1.1.The Priority of Fairness Within the Framework of Unrestricted Practical Reason

5 The theory of restricted rationality on which utilitarianism notably draws leads to a focus on individuals’ well-being. Besides the technical difficulties this creates (interpersonal comparison of utilities, etc.), it is impossible to consider fairness for its own sake within this framework. The classic utilitarian thesis is well known: good is achieved when “the greatest happiness for the greatest number” holds sway (Bentham 1789). While everyone is free to rationally pursue their egoist interests, an invisible hand ensures balance is achieved in the economic sphere, just as in the moral sphere an impartial spectator (Smith 1759) ensures harmony between altruisms. All this is in direct continuity with the harmony which Leibniz supposed to exist between these monads or independent individuals (Renaut 1989). However, the freedom of the individual is now only that of “a spit which, once started, continues to move by itself,” as Kant ironically described it (1788), also following on from Voltaire who, in Candide, derided Leibniz’s doctrine and his best of all possible worlds, thus finding himself (in anticipation, as it were) denouncing utilitarian doctrine. Be that as it may, if individuals are content with optimizing these functions of utility (with or without moral preferences) and if—from the point of view of the collective they form—they attain the greatest possible good when the sum (or the average, see Harsanyi, 1977) of utilities is at its maximum, then the situation is fair, or cannot be more fair. Fairness is thus merely a by-product or a derivative of good. It is judged by its effects on well-being: the distribution of a resource is fair once it maximizes utility. There is one single restriction: in the sum of utilities, each must count for one. However, this condition of impartiality is rather impoverished, as it does not prevent the rights of certain individuals from being potentially sacrificed for the good of the collective. [1] This consequentialism is thoroughly challenged by a Kantian or non-restrictive theory of rationality which instead, alongside well-being, enables reflection on the ethical principles which may not be contravened if we are to talk of fairness.

6 A theory such as this is rooted in the principle that agents act or endorse an opinion for good reasons. Of these reasons, some are neutral while others are purely agent-relative (Nagel 1991). The latter relate to the agent’s economic interests or their moral view of the good. If we accept that there are different concepts of material and moral good and that, given that these are not necessarily congruent, conflicts may arise, it becomes necessary to turn to neutral reasons to find a compromise, or to bring points of view or life projects in line with each other. A reason is neutral when it is acceptable to all those it is seeking to convince. It could also be described as objective, whereas agent-relative reasons are purely subjective. Neutral reasons strive for fairness in the field of legal judgment, just as they strive for truth in the field of factual judgment. At this general level, fairness has not been treated very differently than truth (Boudon 1995a). When a concept of good is acceptable to a group of individuals, it is considered fair or reasonable to this group of individuals, since it enables them to bring their life projects or points of view in line with each other. It therefore follows—and this is vital—that fairness requires unanimous agreement on these principles.

7 Within the framework of his definition, and contrary to the utilitarian thesis, the ethical deontological position consists in accepting the priority of fairness over the good, or of the reasonable over the rational (Rawls 1980), of pure practical reason over empirical practical reason. Optimizing, restricted rationality is only one part of a broader rationality, in particular since if individuals seek to satisfy their material interests or concepts of good life, they must also justify what they think or do. Now, if this is to be a solid justification, it must draw on unanimous and primary principles. For agents, these are principles on which all should agree if they are to act or form an opinion in its place. They therefore embody what agents must do or think in order to be fair. These are convincing reasons, since their position within the framework of what is or could at any time become a public debate means that they cannot be justified by the interests or ethics of an agent who has not necessarily taken the interests or ethics of all others into account. In other words, by giving fairness priority over good, an agent adopts an ethical and non-perfectionist point of view toward the categorical imperative. He forms an opinion on the basis of neutral reasons, as the terms of this opinion are considered acceptable by all and it does not harbor any preference for any individual’s interest or concept of good life.

8 Of course, an agent may still decide to ultimately promote their own rational good above what would be reasonable for all. The resulting behavior or opinion is egoistic in the sense that the reasons are only convincing for the agent himself, or for those who share the same interests. This egoism can prove more complex if it implies a calculation, whether strategic or otherwise, regarding the good or the behavior of others. While still retaining a preference, the agent can thus attempt to include others in their interest by offering them a contract of mutual interest. Nevertheless, for as long as the reasons cited do not attempt to convince everyone, the contract in question remains a coalition against at least one individual whom we are prepared to sacrifice. This attitude may take various forms, notably depending on whether others are an end or a means. The sympathetic show themselves to be sensitive to the reasons of those who do not benefit, or barely so, from the contract. They are affected, on an emotional level at least, by the situation of these others, and it would not necessarily be contrary to their “best” interest for this situation to improve. However, bearing in mind their own other interests this would force them to abandon, they are not prepared to become personally involved. Cynics, on the other hand, are happy to contribute, but only to maintain the terms of the contract or even weight them in their favor. Under the guise of altruism, compassion and cynicism turn out to ultimately be second-level egoisms. It is for this reason that we sometimes make reference to “rational altruism” (Ballet 1998). Rational choice theory does a good job of explaining these simple and sophisticated forms of egoistic behavior, but it is clear that by confining itself to these, it invites us to reduce practical reason to the sphere of the pragmatic, or to the subject’s choice to prioritize “hypothetical imperatives” (Kant 1785).

9 For fairness is only the accidental optimization of a good and, unlike egoism, altruism (pure or actual) can no longer be excluded from the field of rationality. Admittedly, altruists do not seek to convince anyone to follow them, but instead, decide to favor another out of personal ethical conviction. They adopt sacrificial, angelic, or supererogatory behavior, without expecting others to do the same (a pure gift which does not demand reciprocity). Unlike the utilitarian way of thinking, this behavior is only “irrational” to the same degree as pure egoism, since neither is based on reasons which are convincing to all. All in all, and as regards reasons in a public debate, there are three major types of behavior or opinion (egoistic, altruistic, fair), deduced from the three key areas of practical reason (pragmatic, ethical, and moral, to draw on the neo-Kantian terminology and approach used by Habermas, 1991), which absolutely must be taken into account a priori if we are to avoid any drastic and ultimately normative reduction of rationality.

10 A typology such as this is a vital tool in debating whether a particular moral opinion pertains to an optimization of good, or seeks to submit itself to an ethical principle of justice; but how can this second orientation be acknowledged in concrete terms?

1.2. Liberalism in the Face of Perfectionism

11 If every individual had the same interests or the same moral concept of good, there would be very few difficulties. These arise as soon as there are divergent and potentially conflicting ethical orientations or interests, as is undoubtedly the case in mass society or between different communities. In principle, the Kantian ethical position accepts that there can be a plurality of concepts of good. More generally, the label “liberal” is given to those with a philosophical or moral attitude which accepts this plurality. Perfectionism, on the other hand, rejects such liberalism, consisting instead of the idea that there is only one single good. As a consequence, if there are empirically multiple concepts of good, some will be erroneous. Either they eventually discover this by better matching their judgment to their moral intuition (pure intuitionism), or we can help them to do so, and there are two ways in which this can be done. The easy way offers a method (namely dialectics) for discovering good itself. One will recognize Platonism here, which moreover considers this good to exist independently of us. The hard way is quite simply coercion, consisting in forcing an individual’s hand. But whichever method is used, fairness here comes down to providing adequate recompense for virtue, or to ensuring that each person disposes of the good of which it is in their “actual” interest to dispose, even if they do not wish to do so.

12 There is no need to devote long discussion to the coercion solution, for as long as the weakest are merely forced to accept the reason of the strongest, there is no unanimous agreement and consequently no fairness. As has already been keenly observed by Rousseau (1762), the so-called “right” of the strongest is quite simply not a right. It is a simple situation of fact. And yet, returning to fact, it is exactly these perfectionist, non-coercive concepts which present the greater difficulty.

13 Only one of two things can be true: either there is empirically one single concept of good, and thus perfectionism is neutral, or there are multiple concepts which can conflict with each other, and perfectionism consequently amounts to imposing a particular concept, bringing us back to the case of constraint—that is, to seeking agreement on this concept, which seems to reflect an ethical position. However, in a situation of conflict it is not likely that one party would accept the other’s point of view as is without resorting to force. Finding agreement means negotiating a compromise after which all involved have in some way reduced their initial maximum requirements. While this is perfectly possible within an ethical framework, since a reasonable concept of good has no reason to tally with a solely rational concept, it is impossible for perfectionists who are obliged to build on non-neutral reasons to remain coherent. In fact, perfectionists already need to discuss rational means of attaining the “true” good life, whereas liberals are still asking how to reasonably match together different life projects, in other words by taking care only to restrict a liberty if it impedes another. Thus only the latter discuss fairness. The former discuss ways of achieving a good which they hope will be desired by all. However, it is impossible to attain this without banishing projects which do not develop this good, and a theory which thus ends up denying rational life projects also denies the moral personality of the individuals to whom these projects belong. [2]

14 This helps to explain why there is some doubt regarding perfectionism. Perfectionists can always be suspected of partiality. A perfectionist judge, if he wishes to remove any suspicion of doubt among those who are answerable to him, has no other solution than to convince them that he alone is totally in the right. Empirically, as soon as there are multiple rational concepts of good, the exercise is so difficult that there is a significant chance of failure, thus leaving him in a partial position in the eyes of all those who do not share his opinion. When it comes to resolving conflict, this is quite simply not resolved on merit. On the contrary, justice as unanimous agreement on what is fair results in principles (whether standards or rules of law) which, since they are by definition shared by all, enable all judges who so wish to easily remain fair.

1.3. Justice as Fairness

15 Fairness thus plays an important role in recognizing a fair judgment, and will represent our first criterion. It should be noted that, on a procedural level, impartiality amounts to the same thing as fairness or reciprocity. The rule is as follows: similar cases must be dealt with in a similar way. A judgment made on the basis of a pure a priori bias cannot be in accordance with this rule, just as, conversely an unfair judgment is formally identical to what it would have been had it resulted from a bias. Where the principle of justice only defines a single equivalence class, that of all those which are subject to it in identical fashion, this becomes evident. This remains true, although less simple, where the principle of justice defines multiple equivalence classes, which is possible wherever there is unanimous agreement on this division. An equivalence class is made up of all individuals who have the same rights and responsibilities. As a result, the rule of reciprocity (“what is good for one is good for the other”) applies here. To simplify things, let us suppose that unanimity ends up creating only two equivalence classes for individuals with the same rights and responsibilities—as in the highway code, with those who approach a crossroads from the right and those who approach from the left. If there is an accident at this crossroads, it would certainly not be fair to consider the rights of those approaching from the right as the same as those approaching from the left. Impartiality or fairness instead consists of respecting the differences in rights and responsibilities between the different equivalence classes, as this is the only way in which similar cases can actually be dealt with in a similar manner. Not being impartial or fair means denying the principle of justice in question, and thus the unanimous agreement regarding this principle. In short, impartiality or fairness is merely the principle of justice in action, and it could be said that an individual who bases their opinions or action on a quest for fairness is implementing a principle of justice.

16 However, there is no resolution of the conflict between a partial judge (whom we above supposed to be a perfectionist) and those who do not agree with him. In fact, this conflict stems from the fact that the judge and the others do not create the same number of equivalence classes for the problem in question. For example, the judge sees two classes within his concept of a good life, namely the virtuous and the unvirtuous, while for the others there is only one single class. They deny that his criterion for distinction should apply in the context in question. And yet each side considers themselves to be impartial—and in fact they are, provided that they offer no more and no less than what their principle of justice can offer to both parties. In order to escape this confrontation between impartialities, we must find a second criterion for recognizing an opinion based on fairness.

