A Methodological Framework for Relational Research in Sociology and Social Work¹

Pierpaolo Donati University of Bologna, Italy

CORRESPONDENCE:

Pierpaolo Donati

e-mail: pierpaolo.donati@unibo.it

Abstract

How to make the research methodology corresponding to relational sociology operational? The Author proposes a general framework based on five rules that explain how to do «relational analysis». First: distinguish between descriptive and problematizing observation of the social fact Y to explain and understand, and connect them. Second: define the observed fact as a relational social molecule configured according to the AGIL relational scheme. Third: express the social fact Y as the product of a combination of Xi factors (subjective and objective) and their relations in a black box hypothetically having an AGIL structure. Fourth: investigate the black box as a morphostatic/morphogenetic process that generates the social fact. Fifth: if the research is aimed at social intervention, configure the research design as a relational Observation-Diagnosis-Guiding (ODG) system. Finally, it is explained in what sense relational analysis can be value-free.

Keywords

Relational analysis, methodology, research design, critical realism, research rules.

The logic of relational analysis and the research design

The analysis of social facts from the viewpoint of relational sociology, as I understand it, follows a method with its own rules governing empirical research. Here I shall be going no further than examining the logic of relational analysis, that is, its methodological framework, while leaving aside the specification of field survey tools.

Sociological research follows from our examination of the question of why a certain social phenomenon (Y) occurs. For example, we may ask ourselves why youth unemployment exists, or why people commit suicide, or why so many couples get divorced. The analysis can be explanatory (Erklärung), if it aims to identify the «causes» of the phenomenon Y (i.e. which factors generate or constitute that phenomenon, and in what way); or it can aim to provide an interpretative understanding (Verstehen), that is, it can contemplate the meaning of the phenomenon under investigation from a certain point of view.

In explanatory analysis, the researcher’s task is to explain the variations of a social fact (Y) in terms of because motives. The variations in Y are due to the different weights of the generative factors and the dynamics of their relationships (for example: why do suicides increase or decrease at different times of the year within the same society?). In interpretative analysis, on the other hand, it is a question of understanding the different meanings that the phenomenon (Y) and its variations have with regard to the in-order-to motives of the agents of the phenomenon as viewed by the observer (for example: why is the pursuit of individual success in the economic field more widespread in Protestant countries than in Catholic ones?).

From the viewpoint of relational sociology, causal and interpretative analyses are not in themselves mutually exclusive. They can be — and in fact are — complementary. This complementarity is achieved using a particular methodological tool, namely the relational version of the AGIL scheme, which will be discussed later.

For example, it is one thing to ask what the impact is of certain environmental factors (material and cultural) on suicide rates in different societies A, B, C, D (or during different periods of the year in the same society), and quite another thing to attribute meaning to suicide rates based on the observer’s perspective on suicide as a relational phenomenon.

In any case, the aim of relational research is to obtain a meaningful explanation and understanding of the phenomenon Y through the analysis of the processes that generate that phenomenon in a given relational context and under given social conditions, assuming that sociology is that science which deals specifically with social phenomena resulting from interactions between individuals and therefore not attributable to the nature of these individuals. Figure 1 (in Appendix) schematises the sequence of sociological research.

Let us first look at the five basic rules which together lead towards the research design. I will explain these rules in detail, and also point out how they operate in regard to the research design, later on.

(i) The first rule states that one must start from the observation of a social fact, and distinguish between the descriptive observation (e.g. one wants to know the figures for unemployment and how unemployment is distributed in a given geographical area, or know the marriage rate trend in a certain population, etc.) on the one hand, and the problematizing observation (e.g. why is there unemployment? why it is it increasing/decreasing? what does being unemployed mean? why is the number of polyamorous couples growing, and what is the point of being part of a polyamorous couple?) on the other. The process of problematization concerns both the because-motives and the in-order-to-motives.

(ii) The second rule states that the observed fact (considered to be a social problem) must be defined, hypothetically, as a social relationship generated by a relational context (e.g. unemployment as a relationship, suicide as a relationship, etc.). It is a question of expressing the highlighted sociological problem (e.g. unemployment, marriage, polyamorous couples, suicide, the influence of religion on social mobility, etc.) as a phenomenon (Y) consisting of a certain relational process resulting in a given relational structure. The AGIL relational scheme allows us to analyse the structure of the phenomenon as a social relationship. The relational use of the AGIL scheme is designed to clarify both the causal factors generating the phenomenon, and the meaning of the phenomenon itself, since both aspects are interconnected.

(iii) The third rule states that the social fact to be investigated must be expressed as a relationship emerging from a «black box» in which generative processes take place.

(iv) The fourth rule states that the black box must be investigated as a process of morphostasis/morphogenesis (i.e. the reproduction of forms or the generation of new social forms) which produces a relational outcome. It is a question of highlighting the peculiarity of the phenomenon Y as a relational sui generis reality with an emerging structure.

(v) The fifth rule concerns intervention-oriented research to bring about social change; it states that where required, sociological analysis may produce operational guidelines for social actions aimed at modifying the phenomenon Y; in my approach, these guidelines derive from setting up research according to the perspective of a relational system called ODG (Observation-Diagnosis-Guide of the phenomenon Y).

Based on these rules, relational analysis constructs the research design as shown in figure 2 (in Appendix).

According to figure 2 (in Appendix), the research design follows a logical sequence whereby, in practice, the sociologist proceeds to problematize the observed social fact Y (rule 1: the researcher formulates the problem); then to define the research target (i.e. phenomenon Y to be explained), hypothetically, as a relationship configured as AGIL (rule 2); then to identify the Xi factors and their relationships that can be considered key to explaining the way in which phenomenon Y is generated, and to transform the Xi factors into indicators (rule 3); after having taken all of the quantitative and/or qualitative empirical information required in order to understand the Xi factors and their interrelationships, the sociologist then constructs the black box in which the Xi factors and their Rj relationships generate morphostatic/morphogenetic (M/M) processes in an AGIL structure resulting in the emergence of the Y phenomenon (rule 4). If the research is designed for the purpose of implementing social action (relational social work)2 in order to modify the phenomenon Y, rule 5 also applies.

If the researcher realizes that s/he has not clearly identified the Xi and/or their relationships, or that something is lacking in the black box, then the research design will have to be completely rethought. This is normal for all research, which never really ends.

The research design I am proposing uses a «constitutive explanatory» methodology involving an abductive logic, as suggested by Peeter Selg (2022). However, I will not only use the method of abduction, generally used by pragmatists and phenomenologists. As a critical realist, I also want to use the method called «retroduction».3 In my view, as claimed by Ritz (2020), abduction in the Peircean sense and retroduction in the critical realist sense refer to different, but complementary, modes of inference. The abductive conclusions that follow from relying upon Peirce’s method provide the starting point for retroductive inferences. The latter inform the tenability of the former. Together, abduction and retroduction contribute to theoretical explanation.

The use of the relational version of the AGIL allows the researcher not only to analyse how the structural interactions between the elements of AGIL configure the resultant phenomenon Y (erklären), but also to understand the meanings (verstehen) that the single elements of the AGIL (that is: its value-pattern, norms, means and purpose) assume for the agents in the flow of processes and in the modification of structures. The interactions and structural relationships within the AGIL do not exhaust the autonomous contribution given by each element of the AGIL, by virtue of the fact that these elements have a substantial reality of their own, even if the relational games played by the agents/actors can give them different meanings depending on time and context.

