Local Realism: An Analysis of Social Choice Theory

Abstract

This paper basically presents an analysis of a practical case. It addresses the study of social choice theory in relation to scientific realism. In line with the positions defended by scholars such as U. Mäki or H. Kincaid, the question of realism is considered a local rather than a universal theory, which requires a case-by-case response, that is, theory-by-theory. The analysis focuses more on the existence of the entities postulated by the theory, ontological realism, than on the truth or approximate truth of its statements. Given that the theory is characterised by the interweaving of economic, political and moral aspects, the question about realism is posed within each sphere. Furthermore, the issue of realism helps to clarify the theoretical status of social choice theory. The conclusion reveals that the valorative neutrality of the theory and its high level of idealisation render it impossible to maintain a realist stance towards it.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

When referring to realism in science, physics immediately springs to mind, and there are a number of arguments that help to uphold a realist position: the causal relationships established through assays, as defined by I. Hacking (Hacking 1983), the miracle argument formulated by H. Putnam (Putnam 1978) or the empirical success over the course of time postulated by R. Boyd (Boyd 1973). Unfortunately, none of these arguments can be applied to social sciences. Generally speaking, and for authors such as Hacking (Hacking 1999, pp. 31–32), interactive classes constitute a fundamental difference regarding the matter of realism, in the sense that classifications in social sciences are interactive, which is not the case in natural sciences. This means there is a conscious interaction in social sciences between classes and individuals in terms of kinds of classification. Uskali Mäki also identifies differences, in this case involving the fields of economics and physics.¹ A series of questions therefore arise, such as: is it possible to be realist regarding quarks and leptons, and anti-realist regarding social structures and economic agents? Does the issue of realism have to be an all-encompassing matter, regarding science in general, or should we differentiate between social sciences and natural sciences? Further still, should differences be made regarding the different fields, and therefore refer to a realism in physics, a realism in economics, another in chemistry and yet another one in sociology? Is it the same to refer to realism when we are dealing with mature theories, such as the general theory of relativity, as when we are dealing with new theories or one less confirmed, such as the rope theory? What does this imply for social sciences?

Mäki refers to a contextualised and local realism; that is, the question of realism is in fact the question of the realism of a theory and not a question about science in general, and it is in this sense that Mäki posits a local and contextualised realism (Mäki 1992a, 1996, 1998, 2005). The same stance is adopted by, among others, H. Kincaid, for whom “the realism issue in philosophy of science cannot be decided in a perfectly general way that ignores specific issues in specific science” (Kincaid 2000, p. 667). This is the strategy to be followed in this paper, addressing realism from the perspective of a specific theory, namely, social choice theory (SCT). The question underpinning this paper is whether one should hold realism about SCT in the sense posited by Mäki: “By a ‘realist reading’ of a theory I mean an interpretation of the theory as putatively referring to entities, such that the theory has a chance of being either true, close to the truth, or carrying the promise of getting us closer to the truth of what it represents” (Mäki 1992a, p. 38).

This will enable us to clarify certain important issues regarding the theory’s theoretical state and its explanatory capacity.

The first step involves a brief characterisation of SCT, paying special attention to the contribution the philosophy of science may make to the theory. This is followed by an illustration of some of the issues the theory addresses, and finally, the question of realism is considered from a contextualised perspective.

Social choice theory stands at the crossroads between economics, politics and ethics, being all seasoned with a smattering of formal logic. In its most simple interpretation, it is a theory that addresses the manners and procedures for taking collective decisions; that is, how we aggregate individual decisions to reach a collective one.² From this perspective, a formal study is conducted of the different voting rules and the requirements that have to be fulfilled by any method of aggregating preferences in order to be considered a sound method. This is where we encounter one of the theory’s better known outcomes, namely, Arrow’s Impossibility Theorem, which states that any decision method that satisfies certain minimum conditions of rationality (reflexivity, transitivity and completeness) and four more or less trivial ethical conditions,³ taken from the majority voting rule, will provide an intransitive or dictatorial outcome. In short, the conditions are inconsistent with each other. This is the formal part of the theory. Generally speaking, it is an illustration of the use of the axiomatic method, which can be used not only to define a voting mechanism, but also for the definition of any mechanism for collective decision-making, especially in distributive matters.