1.4. A Spectator’s Position

17 The basic idea is a relatively simple, even if it has been adapted into fairly different forms in politics and moral philosophy. A “spectator,” i.e., an individual whose interest or bias for a certain concept of a good life are not at issue in a given conflict, can adopt a neutral, impartial, or fair point of view. In long-lasting conflicts, therefore, a mediator is often sought with no interest or preference for the good in question, as this ensures that they will adhere solely to a rule of justice that is not relative to a good and, in doing so, enables them to bring points of view together on neutral reasons, thus ultimately enabling unanimous agreement to emerge. However, there are many cases where it would be impossible to turn to an effective mediator. An agent who ponders his reasons can of course seek external impartial advice, but may also consider his own point of view from an impersonal perspective, by way of an effort of abstraction, in order to place himself in the shoes of a spectator submitting his opinion to the test of unanimity. In short, an objective point of view is not only possible in the field of knowledge—it is also possible in the field of morality, and by using an identical process of abstraction. It is simply a case of considering your own rational life project or point of view as one among many, no longer identifying with it and thus regarding it as being as applicable as any other, as a pure spectator will do spontaneously. Under these conditions, there remains basically only one neutral fairness able to guide reasoning. As Raymond Boudon notes (2000), in this position all involved attempt to base their reasons on solid justification, such that they can be endorsed by any other but also by all, placing us squarely within the sphere of fairness.

18 Of course, this assumes that an agent can make judgments at a distance from himself, in other words, that he is able to put his own interests to one side (as well as his preferences on both economic and moral good). This process of setting aside can only be achieved via a paradigm (in particular, a “Kantian” one) that stipulates the autonomy of the subject, and cannot conversely be achieved via the paradigm of homo economicus, where the agent is limited to viewing their own interests only, or the paradigm of homo ideologicus, where the agent is totally bound by superstructures which hang over him and also worm their way into his thoughts (via an obscure mechanism).

19 Speaking of a fair spectator or, in other words, a “reasonable and impartial spectator” (Kant 1785), it is clear that we are absolutely not considering this concept in a utilitarian sense. That would introduce something rather contradictory to our approach here; and in any case, there is no point in making additional psychological assumptions in order to define impartiality. In addition, we are not taking the approach of either David Hume (1740) or Adam Smith (1759), who construct impartiality on the basis of natural sympathy. Their approach prevents any consideration of fairness, since sympathy denies any conflict, and thus the conditions under which fairness comes into play. For Hume, individuals enjoy the fact that there is enjoyment for others. Smith is a little more complex, but his position does not change the basis of the problem in any way (the spectator enjoys the fact that individuals sympathize). Sympathy creates depersonalization, in the sense that others’ utilities become part of my utility as a spectator. Taken to the extreme, this line of reasoning results in defining impartial judgment as judgment which maximizes the total of all utilities, weighted equally. In addition, as noted by Spencer (1880), the sympathizer’s altruism cannot ever be pure altruism. It will always conceal a level of egoism, of self-love, as has been more recently observed by Dupuy (1997). Pleasure in another’s pleasure is an eternal mise en abyme which is only halted when confronted with egoism. It is a matter of a narrowly rational altruism.

20 Our spectator is more like that of John Rawls (1971), an agent who hides behind a “veil of ignorance” and who thus no longer has any sympathy for anyone, or even—we would say—for himself. In order to define what is fair, this spectator can no longer rely on any good whatsoever, however noble, effective, or ideal it may be for a perfectionist. He can no longer turn to neutral reasons. In no way does he seek fairness and impartiality as a good. Fairness is thus merely a means toward a perfectionist good. A fair spectator, on the other hand, is in a position where he has no other choice but to consider fairness as an end. Describing him as behind a “veil of ignorance” is for us a way of demonstrating that he only reflects upon a practical principle which is generalized, i.e., not linked to particular circumstances, and universalized, i.e., not linked to a specific individual and thus a specific good. He then asks whether this principle can become a “law,” in other words (building on Kant’s approach), whether it is something which everyone could logically and practically wish to be applicable to all in its general and universal form. If it passes this test of universal agreement (which is fundamentally a test of noncontradiction, although we cannot develop this point further here), it is the principle of a justice where fairness is totally neutral. In a nutshell, a fair spectator is not attempting to promote an ethic of fairness. By virtue of his position alone, he can see that justice as fairness is an end and a priority (or a categorical imperative). Turning to the position of a fair spectator is ultimately nothing more than a way of achieving this “reign of ends” (Kant 1785) where neither fairness nor the individual are simple means which can be sacrificed to optimize a good, since they are always also ends in themselves.

21 In order to understand all that separates this point of view from a utilitarian one, we can also draw on Rousseau’s distinction (1762) between the will of all and the general will. As he wrote in The Social Contract, “There is often a great deal of difference between the will of all and the general will; the latter considers only the common interest, while the former takes private interest into account, and is no more than the sum of particular wills” (CS II, 3). If an individual can only reason according to their own interest, there can be no general will, but only a mere counting of specific wills (although this “mere” brings in the paradoxes brought to light by Condorcet and more recently developed by Arrow regarding the lack of an unequivocal counting procedure for determining the wishes of a group), leaving us with no other option than a utilitarian perspective. The direction chosen by Rousseau does not call for a counting of this kind (“where do one hundred who want a master get the right to vote for ten who do not?” CS I, 5). The Social Contract does not suggest constructing general will via calculation, but instead recognizing it on the basis of a particular position (original position) from which it is easier to see what is fair—in other words, ultimately, from the position of a fair spectator.

22 Like the Kantian distinction between the reasonable and the rational, the distinction proposed by Rousseau offers a framework for interpreting opinion. It is clear that today’s analysis of opinion polls is often restricted to locating the “will of all,” and simply indicates to which particular family of interests or ideological family it belongs. While we are not restricted to this reading, the question remains of how we understand “general will,” or fairness in the sense defined above, or this voice of the fair spectator. As Rousseau also observes, “Each individual can, as a man, have a private will contrary to or different from the general will that he has as a citizen; his private interest can speak to him in an entirely different manner than the common interest” (CS I, 7). How can we be sure that it is the “citizen” speaking? We know the abuses of tyrants who claim to be the only citizen among men.

23 There are two possible solutions. The first, the one envisaged by Rousseau, consists of only considering general rules, “because there is not a man who does not think of ‘each’ as meaning him” (CS II, 4); the second, which could be perfectly implemented by the person analyzing a poll, consists of paying better attention to the opinion of those whose interests are not at issue, namely spectators, or those who are speaking contrary to their interests, namely committed individuals. As far as polls are concerned, forming the hypothesis of a fair spectator means developing a method of reading within which, even if individual interests typically have a strong influence on the opinions of partial agents, an impartiality that is nonperfectionist is nonetheless present, and constitutes an effective motive for opinion.

24 It is precisely this which we will attempt to demonstrate, taking the example of a statistical opinion poll. Our analysis will be split into three stages. We will initially select a broad range of opinions from the poll in an attempt to establish the overall logic and detect whether the economic interest accounts for this logic, or if it seems that there is also a need to turn to the moral dimension and a certain requirement for fairness. The second stage will consist of demonstrating that, for a given opinion, and all other things being equal, not only is a moral model at work here, but it is vital to explain this opinion. Finally, after establishing that this model is linked to the requirements of fairness, all that remains is to confirm that—above and beyond all the good reasons of partial agents—this fairness (when it belongs to a spectator) constitutes the horizon, or, in other words, that opinion tends towards justice and neutral fairness when economic interests and conceptions of the good life are put to one side.

2.1. Judgments on What Is and Should Be

27 A factor analysis of the questionnaire initially reveals that, if opinion is structured, it is only so in a simplistic manner. The first main axis only accounts for 5.6% of the inertia. We must go as far as the 18th axis to reach 50%, and the 40th to account for 90% of the inertia. [3] This does not indicate that French opinions are incoherent, but instead that those surveyed did not establish a single interpretive framework which they knowingly drew upon throughout the interview. It seems that they considered each subset of questions afresh and as new.

28 The first factorial axis (horizontal in chart 1) offers a reasonably clear representation of an opposition between those who consider French society to be reasonably fair and those who do not. The former are reasonably confident in the future, both their own and that of society in general; their interpretation of the French social situation is generally positive, and they mostly consider that inequalities and exclusion are almost constantly dropping. They are also more likely to place greater trust in the state than in individuals. In the political sphere, this satisfaction translates as a certain level of conservatism: here the minority states that “French society should be kept as a state,” with even 22% stating that “there is the right amount of state intervention in economic and social affairs.” The other sociodemographic variables show that such people are more likely to be graduates, professionals, or senior management with a high level of earnings and living in Paris.

29 At the other end of this first axis we can observe the opposite responses. We primarily find those with a pessimistic view of their future and the future of subsequent generations, those wishing for radical reforms, and those who think that France’s general level of health has deteriorated over recent years. This third opinion, which is essentially held by older French people, proves far to be much more structuring that the method of identifying those aged 65 and older. We should also add—although such responses are less characteristic in France since they are almost commonplace— that these people are more likely to say that the level of state intervention is not sufficient, that the level of social protection funding is insufficient, society is unfair, or that there has been no exaggeration of the issues of major health crises (mad cow disease, dioxin in chickens, asbestos, etc.). They are also more likely to state that exclusion can be explained by the fact that there is no longer enough work for all, an opinion which goes hand in hand with the assertion that there are too many immigrants in France. The population categories appearing most clearly here are those in insecure situations (single parents below the minimum wage, the unemployed, pensioners not in work, etc.) as well as those with fewer qualifications, the less rich, and those who are older.

30 The second axis (vertical in chart 1) accounts for 4.2% of inertia and displays a dual contrast. Firstly, it reflects an opposition that appears in numerous other surveys (Forsé 1999) between an interventionist vision and an economically liberal vision. The liberal pole is fairly unambiguous. The individuals belonging to this pole (at the lower end of the chart) clearly think very clearly, in a way that marks them out, that social security funding is excessive and that state intervention in economic and social affairs is too great. They consider the monopoly of social security to be a bad thing, and believe that the current retirement system should be replaced with insurance or a system of individual savings. We also find the view that those who are excluded are people who do not wish to work, meaning that we should be more selective in allocating aid. Solidarity is not rejected, but should be accompanied by a quid pro quo from the beneficiaries. The opposite interventionist pole is trickier to interpret, as the three ways of thinking—“too much state intervention,” “sufficient state intervention,” and “not enough state intervention”—do not follow the expected order along the vertical axis, with the latter two of these opinions appearing in around the same place. Those who are excluded are not viewed as people demonstrating bad faith or who do not wish to work. The most distinctive responses, which are thus most relevant in describing the interventionist sphere, particularly stress the idea of unconditional solidarity: for example, a refusal to require the unemployed to perform community work or a refusal to require a quid pro quo for a minimum income program.

31 The second axis displays the customary opposition between economic liberalism and “cultural liberalism” (Grunberg and Schweisguth 1990). Moreover, we are talking more of a moral liberalism here, since on the one hand we primarily locate this position using opinions that involve a moral dimension, and, on the other hand, because it is substantially morals that are at issue here rather than cultures in the anthropological sense. In principle, morally liberal individuals (found at the higher and of the factorial chart) reject any differentiation norms which take into account ethnic criteria or sexual practices, such as homosexuality (73%). Moral liberalism corresponds to a refusal to judge anyone’s personal life, to the idea that every individual has the right to conduct the life projects that they wish provided that they respect those of others. This position can even go as far as denying that the economic sphere may constitute an obstacle to these life projects. The opposite perspective on the vertical axis, however, makes a judgment on the link between economic resources and these life projects. On the one hand we have individuals who consider economic resources to be practically a right “to live,” a right to conduct their own projects well; and, on the other, those who consider these resources to primarily be remuneration for merit (or risk)—personal development versus merit, in a manner of speaking.

32 On the whole, the superimposing of two contrasts on the same factorial axis demonstrates the high degree of congruity between economic and moral opinion. Moral particularism is associated with economic liberalism, as it results in a greater confidence in one’s own standards rather than those of a higher authority, when this authority is considered not to sufficiently ensure selectivity. Conversely, the refusal of a selective (or overly selective) norm, based on moral principles rejecting discrimination on the basis of lifestyle, tends to seek solidarity for all, with no selection, including in the economic sphere. Table 1 offers an especially indicative example among many others. Having said that, while (as we have already mentioned) economic liberalism and moral particularism form a fairly unimodal pole, the interventionism associated with universalism splits into two, as it is possible to consider the situation either satisfactory (since there is sufficient state intervention in economic and social affairs) or unsatisfactory (in particular due to insufficient state intervention).