A relational gaze is adopted whereby entities4 and relations are co-principles of reality. It is true that a knife can be used to cut food at the table or to injure or kill a person, but it is always a knife; and, in any case, one cannot caress a person with a knife. it is true, as Shaikh (1990) states, that a loom can be used by a family to make its fabrics, or by a capitalist to make a profit, but it is still a loom; what changes is the relationship of use, not the entity as such.

This allows for a deeper understanding of the involvement of agencial and non-agencial elements in the processes giving rise to the outcome Y. Here we see the original contribution of the critical realist relational approach. In fact, in most sociologies there is a discrepancy between the understanding of the phenomenon, which is generally sought at the micro level of the phenomenon as a transaction between agents (see, for example, Emirbayer, 1997), and the explanation of the phenomenon at the macro level, which is most often investigated as an effect of automatic or mechanical feedbacks (for example Marsden & Lin, 1982). In this case, the micro and the macro, in effect, can appear as irreconcilable dialectical oppositions, while they are «polar oppositions» (Gegensatz) in the sense of Romano Guardini (1925), which means that they define a field of tensions generative of an emerging entity and not a conflicting duality of mutual exclusion. What relational analysis does is to connect the micro and macro levels through the meso level, analysed on the basis of a stratified social ontology and an analytic epistemology (see figure 3 in Appendix). In this way, the relational approach seeks to avoid both constructivist subjectivism (at the micro level) and mechanical structuralism (at the macro level).

Let us now examine in detail each of these rules and the ways in which they are used to construct the research design.

First rule: «distinguish between descriptive and problematizing observation, and connect them»

An empirical sociological study, whatever the reason behind it, aims to comprehend a social fact in terms of its determinants and its consequences. But what is a «social fact» exactly?

Let me explain. A social fact is not a material «thing» that exists independently of human action. A fact is social insofar as it is generated by the relations between the social agents/actors. Certainly, the social fact consists of a reality in itself; however, from the point of view of sociological analysis, it becomes a sociological object if, and insofar as, it is observed «in a certain way«. What is «social» in poverty? What is «social» in suicide, unemployment or the breakdown of marriage?

The sociologist constructs the social problem, and gives importance to certain aspects of the social fact that are of interest to him/her. He/she «indicates» it in a certain way, using qualifications which must be given a name. The social fact therefore has two aspects: on the one hand, we have what happens in objective reality (the phenomenon lying outside of the observer); on the other hand, there is the way in which that phenomenon is observed (i.e. what the observer defines as relevant for an understanding and explanation of the social nature or quality of the fact). Sociological knowledge is a way of relating these two aspects.

For example: official statistics show that the birth rate is lower in country A than in country B, and one wonders «why is it lower?». Before collecting data relating to the phenomenon of births, relational observation problematises the social question inherent in the empirical fact of differences in birth rates, in the sense that the researcher poses the problem of data collection in the light of the question: «why the phenomenon Y (having children) occurs, and what is its meaning, in the various contexts?». Since the answer is not immediately evident, the social fact described must be expressed as a sociological (not demographic) problem and investigated as such. This can be considered where «real» sociological analysis begins, i.e. a kind of reasoning that begins with abduction and continues with retroduction.

From the very outset, the problematizing observation conditions the way in which the investigation of the phenomenon will be organised, because it is the type of question (the «why?», i.e. the because and in-order-to motives) that determines the direction and formulation of the research design.

A question is trivial if the answer is immediate and obvious. On the other hand, it is all the more «instructive» (and, in a certain sense, «intelligent» — that is, it helps to intus legere) the more it aims at resolving an apparent paradox or a counterfactual expectation. For example, if we observe a statistical increase in the number of households, we also expect to see a parallel increase in the number of the target population; however, demographic data indicate that the population is decreasing. Hence the question: why is the population falling while the number of families increases? Other examples include the questions: if there is economic growth, why is it that unemployment is on the increase?; and why is it that people greet each other when they meet strangers on mountain trails, and yet on city streets they do not?

The social fact comprises both objective and subjective dimensions: e.g. poverty has material aspects (the lack of means) and subjective aspects (the perceptions and experiences of what people lack, the meaning and value of what is lacking). The quality of what we call the «social» aspect of a fact is inherent in the relational characteristics of that fact; in other words, it consists in what connects an observed state of affairs (the number of suicides, thefts, unemployed people, etc.) to the actors and their context. The social aspect is the embedding of actions within a relational context; the phenomenon Y is enmeshed, entangled, embodied, embedded in a network of relationships (Manterys, 2021), and yet it must be traced back to the reciprocal actions of the agents, because the influence of the structural conditioning of the network is mediated by the reflexivity of the agents as subjects (Donati, 2021).

In short, relational analysis aims to define social facts in terms of relationships and relational contexts. If I ask myself «why is there poverty in this region?», the question must be formulated in such a way as to suggest that poverty is a relationship generated by a certain specific framework of relationships which need to be investigated; likewise with unemployment, the perception of immigrants as a danger, marital instability, and so on.

This being the case, the description of a social fact is not immediate, but reflects the need to understand what is social in the phenomenon observed, as said phenomenon refers to a relational context that generates and constitutes it, and determines its own consequences.5

The first rule of relational analysis therefore suggests that a distinction be made between the descriptive observation and the observation problematizing the social fact by linking the former to the latter. Real social research begins when we ask ourselves why the social fact occurs, and what is social about it, which presupposes that the description of the phenomenon responds to the problematization of the social fact.

Second rule: «define the observed fact as a relational social molecule configured as a relational AGIL scheme»

Once the social fact Y has been expressed as a problem, we can move on to the second rule. This states that the observed fact Y must be hypothetically defined as a relational product, that is, as a social molecule generated by a specific social context (a relationally contested context).

To say that the phenomenon Y must be expressed as a relational product (i.e. a social relationship having its own structure, albeit produced by transactions and flows of interactions) means that the sociological problem raised (e.g. why unemployment increases? Why does drug use increase? Why fewer people get married?) has to be expressed as a phenomenon consisting of the relational way of being of social reality (which has its own «consistency»). The questions of unemployment, drug use, birth rates and so on need to be defined as social relations generated by a relational context (not as reified or reifiable «things» or mere «events»).

To define the phenomenon Y as a social relationship, i.e. a molecular relational structure, I propose to use the relational AGIL scheme.6 What does this scheme consist of? Here we get to the most challenging, difficult and important part of my relational analysis. The researcher needs to be able to define phenomenon Y, by way of hypotheses, as a social relationship composed of factors that can be boiled down to four different dimensions: the means (A from adaptation), the situated goals (G from goal-attainment), the norms (I from integration), and the value-pattern (L from latency), which together constitute the relational form AGIL (figure 4 in Appendix).

AGIL is the acronym — a recursive acronym7 — of these four dimensions, which should all be present in order for a social relationship to exist. Indeed, it is their relational combination that produces the social fact as a social relationship having a molecular structure. The four dimensions A, G, I and L can be defined as analytical and relational components.

For example, we can define the AGIL dimensions of poverty in a population or social group in the following way: poverty depends on the means (A) available within the group in relation to a certain standard of living desired as an aim (G), which in turn depends on the value-pattern (L) attributed to goods prevailing in the group, and on the norms (I) that combine means, purpose and value attributed to goods. These elements or dimensions of poverty as a social relationship are based on the interactions of the group with the external social context (e.g. local, national and international rules of exchange). Each of these elements is related to its own external environment (Donati, 2021: 84-86). Here, in figure 4 (in Appendix), I limit myself to pointing out that the means depend on interaction with external material conditions, and that the value-pattern depends on the system of ultimate values (in a broad sense) prevailing within the group.

