issue agenda of the participants and the message expressed by their participation” (2011:80) when it comes to manifestations and contacting.

Thus, motivations related to the outcome ought to be more significant for these modes of participation. With this said, collective incentives are not necessarily completely irrelevant in explaining voting and party activity, hence the multiplicative term is included in all four equations.

Furthermore, Bäck, Teorell and Westholm (2011) distinguish between different forms of selective incentives, or what has so far been referred to as positive effects that are

independent of the individual’s contribution to the outcome. Building on previous research they choose to include social norms (sn), strength of party identification (pi), radicalism (r), and entertainment value (e) in their theoretical approach, notated by subscripts to the D-term.

Yet, as was considered the case with collective incentives, these selective incentives are also theorised to have different effects on different modes of participation. Hence, for example the strength of party identification is argued to be significant only in explaining voting and party activity because these modes of participation “ordinarily provide far more direct ways of expressing party loyalties than either contacting or manifestations” (2011:81).

(25)

22 Table 1 gives an overview of what has been said so far, gathering all the proposed

specifications of rational choice theory. Looking at this, it can be concluded that while these approaches all vary to some extent, there are some features that are salient in terms of

providing a solution to the paradox of participation. Firstly, there are the selective incentives, represented in the calculi by the D-term. The precise content of this category of incentives varies between the different accounts but a common definition is given by Riker and

Ordeshook (1968) identifying them as positive effects that are independent of the individual’s contribution to the political outcome. Secondly, when considering various forms of collective action, the p-term can be extended to contain both individual and collective political efficacy, thus changing the character of the multiplicative term and improving its potential to be non- zero. In the next section, these features will be analysed and evaluated using the analytical concepts discussed above of falsifiability and leverage. However, before commencing what is the core activity of this thesis, some remarks concerning the limitations of the analysis should be made.

4.2. Further notes on the selection criteria

As mentioned in the introduction to this chapter, the selection of theoretical approaches has been restricted in several ways. The first criterion, that the theory must aim to explain political participation as it is defined in this study, is straightforward. The purpose of this evaluation is to say something about rational choice theory as it applies to political

participation, hence rational choice accounts applying to other fields of study are irrelevant.

Equally, the criterion that the theoretical accounts must explicitly and exclusively belong to the rational choice tradition is a means by which to ensure that the evaluation remains focused.

It is also important to restrict the analysis by only including theories that are built on the same underlying assumptions. On this point, Bacharach (1989) argues that theories cannot be compared on the basis of their underlying assumptions because such assumptions are always value laden. Such comparisons will therefore always be ideological to some extent and thus belong to the realm of political theory, not to that of theory evaluation. Hence, while attempts have been made at solving the paradox of participation by applying a minmax regret

framework to the calculus such approaches are not comparable with those that apply a utility maximisation logic. Seeing as the vast majority of research over the last sixty or so years has retained the expected utility as the dependent variable in the calculus, so will this study.

(26)

23

(27)

24

(28)

25 It should also be mentioned that, for the subsequent analysis, the calculi will be viewed more categorically depending on which elements they are built of. Thus, the focus will not be on whether Riker and Ordeshook’s approach is preferrable to that of Finkel and Muller, but rather whether explanations using selective incentives as an explanatory variable are preferrable to those that do not. Therefore, the evaluation will not be restricted to the

arguments brought forth by those contributors hitherto cited but will extend to insights gained elsewhere in the literature on rational choice theory and political participation. This way a wider range of perspectives can be allowed into the analysis without making the evaluation too extensive or unfocused. Hence, by limiting the number of theoretical approaches under investigation the evaluation is kept focused and succinct while not unnecessarily excluding insights from a wider range of research from the discussion of them.

5. Evaluating Rational Choice Theory and its Extensions

In this section of the thesis, the various approaches to solving the paradox of participation will be evaluated using the criteria of falsifiability and leverage as discussed above. Since leverage can only be considered in a comparative analysis, this venture will take place in the last part of this evaluation. As mentioned before, falsifiability poses a sort of threshold that theory must pass to be admitted as theory. Evaluating leverage would thus be pointless if the falsifiability of the theoretical approaches under comparison had not first been established.

The theoretical approaches will be evaluated not primarily as distinct theories but rather as examples of theories containing certain elements. Hence, part two of this evaluation will look at those approaches that employ collective incentives and collective rationality as explanatory factors allowing rational choice theory to move beyond the paradox and argue for how the inclusion of this theoretical element affects the theory’s falsifiability. The same will be done for selective incentives in section 5.3 before moving on to the issue of leverage in the fourth section.

5.1. The standard elements of the calculus

Most of the theoretical approaches with which this thesis is concerned include the theoretical elements which were first proposed by Downs (1957) and formalised by Riker and

Ordeshook (1968) in the original calculus of participation. Hence, before looking more closely at the various attempted solutions to the paradox of participation, some general observations can be made in relation to the evaluative criterion of falsifiability. This might

(29)

26 appear redundant seeing as the standard form of rational choice theory has very much been falsified, ergo the paradox. Yet, a point can be made in the name of transparency and

thoroughness for an evaluation even of these basic theoretical elements. Furthermore, it could potentially be argued that rational choice theory has been falsified on false grounds.

Investigating how the concepts in the theory have been operationalised and measured can be seen not only as a means of evaluating the falsifiability of the original calculus but also as a way to establish whether the theory has been reasonably applied in previous research. In order to be as transparent as possible and to avoid skipping any steps and thus attract very reasonable criticism, this first part of the analysis will evaluate the falsifiability of the original elements of the calculus.

The very first issue to consider is that of operationalisation and measurability. Beginning with the dependent variable of political participation, it will in most cases be quite straightforward both to operationalise and to measure. Of course, the operationalisation will vary depending on which form of participation the theory concerns and while it is quite easy to operationalise electoral participation as having voted in a given election some forms of political

participation such as engaging in political organisations can be less easily operationalised.

Some organisations are overtly political, such as political parties, while others might not have as their main motive to influence public policy yet take a stand on certain political issues that they find concern them. This has potentially been a consideration for researchers and theorists in the field in deciding which phenomena to study. By choosing to look at party activism rather than for example membership in political organisations, the ambiguity concerning which organisations should be seen as ‘political’ is avoided. The fact that political participation is a concrete activity, rather than a subjective attitude, also make it easier to measure it with good reliability. As a general conclusion it can therefore be said that, at the theoretical level, operationalisation and measurement of the dependent variable will not be a significant obstacle to falsifiability for rational choice theory.

Looking instead at the independent variables of the original calculus, the issues of

operationalisation and measurement becomes less straightforward. For example, Riker and Ordeshook define costs as negative effects “for which the magnitude is independent of the individual’s contribution to the outcome” (1968:27). Usually monetary costs, time costs and potentially also some cognitive costs or physical costs are included in this category of negative effects. Yet, it is not inherently given that there are no other potential negative effects. As has been shown in the previous section, social norms are in some theoretical

New Perspectives on the Paradox of Participation

New Perspectives on the Paradox of Participation

A Theoretical Evaluation of Rational Choice Theory as it Applies to Political Participation

Table of Contents