Table 1

Moral Opinion and Unconditional Solidarity (in %)

Homosexuals are no different to others	The unemployed should be asked to participate in community work
Yes	No	Total
Generally agree	75.7	24.3	100.00
Generally disagree	83.8	16.2	100.00
All	77.9	22.1	100.00
Chi-squared = 28; degree of freedom (d.f.) = 1; probability of higher chi-squared (p) < 0.0001

Moral Opinion and Unconditional Solidarity (in %)

33 This prompts an initial overall interpretation using the opposition between insiders and outsiders—in other words, between those who are well covered by the welfare state and those who are instead on the margins of its benefits. The first axis offers a clearer indication of this feeling of being or not being an insider. This is a degree of satisfaction regarding the deployment of social aid, a judgment on the gap between needs and aid actually received. The second axis displays the same opposition, but reformulated as a judgment on what should be the case. The economically liberal pole puts forward the idea that we should be more selective in assigning aid, i.e., that there should be fewer insiders and they should be “better” selected, while the interventionist or morally liberal pole rejects the very concept of selection, suggesting that state aid should be deployed to the benefit of all. It should be noted that this opposition between restricted and extended solidarity is reflected in a factor analysis carried out on a survey performed in Belgium in 1993, containing some questions fairly similar to those used here (Jacquemain 1995).

2.2. Three Opinion Classes

34 We will not be studying the third axis, which is more difficult to interpret. It may demonstrate an opposition between radicals and moderates, but it is uncertain. To venture further, we will create an ascending hierarchical classification (using Ward’s method) on the basis of the first five factorial axes. This approach consists of grouping individuals together according to their proximity in terms of opinion (judged by their factorial scores). It should be noted that by confining ourselves to the first five axes, representing 20% of total variance, we are indeed favoring more structuring opinions. The others are filtered out by virtue of their lack of relevance; they are considered as “white noise” and thus as inadequate for establishing affinities.

35 We will be using a classification in three groups, as this number appears to be both necessary and sufficient to correctly describe opinion structure. These three groups are rather well situated around the three points of the triangle formed by the three viewpoints of “too much state intervention,” “sufficient state intervention,” and “not enough state intervention” (see chart 1). Cross-tabulated results (chi-squared = 599, d.f. = 4, p < 0.0001) confirm that the first group has a higher than average trend towards “not enough state intervention,” while the second considers that it is sufficient, and the third that there is too much state intervention in economic and social affairs.

Chart 1

Extract From Factor Analysis and Meaning of Axes

Extract From Factor Analysis and Meaning of Axes

36 Note for the reader: without significant magnification, the chart containing all points (over a hundred) is extremely difficult to read in this factorial design. To mitigate this issue, only a smaller number of points are shown here, although without hiding the logic of the overall analysis. The meanings which can be attributed to the axes are written in italics. The points corresponding to opinions are only active within the formation of these axes. The others, describing sociodemographic positions, are merely illustrative. To facilitate location of the three typological classes, their projections (as additional variables) are linked using lines.

37 The underlying logic of these three groups is fairly easy to find. The first (at the top left of the chart) considers itself to be not well protected by the state, and would like both an increase in the state budget and a reallocation targeted more in their favor. To some extent, the second (at the top right) denies inequalities, without seeing them. The idea that economic resources could be a hindrance to life projects is strongly rejected. At the same time, they are not far from thinking that current benefits are a fair response to this principle. Injustices, where tolerated, are considered to be “residual,” with little (and not much additional expense) required to remedy them. The individuals in the third group (at the bottom of the chart) consider themselves victims less of the economic system than of the redistribution system—the “cash cows” of the state, according to caricature. If there is “too much state intervention,” the important thing is to make the “right” choice between outsiders and insiders. Unsurprisingly, they are more likely to state that they are “happy to accept a reduction in social benefits with a corresponding reduction in their contributions and taxes.” More surprisingly, however, they are also more likely to claim they are “happy to pay greater contributions to maintain the level of social benefits.” They undoubtedly have the strong feeling that the social protection system depends on them and that what they pay comes back to them in equal measure, hence this insistence on selection.

38 In short, at the top left of the factorial chart we have people primarily living in an insecure situation, outsiders (neither selective nor satisfied) who would like to become insiders. At the top right, we find non-selective or universalist insiders who are reasonably satisfied with their current protection and convinced that there is something for everybody without major system change, i.e. there is no need to split hairs. At the bottom we have selective or particularist insiders who have no desire to allow outsiders in and who are very concerned with the distribution of the “cake.” They are generally unsatisfied with the system for which they are close to desiring radical change, as some outsiders also do, but in an entirely different economically liberal sense.

39 These three groups can be distinguished from one another, not simply by their universalist or particularist moral position or their interventionist or liberal economic opinion, but also by their social position. Sociodemographically, outsiders are more often women aged between 30 and 50 or over 65 and living in a medium-sized town or rural community; they may be white or blue collar employees, but this is often part-time, and they are more likely to be looking for work; they have low income and fewer qualifications. Universalist insiders are generally male, often young, living in Paris or another city, with a high level of qualifications and income; they most often work in the public sector as senior management or even in intermediate occupations. Selective insiders are also generally men, but middle-aged and living in small towns; they are generally self-employed with no employer or, if employed, often blue-collar workers in the private sector; they have a low or average qualification level. However, the income gaps within this group are not significant.

40 It is important to highlight the fact that the chi-squared distributions associated with all the tables linking typological classes and sociodemographic variables are significant. These demonstrate that this classification is more than just a good summary indicator of the opinions emerging in the analysis; it also indicates the social position measured against common variables. Economic opinion, already linked to moral opinion, seems to be strongly influenced by social position, which gives rise to divergent interests and perceptions.

41 There is no question that this is a matter of two partial agents whose judgments are subject to positional biases, which moreover themselves involve cognitive biases. This in no way means that the partial agent does not have good reasons for thinking what they think. Given their social and cognitive context, it is entirely rational that they endorse the economic opinion that they do. They are rational agents, but this rationality is a situated rationality. In other words, their opinion is the result of a calculation of interest as they understand it. Outsiders have an interest in expansion of state aid, as they can hope to receive additional benefit. Universalist insiders have no interest in this aid being either increased or reduced, as they are already well protected. They will not gain much from an increase which would also result in their paying greater contributions. Besides the fact that it would clash with their universalist morals, a decrease would also potentially affect and come back to them. Selective insiders, who are often self-employed, feel burdened (if only because in such cases they pay both the employer’s and employee’s share of contributions, and are subject to professional taxes on top of other taxes, etc.) by levies which weigh upon their professional activities and net income. Quite naturally, they would like these to decrease. It is not necessarily a matter of removing all allowances, [4] but of moving towards a more selective distribution which is both more parsimonious and more in accordance with their moral principles. This is in their (short term) interest as they see it.

2.3.A Shift Towards Universalism As Immediate Economic Interest Decreases

42 Does this mean that there is no place here for a fair spectator? Universalists, like particularists, can certainly argue that their economic opinion is favorable to the common good, and that by adopting it they are favoring this good over their own interest. The former may think that the benefits of growth should be subject to a more egalitarian distribution, that no-one should be left out, as aid and benefits of all kinds support demand, which is itself the key to common economic wealth. Particularists, on the other hand, may argue that a reduction in expenditure, debts, or taxes will benefit all since it favors economic activity and thus employment, wealth, and so on. However, resorting to such theories strongly resembles a form of a posteriori rationalization barely separable from the short-term subjective interest of all those involved. In terms of morals, we are confronted with an altruism which is difficult to detach from self-concern: common interest (or the interest of others) is much better served if, in either case, it comes down to serving self-interest. There is little place for impartiality here.

43 However, the position of universalist insiders seems much more ambivalent. Admittedly, we have observed that they have a greater interest in the status quo, but one question still remains: how can we account for these rather solidaristic opinions on topics which barely affect them on a personal level? Why are their coordinates on the second axis as high as those for outsiders? Why, for example, are they in favor of a minimum income program (in France called RMI) without obligation, or an RMI extension to those aged under twenty-five, or even an increase of the amount (see table 2), despite the fact that they, their children, and their loved ones are in all likelihood not going to benefit from it (see table 3)? And furthermore, if they feel that there is a risk, why do they not undertake to establish selection criteria favorable only to them? Their “best” interest does not seem to be sufficient in explaining their opinions. Of course, if this interest goes against their moral universalism, the latter would most likely be diminished. Nevertheless, a reading such as this only implicitly justifies their moral convictions regarding what is fair.

Table 2

Typological Class and Opinion on the Level of the French Minimum Income Program, or RMI (in %)

	Opinion on the level of social welfare
	The level of RMI should be increased	The level of RMI should be reduced	RMI is at a good level	Total
Outsider	80.0	0.9	19.1	100.0
Universalist insider	78.7	1.5	19.8	100.0
Selective insider	70.1	6.2	23.7	100.0
All	75.9	3.1	21.0	100.0
Chi-squared = 94.16; d.f. = 4; p < 0.0001.

Typological Class and Opinion on the Level of the French Minimum Income Program, or RMI (in %)

Table 3

Typological Class and Personal Knowledge of Someone Receiving RMI (Including the Person Being Surveyed) (in %)

	Do you know anyone receiving RMI?
	Yes	No	Total
Outsider	42.5	57.5	100.0
Universalist insider	35.6	64.4	100.0
Selective insider	40.1	59.9	100.0
All	40.0	60.0	100.0
Chi-squared = 11.23; d.f. = 2; p = 0.004.

Typological Class and Personal Knowledge of Someone Receiving RMI (Including the Person Being Surveyed) (in %)

44 There seem to be three possible interpretations. Firstly, given that an economic interest is not directly in play within the questions being asked, the person being surveyed can display their “passion for equality”—Tocqueville’s democratic passion if ever there was one—without attempting to impose it, in order to substantiate their opinion with neutral (and thus fair) reasons. Secondly, this moral judgment remains at work, but skewed by their social position: their proximity to and satisfaction regarding the welfare state prompts them to support its universalist mission. Finally, we must turn to a Marxist interpretation of ideology: universalist opinions on questions which barely concern such satisfied people are simply part of a rhetorical or cosmetic veneer concealing their particularist interests. Multiple correspondence analysis does not of course enable us to decide between these three options, so we will have to turn to other methods.

45 To end this descriptive section, we should emphasize that the remarkable macroscopic congruity between moral judgment, economic opinion, and social position emerging here would not be quite so easily observed if it were not based on solid microsociological realities. If we accept that the ternary logic of the opinions we have arrived at is a correct summary of the various attitudes relating to the welfare state and its programs, it may be inferred that the underlying rationality may not be as “uniform” in principle as a restrictive theory such as that of homo economicus might lead us to believe. Once interest is strongly involved, there is no question that this rationality comes from a traditional optimizing utilitarianism. On the flip side, it is at least in part based on convictions regarding what is fair which do not serve an agent’s particular interests, as suggested by one of the options in the previous paragraph, meaning that optimization can no longer form part of the discussion. Since the opinion of such emerging non-partial agents is not irrational, it is important that we understand what justifies it.

3.1. Conventional Interpretations of Opinion

47 A frequent treatment of opinion polls consists of not taking opinion seriously, instead focusing on variables referring to the supposed context in which the agent is embedded. On a statistical level, therefore, what is tested are various well-known “mechanisms.” To explain the fact that individuals are in favor (or otherwise) of increasing RMI, we thus verify that the risk of one day having to rely on it weighs the scales in favor of an increase. It is quite easy to identify this calculation of interest on the basis of variables such as the respondent’s assets and level of income. This reasoning can also be extended to other sociodemographic variables, such as their being employed.