A more complete conceptualization of the AGIL defines these four dimensions as specific functions of subsystems having their own symbolic code (together with generalized symbolic means of communication and interchange), to which the researcher must attach empirical correspondents (at the various micro, meso and macro levels) (see table 4).

The AGIL scheme (figure 4 in Appendix) has a «molecular» structure, insofar as it is what lends quality to each specific social fact to be investigated as a relational product. Unemployment, drug use and birth rates are examples of relational forms. This form is precisely «the social», i.e. it constitutes what we call «the social» as it indicates that sui generis reality that links the four elements of the AGIL relationally.

Each relational product (a certain birth rate, a certain type of poverty, etc.) possesses a molecular structure, insofar as it is made up of simpler elements that must exist and interconnect in a certain way; that is, these elements must be combined relationally to generate a certain phenomenon. The AGIL scheme is tasked with grasping this molecular structure of the studied phenomenon.

The molecule can be seen as the result of processes that combine two axes: the means-norms axis (A-I), which is the religo (bond) of the relationship, and the value-goals axis (L-G), which is the refero (symbolic reference) of the relationship (with regard to such concepts, see Donati, 2021).

The phenomenon Y (figure 2 in Appendix) is the product of the combined effect of these two axes which, in operating together, generate the phenomenon Y. For example, if we talk about «taking drugs», the relational analysis starts from the assumption that this behaviour is an expression of a way of relating to those who take drugs with respect to themselves, to others, to the world. If, in a certain population or social group, drug use is observed to increase, relational analysis will consist of investigating the structure and dynamics of these relationships in the social context of drug users, on the assumption that the increase of consumption is an emerging effect of changes in the combination of means-norms and values-objectives that characterize the relationships that lead to drug use in that context.

What I want to highlight is the structure and dynamics of the relationality that characterizes a certain phenomenon or behaviour, which can be observed as a social molecule of its own kind. For example, the doctor-patient relationship has a different «molecular» structure from that of the mother-child relationship. Going to the doctor means activating a relationship according to an AGIL (a social molecule) which is completely different from the AGIL of treating a child. The sui generis quality of one relationship differs from that of another in terms both of the elements that compose it and of the links between the specific elements of that relationship. The differences can be analysed in terms of both the internal structure of the relationship and the way in which it differs from other, diverse external relations.

Each relational form can be considered from the point of view of how the social molecule (figure 4 in Appendix) organizes itself internally and/or from that of how it organizes itself in relation to the environment (table 1).

In the first case, the AGIL is called the MKSU, since it is a specific combination of internal means (M), self-preserved goal (K), self-determined norms (S), self-referenced values (U). In the second case, the AGIL is observed from the point of view of how it is organized in relating to the outside world; I call this DRCP, an acronym that indicates how the molecule is: (i) more or less dependent (D) on outside means, (ii) more or less responsive (R) to environmental pressures, (iii) characterized more or less by external constraints (C) in regard to its regulatory rules, (iv) more or less oriented towards pro-social values (P) in favour of the outside world.

Table 1

The components of the AGIL relational scheme in the MKSU and DRCP modes.

The four dimensions of AGIL:	M-K-S-U The analytical components of the internal relationality of the social molecule	D-R-C-P The analytical components of the relationship that the social molecule has with the external environment
A Adaptation (means)	M = The means (tools and resources) suitable for achieving the goal G depend on the inside	D (means dependency) = The extent to which the means (tools and resources) suitable for achieving the goal depend on the outside world
G Goal-attainment (target)	K = The goal pursed reflects an orientation towards organizing the relationality of the social molecule for its own self-preservation	R (goal responsiveness) = The extent to which the goal pursued responds to environmental pressures
I Integration (norms)	S = The regulatory norms internally governing the relational social molecule are self-determined	C (normative constraints) = The extent to which the rules for governing the internal relationships are constrained by the outside world
L Latency (value-pattern)	U = The value attached to the goal pursued, which serves as a directive distinction] of the way in which the internal relationality of the social molecule is organized, is a self-referenced interest	P (pro-social values) = The extent to which the value attributed to the aim pursued, and which acts as a directive distinction of the way in which the relationality of the social molecule is organized, is the expression of pro-social values in favour of the outside environment

For example. Suppose we want to explain and understand why one social group has a different birth rate than another group. The analysis will be conducted by analysing the MKSU and DRCP components of each group in order to compare them and thus understand why one group has a lower or higher rate than another. The definition of the MKSU and DRCP components is summarised in table 1. To understand and explain outcome Y (birth rate), it is necessary to know: (i) as regards the means component (A-D), to what extent they depend on internal or external resources; (ii) as regards the goal (K-R), to what extent it is geared towards self-preservation, or instead is responsive to the external environment; (iii) as regards the rules that regulate the relational social molecule (S-C), to what extent they are internally self-determined, or alternatively are heterodirected (constrained by the environment); (iv) as for the value-pattern (U-P) to what extent it is a self-referenced interest, or rather an expression of pro-social orientations towards the external environment.

Of course, all the examples I give here are greatly simplified due to the limits on the permitted length of this article.

One very important thing should be noted here. In my view, the relational analysis, precisely because it highlights purposes and values in the structure of the relationship (for instance having a child), points to what is specifically human in that relationship, while at the same time combining this dimension with the factors and processes that can comprise automatisms that are not dependent on intentional subjective motivations or on cognitive-symbolic-expressive abilities. In this way, relational analysis combines the deliberate, conscious qualities of human action with social mechanisms, without detracting from the reality of each relational element and level. That is, it respects the emergent properties of each layer of reality.

In short, the second rule of relational analysis requires sociological problems to be studied in the following manner: (i) the problem is defined as the existence of a phenomenon with a relational molecular structure that gives rise to a certain problem, and this problem comprises four dimensions — A, G, I, L — whose connotations must be empirically verified each time with regard to each social phenomenon; (ii) precisely by observing how the four components are present and how they are configured in fact (i.e. in more or less appropriate and consistent ways), can the cause of the problem (deficit or other) in each dimension be highlighted; (iii) the exchange relationships and inter-actions between the four dimensions is investigated; (iv) in the end, the problem to be dealt with must be formulated as the generation of a specific (sui generis) relationality that is considered to be problematic.

The usefulness of the relational analysis is already evident in the first phase of research (i.e. the interpretation of Y as a molecular structure AGIL) which, after all, determines the nature of the rest of the investigation. Take the following three examples.

Example (a): the identification of those social goods not conceived of by lib/lab sociology (i.e. methodological individualism and holism)

When dealing with the question of social goods, modern social science divides them into private goods (lib) and public goods (lab) (Donati, 2021). In short, it is argued that social goods have the relational structure of one or the other such category. The question then arises: are there no other types of goods? If my problem is to understand a «community» (or a non-profit Third Sector organization), how can I do so using the private/public distinction, that is, using the methodologies of individualism and holism?

In investigating the types of goods that serve people’s lives, Antonine Wagner (1991) proposed a typology comprising three types of goods: public goods, private goods and collective goods. Their definitions were based on the intersection of two distinctions: sovereign/non-sovereign consumption, rival/non-rival consumption by the users. One space remained empty (see table 2).

Table 2

Types of goods according to Wagner (1991).