The economic part has more to do with individual models and the theory’s mathematical tools. There is one field that has been characterised by an economistic approach since its inception, and that is the field of social choice theory. Indeed, Arrow himself singles out market and vote as two special cases of the more general category of collective social choice (Arrow 1963, p. 5). We may therefore refer to the ordering of preference, indifference curves, individual utility and social utility, alluding to homo economicus as the protagonist of these decisions that are to be aggregated, etc. The key question here is that the two theories, consumer behaviour theory and rational choice theory, share an essential part of their ontology. Furthermore, the bulk of these decisions that are addressed are decisions of a distributive nature. Strictly speaking, the alternatives presented in social choices are social states that are defined as:

A complete description of the amount of each type of commodity in the hands of each individual, the amount of labor to be supplied by each individual, the amount of each productive resource invested in each type of productive activity, and the amount of various types of collective activities, such as municipal services, diplomacy and its continuation by other means, and the erection of statues to famous men (Arrow 1963, p. 17).

It stands to reason that when ordering the alternatives, or social states, at both an individual and collective level, criteria are called for to indicate which alternatives are socially preferable. These criteria, whether they involve equality, merit, needs, utility, etc., are ethical criteria, yet the theory does not distinguish between them. “A member of Veblen’s leisure class might order the states solely on the criterion of his relative income standing in each; a believer in the equality of man might order them in accordance with some measure of income equality” (ibid.).

Finally, politics envelops everything insofar as it is defined as decision-making for achieving a group’s goals in public affairs.

Choice theory is, or should be, a vast interdisciplinary field, where economists, politicians, moral philosophers and mathematicians come together, being furthermore a field that provides philosophers of science with an extremely fertile field of work to which they can apply a different treatment. Unfortunately, collaboration is scarce, and within this field philosophers of science are considered outsiders. My interest, in what follows, is to highlight some of the problems that may be addressed from the philosophy of science in this field of work and whose discussion will ultimately lead to the consideration of realism within social choice theory. These questions will also serve to illustrate how this theory works, over and above well reported results such as Arrow’s Impossibility Theorem.

The first of these questions involves conceptual definitions and how these are applied in the theory. We shall begin by studying aspects of the axiomatic method. Generally speaking, the axiomatic method may be defined⁴ by the design of a functional mathematical figure that depicts, schematises or models the aggregation and decision models. Once this figure has been designed, the analysis goes on to focus on the sets of properties, or axioms, that these processes may, or should ideally, satisfy. There are sundry ways of applying the axiomatic method; first, there are the impossibility outcomes, of which Arrow’s theorem is the paradigmatic example, where it is denied there is any function of a specific class that satisfies a specific set of conditions. A second approach would be to investigate the properties of, or the conditions fulfilled by, a specific defined function, and third, given a set of properties, find out which functions fulfil them. In each case, the mathematical function seeks to order the different social states according to their social desirability, with this concept being defined through the conditions the function has to satisfy. This latter point is the key one to be stressed; in other words, the ordering criterion is defined through the function’s conditions. It is not a prior criterion but rather one that is built based on such conditions; what we want to say by this. It may be expedient to develop it through an example related to egalitarian distributions. The example presented forthwith is by B. Tungodden (2003). The author defines moderate egalitarianism (the criterion for ordering social states) as the combination of an interest in fostering equality with the principle of personal good. The conditions defining it are as follows: anonymity, conditional contracting extremes, principle of personal good and the strict priority to equality promotion. There now follows a definition of these conditions.

• Anonymity: For all alternatives x and y, if x is a permutation of the values of y, then x is equally as good as y.

• Strong Conditional Contracting Extremes (or betterness): For all alternatives x and y, if (1) all the best-off persons in x are best-off persons in y and their well-being level is strictly lower in x than y; (2) all the worst-off persons in x are worst-off persons in y and their well-being level is strictly higher in x than y, and (3) the well-being of everyone else is the same in x and y; then x is better than y.

• Strict Priority to Equality Promotion: For all alternatives x and y, if (1) there are persons with higher well-being in x than y and persons with higher well-being in y than x, and (2) x is more equal than y, then x is better than y.

• Principle of personal good: For all alternatives x and y, if everybody is as well off in x as in y, and someone is strictly better off (in x) then x is better than y.