48 Besides mechanisms in terms of interest, “ideological” factors are often considered. Political positioning generally plays this role, but this information is lacking in the survey being analyzed here. Nevertheless, we can partly compensate for this lack using a characteristic theme or, if our interpretation of the preceding MCA is correct, using the second axis which was a good indicator of economic liberalism. In any case, it is more interesting to shed light on the cognitive factors which are generally hastily subsumed under the heading of “ideology.” Thus, being situated at a certain point on the social pyramid results in a relatively detailed knowledge of this section of the pyramid, and also results in generalizing this for the rest of society, i.e., by overestimating its weight or importance. Likewise, it should be borne in mind that our opinions are formed on the basis of what we see and hear around us. The group to which we belong can also play both a cognitive and normative role. Whether this is a group of loved ones or a wider group, such as officials, or the “left family” versus the “right family,” opinion may reflect not a judgment but a claim to belonging. Alternatively, it may be an attempt to mark oneself out: in the previous section, we saw that selective insiders are particularly concerned with emphasizing their differences from outsiders, and do all that they can to ensure that the barrier remains in place.

49 The opinions of “embedded” agents can thus be easily modeled using these mechanisms. However, if we wish to avoid arbitrarily excluding anything, we should consider that individuals follow moral principles reflecting either a concept of good life or a concept of fairness. In order to empirically demonstrate this effect, we must formulate hypotheses regarding the moral principles upon which those surveyed draw in order to answer the questions posed to them, then link these hypotheses to the variables of the survey, and finally demonstrate the explanatory power of this dimension for a question such as the level of RMI. In other words, we must reconstruct the implicit moral argument and demonstrate its effectiveness.

50 These hypotheses explaining opinion are of course not mutually exclusive—we can consider them to be all at work. If we accept this causal pluralism, it tempting to assess the explanatory weight of each of these hypotheses using a logistic regression. This will enable us to ascertain for the specific case in question whether the moral dimension plays a marginal role vis-à-vis various forms of interest and, thus, whether ignoring it will have little consequence or, alternatively, will represent a significant loss in understanding the opinion. However, it is important to specify that there are two limits to this approach. First of all, it requires the construction of an ad hoc model for the moral reasons that come into play for a given question. In order to validate the importance of the moral dimension in general, this model cannot be transposed as is to another problem, since there is no guarantee that these principles will remain relevant to the subsequent question. The moral model which we will be presenting thus has no claim to universality. Secondly, given that we will be constructing an extremely summarized statistical modeling, some variables will remain partially ambiguous. This modeling has more of a heuristic than a definitive status, but the resulting hypotheses—in particular as regards the moral principles at work—will be subject to more advanced validation in the final part of this study.

3.2. The Two Principles at Work in the Moral Model

51 Our proposed moral model can be summed up in terms of two strong arguments which individuals must combine together in whatever way they can. What distinguishes this model from an ideological reading is the fact that these arguments are relevant. The first principle (P1) consists in saying that people refuse to sacrifice others or allow them to live in a clearly intolerable situation. This principle is strong enough to be reasonably clear. For example, it can be seen in the fact that an overwhelming majority of those surveyed in the poll condemned poverty, and those who accepted it simply wanted to say that there will always be rich and poor. The second principle (P2) holds that people refuse to allow others to live at their expense. This principle is much less obvious than the former. It does not mean that people have absolutely no spirit of charity, but instead that their wish is, in a manner of speaking, “not to be cheated.” If this principle does not seem to be as immediate, this is primarily due to the position of the welfare state in the West, and more specifically to the concern shown for the poor which this welfare state makes possible. This is reminiscent of the conclusions made by Abram de Swaan (1995, 255) in In Care of the State: “‘Something ought to be done about it.’ If there is misery, ‘it should be taken care of’: not by the beholder, but by something else, by ‘it,’ by the hidden subject of all these phrases in the passive mode: the state. The state is the abstract, universal, and anonymous caretaker of all members of society. No one is under any immediate obligation to the stranger who may implore his aid. The modern concern for the sick and the poor beyond one’s intimate circle can be more intense, because it is certain not to cost anyone too much in particular—at most a small increase in fees and taxes. And once they could afford this infinitesimal share of the collective burden, people allowed themselves to be more easily moved by the sight of suffering. The price of empathy has gone down so much that even common people can afford it.”

52 The development identified by De Swaan explains why P1 is so evident and P2 passes increasingly unnoticed. Even so, the second principle remains at work, and where P1 is confronted with P2 and a strong opinion is called for, individuals favor the first principle all the more since they are convinced that the state is successfully fulfilling its mission to collectivize charity. The relationship between these two principles can also be understood by drawing on the concept of responsibility: the more collective responsibility is considered to be, the easier it is to accept P1; conversely, the more responsibility is individualized, the greater the emphasis on P2 and on the limits of charity, in order to preserve the moral element.

53 So how can we now link these principles to the empirical data? P1 must be based on a judgment regarding the minimum acceptable living standard. Those surveyed were precisely asked to offer their opinion on the minimum income they considered necessary in order to live—what is known as a “living” wage. [5] Around 99% of respondents felt that it would be impossible to live on less than €381.12 (2,500 F) per month, and 72% that RMI was too low. Apparently, some people feel that RMI should not be sufficient income to live on. Indeed, RMI is only one benefit among many used to support those in insecure situations, but this only accounts for the majority of the gap found between the desired level of RMI and the level of “living” wage granted. In fact, under the first principle, we can expect that those who claim it is impossible to live on less than €762,25 (5,000 F) are also more likely to claim that RMI is too low, compared with those who believe it is possible to live on €457,35 (3,000 F) or even €304,90 (2,000 F). If this hypothesis is correct, notwithstanding idiosyncrasies it is to be expected that the average ideal level of RMI would be higher for individuals who also give a higher figure for income necessary for (a decent quality of) life. A simple cross tabulation supports this hypothesis (see table 4).

Table 4

Opinion on the Level of RMI and “Living” Income

	RMI should be increased	RMI is at a good level	RMI should be decreased	Total
“Living” wage <€762.26 (5,000 F)	67.5 (–***)	27.6 (+***)	4.9 (+***)	100.0
€762.26 to €914.68 (5,000 to 5,999 F)	76.1	21.2	2.7	100.0
€914.69 to €1,067.13 (6,000 to 6,999 F)	78.1 (+*)	19.4	2.5	100.0
“Living” income > €1,067.14 (7,000 F)	78.5 (+***)	18.6 (–***)	2.9	100.0
All	75.8	21.1	3.1	100.0
Chi-squared = 37.14; d.f. = 6; p < 0.0001.
Note for the reader: the adjusted residual is a deviation parameter for the hypothesis of independence, in the link between two types of response, which asymptomatically conforms to a standard normal distribution. A value outside of the interval [–1.96, +1.96] thus indicates an independence deviation that is significant at the 5% level (Haberman 1973). In this table, as in those that follow, we only reproduce the sign of this residual where it is significant at the 1% (***), 5% (**), or 10% (*) bilateral level.

Opinion on the Level of RMI and “Living” Income

54 However, the “living” wage is not the only key to understanding the formation of opinions regarding RMI. If we consider this key to be a factual judgment on the basis of which P1 can be applied, we still need to find an empirical base for the second principle. The fact that many French people apparently think that RMI should be less than the “living” wage implies this principle. The hypothesis here is that individuals offer a judgment on the level of RMI which as far as possible reconciles a sense of solidarity, where all have at least a vague desire for others not to die (principle P1), with the idea that this RMI is all the same barely decent, even sufficiently indecent for others to be dissatisfied with this situation (principle P2). This internal negotiation results in a sort of ideal level of RMI, which is compared to the current level in order to decide whether RMI is too high or too low. This is a search for a satisfactory balance between providing sufficient income to live on and preventing anyone from contenting themselves with it. This arbitration depends on individuals’ convictions, either regarding the risk that the person being helped will “profit” from the situation or regarding the “unconditional” nature of this aid since it is not the “fault” of such people, or even since in some way greater importance is being attached to personal fulfillment than to merit. We are not attempting to untangle all of these reasons here. Instead, we can locate this split using the corresponding variable on the second axis of our MCA (even if, when regarded in isolation, this does not enable a decision to be reached between reason and ideology).

3.3. Estimating the Impact of Different Interpretations of Opinion

55 We now have enough elements to perform an ordinal logistic regression on the level of RMI and to assess the different factors influencing the formation of an opinion regarding this level. We have adopted the following explanatory variables: sex, age, qualifications, employment or unemployment, income, property assets, “living” wage, size of urban area, personal knowledge of someone receiving RMI, and finally—as a continuous variable—the coordinates on the second axis of the MCA. In order to enable an appreciation of the respective merits of the effects of each of these, most variables are linked with a single parameter for analysis.

56 This regression (presented in detail in the appendix) clearly shows that “living” wage and the second coordinate of the MCA (hereafter labeled F2) are the most determining factors. Our model thus appears to be a particularly relevant one. It should be emphasized that opinion regarding the level of RMI was simply an ideological consideration, and the effect of “living” wage was not designed to come into question. In so far as there is such an effect, i.e. in so far as there is an argument to be made based on this “living” wage, we can conclude that F2 is the missing link. Similarly, if (as discussed above) F2 can be viewed as an ideologically-oriented arbitration, it would seem that this cannot be its only meaning here given that opinion is formed on the basis of reasoning which combines “living” wage and a line of argument linked in some way or another to F2.

57 Before returning to this key point, let us briefly review the effects of the other variables incorporated into the regression, even if they are less important. We will therefore initially turn to assets, which can be interpreted in terms of interest or, more specifically, risk: those with a safety net have less desire than others for the level of RMI to be increased—i.e. home owners and those with free accommodation less than homebuyers, and the latter less than tenants. The effect of income is not significant, but the parameter tends is in the same direction, suggesting that the richest are more reticent than others regarding an RMI increase.

58 In addition, there is an age (or generational) effect. The proportion of opinions in favor of increasing RMI decreases with age. This is as expected: young people are more sensitive to the difficulties of integrating into the world of work. However, once people become economically active they take a less favorable view regarding such an RMI increase.

59 Finally, we can detect a moderate effect (only significant for a confidence interval of 90%) in the size of urban area and personal knowledge of someone receiving RMI. The former of these can be interpreted as a difference in living standards: those living in towns require more than those in the country. The effect of an immediate social entourage, on the other hand, is much more complex. While those receiving RMI and those with family members receiving RMI are of course in favor of increasing the level of RMI, the most reticent are not those with no personal knowledge of people receiving RMI but instead those who know such a person outside of their family. These recipients of RMI within an entourage are in all likelihood “acquaintances” or those with weak ties rather than friends, i.e., easy to stigmatize. However, the importance of this should not be underestimated: those who wish to reduce the level of RMI, while under-represented in this group, remain rare.

60 Finally, two variables appear in our model without causing any effect: qualifications and sex. As regards qualifications, even cross-tabulated results do not show any effect. By contrast, women are strongly in favor of increasing RMI. We have to assume that this opinion is due more to their over-representation among those in insecure positions than to genetics or depth psychology (which would contrast the “supportive mother” with the “warrior father”).

3.4. “Living” Wage and Risk Aversion

61 However, we should remember that all of these effects are still less important than that of “living” wage and F2 in explaining opinion regarding RMI, demonstrating the importance of a moral line of argument in the face of interpretations in terms of ideology or interest. However, can the judgment on “living” income not be integrated into these interpretations? First of all, it could be argued that individuals calculate this “living” income on the basis of their convictions regarding the very principle of RMI. Opinion regarding RMI is thus for example determined by political positioning: those on the right are more likely to be against RMI in principle, prompting them to consider that “living” income should be low as a way of reducing their “guilty conscience”—and the opposite applies to those on the left. Since the question of the minimum amount an individual requires to live is posed in the questionnaire we are analyzing, before any consideration of RMI in its proper sense, it is important to formulate a reasonably strong hypothesis regarding individuals’ ability to anticipate. In fact, it should be recognized that these two questions are so intermingled in people’s minds during their daily lives that they base their response on “living” income before any reference is made to RMI.