	Non-rival consumption	Rival consumption
Non-sovereign consumption	Public goods	Collective goods
Sovereign consumption	?	Private goods

I thought that by applying the AGIL scheme, I could fill the aforesaid void, and from there give a more detailed and complete meaning to Wagner’s typology. Wagner had identified goods corresponding to the political system (public goods, G), those corresponding to the market (private goods, A) and a category of goods he called «collective» that are produced by voluntary sharing, as in voluntary associations (in the AGIL scheme it was clear to me that they should be placed in I). The letter L in the AGIL scheme remained uncovered, and so I then tried to fill in the gap with what I have called «primary relational goods», consumed by sovereign and non-rival consumers.

The goods that Wagner called «collective», without being able to distinguish them from those contained in the blank cell, were actually relational goods of an associative character, which I have called secondary relational goods to distinguish them from the primary goods pertaining to families and informal networks (see table 3) (Donati & Archer, 2015: 210). There are many examples of analysis of common goods conducted with the relational methodology set out here, which I do not have enough space to explain in detail (e.g. Donati, 1997; Guadarrama Sánchez, 2018).

Table 3

Donati’s typology of social goods.

	Non-competitive consumption	Competitive consumption
Non-sovereign consumption	Public goods (governed by the political-administrative system) (G)	Secondary relational goods (associative, as in the organized third sector) (I)
Sovereign consumption	Primary relational goods (families & informal networks) (L)	Private goods (free market) (A)

Example (b): Unseen citizenship rights

The debate on citizenship rights has not so far clarified the difference between human rights and the other rights that make up the complex of rights of citizenship rights (i.e. civil, political and social ones). To clarify the differences, I have applied the AGIL methodology to the theory of T.H. Marshall (Citizenship and Social Class, 1950), which for a long time was regarded as the leading theory in this field. According to Marshall, citizenship rights are of three types: civil rights (those relating to individual freedoms, freedom of speech, the right of association, the right to due process when accused of a crime, and so on); political rights (i.e. electoral rights in a democratic political system); and social (or welfare) rights. Once again, the accounts did not add up, because this theory included A (civil rights), G (political rights) and I (social welfare rights), but the L was missing (L is always the most problematic dimension). Hence the question: what could the L of citizenship rights contain? The answer I gave was that the L included human rights, which therefore had to be distinguished from civil rights under the AGIL scheme (Donati, 1985: 405-437).

The relational approach also leads to criticism of the idea of human rights proposed by Parsons. Parsons had identified them as cultural rights in the form of rights to higher education. In fact, from the viewpoint of relational analysis, rights are to be seen as social relations, and human rights are not the rights to a performance (e.g. obtaining an educational qualification), but are precisely those rights to the social relations that make humans flourish; for example, the right of children to be raised in a family rather than on the street or in an institution.

However, that was not all. According to Marshall, citizenship rights had historically emerged «in single file»: that is, first civil rights, then political rights, and then social rights. In light of the AGIL scheme interpreted relationally, this may not be true, since on an historical level, the four dimensions of citizenship rights (A, G, I, L) may have developed following very different, and not necessarily linear, paths. In fact, historical research conducted after Marshall showed that this was in fact the case. For example, in some countries political rights were established before civil rights, while in other countries political democracy was not the consequence of liberal civil rights but of a revolution in those countries. In other nations social rights exist but civil rights are scarce, and political rights are a mere facade (e.g. in dictatorial regimes).

Example (c): the identification of civil welfare

When pondering the problem of coming to grips with the various types of welfare policies, I examined Richard Titmuss’ theory on welfare state models (Social Policy: An Introduction, 1974). He developed three models of the welfare state: the residual welfare state model (which leaves human well-being primarily to individuals and their families); the acquisitive-performative welfare state model (which links human well-being to the «merits» that each person has in carrying out their work, and to the social security that can be obtained on a voluntary contractual basis); and the institutional-redistributive welfare state model (based on the idea that citizens must be guaranteed social well-being by the state according to their needs, regardless of whether they are «deserving» or not). Applying the AGIL approach to this theory has always resulted in an empty AGIL cell. The acquisitive-meritocratic model corresponds to the primacy of the market (A), the institutional model to the primacy of the state (G) and the residual model to the family (L): but what corresponds to the I?

Clearly, Titmuss had forgotten the third sector organizations that care for the frail, the poor, the sick and the various fringes of society, which he evidently did not regard as generators of social well-being. To put it briefly, there was no civil welfare, which I identified in the social integration cell (the «I» of AGIL).

Applying the relational methodology, it was clear that the welfare state could assume many more very different forms, precisely due to the fact: (i) that it is necessary to take into account the role of the third dimension (the third sector, the community organizations) in the creation of welfare, (ii) and this entails redefining the other three dimensions (residual, acquisitive-meritocratic, institutional) due to the need to consider the third dimension and its relations with other welfare producers. In short, welfare systems could, and still can, be configured in a more flexible, rich and complex relational manner than was indicated by Titmuss.

Many other examples could be given here. In the past I have specified such examples concerning, for example, the analysis of health as a social phenomenon, the study of risk in social networks, the prevention of certain diseases, the effectiveness of social support, and the dynamics of social capital.

Third rule: «express social fact Y as the product of a combination of Xi factors in a black box»

The third rule states that the phenomenon Y (already interpreted and expressed as a relation) must be reconstructed as the effect of a set of factors and processes which, combining in certain ways within a «black box» inaccessible to the naked eye, bring out Y (as shown in figure 2 in Appendix).

We now enter the black box, i.e. the configuration of the space-time mode in which the objective and subjective factors Xi are supposed to combine with each other to generate the outcome Y.

First of all, the black box can be both «trivial» and «non-trivial». It is considered trivial due to the fact that the inputs into the black box always produce the same output. It is considered non-trivial or complex as a result of its degrees of internal contingency, which generate potentially different effects from the same inputs. If the answer to the question «why does Y happen?» is immediate and obvious, then the black box is deemed trivial, as it operates through a simple automatism. If, on the other hand, the question does not elicit any immediate, evident answer, it means that there is a complex black box operating interactively (rather than automatically), since the input factors and the processes that produce phenomenon Y are interrelated in ways that are invisible to the naked eye, and which vary and thus need to be investigated by adopting ad hoc procedures.

For example, if the figures for the alcohol consumption of a controlled social group indicate stability over time, we can deduce that the black box works in a morphostatic way. If, on the other hand, the figures repeatedly indicate levels of consumption that are significantly different from one another, then the black box is operating in a non-trivial way, i.e. the relational networks characterising alcohol consumption are of a morphogenetic type.

The structure of the black box can be designed according to the configuration of AGIL. Let us examine the individual dimensions concerned.

(G) The black box, observed as a process, is a «system» that establishes its internal relationships on the basis of a situated goal. If I observe a person going to the bar, the person’s aim will presumably be to get some coffee or other drink or item that can be found at the bar, while the bartender is there to serve such items to that person. The relationship that connects one subject to another (customer-bartender) is first of all characterized by the situated purpose for which a subject has acted. The form of the relationship is characterized by the complementarity between that purpose («get something at the bar») and its contact person (the bartender in his bar). If the individual went to ask the butcher for a coffee, that relationship would not arise, and the butcher would probably reject any such relationship. Likewise, if the person went to the bar only to meet a friend, the relationship sought would have another form. As obvious as this consideration may seem, it is a fact of daily life that in the face of a certain need or desire, people do not know which relationships to activate. And even when they activate them, they never quite know how to qualify them. The case of the bar is trivial, but we could make non-trivial examples, such as looking for a partner or a job, not to mention the needs relating to existential problems.