The problem is that the conditions described generate an intransitive result. Let us now consider an example. “Suppose that y = (1, 100, 100) is considered more equal than x = (2, 10, 100), and hence strict priority to equality promotion implies that y is better than x. Compare x with z = (2, 10, 10). From the principle of personal good, it follows that x is better than z. By transitivity, we now have that y is better than z. However, this violates strict priority to equality promotion according to the minimal requirement of strong conditional contracting extremes on equality” (Tungodden 2003, p. 14). The option is therefore either to modify any one of the conditions or renounce transitivity. Tungodden’s solution is to modify the conditions, defining equality according to the maximin principle, which selects the situation that maximises the state of the worst-off person. This allows affirming that x is better than y, according to the definition of equality, and avoiding the intransitivity of the outcome.

This means the formal requisite of transitivity informs the concept’s characteristics that are chosen to order the different social states. Not just any characterisation of equality will be valid, but instead only those generating quasi-orderings. For some philosophers, on the other hand,⁵ the concept of equality does not generate transitive orderings, so this structure, and therefore the axiomatic method, needs to be renounced for ordering social states. What is at stake here is that a technical requisite informs the concept’s characteristics that may or may not be chosen. This is one of the reasons for the clash between the advocates of the axiomatic method and its detractors, or, as some have put it, between economists and philosophers.⁶ In simple terms, philosophers have an intuitive understanding of the nature of inequality, with their definition being embodied in principles that very often generate impossibility outcomes. Economists, for their part, define the concept according to the ordering generated by some or other set of principles. The concept is therefore designed or constructed by theoreticians through the measurement process.⁷ What is ultimately at stake is the explanation of the conditions that order social states, and when it is undertaken in economic terms it is related to logical compatibility, whereas for philosophers the ordering should be decided by their notion of equality, regardless of possible inconsistencies or the possibility or not of ordering all the social states.

A further issue dividing economists and philosophers refers to the possibility of using experimentation as regards distributive theories.

The traditional role assigned to assays according to the standard conception is to test theories. This implies, as has been sustained, that experimentation has no relevance in the context of the theories of distributive justice, given that these are normative theories based on principles or ethical judgements. Since Hume’s famous dictum, regarding the impossibility of inferring is-ought judgements, theoretical knowledge is generated by scholarly introspection and deduction from the general principles so founded. The general opinion is that popular beliefs about justice are one thing and a correct theory about it is quite another. The data obtained through empirical research are not relevant for the correction or validity of the theories. However, in recent years there has been a growing interest in research in this area, which seems to indicate that this view is not widespread and experimentation does have a role to play in this field.

I have posited elsewhere (Torres 2010) different roles for experimentation as regards distributive theories, so I shall focus here solely on the part it plays in conceptual clarification, in the sense it allows a conceptual exploration of the consequences that may be forthcoming from the definitions of the concepts.

One of the most important fields of research in egalitarian theories is the definition of the measures of inequality. The concept is highly complex, and a wide range of measures have been suggested for ordering social states in different ways. The research conducted by Yoran Amiel and Frank Cowell in 1991 (Amiel and Cowell 1992)⁸ sought to prove experimentally whether the suggested axioms, as properties of the Lorenz curve, fitted the definition of the concept of inequality that individuals hold. In short, the Lorenz curve, one of the more popular mechanisms among economists for evaluating the inequality of different income distributions, is defined by the following axioms:

• Anonymity: for all the alternatives x and y, if x is a permutation of the values of y, then x is equally as good as y.⁹

• The principle of population: Replication of the population and its income should not affect the index of inequality.

• The principle of scale invariance: The only important thing is the relative benefit and not the absolute one. This means the index is not sensitive to proportional increments of the relevant benefit.

• The Pigou-Dalton transfer principle: Inequality always decreases when there are transfers from better-off individuals to worse-off ones, as long as the mean income does not decrease and the order among the individuals involved remains the same.

It should be noted that accepting the Lorenz curve means accepting every one of these axioms. The first three may be more or less questionable as they involve a choice between absolute measures and relative measures of inequality. The same does not occur with the Pigou-Dalton principle, which is almost unanimously accepted by theorists. By way of example: let us assume the following distributions: A = (1, 4, 7, 10, 13) and B = (2, 4, 7, 10, 12). According to the principle, inequality decreases in the step from A to B given that, first, there has been a redistribution from better-off to worse-off; second, the mean income has not changed, and third, the order among the individuals involved remains the same. Let us now consider the following distributions: A = (1, 4, 7, 10, 13), B = (1, 5, 6, 10, 13). If you consider that A is more unequal than B, this means you are a proponent of the Pigou-Dalton principle, whereas if you consider B is more unequal, then you do not support said principle. The fact this principle is not supported in this specific example may be due to two reasons. First, the transfer does not take place between individuals that we may associate to the situation of better- and worse-off, as they occur in the middle of the vector. Assays appear to show that agreement with the Pigou-Dalton condition depends on who transfers and who receives. When the transferor is the best placed individual and the receiver the worse placed one, the principle receives high support, whereas when the transfers occur at the top of the table, that is, between better placed individuals, support for the principle decreases (Amiel and Cowell 1992, pp. 15–17). On the other hand, it can be affirmed that with the transfer, although individual 2 sees an increase in their income, this implies the distance between the third worse-off and the fourth has increased, and this may be seen as an increase in inequality.