62 “Living” income can also be reduced to risk-based reasoning. Even though the richest have little likelihood of needing to rely on RMI, on the one hand they feel that “living” income should be increased (since, if the poor have difficulty imagining living on few means, the rich find this even more difficult to imagine), and on the other hand they are strongly risk-averse. All in all, therefore, they would like to increase their safety net. Reasoning of this kind is unstoppable when it is not provided with content, as the risk withstands any empirical results: if the rich wish to reduce the level of RMI, it is because the risk to them is low, and if the rich wish to increase the level of RMI, it is because they do not care whether the risk is low, as the relevant factor here is that they are risk-averse. They want to minimize their maximum loss, even if it is unlikely (the term used to describe this logic (Peretti-Watel 2000) is a minimax strategy).

63 How can we empirically confirm or invalidate this concept of risk-aversion? By drawing on our typology distinguishing between outsiders, selective insiders, and universalist insiders, and considering whether the last of these categories is more risk-averse than the others. In the survey, we have a question regarding confidence in the future as regards reducing poverty and exclusion. If they are pessimistic, are universalist insiders more likely to want RMI to be increased? Table 5 shows that this is not the case. For them, confidence in the future has less of an impact on opinion regarding the level of RMI than it does for outsiders and selective insiders—even their optimist or pessimist nature does not alter their attitude regarding RMI. Nevertheless, the counter-argument could be made that this question regarding confidence is too general and does not really reflect self-concern regarding the risk of poverty.

Table 5

Opinion on RMI and Confidence in the Future as Regards Poverty, by Type (in %)

		Opinion on level of RMI
Type	Poverty and exclusion are likely to…	The level of RMI should be increased	The level of RMI should be reduced	RMI is at a good level	Total	Phi Chi-squared
Outsider	Reduce	77.9	2.9 (+**)	19.2	100.0
	Increase	82.2 (+***)	0.9	16.9 (–***)	100.0	0.12
	Stay the same	68.4 (–***)	0.0	31.6 (+***)	100.0	18.03 (***)
	All	80.9	1.0	18.2	100.0
Universalist insider	Reduce	78.5	2.0	19.5	100.0
Increase	79.6	1.7	18.8	100.0	0.04
	Stay the same	76.8	1.1	22.1	100.0	1.37 (n.s.)
	All	78.6	1.7	19.8	100.0
Selective insider	Reduce	69.0	7.1	23.9	100.0
Increase	72.4 (+**)	6.6	21.0 (–***)	100.0	0.09
	Stay the same	61.5 (–**)	4.9	33.6 (+***)	100.0	11.67 (**)
	All	70.7	6.5	22.8	100.0

Opinion on RMI and Confidence in the Future as Regards Poverty, by Type (in %)

Table 6

Opinion on RMI and Concern About Poverty, by Type (in %)

		Opinion on level of RMI
Type	Personally concerned about poverty	The level of RMI should be increased	The level of RMI should be reduced	RMI is at a good level	Total	Phi Chi-squared
Outsider	Very	82.5 (+***)	0.7	16.8 (–***)	100.0
	Fairly	77.1 (–*)	1.2	21.7	100.0
	A little	75.0	1.9	23.1	100.0	0.13
	Not at all	47.6 (–***)	0.0	52.4 (+***)	100.0	22.69 (***)
	All	79.9	0.9	19.2	100.0
Universalist insider	Very	83.4 (+***)	0.6 (–**)	16.0 (–***)	100.0
Fairly	74.4 (–**)	2.2	23.4 (+**)	100.0
	A little	68.0 (–*)	6.0 (+***)	26.0	100.0	0.16
	Not at all	(55.6 (–*)	0.0	44.4 (+*)	100.0	23.03 (***)
	All	78.6	1.6	19.8	100.0
Selective insider	Very	74.2 (+***)	4.1 (–***)	21.7 (–*)	100.0
Fairly	68.6	7.3	24.1	100.0
	A little	54.6 (–***)	12.4 (+***)	33.0 (+**)	100.0	0.17
	Not at all	42.4 (–***)	21.2 (+***)	36.4 (+*)	100.0	41.08 (***)
	All	70.2	6.2	23.6	100.0

Opinion on RMI and Concern About Poverty, by Type (in %)

64 Another question proves to be more accurate, since those surveyed were also asked if they were personally concerned with poverty. Table 6 shows that in this situation, self-concern naturally plays a role in attitude towards RMI. However, the link between this concern and opinion on the level of RMI is not as strong for universalist insiders as it is for particularists, and is barely stronger than for outsiders.

65 The hypothesis that a particularly strong risk aversion among those who are protected compensates for an objectively low level of risk has therefore not been proven. This can be demonstrated in even greater detail. It would seem that universalist insiders are more likely than others to show greater concern for their own poverty than for France in general. This may be a sign of their risk aversion, but when this difference in concern regarding poverty is combined with opinion on the level of RMI, it would seem that those more concerned with themselves than with France as a whole are under-represented among those desiring a reduction in RMI. If they truly were this risk averse, the opposite should be the case.

3.5. A Moral Model with Several Possible Readings

66 Now that the “risk aversion” hypothesis can be ruled out, let us return to our “living” wage interpretation. Ultimately, the most robust hypothesis consists of combining this variable with P1, and this comes down to a reading that highlights the moral lines of argument. However, principle P2 should also be used to create an operating argumentative model. Variable F2, which is a composite and on which we have drawn to isolate the effect of this second principle, as previously mentioned also supports the hypothesis of an ideological conflict between particularists and universalists, but each protagonist makes use of good reasons within this conflict. We will demonstrate in greater detail that a holistic reading which fails to interpret F2 in this way but instead does so in terms of restricting superstructures is not empirically valid. Nevertheless, it is now extremely clear that such an interpretation does not enable a correct reading of what appears to be a moral model. Since the dimension of interest has been broadly taken into consideration within the regression, we also know that conflict between particularists and universalists draws less on motives than on reasons linked to a concept of what is good or fair. Moreover, it is not sufficient to immediately form a conclusion regarding the level of RMI, since this is only one part of the overall reasoning and should be linked with P1 in order for such a conclusion to be reached. Ultimately, we can conclude at this stage that concepts regarding what is moral or fair are in any event far from insignificant compared to interest when it comes to explaining an economic opinion such as that regarding the level of RMI. The moral model we are proposing thus has the advantage of being the only model to explain this extremely significant effect in combination with “living” wage and F2.

67 Furthermore, it also displays the key role played by neutral reasons, as compared with agent-relative reasons. Those surveyed are more reasonable than rational, in that their socioeconomic position does not automatically indicate their opinion. In particular, we have observed that the concept of economic risk was not a decisive factor. Their reasoning cannot be purely consequentialist, since an approach such as this reduces the welfare state to an insurance system where there is nothing to do but evaluate risk. The debate is wider than this, and thus undoubtedly has an ethical dimension. P2’s degree of reciprocity is not merely the result of a calculation—it is the subject of in-depth debate aiming to ensure the agreement of all, and not just those who are “well” insured.

68 Within an ethical reading such as this, the first principle of our moral model is a principle of a minimum level for resources applicable to all, in particular the very poorest, in order to judge the correct level for RMI. The second principle is a principle of reciprocity, which goes some way towards “balancing” P1. It requires reciprocity in the sense that, if I were in the position of the very poorest, I would expect the level of aid that I claim to support me but no more. Any more would be what is commonly labeled as “being cheated.” It should be noted that reciprocity imposes a limit on charity, avoiding the pitfall of angelism (which will be dealt with later in this study). The challenge is to ascertain whether this reading is the only acceptable one, since for the time being we do not know whether P1 and P2 are principles of fairness in an ethical sense or whether they are simply perfectionist principles. They apply for the entire population, all things being equal, but this exact state of affairs is only possible because they are open to multiple, divergent interpretations.

69 From the outset, let us dismiss two of these which consist of attaching an a priori orientation to the model, although for different reasons. P1 strongly resembles Rawls’s difference principle, and P2 tends towards a principle of equal liberty, or even equality of opportunities, which together (in Rawls’s opinion) make up the two key principles of justice. This potential relationship should not lead us to conclude that P1 and P2 are thus ipso facto ethical principles of justice. Furthermore, these principles do not result from reasoning over the entire distribution of income—we are restricted here to merely considering the problem of RMI in a community, and we should reiterate that the model we have found does not allow any claim to be exported into another context. In our survey we do not know whether, above a satisfactory base level (P1), individuals would, for example, seek to optimize the average or standard deviation of the distribution (Boudon 1995b). In any case, determining whether an ethical interpretation is the correct one represents the core decisive factor of the issue, and requires a solution that has been validated by empirical data. A second possible reading should also be dismissed—it consists of making P1 and P2 into principles of pure economic efficiency, presumably borrowed from some doctrine, with no link to individuals’ economic or moral interests. This brings us back to homo ideologicus, which we have already proven to have little relevance in interpreting the model. We will nevertheless demonstrate below that a holistic interpretation contradicts the data. It should incidentally be noted that while P1 can cope with moderate economic liberalism, advocating a low level of state intervention, it is equally incompatible with a totally liberal economic doctrine and classic normative utilitarianism alike. On the other hand, there is no reason to rule out an interpretation of P1 and P2 which links arguments of economic efficiency with the moral and economic interests of those involved. Such arguments are examples of the justification of common welfare, but with the particular characteristic of introducing a reality constraint (in the form of “not everything is possible”). In a manner of speaking, they result in an intermediary position between the two extremes represented by a strongly selective, particularist interpretation and a non-selective, universalist reading. Let us consider these three readings of P1 and P2 by traversing the F2 axis, without turning to a justification that stems solely from being personally concerned with RMI or knowing someone in such a situation, since the moral model holds true even when this condition is controlled in the regression. Each time, we will consider how many equivalence classes are formed by a given interpretation.

70 As was established in the opening section, fairness is unanimous agreement regarding rights and responsibilities. However, this agreement may be reached by recognizing that there exist one or many equivalence classes of individuals who share the same rights and responsibilities. More concretely, a consideration of the number of equivalence classes here leads us to question the limits of reciprocity which (from a particular perspective) could garner the support of all. This number is therefore crucial.

71 Thus, at the bottom of the F2 axis, “pure” particularists reject any plurality of concepts of the good and apply the moral model by creating (at least) two equivalence classes. Using their own personal criteria, they distinguish between the virtuous and unvirtuous, or the deserving and the undeserving. This criterion may be the fact that individuals demonstrate a desire to work, but it can go as far as making reference to lifestyle. Of course, as regards P1 this does not answer the question of knowing what happens to those who are excluded, i.e., those who (in particularists’ opinion) do not have reasonable life projects. Particularists may feel that the deserving should be the only people to benefit from public aid financed by a mandatory contribution, and that for the others it should suffice to address emergencies and offer daily aid to the charity sector. Halfway up the F2 axis we find individuals who accept a plurality of life projects. They share this moral liberalism with universalists, but also feel that there should be a limit to reciprocity (to P2) in the name of common good, understood as economic efficiency. Like “pure” particularists, they therefore also create (at least) two equivalence classes and consider some life projects to be unreasonable (laziness, for example). This distinction criterion is of course less strict, but it still falls under the auspices of perfectionism. The limits of reciprocity are no longer justified in terms of being “cheated” by helping those who live “bad” lives, but simply by those who do not make any productive effort. In their eyes, RMI is justified as a kind of insurance that favors productivity. By refusing to economically insure the undeserving in the name of economic effectiveness, they may be cynically worrying about their risk premium. However, they may also be convinced that it is fair. They are therefore thinking less of themselves than of all those considered deserving, and find themselves “obliged” to prioritize economic efficiency in life projects in order for them to be insured. Even if the term is ambiguous here, this is in the “best” interest of all. It is also in their own interest, as we cannot ignore the fact that it ultimately reduces their level of contributions. In other words, their opinion is a matter of compassion—“it is a great pity, but no more can be done”—or, if one prefers, where the criterion of economic effectiveness is held to be true, an increase in the well-being of the deserving increases their own well-being as pure externality. Finally, at the top of the F2 axis we find universalists who are pure liberals in a moral sense. As well as accepting a plurality of life projects, they also consider all to be equally reasonable as regards eligibility for RMI. They do not consider reasonable life projects to be distinct from each other via a restrictive criterion of perfection, even if this criterion matches their own idea of happiness, or in the name of economic effectiveness. They therefore only create one equivalence class. Whilst they do not consider the principle of reciprocity to translate into a limitation on a person’s freedom to choose their own rational life projects, this does not make them purely altruistic. Reciprocity simply means that if I were in the position of the very poorest, I would expect the level of aid that I claim to support (at a bare minimum above living income) but no more. Any more than this would mean “being cheated.” In other words, the reciprocity of P2 imposes an upper limit on minimum income, while leaving each person free to give more than others without reproaching others for not doing the same (supererogation).