(A) The system observed as a black box contains more or less suitable means in relation to the purpose for which the system is implemented. What are they? This obviously depends on the case in question. In the customer-bartender relationship, for example, there are specific means concerned (money is what you need to get a coffee in a bar); in the doctor-patient relationship, on the other hand, there are different means involved (the proper means are the details and information that the patient gives to the doctor in order to receive a correct diagnosis and treatment), and so on. The banal example of the bar scenario should not lead us to assume that the means are easily identifiable: what means can be used to recoup the esteem of colleagues, or lost love?

(I) The system seen as a black box has norms that are specific to that relationship. When we are at the bar we observe certain rules, and these rules differ from those that apply when we visit the doctor for a check-up.

(L) The system observed as a black box possesses a cultural model of the value of the desired thing; that is, a criterion (value) on the basis of which the actors judge the relationship created. The value criterion when going to the bar is the customer’s wish for a moment’s relief by drinking or eating something, and the bartender is supposed to behave accordingly, and thus to try and offer a good degree of relief (if this were not the case empirically, the relational analysis would serve precisely to highlight how and why it is not). The value criterion of going to the doctor consists in the treatment of ill-health, both on the part of the patient and, one presumes, on the part of the doctor. The criterion used to evaluate the relationship is a value in the sense of a model of what is desirable (over time, and thus for an improved situation in the future). Of course, the value that the relationship has may differ between the doctor and the patient, and in fact most of the time it involves negotiation between them, which in turn can be either of a conflicting or cooperative nature.

To give another example, when considering the problem of unemployment, we may ask ourselves why it is that some people cannot find a job. In an abstract, hypothetical way, the reason may depend on how their internal AGIL (MKSU) and/or the AGIL (DRCP) of relations with the outside world (specifically the labour market) operate(s). As for MKSU, it may be that the type of job sought (the unemployed target) simply does not exist, or for which s/he has inadequate means (for example, s/he does not possess the appropriate qualifications); or the unemployed follow rules that are unsuitable when looking for a job, or they may have in mind an ideal (self-fulfilling) type of work that is wholly imaginary. As for DRCP, it could be that the labour market does not offer enough job opportunities, or that employment contracts are non-existent or unfair, etc. Unemployment (Y) can be accounted for by verifying the lack or inadequacy of one or more of these factors (Xi) and by the relationships between these various components of the problem.

One may wonder what the «ultimate value» is (see L in figures 4 and 5 in Appendix) when we go to the bar or to see the doctor, or when we look for a job. What do transcendence and ultimate realities (even religion) have to do with these everyday, apparently «banal» and «immanent» gestures? Let us now examine figure 5 (in Appendix), which shows the space-time structure of the black box as AGIL.

Along the vertical axis (L-G), the analysis of the «deeper» values that legitimize the situated goal leads to an analysis of the social relationship in terms of immanence (here and now) and transcendence (there and after). For example, the relationship with one’s professional occupation can be the basis of a purely immanent value (the search for material well-being in the present) or it can refer to a transcendent value (work as a vocation or mission that transcends its immanent value) In the relational interpretation of the AGIL scheme, this analysis is placed along the vertical axis representing the time dimension (the distinction between before/after or present/future). This axis (or temporal dimension of the social relationship) shows that relationships are not exhausted (not consumed) in the present, but imply a potential future state of affairs (whether or not this state of affairs actually comes to be is another matter of course).

However, the temporal dimension must be connected to the spatial one, that is, it must be referred to a situated context. The horizontal axis (A-I) in figure 5 (in Appendix) represents the spatial axis, which refers to the internal norms (I) used to regulate the relationship in that context and to the search for the appropriate means (A) available in the surrounding environment.

One aspect I would like to emphasize here is the fact that the AGIL scheme serves to see what is not perceived by purely behavioural sociology. In fact, going to the bar or to the doctor, or looking for a job, is a form of behaviour indicating not only a certain cultural model (respectively, in regard to refreshment, health or the appreciation of work), but also a connection to a deeper meaning that particular behaviour has for the social actor. The sociologist cannot ignore the importance of people’s ultimate values in influencing the latent value underlying (legitimating) the model of well-being, health or work being pursued as a relationship to one’s life and to the world. This is, of course, a difficult task to accomplish, since a latent value always refers to another value that justifies it. This is why even the most banal everyday gestures (consumption, way of eating or dressing, home furnishings, etc.) refer to «ultimate realities»: they do so, as it were, through a chain of symbolic values motivating action.

In theory, this process can go on indefinitely, but it actually reaches a limit when the relation generates eigenvalues.

For example, those looking for a job do so in a way that to a lesser or greater degree reflects profound values such as self-esteem, a life project, the sense of one’s future, and certain fundamental reasons for living, which are very different from those that are at stake when we go to the bar. The importance of values in any social relationship should not be underestimated, regardless of how obvious, routine or banal that relationship may seem. The sociological analysis conducted through by means of the AGIL approach, allows us to design an analysis on the basis of which we can go in search of those invisible, implicit, unspoken, perhaps unconscious realities, which nevertheless exist and are at stake in social relationships.

The «ultimate» reason why an actor activates a relationship, lies within the sphere of latency (L) (we should say, within the latency of the latency of the latency ...);8 however, this is not isolated from the other dimensions of the AGIL scheme. The «value» that sets the relationship in motion is conditioned, and must in turn relate to the other dimensions, that is: the norms that characterize that type of relationship, its situated goal (e.g. to satisfy hunger or thirst when going to the bar for refreshments, to cure a certain illness or relieve discomfort when visiting the doctor for treatment), and the means by which to achieve the goal.

The arrows linking the four dimensions in the AGIL diagram (figures 4 and 5 in Appendix) indicate this: they highlight the fact that each dimension relates to the other. Means, ends, norms and values draw meaning from one another. Their real distinction lies in the connective reference of one to the other (and not in mutual negation, as in the binary distinction according to Luhmann). The reference implies two operations: ad intra (the re-entry of what distinguishes an element based on its own internal guiding-distinction) and ad extra (dependence or hetero-reference, from which inter-dependence arises). The relationship always has this dual value, which does not mean uncertainty or radical opposition between one and the other, but rather the «dual value» of the superimposition of the inside and outside (which can naturally generate uncertainty).

Adopting a more sophisticated approach, the two sets of operations that define the AGIL social molecule — one facing inside the relationship (e.g. looking for a satisfying job), and the other facing outwards (e.g. adapting to the heteronomy/autonomy of working conditions) — can be conceptualized in terms of time and space: respectively, on the basis of the temporal distinction «before/after» and the spatial distinction «internal/external» (see figure 5 in Appendix).

The temporal dimension is vital to the morphogenesis of the relationship (knowing at what point one is in the search for that job, what was missing before and what is missing now or will be missing later), while the spatial dimension is necessary in order to preserve the identity of the relationship with respect to its environment, i.e. other types of relationship (am I looking for the same job as before, or am I looking for a different kind of job?). Even when the relationship changes, it maintains these two needs.

Stating that the social phenomenon Y (the observed initial fact to be understood and explained) is a relationship means seeing it from the perspective of an AGIL relational system that emerges from a structural network-like system of relationships. This is what relational sociology means when it argues that society consists of a network (or networks) of relationships.