What is the significance of these findings for the concept of inequality? Should theorists take them into account in their definition of the concept of inequality? Should the transfer principle be modified so that it accepts only transfers between the better- and worse-off, while rejecting transfers in the lower part of the table? Or are theorists in a privileged position for evaluating inequality? Can we associate the empirical validity of the theory to the extent to which it reflects the understanding the majority of individuals have of equality or justice? Accordingly, can an individual’s moral intuitions play the same role as observation in science; in other words, can they help us to verify the theory? If that is the case, can that provide the pillars for upholding moral realism? In one sense, the moral intuitions of individuals we discover through experimentation seem to support a certain moral realism, in another, given their diversity they seem to reflect the particular moral theories dependent on their tradition or culture, which would support an antirealist position.

The question of moral realism is beyond this paper’s scope, but to the extent there is an ethical dimension involved there are certain points to be made. Following R. Boyd, we understand moral realism to be the doctrine that maintains the following:

Moral statements are the sorts of statements which are (or which express propositions which are) true or false (or approximately true, largely false, etc.): The truth or falsity (approximate truth…) of moral statements is largely independent of our moral opinions, theories, etc.; Ordinary canons of moral reasoning – together with ordinary canons of scientific and everyday factual reasoning- constitute, under many circumstances at least, a reliable method for obtaining and improving (approximate) moral knowledge (Boyd 1988, p. 307).

Boyd’s strategy is to show that moral methods and beliefs are much closer to our scientific methods and beliefs, in the sense of external, empirical, inter-subjective and objective, than we are inclined to believe. In this sense, Boyd gives moral intuitions the same role as intuitive judgements in science that individuals learn through study and training in the discipline.

But it seems overwhelmingly likely that scientific intuitions should be thought of as trained judgements which resemble perceptual judgements in not involving (or at least not being fully accounted for by) explicit inferences, but which resemble explicit inferences in science in depending for their reliability upon the relevant approximate truth of the explicit theories which help to determine them (Boyd 1988, p. 319).

The problem, as we saw earlier, is that economists and philosophers not only have different academic backgrounds but also different intuitions regarding what equality is, for example.

In view of the problems selected and the drift taken by the preceding paragraphs, the reader will have surmised we are heading towards a discussion of the theory’s realism. This, too, is a field that confronts philosophers and economists because of the completely different use these two groups make of the term realism. Let us once again cite Mäki:

Economists use the term for the purpose of attributing properties to their representations, such as models and their assumptions. (…) In contrast, philosophers use the term realism to denote various philosophical theses or theories, such as existence (…) relations between the words and the world (…) justified knowledge claims (…) the goals of science (…) and so on (Mäki 1998, p. 301).

We are facing a three-pronged difficulty. First, the varied definitions of realism and the different spheres of application, whereby realism has different classifications. Second, the theory’s level of abstraction and the importance of the formal component. Finally, the theory interweaves economic, political, moral and mathematical components.

The issue of realism becomes seriously complicated when an attempt is made to find a definition of realism or an appropriate classification. A differentiation should be made between ontological, epistemological, semantic, theoretical or progressive realism (Dieguez 2005), or there is one realism, but we should distinguish between the metaphysical, epistemic and semantic approach, as propounded by S. Psillos (Psillos 2003, p. 60), D. Hausman, in turn, formulates realism on the basis of four postulates:

Goals: science aims to discover the truth about its subject matter as well as to assist human practice… Truth: the claims theories make, including the claims involving unobservable, are true or false and should be true. Existence: the unobservable entities referred to by true theories exist. Knowledge: It is possible to have good reason or evidence for scientific theories, including theories that talk about unobservables (Hausman 1998, p. 191).