72 All in all, where P1 and P2 are applied the first two types of individuals create two equivalence classes for somewhat different reasons. While the latter are liberal, they agree with particularists on this key point. Universalists, on the other hand, contrast with these by only recognizing one single equivalence class. And yet, all of them—particularists and universalists alike—can claim that, based on their own reasons, their reading of the moral model is the correct, impartial one set against the other partial versions, or (which amounts to the same thing) that their reasonable concept of good life is fair and the others are not.

73 In order to move beyond this observation of multiple perfectionisms, we know that we must simply consider what a spectator’s position would be. We have a moral model, a concept of the good subject to multiple different readings, but the fair one—or the one that makes P1 and P2 principles of fairness—is that of a spectator whose interest in relation to their material or moral good is not affected. Behind this “veil of ignorance,” their point of view can only draw on neutral reasons. Whether this is termed general will (Rousseau), pure practical reason (Kant), discourse ethics (Habermas), or quite simply justice as fairness (Rawls) makes little difference, as all have the same intention at heart. A spectator’s position expresses a neutral fairness that owes nothing to the particular interests of any of those involved. In this sense, this position is the only fair one. The question is therefore “simple”: how can we isolate such a spectator within our opinion data? Once this opinion has been identified, it is enough to know whether it involves forming one or many equivalence classes in order to fairly apply principles P1 and P2 (to the problem of allocating minimum resources to the very poorest).

4.1. The Effects of Morals and Economic Interest

74 In order to investigate the position of a fair spectator, we will reexamine opinions on RMI. However, we would obtain identical results to those we are about to present by examining questions posed in similar terms regarding unemployment benefit. The preceding regression can now be brought back into play, reducing the number of variables and also the number of modalities for each. Given that we are already aware of the role played by variables indicating sociodemographic position, we will no longer be paying them any attention, and for each variable—whether explanatory or to be explained—the responses will simply be grouped into two modalities. Opinion regarding the level of RMI will thus be recoded into two categories: “It should be increased” (76%) versus “It is at a good level” or “It should be reduced.” Likewise, as regards requesting a quid pro quo from benefit beneficiaries, we will simply differentiate between those who are extremely in favor or generally in favor, and all others.

75 As regards the moral dimension, this strategy leads us no longer to consider a continuous and complex variable such as F2, but instead to search for a variable that enables us to clearly distinguish between a universalist attitude and a particularist attitude. Any question designed to isolate a selective orientation—whether as regards lifestyle or in the economic domain—could be a candidate here, and, given the significant correlation between all variables, could in truth be used without greatly affecting the results. We have chosen the question regarding homosexuals (“Are they any different from others?”), as this seems a priori to be very distant from the issue of RMI. We cannot prove that the interpretation of responses to this question as indicating a universalist versus particularist moral orientation is definitively correct, but this is more than likely given the aforementioned significant correlation with the responses to other questions, implying the adoption of a universalist or particularist point of view in the economic domain and as regards lifestyles. Table 1 provides one example, that of a quid pro quo for unemployment benefit; and we can draw on another here by observing the close links between responses to this question and those regarding the ideal role of women or regarding immigrants. Cross-tabulations unambiguously show that people who consider homosexuals to be no different to others also do not think that there are too many immigrant workers, nor that women’s ideal role is to stay at home and raise their children (the chi-squared figures are extremely significant). The responses to these three questions are therefore not specific to the topic being covered: they are guided by a general moral attitude contrasting a universalist orientation with a particularist interpretation which can be safely summarized in the question regarding homosexuals.

76 We already know that F2 has its own major effect on opinions regarding RMI. Nevertheless, given the specific nature of the question that has just replaced it, it is important to verify that all things being equal this is still the case. As regards homophobia it is particularly important to verify whether the consequences of morals, measured against this new indicator, are unaffected by age (moreover, it is strictly speaking more likely to be generation rather than age that comes into play here). If this is not the case, we would have to conclude that the responses regarding homosexuality reflect a difference in appreciation between generations rather than a moral attitude in itself, in other words, one that could potentially appear in any age group. However, the result is quite clear: including age in a logistical regression has no effect on the significance of the link between attitude towards homosexuals and opinion regarding RMI. [6] The effect of this interaction is also not significant (at the 10% level). In other words, on a similar scale, universalism is the leading factor here, to roughly the same degree it is in preferences, regardless of age, for favoring an increase in RMI or for no quid pro quo.

77 Whether in terms of this increase or a quid pro quo, those receiving RMI or with a family member receiving RMI (13%, and this is generally a spouse, child, or parent, as demonstrated by other surveys or questions of this type previously posed) are of course directly affected, and there is no need for lengthy discussion on the fact that this represents an indicator of direct and immediate interest affecting the responses to these questions. [7] However, it may be assumed that receiving RMI or being close to someone receiving RMI is too restrictive to make a judgment—the prospect or risk of one day receiving or having a family member receiving RMI can also affect whether or not a person feels affected. In order to take this into account, we constructed variables indicating economic interest that also involve personally being or having a family member who is unemployed and not receiving benefit, unemployed and receiving benefit, or in an insecure situation, whether as separate variables or as a gradual accumulation of all these situations, as they are not mutually exclusive. Now, while the results are not exactly identical, the general tendency remains unchanged. Therefore, in what follows we shall limit ourselves, with no loss of generality, to summarizing the dimension of direct economic interest in terms of whether one receives or knows someone who receives RMI. Not being thus affected thus makes one a spectator in regard to this economic interest. In truth, only a “partial” spectator, since interests relating to a concept of the good life have not yet been controlled for; but in order to prevent burdening the expression we will nevertheless use the term “spectator” from this point onwards to designate someone not affected.

78 Cross tabulations confirm what we might expect in the light of the previous analyses. Just like differences in moral orientation, being or not being a spectator has a very significant effect on opinions regarding RMI. Unlike those affected or particularists, spectators or universalists would like RMI to be increased and are opposed to a quid pro quo considered restrictive for the beneficiary. For either of these questions, logistical regressions show that this extremely significant effect for morals remains when economic interest is controlled for, just as the significant effect of interest remains for identical moral positions (table 7). The effect of morals is however stronger than that of interest (as already emerged in the regression for the previous section).

Table 7

Logistical Regression on the Question: “Should those receiving RMI accept any work offered to them?”

Effect	D.f.	Chi-squared	Prob. of higher chi-squared
Those who receive or know someone who receives RMI	1	12.3	0.0004
Moral opinion	1	22.2	0.000002
Interaction between these two effects	1	8.6	0.003

Logistical Regression on the Question: “Should those receiving RMI accept any work offered to them?”

79 This extremely clear effect of morals explains why universalist insiders are higher than one might expect on the first two axes of the factor analysis previously presented. Outsiders, often in insecure situations, are affected (either personally or via someone close to them) by many of the questions selected in order to perform this factor analysis. We might therefore expect to find them in close proximity to items implying an improvement to the lot of the very poorest, who are at the top of the chart. Morals do not affect this significantly. On the other hand, universalist insiders—who have no direct interest—might have been situated at the bottom of axis 2 (vertical). Let us not forget that they are often public sector employees and/or senior management—in other words, people unlikely to be or become beneficiaries of RMI or to know someone who is. Although they appear at the same height as outsiders, this is because their universalist moral opinion leads them to prefer “solidaristic” type responses, to roughly the same extent as outsiders’ interest does. However, the question remains of whether this opinion reflects a striving for fairness, or the promotion of a concept of the good life.

4.2.A Universalist Spectator Tends to Be Committed Rather Than Compassionate

80 Among spectators, to whom we shall now limit our enquiry, so as to answer this question, universalists are unsurprisingly clearly in favor of an increase in RMI or lack of quid pro quo, while particularists take the opposite view. Significantly and predominantly, universalism is therefore linked to “solidaristic” opinion (see table 8). However, we have only eliminated the most obvious form of interest (whether or not one is personally concerned). But can we be so certain that this solidaristic opinion does not conceal a second-level egoism that desires an improvement to the good of others only because it has a positive effect on one’s own well-being? It can also be observed that it is easy to express a solidarist opinion, and that sometimes there is many a slip twixt cup and lip. An individual’s true actions do not always match what they say they believe in a survey. When it comes to the level of RMI, this means that we must consider whether the costs of any potential increase being called for have been taken into account. If the hypothesis of a “free opinion” is confirmed (among spectators), what we believe we have identified as a solidaristic response oriented by universalist morals would in fact merely be plain compassion. And at the same time, we would be reduced to the aspect of interest alone, and would be as far as ever from the position of a fair spectator.

Table 8

			Those receiving RMI should accept any work that is offered to them
Personal knowledge of someone receiving RMI			Yes	No	Total	All
Yes	Homosexuals are no different to others	Yes	79.0	21.0	100.0	74.2
	No	76.2	23.8	100.0	25.8
	All	78.3	21.7	100.0	100.0
No	Homosexuals are no different to others	Yes	82.8 (–***)	17.2 (+***)	100.0	72.5
	No	90.3 (+***)	9.7 (–***)	100.0	27.5
	All	84.8	15.2	100.0	100.0
Chi-squared of upper table = 0.44; d.f. = 1; p = 0.51; phi = 0.030; absolute value of adjusted residuals = 0.7.
Chi-squared of lower table = 28.32; d.f. = 1; p < 0.0001; phi = –0.094; absolute value of adjusted residuals = 5.3.

Those Receiving RMI Should Accept Any Work Offered To Them, by Personal Knowledge of Someone Receiving RMI (or Being Said Person) and Opinion on Homosexuals (in %)

81 An individual feels compassion or sympathy when the misfortune of others has a direct effect on their own well-being, and in the words of Amartya Sen (1977, 326), “It can be argued that behavior based on sympathy is in an important sense egoistic, for one is oneself pleased at others' pleasure and pained at others' pain, and the pursuit of one's own utility may thus be helped by sympathetic action.” In order to move beyond this stage, we must approach what Sen calls “commitment.” This commitment may be an interested one—for example, when an individual shows commitment to collective action designed to protect or promote their own interests—but this is not always the case: “If the knowledge of torture of others makes you sick, it is a case of sympathy; if it does not make you feel personally worse off, but you think it is wrong and you are ready to do something to stop it, it is a case of commitment.” (Sen, 1977, 326). It should be noted that sympathy has no economic cost, but is rather pure externality: it is enough that an individual believes another’s well-being to be increased for their own well-being to also see an increase. Commitment, on the other hand, does have a cost, since in this case a person chooses “an act that he believes will yield a lower level of personal welfare to him than an alternative that is also available to him” (Sen, 1977, 327). However, commitment can be guided by two totally divergent reasons, between which Sen unfortunately fails to distinguish: altruism and justice. The ambiguity may be left in place for the time being, but it will certainly be necessary to resolve it at some point.

82 To say that one supports an increase in RMI is not enough to identify oneself with a search for neutral fairness. Whether this opinion stems from a direct interest or from compassion or sympathy, as just discussed, we are in fact here dealing with a partial spectator. It must be shown that unaffected individuals who support this opinion are “committed”—in other words, that they are aware that such a measure could mean an increase to their mandatory contributions and that they are prepared to contribute more. In that case, they would be supporters of a measure which they consider to be in clear opposition to their immediate “egoist” economic interest, and we will be justified in explaining their opinion by invoking the stance of a fair spectator.