One crucial problem of sociological research is linked to this: at what point should a researcher stop his/her analysis of the black box, given that within that black box, one system of relations always refers to another such system in a potentially indefinite sequence? Ending the analysis is in principle always problematic. This is because one value refers to another value, and this to yet another (L  L  L… ∞). The same goes for the other dimensions (A  ∞, I  ∞, G  ∞), obviously each according to its own specific guiding distinction (value is differentiated from other values, means from other means, rules from other rules, and aims from other purposes).9

The analysis may be interrupted for practical reasons (lack of time and means to proceed further), or because, as one proceeds to explain a system of relations by means of another system of relations, co-relations always become increasingly less significant. It is clear that scientific research is of this second type. It can be brought to an end when the black box analysis has stabilized, meaning the analysis no longer produces any additional information that is meaningful or useful to the researcher.

The structure of the black box (figure 5 in Appendix) according to the AGIL scheme, reminds us once again that the social is what unfolds in time and space, remaining external to the physical-material environment on the one hand, and to the environment of transcendent values on the other, while at the same time having to relate to them nevertheless. The acting subject is both internal and external to this relational field. Said subject comes before it and goes beyond it. In this sense he/she transcends it. The human person is the relational field’s agent-actor, and in this sense lies within it, but at the same time is external to that context precisely because she can relate to something else. This is why the individual «subject» is not the centre of sociological analysis.

The sociologist obtains important information from the individual subjects, but the sociologist’s analysis does not concern the interiority of the subject as such, but rather the social relations in which the subject expresses him/herself, by which s/he is more or less conditioned, and which s/he accepts or changes with her/his reflexivity. Sociology does not look «into the minds of individuals». Since people are individuals-in-relationships, and the relationship is constitutive of the making (to flourish or not to flourish) of the human person, relational analysis has to consider the reflexivity of the person in a certain way, that is, insofar as it is in relationship to others and contributes to the relationality of the context. The biological nature and the inner psychic and moral consciousness of the person remain outside of the scope of sociological analysis, and is dealt with by other disciplines.

If, for example, the sociologist investigates electoral behaviour, s/he certainly does not investigate what happens in the psyche and moral conscience of individuals, but only studies how political opinions are formed and expressed as social relations in a certain contested relational context.

The example of the analysis of electoral behaviour offers us the opportunity to clarify the difference between understanding and explaining the phenomenon Y.

On the one hand, understanding the phenomenon concerns the subjective motives of the actors, and the meaning that the situations they create or find themselves in, and the actions they perform, have for them. These aspects constitute the refero — that is, the L-G axis — of the relationship. On the other hand, explanation concerns the mechanisms by which subjective motives (values and purposes) combine with situational factors (means and norms) — the religo or A-I axis of the relationship — so as to generate the phenomenon Y (electoral behaviour).

Understanding and explanation therefore pertain to the way in which the black box operates its internal morphostasis/morphogenesis. It is there that the specifics of relational analysis are exercised, as I shall explain in the following section. The result of a well-executed relational analysis offers a «comprehensive explanation» of the emergence of Y (as a social relation) from the Xi factors and their relations Rj.

The construction of the research design up to this point can be summarised as follows. We start from the observation that there is a social fact Y. It is thematized as a problem (why does it exist? How is it possible?). It is defined as a relationship (AGIL). Then one wonders why such a relationship has emerged. The design is constructed by identifying the subjective factors (value orientations and attitudes) and objective factors (adaptive conditions, i.e. means and norms, independent of subjective will) that are deemed key to the generation of the observed fact. The Xi factors are then placed in a system of relations (AGIL) constituting the «black box». This black box must offer a comprehensive explanation of how the phenomenon Y is produced. The question is: how does the black box work? We shall see in the next section.

Fourth rule: «investigate the black box as a morphostatic/morphogenetic process that generates the social fact Y»

The fourth rule says that the black box can be conceptualized as a morphostatic/morphogenetic process of structures and relationships interpreted according to the AGIL scheme. The morphogenetic analysis of the black box aims to highlight the sui generis character of the emerging phenomenon (Donati, 2021: 71-110). Let us see what that means exactly.

The Xi factors that the researcher identifies as relevant information are, at this point, placed in one of the four dimensions of the AGIL configuration in the black box. Whether they are placed in A or G or I or L depends on the meaningful position they have in relation to the other factors. For example, if the researcher assumes that the disposable income of individual Z is a relevant factor in explaining-understanding his/her way of life (life opportunities, conceived as possible social relationships), then the researcher will reflect on where to place the income of the individual Z in his/her AGIL relational system depending on whether this income is considered by him/her as a means with which to pursue another purpose (and therefore to be placed in A) or rather as an end in itself (and therefore to be placed in G).

Once the researcher has «filled» the AGIL of the black box with the Xi factors, then s/he has obtained the initial (hypothetical) structure of the Y-relationship phenomenon to be investigated. The configuration given to the black box must be such as to permit the interactions between the factors Xi over time/in space to be highlighted, so as to be able to describe the processes of morphostasis/morphogenesis that generate Y as a sui generis social reality (relationship) having its own peculiar structure (see figure 6 in Appendix).

Figure 6 indicates that social fact Y is understood and explained as an emerging phenomenon, starting from the initial (hypothetical) structure of the phenomenon (which we call the «Y1-relation», the initial AGIL) which has been subsequently modified over time by the interaction between the Xi factors, acted on by the actors, who have produced a new relational structure, the «Y4-relationship» phenomenon (the final AGIL). If Y4 is equal to Y1, then the process is morphostatic. If they are significantly different, then the process is morphogenetic.

It can be said that in static societies (such as simple or primitive ones), morphostasis is prevalent, while in modernizing societies morphostasis (or pure reproduction) is something of an exceptional, borderline case.

The relational analysis emphasizes that the social fact always has an emergent character. What does «emergent» mean? The time has come for us to clarify the difference between additive and emergent phenomena.

The distinction arises from the fact that the logics governing social relations are incapable of being reduced to the logics governing individual actions, since relational logics are the expression of another order of reality. When relational logics are such as to create social forms endowed with the characteristics of irreducible systems, we speak of emergent effects. What are the characteristics of the emergents? They are: non-aggregativity, non-decomposability (or quasi-decomposability), high diffusivity or low localization, and the high complexity of interaction.

(i) First difference: the emerging phenomenon is constituted by interactive processes and not simply additive combinations. Aggregative phenomena arise from the sum or juxtaposition of the elements and properties of the component parts, and can be explained in terms of those, while a phenomenon is considered emergent when it is a relational form that arises from interactions which generate an entity that possesses its own causal qualities and properties independent of the qualities and properties of the individual elements that compose it. Aggregative systems. On the contrary, non-aggregative systems are formed and operate in the opposite way, because they are constituted interactively.

(ii) Second difference: the emerging phenomenon is not decomposable. Aggregative systems are modular because they can be broken down into parts that operate through their intrinsic functions and exchange only their inputs / outputs; they can replace parts, remove them, break them down and re-aggregate them, since there are no substantial relationships between the parts that act as load-bearing structures. Emerging systems are governed in the opposite way, because the relations between the parts are qualifying for the system itself. Parsons’ AGIL is modular, while relational AGIL is non-modular. For example, a bureaucratic organization (such as a municipality) is made up of parts that have their own functions and have to exchange their inputs/outputs. Anything that goes beyond mere bureaucratic exchanges represents a problem because the system has difficulty seeing and managing the effects of the relations between the parts. An organization (for example non-profit) that is established and operates in view of relational goods, on the other hand, is of the second type in that it operates on the relationships between the parts, and governs its effects by operating on these relationships; here, the modular characteristics are not valid because the internal and external interactions continually modify the way the parts operate, and, if a part is missing, the emergent effect is not produced. This is why we can say that third sector systems producing relational goods are non-decomposable (or scarcely decomposable).