And Mäki distinguishes between an ontological realism, a referential one, a representational one and a veristic one. The crux of the matter is that the manner in which realism is defined will provide one response or another to the differences between natural sciences and social sciences. For example, if we consider the question of realism by posing the question: “is there a domain of objects that is independent of the mind – individual or collective – that can be expressed properly through language and reliably captured by knowledge?” (Gonzalez 1993, p. 14) The answer on the realism of the theory will depend on the definition provided on what it means that objects are independent of the human mind, although if we take that independence in the true sense of the word, there appears to be no place for realism in social sciences.¹⁰ In this paper, when considering the realism of SCT, it is affirmed that the objects the theory refers to exist and the properties attributed to them are true or tentatively true, although our interest is focused mainly on ontological realism.

The introduction to this paper indicated that social choice theory includes aspects of human choice that interweave economics, politics and ethics, and that the theory shares part of its ontology with consumer behaviour theory. The first issue regarding the realism of the theory being considered here is whether the question of realism should apply to all these aspects, or refer instead solely and exclusively to human choices and their component parts; that is, desires and beliefs or preferences and expectations. Our first question, therefore, is: what is the ontology of the theory?

If we focus solely on the economic sphere, we find that the ontology of the theory is exactly the same as that we would postulate for microeconomic theory, which in the case that concerns us would be limited to people’s system of choice and the properties advocated for it; basically, rationality as consistency and the maximisation of utility as selection criterion. Hausman provides a realistic response to the problem in hand. According to Hausman, subjective probabilities and preferences are idealised variations of the notions of desires and beliefs with which we function in everyday life. It is therefore difficult to affirm that all other human beings have desires and beliefs, yet deny there are systems for ordering subjective probabilities or preferences (Hausman 1998, p. 199). Along these very lines Mäki affirms:

the existence of the objects of the scientific realm should not be a major issue in economics. The referential realisticness of the fundamental elements of economic theories is more often than not beyond doubt: since the terms of economic theories seem to refer to entities with which economists and others are familiar on the basis of ordinary experience (Mäki 1996, p. 434).

We can, therefore, provisionally provide a realistic answer to the question of what Mäki refers to as the “ontic furniture” of our theory, in the sense that the objects the theory refers to do exist, even if it is only commonsense realism as defined by the authors. There are two important issues here: idealisation, mentioned both by Hausman and by Mäki, and the implications of commonsense realism. Commonsense realism means realism regarding the objects and representations of common sense, such as green cucumbers and fat politicians. This realism is different to the scientific realism that involves realism regarding scientific objects and their representations (Mäki 1992b, p. 174). In our analysis, in keeping with these scholars, we would have a commonsense realism regarding the behaviour of the actors that is conceptualised in terms of folk psychology, that is, as intentional actors with beliefs and desires and whose actions are governed by them. This characterisation is replaced in the theory by the rational agent, with an ordering of preferences that is reflexive, transitive and complete, with perfect information and capable of maximising a utility function. This replacement occurs through abstraction and idealisation (Mäki 1998). In fact, staying with Mäki, there are two ways in which theoretical representations stray away from commonsense representations, namely, modification and reordering. Modification means the reformulation of language expressions in terms of technical vocabulary and a formal register. Other more substantial modifications include selection, abstraction, idealisation, exaggeration,¹¹ projection, aggregation and their multiple combinations (Mäki 1996, pp. 433–436). We have referred to the minimum version of commonsense realism, the radical version means realism regarding the objects of common sense and its representations and antirealism regarding scientific objects and their representations (Mäki 1992b, p. 174). The minimum realism of common sense would affirm that individuals have beliefs and desires, while radical realism would say there are no reflexive, transitive and complete orders of preference, nor individuals that maximise utility. This may have two readings: the one posited by Mäki, when he affirms that an economist’s stance would be to consider that the supposed events involved in the agent’s behaviour are unrealistic¹² and a second reading that simply involves being antirealist regarding the rational choice theory.

The preferences of consumers are represented by the axioms of standard neoclassical theory as complete (…), reflexive (…), and transitive (…). It would be a mistake to conclude that if consumers do not have preferences with these characteristics, they do not have preferences at all. There may be other reasons to doubt the reality of preferences (along with the rest of the folk psychological realm), but, say, the intransitivity of preferences should not be one such reason. The axioms of consumer theory may refer to real entities irrespective of how these entities are represented (Mäki 1996, p. 436).