83 Questions focusing on this topic, such as those posed in the survey, do not enable a full, unambiguous verification of this hypothesis. For example, we only have one question of the type “Would you be prepared to pay higher taxes in order to finance an increase in RMI?” However, certain questions can be used to form good approximations.

84 The four questions dealing directly with the issue of costs aim to determine whether the interviewee is prepared to contribute more in order to ensure maintenance of health insurance, pensions, family allowances, or unemployment benefit. Correlation between these four items is high—in other words, the majority of those surveyed did not attempt to distinguish between these four forms of benefit. Furthermore, the same largely applies to those in favor, namely insiders. The fact that they did not openly attempt to orient their responses in favor of one type of benefit over another suggests that they primarily wished to demonstrate that they were prepared to offer a commitment on a general level. This therefore at least partially validates the choice of one of these questions as an indicator of commitment. We have selected the question relating to unemployment benefit since (like RMI, but unlike other benefits connected with health, family, or retirement) this benefit is undoubtedly the one least likely to directly or eventually affect universalist insiders. Furthermore, there is a strong correlation between responses to the various questions regarding RMI and those posed in comparable terms regarding unemployment (such as that regarding any potential quid pro quo). This suggests that, while not identical, the two situations were regarded in a relatively uniform fashion by those surveyed. Like the variable relating to an RMI increase, that relating to a rise in unemployment contributions will be recoded into two categories, with those entirely or generally in favor (30%) set against all others.

85 When it comes to rough links across the entire population, we should note that being prepared to contribute more and also considering homosexuals to be no different to others both correlate very strongly with support for an increase in RMI. Furthermore, there is also a strong link between acceptance of a rise in contributions and a rejection of homophobia. Having said that, the question we are considering does not concern the entire population. It involves ascertaining whether the universalism of those who are not affected (or hardly so) by the question of minimum social benefits is a committed universalism. The problem is not relevant to others. In addition, even among those not involved, in order to simply deal with the commitment/compassion choice we must restrict the investigation to those who support an increase in RMI. This is because, for these individuals alone, not being prepared to accept a rise in contributions can be seen as an indicator of compassion or sympathy. They want to see an improvement in the lot of the very poorest, but do not want this to have any potential effect on their own well-being. Conversely, those prepared to contribute more are committed in the sense that they accept the idea that the RMI increase they are calling for could ultimately be contrary to their immediate economic interest. By selecting a subsample of unaffected people in favor of increasing RMI, we will thus be able to determine whether the universalism displayed by spectators looks more like commitment or compassion, when compared with the average of solidaristic spectators.

86 A contingency table intersecting our moral opinion indicator with our commitment indicator for the previously defined subpopulation will suffice for this purpose. Particularists, both spectators and solidarists, are primarily compassionate. This also applies to universalists in the same situation, but to a lesser extent. However, it is precisely this deviation from the trend, or deviation from the average or from independence (which here amounts to the same thing) that is of fundamental importance to us in answering the question that has been set. From this perspective, the result (table 9) is perfectly clear: unlike particularists, universalists tend to be committed rather than compassionate. This trend is extremely significant, whether as regards the chi-squared, adjusted residual, or odds ratio. In a nutshell, whatever statistical test we use, unaffected universalists have a strong inclination toward commitment over compassion.

Table 9

	Happy to contribute more
Homosexuals are no different to others	Absolutely or generally agree	Absolutely or generally disagree
Generally agree	+3.7 (***)	–3.7 (***)
Generally disagree	–3.7 (***)	+3.7 (***)
N.B. – For a definition of adjusted residuals, refer to the note for the reader on table 5 (p < 0.001, two-tailed test) Chi-squared = 14; size (n) = 1,639; d.f. = 1; p = 0.0002. Odds ratio = 1.6; p < 0.001 (two-tailed test).

Commitment and Moral Opinion Among Those Who Would Like RMI to be Increased but Are Not Affected (Either Personally or Via a Member of Their Family), by RMI or Unemployment Benefit (Adjusted Residuals)

4.3. Universalist Spectators, Particularly Committed Ones, Are Not Cynical

87 Let us take stock. The solidarist opinion (which supports an increase in RMI, or does not support a binding quid pro quo) is found among those affected, but is also held by a large majority of those unaffected, namely universalists; and the latter tend to be committed. However, another form of indirect interest arises out of this very trend. Whereas unaffected universalists are prepared to contribute more to enable an RMI increase, is this not quite simply in order to “purchase” social peace? Recognizing that the very poorest will get together to demand an increase in minimum social benefits, they would prefer to pay the price needed to prevent the continuation of the conflict—which may for example translate into vacant accommodation being occupied by the homeless or badly housed—or quite simply to keep their psychological peace of mind intact. This time, universalists’ solidarism conceals a form of cynicism.

88 It is important to refute this Hobbesian line of argument, not simply because it goes against the hypothesis we are attempting to verify, but also more broadly because it results in a concept of a social contract linking every citizen to the welfare state, which is incompatible with the Kantian approach we have chosen. For Hobbes (1651), as for all followers of the tradition he began, the cooperation required by this social contract is rational in that it simply translates each person’s personal interest. It is rational for a person to agree to part of their resources being deducted and redistributed among the poorest, as this preserves social peace. Each person makes the same calculation and sees that they cannot preserve their long-term interests, namely social peace, without sacrificing their short-term interests, namely keeping their income intact. It is therefore in everyone’s interest to agree to common rule—in other words, to accept the deductions of the welfare state, and even their increase, should they prove insufficient to ensure that no conflict arises.

89 The perspective of justice as fairness leads to an entirely different theory. As we have seen, a spectator by definition addresses an unspecific or generalized other. If I am myself included within this generalized other, I can only acknowledge a judgment or opinion as being fair if I am not put in the position of the other targeted by the judgment or opinion, and if I have not verified whether I would accept this judgment or opinion in the particular situation. The judgment is fair not only because I am not involved, but also because, were I in the position of those involved, I would accept it. A fair spectator aims for unanimity through this (first) form of reciprocity, which generally consists in basing the contract on each party’s acceptance of the others’ point of view. While the question raised by the competing theory is that of which institution or principles of distribution would maximize everyone’s interest, the question here is that of which institution or principle would be unanimously satisfactory were I in the position of any other person involved, including the very poorest. As observed by Kant, this perspective alone can guarantee that no party will ever be sacrificed to the interests of another: “When someone prescribes for another, it is always possible that in doing so he may do the other an injustice, but this is never possible with respect to what he decides for himself (since volenti non fit injuria)” (Kant 1797, §46). A principle, a rule, an institution, a measure, an opinion etc. is only legitimate if “each person decides the same for all and everyone decides the same for each person” (Kant 1797). In other words, justice as fairness requires unanimity based on generality and universality.

90 We therefore need to demonstrate two things: firstly, of course, that the Hobbesian argument does not stand, so as to eliminate a third form of interest which could be hidden within the solidarist opinion of committed universalists, but secondly—and this time positively—that the competing model of neutral fairness is in fact the stance taken by such universalists.

91 As it stands, the questionnaire does not contain any questions directly broaching the topic of social peace. We must therefore agree to turn to indirect evidence. If fear of conflict is a motivation, it is logically to be accepted that those who would be happy to contribute more have a greater perception than others of poverty and unemployment being problems that concern them or France in general. Likewise, they should be more disposed to think that inequality, poverty, or exclusion have increased (“over the last five years” in the questions asked) or at best remained the same, but not that they have decreased. This reasoning should of course hold true as regards assessing the future of inequalities or poverty. However, this is absolutely not what is observed, even at the level of the population as a whole.

92 Whether personally or for France as a whole, being concerned with poverty is in no way linked to being prepared (or otherwise) to contribute more (the chi-squared figures in the corresponding cross tabulations are entirely insignificant). As regards the development of inequalities or poverty, there is a significant link, but this is the opposite of what a Hobbesian argument would suggest. Those who are happy to contribute more feel that inequality, poverty, and exclusion have decreased. The result is exactly the same if, instead of focusing on past developments, we deal with the future developments predicted by those surveyed for it could be conceded that the Hobbesian argument is based less on perception of the past than on fear of the future. However, once again, this appears not to be the case: those prepared to accept an increase in contributions think (in large numbers) that inequality, poverty, and exclusion will lessen in the future.

93 That said, these trends concern the entire population, and do not answer the question of whether someone hardly or not at all affected by RMI and supporting its increase is in fact “purchasing” peace of mind when they say that they are prepared to pay more. In order to ascertain this, we must simply reexamine all of the preceding intersections in the subsample of those in favor of an RMI increase but not affected by it. However, these results are absolutely identical to the previous ones. The solidarism of spectators who accept commitment cannot therefore be accounted for by their interest in preventing a situation considered to be potentially worrying, serious, or even dangerous.

94 Logically, these results hold true if the subsample is further restricted so as to contain only universalists. Consequently, the commitment of a universalist spectator, leading them towards a solidaristic opinion, is not based on an interest that is cynical in nature. From this perspective, it is also interesting to ponder whether, among spectators, there is any difference between universalists and particularists, or between committed and compassionate universalists. In regard to their preoccupations, the link with commitment is equally insignificant for universalists and particularists. On the other hand, this link most often differs significantly when broaching the question of past or future trends in inequality or poverty. The clearest case concerns feelings regarding the future of poverty. The trends are the same in other cases, although slightly less accentuated. As can be seen in table 10, judgment on the development of poverty has a greater impact for universalists than for particularists. For the latter, the choice to contribute more or otherwise is not based on an assessment of the situation in terms of poverty. However, this assessment forms part of a universalist’s choice and thus creates a clear difference between the committed and the compassionate. While a compassionate universalist spectator fears the future and is thus, entirely logically, not prepared to contribute more, a committed universalist spectator is optimistic. It is precisely to these latter individuals, and here more than ever, that the Hobbesian argument particularly fails to apply! The opinion of committed spectators, and in particular universalists, favors an increase in RMI which is linked to their optimism and not to fear of a future which—in terms of poverty—could disturb their social or psychological peace of mind. There is no cynical interest, particularly for committed universalist spectators. Whatever the level of analysis, the Hobbesian argument can thus be entirely rejected.

95 If this indirect reasoning is accepted, given the lack of strictly topical questions, there can be no doubt that we have been able to eliminate immediate (directly affected) or secondary egoist economic interest in order to account for a spectator’s solidaristic opinion. This opinion is based on a universalist concept which is (tendentially) neither compassionate nor cynical. Of course, it could be argued that we would have to rely on other forms of interest in order to gain a greater level of certainty. In addition to the fact that we are limited by the questionnaire and that, from this perspective, we feel that we have gone as far as we can, we fear that arguments such as this will simply result in a never-ending investigation. Let us therefore turn to this positive line of argument, as there is only one key question remaining unanswered: If the solidaristic opinion is that of a spectator whose material well-being is not in question, in what way is it fair?

Table 10

			Poverty and exclusion in France in the future
Homosexuals are no different to others		Will decrease	Will increase	Will remain the same
Yes	Happy to contribute more	Yes	2.7 (***)	–1.7 (*)	–0.9
	No	–2.7 (***)	1.7 (*)	0.9
No	Happy to contribute more	Yes	0.3	–0.5	0.3
	No	–0.3	0.5	–0.4
N.B. – For a definition of adjusted residuals, refer to the note for the reader on table 5. Chi-squared of upper table = 7.7; n = 1,181; d.f. = 2; p = 0.02; phi = 0.081. Chi-squared of lower table = 0.2; n = 382; d.f. = 2; p = 0.89; phi = 0.025.