(iii) Third difference: the emerging phenomenon is diffusive, that is, it is weakly localized. Localizable systems are those in which each function corresponds to a physical place (a structure, an institution) in which it resides. Non-localizable systems, on the other hand, are those in which the functions are not physically localizable in a role, in a specialized structure, in a circumscribed place. Their functions are «widespread», that is, participants share in them to a considerable degree. For example, social welfare systems that operate as bureaucratic organizations are of the first type. The systems that provide social assistance through informal relationships (informal helpers), on the other hand, are of the second type. More generally, third sector welfare organizations have the properties of systems with a high degree of diffusivity (i.e. weak localization), since their functions are not spatially or physically localizable, not being specialized to the point where they can be located in a precise place. Unlike state institutions and more formalized market organizations, third-sector movements or networks possess social properties that are not restricted or circumscribed to certain roles (they are not «localizable»), i.e. they are not attributable to this or that component (or function specification), because they are distributed in a supra-functional way.

(iv) Fourth difference: the emerging phenomenon is characterised by its highly complex interaction. Social systems have a different complexity due to the fact that the interaction can be linear or non-linear (linear insofar as their response is directly or inversely proportional to a given variable; non-linear meaning that there is no predictable proportionality). Aggregative systems generally behave in a linear manner, while non-aggregative ones display non-linear behaviour. For example, hospitalization of a patient typically follows linear procedures. On the other hand, assisting a child or an elderly person suffering from a chronic disability on a daily basis, is an activity characterised by non-linear interactions with often unpredictable outcomes. More generally, in third-sector organizations the interaction is predominantly non-linear. This is like saying that the third sector is more complex than other societal spheres such as the state and the market.

The case of the third sector is particularly significant for its differences with respect to the behaviour of both bureaucratic systems and market organizations. In summary, the peculiarity of the third sector sphere lies in the fact that the organizations that are part of the third sector are characterized to a greater extent by non-aggregation, non-decomposability, limited localization and complex interaction. These types of systems are also those that generate the most emerging effects, which makes us understand why relational goods represent a phenomenon produced above all by the third sector (Donati, 2021).

The difference between additive and emergent social facts is crucial here. So-called collective phenomena need to be distinguished into the first and second types. Collective social movements are formed on the basis of a reticularity that additive phenomena do not possess. However, reticularity can have different characteristics depending on the case in question. Sometimes it results in the creation of collective subjectivities, while sometimes it does not.

Fifth rule: «if the research is aimed at social intervention, configure the research design as an ODG system»

As I have already said, relational analysis can be directed towards knowledge (the understanding and explanation) of the social fact, or may go further and also aim to support social action to change an unwanted phenomenon (e.g. poverty traps, vicious circles of social deprivation, unemployment, drug addiction, etc.) understood as a product of a relational context. The fifth rule provides support for social interventions designed to modify phenomenon Y by operating on the processes that generate it. It is a question of understanding how to proceed in this case.

To apply this rule, research should be designed from the outset in such a way as to foresee its practical impact. This presupposes the adoption of a research approach entailing a reflexive sequence of three operations:

(i) The relational observation (O): by relational observation I mean the ability to interpret a social fact as a problematic relationship, that is, as an undesirable state of affairs consisting of certain relations generated by a potentially modifiable relational context.

(ii) The relational diagnosis (D): by relational diagnosis I mean the introduction of an evaluative distinction (e.g. between satisfactory or unsatisfactory, physiological or pathological) in the observed state of affairs. These distinctions are obviously cultural elaborations. The distinction can be made by the observer, by the actors involved, or by a third party.

(iii) Relational guidance (G): by relational guidance I mean the social action designed to change a state of things considered undesirable (e.g. deviant, unsatisfactory, pathological) through the modification of the relationships that give rise to it, that is, by designing and implementing a steered morphogenetic process that produces a change in the relational context so as to achieve a desired state of things.

The above sequence, called the ODG system, allows the sociologist to go from the cognitive level (why phenomena occur) to the operational level (modification of phenomena), which is the level of professional application. ODG systems are the specific systems that relational theory offers to participatory action research and practice.

The above sequence is reflexive, in the sense that it must continually operate on itself by starting over again. In short, the sociologist starts from an ODG system at time T1 and re-applies it to the T4 structure that the operating system has produced through the interaction of the actors (as in figure 6 in Appendix).

For example, if I observe that problematic behaviour (such as drug abuse, violence, bullying, etc.) is on the rise among the young people of a specific neighbourhood, and I diagnose this phenomenon as a social pathology emerging from the relational context in which they live, relational guidance will consist in involving the actors from that neighbourhood (first of all young people and their families, then neighbours, schools, social workers, etc.), guiding them using appropriate non-directive support methodologies in order that they may reflexively interact with one another in order to generate a morphogenetic process that can lead to the emergence of social relations favourable to the avoidance of anti-social misbehaviours. A concrete example of social analysis and intervention to prevent deviant youth gangs and make them desist from offending was provided by Weaver (2016) (see figure 7 in Appendix).

Relational guidance is characterized by changing relationships through and together with the subjects who experience them. To design the social action in question, AGIL is required. The action must, in fact, fit into a context that we can configure as AGIL, and must lead to another AGIL that is closer to the desired state of affairs. The action-intervention system should modify the four components of AGIL and their relationships in such a way that the specific functions of each component operate in view of the result(s) pursued.

These functions (in the analytical sense) have correspondents in the subsystems (again in the analytical sense) of the societal system (see table 4): the efficiency function (A) is entrusted to the economic subsystem, the efficacy function (G) to the political subsystem, social integration (I) to the subsystem of civil culture, while the evaluative criterion of the relationships to be promoted (L) is entrusted to the subsystem of inter-subjective trust relationships.

Mobilizing these analytical sub-systems means making use of the generalized symbolic means of exchange and communication that characterize them (see table 4): money for the economy, power for politics, influence for public opinion, moral responsibility for civil culture, and commitment to values for trusting relationships. These generalized symbolic means are symbols generated within a sub-system of their own (e.g. money in the economy), but as means they are recognizable by the other sub-systems and are in fact used by them. Furthermore, to the extent that they circulate throughout the overall system, they can be considered to contribute to the cohesion of the relational system. Generalized symbolic means are used for the purposes of communication between the various social spheres, and for exchanges between those same spheres. In operational terms, the action needs to identify the empirical referents (see table 4) involved in the action network: that is, market undertakings, political-administrative institutions, secondary socialization agencies (formal ones such as schools, and/or associative and informal organizations such as non-institutionalized associative groups), families and primary networks.

Table 4

Social action (relational social work) conceptualized as an operative relational AGIL scheme.