We have established the possibility of being realists, at least a minimum ontological realism, regarding the theory’s economic sphere, but what occurs as regards the political component or the ethical component? What is the “political furniture” of our theory? There are voting rules, decision-making committees, political options to decide between, distribution rules, political constitutions, Pareto optimal outcomes… On a trivial level, all these items are constructed, in the sense that if human beings did not exist they would not exist either, but that does not stop them being real. A useful distinction along these lines is the one that I. Hacking takes from J. Searle among items that are ontologically subjective but epistemologically objective (Hacking 1999, p. 22).

What happens with the theory’s moral component? Although the two cases being analysed have a constructivist flavour, care needs to be taken when referring to realism regarding social choice theory. The notions of equality or inequality analysed here pertain to the sphere of people’s beliefs regarding what is, socially speaking, a fair or preferable state. Regarding ontological realism, there is no difference between an individual ordering alternatives according to a criterion of personal interest and ordering them according to their equality, or by those that are socially fairer. As we have already noted, the theory is neutral regarding the values used in the ordering, and what’s more, this is the true performance of the theory’s characteristic axiomatic method, but R. Hardin has made a point that is relevant here. We are referring to the ordering of social states by individuals according to the relevant ordering criterion; that is, we are returning once again to the model of rational choice that underpins the theory. According to Hardin, the theory’s neutrality regarding the underlying ordering value converts the theory into a conceptual instrument rather than an explanatory one. Whereas the theory is a suitable explanatory instrument when it is based on individual self-interest, it is merely an exploratory tool when it lacks that value or is neutral among values, given that Arrow’s result is forthcoming regardless of the value used to order social states. Hardin affirms:

To a large extent, unless we can carry out a program such as Becker’s, the difference between assuming some value theory for the actors and not assuming any is the difference between explanatory and conceptual analysis in individualistic political economy. If we impute certain substantive values to the actors in our theory, we can say what behaviors or choices follow from those value commitments, If we do not, we cannot say much (Hardin 2001, p. 68).

Valorative neutrality appears to contradict the fact that social choice theory and consumer behaviour theory share part of their ontology, that is, rationality as consistency and the maximisation of utility as an ordering criterion, with it all depending on how we define the utility. The contradiction arises if the maximisation of utility is interpreted as self-interest, although in theory utility does not imply self-interest.

We should at this point assess the possibility of making a realistic reading of SCT, and we have to admit that it is difficult to maintain a realistic stance towards it. The most we have managed to establish is a commonsense realism regarding the agents that, in its radical version, is compatible with a scientific antirealism or with a heavily idealised and abstract version of the theory. The question is that with the theory characterised in these terms, the analysis also concludes in an antirealist position. Let us see why.

B. Ellis distinguishes between different types of explanations and theories as a relevant issue in the question about realism (Ellis 1985).¹³ We therefore have causal, functional, model and systemic theories as well as causal, functional, model and systemic explanation.

A model theoretic explanation is information about how (if at all) the actual behavior of some system differs from that which it should have ideally if it were not for some perturbing influences may be causing the difference. (…) Model theories define norms of behaviour against which actual behaviour may be compared and (causally) explained (Ellis 1985, p. 55).

It seems fairly clear, as we have been affirming throughout this paper, that SCT is a model theory and the type of explanations available to us are theoretical model explanations. The most important question is, according to Ellis, that realism is applied to causal theories and not to theoretical models. This is because postulating that A causes B is to accept that both A and B exist. Nevertheless, the hypothetical entities of the model theories are not the cause of anything and so there is no reason to believe that the entities they refer to exist. This would be applicable not only to the core of the theory related to the choice of the agents¹⁴ but also to the distributive mechanisms described here. These mechanisms explore the outcomes that would be forthcoming in a distributive sense applying the different notions of equality and inequality. A theory of these characteristics has no place for realism. Therefore, the question of the theory’s realism is a question about its theoretical status and its explanatory capacity. This means the theory is not explanatory, it does not shed light on the agents’ true choices, nor does it predict the results of those choices.

Both the neutrality in terms of values and the idealisation of the theory head in the direction indicated by Ellis, the theory is not explanatory, it is a useful tool, a mechanism for conceptual exploration, but its entities do not exist nor are its propositions susceptible to be true or false given their ideal nature. The conclusion, therefore, is that it is not possible to make a realistic reading of SCT, which does not imply, nonetheless, that there is no realist positioning regarding other social theories, with their status regarding realism being something that will have to be clarified on a case-by-case basis.