Commitment and Feelings on the Future of Poverty or Exclusion, by the Moral Opinion of Those who Would Like RMI to be Increased but Are Not Affected (Either Personally or Via a Member of Their Family), by RMI or Unemployment Benefit (Adjusted Residuals)

4.4.Unlike Particularists, Universalist Spectators Never Reason According to Their Economic Interest, Nor According to Their Concept of a Good Life, in Judging the Lot of the Poorest: They Are Fair

96 Our spectators do not have ulterior motives, but their opinions are not established without foundation. In order to make this assessment, we must simply take into account the concepts of the good life, but also considerations regarding justice. [8] A host of questions posed in the survey enable us to provide an idea of the difference that separates universalist spectators from particularist spectators. We already know that, unlike the latter, the former believe that RMI should be increased and that there should be no restrictive quid pro quo. However, they also believe (and there is still a significant difference compared with particularists) that it should continue until the beneficiary finds sufficient income. They are also in favor of extending RMI to cover those aged under twenty-five. As regards causes of poverty, universalists—once again demonstrating a significantly different approach to particularists—believe that it could be due to bad luck, and definitely not to the fact that those in this situation do not wish to work. It is therefore clear that they feel that aid should absolutely be provided to the very poorest, and even that it should be improved, not out of personal interest but because their concern should quite simply be a generalized concern for all others. If RMI must exist as a form of permanent right, without quid pro quo, until the beneficiary’s lot has improved, it is because universalist spectators refuse to distinguish between the good and bad poor. In other words, they only create one equivalence class. They adopt a position that consists in taking the judgment of others into account without dictating to them (whatever their own thoughts may be).

97 We can go further and demonstrate how many particularists and universalists conflict with each other regarding this attitude. While they at least agree on one aspect of the good life, they do not draw the same conclusions regarding fairness. For they agree on the Aristotelian principle that individuals want (or should want) to exercise their skills—more simply, that it is a good thing to work when the opportunity is available. And thus, the vast majority advise those receiving RMI who are able to work to strive for social inclusion (94.7%) and to look for work (96.7%). The formulation of the questions (“If people receiving RMI are able to work, do you think it is acceptable for them to be asked in exchange for the RMI they receive: […] (iv) to strive for social inclusion; (v) to look for work?”) takes it for granted that those receiving RMI are simply lacking in will, which of course favors the demand for a quid pro quo, and all the more so where it is suggested that RMI requires an exchange. Nevertheless, it is typical that requests which stand as advice elicit the same support from particularists and universalists, whereas when advice becomes coercion (with those receiving RMI required to accept the work offered to them in return for the benefit), universalists are significantly more reticent than particularists (see table 11). This demonstrates that the former prioritize the fair over the good, and are much more concerned with respecting the person receiving RMI than the latter are. In their opinion, what they consider to be a good life has no legal force. It therefore seems that unaffected universalists best display the voice of the fair spectator.

Table 11

		Homosexuals are no different to others
Ask those receiving RMI…		Yes	No
To accept the work that is offered to them	Yes	–5.3 (***)	5.3 (***)
No	5.3 (***)	–5.3 (***)
To strive for social inclusion	Yes	–1.4	1.4
No	1.4	–1.4
To look for work	Yes	–0.1	0.1
No	0.1	–0.1
N.B. – In this table, as in the previous ones, the absence of an asterisk signifies that the adjusted residual is not significant, even at the low (bilateral) level of 10%. On the other hand, (***) here signifies that this residual is significant at the 1% level (and beyond). N.B. – first table: 3,224 individuals, chi-squared = 28.3; d.f. = 1; p < 1%; phi = 0.094; second table: 3,294 individuals, chi-squared = 1.9; d.f. = 1; p = 16%; phi = 0.024; third table: 3,304 individuals, chi-squared = 0.05; d.f. = 1; p = 94%; phi = –0.01.

What Can Be Requested of Those Able to Work Who Are Receiving RMI, by Opinion on Homosexuality Among Those who Are Not or Are Barely Affected by RMI (Adjusted Residuals)

98 In order to conclude this demonstration, we must simply reverse our approach and construct a fair spectator without drawing on the universalist/particularist distinction. It should once again be reiterated that two conditions are sufficient (but not necessary) in order to be a fair spectator: having no material interest in play, and not allowing a personal concept of the good life to express itself. By definition, a fair spectator is disinterested and is not a perfectionist, otherwise they will always be suspected of attempting to impose their own non-neutral reasons, i.e. of being partial, which would be contradictory. For us in this case, this means (i) someone not affected by RMI, whether personally or via a family member; (ii) someone who considers working to be a good thing; but also (iii) someone who refuses to dictate to others, in other words who is not a perfectionist. These criteria together fulfill the empirical conditions making them a fair spectator. The question is therefore: Who is this fair spectator? First of all, we can see that a great majority are universalists (83%, compared with 73% in society as a whole). However, we can move beyond this by comparing fair spectators with perfectionist spectators, i.e. those who would require employment to be accepted. This is set out in table 12, and the result admits of no ambiguity: the vast majority of fair spectators are universalists.

Table 12

		Homosexuals are no different to others
		Yes	No
Accept employment offered	Yes (perfectionist)	–5.3 (***)	5.3 (***)
No (liberal)	5.3 (***)	–5.3 (***)
N.B. – In this table, (***) signifies that this residual is significant at the 1% level (and beyond). N.B. – 3,129 individuals, chi-squared = 27.7; d.f. = 1; p < 1%; phi = –0.094; odds ratio = 0.5; p < 1%. It should be noted that by offering a stricter definition of a spectator, and thus excluding those who are affected (either personally or through someone close to them) by unemployment or insecurity, the adjusted residual remains strongly significant, with an absolute value of 4.4. The same applies to all other significance indicators.

Those Receiving RMI Should be Asked to Accept Employment, by Opinion on Homosexuality Among Those Not Affected (Either Personally or Through Someone Close to Them) by RMI, and Who Consider Working to Be a Good Thing

99 If we now compare those affected by the issue of RMI with those not affected, we can see on the one hand that those affected are, of course, less coercive, and on the other, that there is no longer any distinction between universalists and particularists. It was thus indeed important to remove the two biases of interest and perfectionism in isolating the position of the fair spectator. The fact that this position emerges once these biases have been removed demonstrates that justice as fairness is indeed a horizon.

100 1Ultimately, we have now established that an individual in the position of a fair spectator will tend to opt for universalist morals—in other words, to refuse to accept different rights and responsibilities among different individuals (as regards minimum income). This leads to a solidaristic opinion not out of compassion or cynicism, but as a result of arguments that are flexible enough to be acceptable to any other. Such a person has their own idea of happiness, but does not seek to impose it. Their arguments are those of justice as fairness, which is prioritized over good.

4.5. A Holistic Interpretation Fails to Account for the Data

101 This result—fairness as a priority—is by definition beyond the reach of a theory drawing solely on limited rational choice, but is a holistic explanation not an option? The most relevant objection—since it is based on interpreting the strong overall correlations—consists in maintaining that, while the argument up to this point may be appealing, it is superfluous, and we need only note that there is a strong correlation between employment status and opinion on RMI to account for what has been empirically observed. In fact, the majority of universalists (universalist insiders, in the terms of the automatic classification) are public sector employees, and the thesis of homo ideologicus (supported by a simple cross tabulation demonstrating the significance of the relationship between professional status and a variable representing opinion on RMI) could have some validity: fair spectators would simply be public officials who strongly favor the state’s universalist mission, by dint of their social position or status. In particular, the result seen in table 12 would be biased owing to the underrepresentation of public sector employees among our universalist fair spectators. Unfortunately for holists, there is no such bias. If public sector employees are excluded from table 12, there remains a strong link between universalism and solidarism. The adjusted residual changes to 4.7 (from 5.7) and the chi-squared to 22.3 (from 27.7) per degree of freedom, which is still extremely significant. In addition, this relative fall in significance is simply the result of the fact that the sample size being considered is smaller (changing from 3,129 to 2,643 individuals). We can account for this by comparing phi, which changes from –0.094 to –0.092. Suffice to say, being a public official has no innate effect on the results of table 12. It should be noted that the exact same applies if we remove senior management and professionals from the equation (the adjusted residual changes to 4.9, chi-squared to 23.5, and phi to –0.090). Proximity to the welfare state, modeled through the status of public sector, private sector, or self-employed, cannot alone absorb a concept of the good life, nor the horizon of fairness when interest is lower. A methodology in terms of good reasons accounts for this perfectly, whereas a holistic reading quite simply fails. This is because a quick reading of overall correlations, which in addition assumes breathtaking black boxes when it comes to individuals, results in an overly simplistic vision when its only basis is the idea of a social position or ideology that seeps into individuals’ minds making them crude “spits.” And moreover, if there are always exceptions, these can always be attributed to false consciousness (if not irrationality) in order to maintain the explanation. Fortunately, taking into account the idea that reasoning such as this results in free will for men and women, only the competing thesis which considers social position to be a bias (one among others) relating to the horizon of justice as fairness is compatible with the empirical data. Without further attacking the old demons of strong determinism, the results we have obtained enable us to emphasize that, even if the intervention of ideology is conversely explained by drawing on the white box of good reasons, it is vital not to underestimate the role of the truly moral dimension, since it prevails in accounting for opinion.

4.6. Universalist Spectators Are Not Angelic

102That said, a final hypothesis could still contradict our interpretation of the results of table 12. Universalist spectators are admittedly fair, but are they so (as previously discussed) out of an ethical concern for justice or out of altruism—out of a love for humanity, for instance? We know that they are not egoistic, since they are not affected; and they are also neither cynical nor compassionate. However, of the three possibilities—namely egoism, altruism, and justice—only the first has been eliminated. The choice between the latter two persists since our spectator’s veil of ignorance is almost total, which is not the case in other theories, notably that of Rawls (1971), which holds that individuals are still motivated by risk aversion. In such cases, the principles of justice behind the veil of ignorance clearly cannot be the result of supererogation. However, this results in an “overthrowing” of altruism which is subsequently eliminated by construction. This overthrowing is, of course, not unattractive, since it enables rational choice theory to be turned against itself, by demonstrating that—even coming from a narrow rationality behind the veil of ignorance—the response to the question posed (by Rawls) cannot fail to give rise to the dimension of justice as fairness. In our case, however, having eliminated first and second level egoism to attain a truly spectating spectator, we can no longer claim a priori that their position cannot be that of an altruist. And there is little point in turning to the theory of fairness for any lines of argument here. Indeed, it can be noted that fairness implies reciprocity, since by definition it means believing that what applies to one person applies to another, which runs against altruism that is constructed precisely without any such concern for reciprocity. Sacrifice is agreed to as a personal measure, and is no way imposed on others. However, this argument does not hold true here, since at this particular point we are attempting to determine whether our spectator’s good reason is an ethical requirement of neutral fairness or, alternatively, an angelism which could give rise to neutral fairness in something resembling a deadweight effect.

103 We cannot escape this dilemma without an idea of the moral model empirically at work. We have such a model. It was established during the third part of this study in order to explain opinion regarding the level of RMI, and of course applies to universalists. As a reminder, this consists of two principles: one a minimum level, P1, and the other reciprocity, P2. It is therefore enough to note that while P1 may be compatible with altruism, P2, on the other hand, is by definition totally contradictory to such altruism. This is, moreover, something we have already observed. Consequently, universalists are fair by dint of their neutral reasons, and not as a result of angelism. With all other possibilities already excluded, given that this also represents the position of a spectator, there can be no doubt that universalists claim no more than the priority of fairness over good.

104 The moral model linking P1 and P2 is thus strictly a model for fairness where only one single equivalence class is created, since it is this which makes a fair spectator as regards the question of minimum income. It has a universalist bent and does not consider two classes using a principle of merit. Also, if all on average draw on the same moral model, an interpretation that creates two equivalence classes is only subjectively fair. It fails to detach itself from a concept of the good life or economic efficiency, which always ultimately comes down to greater concern for oneself than for others. This result, far from normative, has been empirically demonstrated: we are content to listen to individuals in a position that leaves them no other possibility than to use neutral reasons to form their opinion. In theory there is nothing preventing two classes from being formed, one of the deserving and one of the undeserving; however, it is clear that they freely made the opposite choice since they have no motives to skew their opinion. In the vocabulary used by Rousseau, they speak as “citizens”; and as is often the case, Kant (1798) follows suit in summarizing their position: “The opposite of egoism can only be pluralism, that is, the attitude of regarding and conducting oneself not as the whole world, but as a citizen of the world.”