The four dimensions of social action configured as AGIL:	Specific function (which can become prevalent in the AGIL system)	Sub-system having its own symbolic code (and the relative generalized symbolic means of interchange and communication)	Empirical correspondents at the macro-sociological level
A Adaptation	Efficiency (adaptation)	Economy (money)	Market
G Goal-attainment	Efficacy (goal achievement)	Politics (power of law)	State or political-administrative system
I Integration	Social integration (social rules disseminated and modified through communication)	Civil culture (influence)	Secondary socialization agencies (formal and informal)
L Latency	The value criterion that acts as a guiding principle of the relationship (e.g. trust or usefulness)	Intersubjective relationships of a fiduciary and cooperative type (commitment to values)	Primary informal networks and families

The ODG systems operate through actions called «networking actions» as they aim to involve all those who are interested in solving a social problem. The effectiveness, efficiency, regulatory and value-cohesion capacity of each specific network will depend on how the AGIL performs its functions in a super-functional way: that is, how it responds to needs that are not limited to explicit and formalized objectives, but also include the potential needs that can develop over time and in space; in other words, needs that go beyond mere functional performance, because they are capable of emergence. A typical emergent product is the relational good, defined as a good that consists of social relations, and is neither strictly public nor strictly private, but is shared on the basis of primary and/or secondary associative relations.

What does we mean when we say that sociological analysis must be value-free

After Max Weber, sociological research was dominated by the idea that the sociologist should adopt an agnostic methodology. In practice, this means that sociologists should only provide data and knowledge, while refraining from any personal reference to ethical values or criteria (Wertfreiheit). This is usually called «non-evaluative sociology». What does relational sociology have to say about it? My answer is that sociology is very rarely non-evaluative, and in principle cannot avoid problems of value, starting from the very formulation of the research design. The idea of a value-free methodology responds to just demands, but it needs to be discussed as it can be a source of error and misunderstanding.

Indeed, if taken to the extreme, ethical agnosticism «neutralizes» sociology, that is, it condemns it to total sterility from the point of view of its practical influence. It is no coincidence that sociology suffers from a permanent crisis and is often condemned to uselessness. This permanent crisis is to the result of the ambiguous relationship that sociology has always had with ethics and morals. Said relationship is both sought after and denied at the same time, while in fact the agnostic methodological criterion favours the lack of any ideal inspiration of sociology (Holmwood, 2002; Burawoy, 2005).

How does the relational analysis I have presented so far respond to the problem of the relationship between ethics and sociology?

Well, first of all we need to distinguish between positive and positivistic science. Undoubtedly, sociology is a positive science since it «stands by the facts»: it cannot and must not manipulate the facts at will. It would be scientifically incorrect for the sociological observer to evaluate phenomena based on his/her own values. However, this does not mean that sociological research must be a positivist science, if by positivism we mean an approach that requires the sociologist to treat social facts as having no ethical value.

If the sociologist behaves in this «positivist» way, then s/he is merely following a fashion: s/he specifically denies the «stuff» of the social, which is its «morality», that is, the fact that social phenomena are imbued with values and norms. In other words, the researcher ends up by treating social facts as if they were determined by mechanical automatisms, when on the contrary they depend on the strongly contingent value choices of the subjects who create social relations.

If at a given historical moment certain social fashions or trends take root, this does not mean that social evolution must necessarily follow in that direction. The criterion of non-valuation indicated by Max Weber (1958) must be understood in the sense that while the sociological observer ought not to make his/her own judgments or introduce his/her own values (Wert) into the analysis, nonetheless he/she cannot ignore the fact that the matter dealt with, that is, social relationships, always contain a «relation to value» (Wertbeziehung) (Ringer, 1997).

While relational sociology must be careful to avoid unduly introducing value judgments, it cannot avoid relating to values for the simple reason that it adopts a relational perspective (Donati 2019). Relationships always contain a value component. Non-valuation can be dealt with more clearly and analytically if we use the relational version of the AGIL scheme correctly. Non-valuation is certainly a requisite of the analytical methodology adopted, and in this sense it is situated in the cognitive subsystem (the A of the AGIL) applied to sociology as a discipline. However, the A of the AGIL is not disconnected from the other dimensions of the relationship between the sociologist and reality in the practice of sociology. If we consider this relationship as constituted by AGIL, even the so-called non-valuation is necessarily related to a criterion of the underlying value: can we deny this? For many the answer is yes, and in fact this gives rise to a nihilistic, paranoid sociology (Teubner, 2001). For relational sociology, on the other hand, values are unavoidable.

As was clarified by Lidz (1981), precisely in reference to AGIL, the choice of the means of investigation is never exempt from cultural references, that is, it has necessary (non-avoidable, albeit contingent) connections with the latent value dimension (the L) of the cognitive relationship. Such cultural references lie on the border between reality and its environment of ultimate values (see figure 4 in Appendix).

In other words, the sociological observer cannot reduce sociological theory to a mere method: whoever does so (as most old and new functionalists do) claims ethical neutrality when instead they are making a precise choice of latent values, or are negating the moral character of reality and the ethical character of knowledge (Hałas, 2016).

Relational sociology is therefore a positive science, but not a positivistic one. This means that it refrains from introducing value judgments when carrying out an analysis, but at the same time it acknowledges that the material it is dealing with is strongly loaded with value. The relational sociologist cannot be an «empiricist»; he/she cannot practice the abstract empiricism criticized at the time by Charles W. Mills (1959), who instead pointed to the need for «sociological imagination». With this, he was inviting sociologists to consider the ethical point of view as a stimulus for a more critical analysis, since it is related to values like justice, peace and non-discrimination. Of course, the imagination cannot consist of value judgments that invalidate the collection and analysis of data. What imagination, then, are we talking about? From the perspective of relational sociology, this sociological imagination lies in problematizing social facts starting from their own, internal ethical value, and in bringing out, rather than obscuring, the ethical connections (references) that they imply for the same social dynamics.

Understanding Wertfreiheit as ethical neutrality or agnosticism is thus a myth when it comes to research aimed at social action. The relationship with ethics is particularly evident at the moment of diagnosis (that is, when defining a state of affairs as unsatisfactory or pathological) and in regard to relational guidance (where to go from the existing situation).

The fact is that sociology deals with intrinsically moral questions, since social facts always refer to values, norms and ends that cannot be removed from the analysis. Hermeneutics can try to «objectify» them, but cannot eliminate them. Hermeneutics can, and indeed must, be non-evaluative when it comes to evoking the intentional sense (that is, the meanings) of acting. If the sociologist treats relationships as ethically neutral, then the resulting sociology will be a reflection of what is simply already given, without any possibility of either critical distancing or a search for sensible solutions to the social problems afflicting society, such as social inequalities or human trafficking.

References

Burawoy, M. (2005). 2004 American Sociological Association Presidential address: For public sociology. The British Journal of Sociology, 56(2), 259-294.

Donati, P. (1985). Considerazione conclusive: vecchie e nuove politiche sociali. In P. Donati (Ed.), Le frontiere della politica sociale. Redistribuzione e nuova cittadinanza (pp. 405-437). Milano: FrancoAngeli.

Donati, P. (1997). Family Associations in Europe: A General Outlook and Typology.’ Associations. Journal for Social and Legal Theory, 1(2), 235-255.

Donati, P. (2011). Relational Sociology. A New Paradigm for the Social Sciences. Abingdon: Routledge.

Donati, P. (2019). The sociological gaze: When, how and why is it relational?. In E. Carrà & P. Terenzi (Eds.), The relational gaze on a changing society (pp. 11-44). Berlin: Peter Lang.

Donati, P. (2021). Transcending Modernity with Relational Thinking. London: Routledge.

Donati, P., & Archer, M.S. (2015). The Relational Subject. Cambridge: CUP.

Emirbayer, M. (1997). Manifesto for a Relational Sociology. American Journal of Sociology, 103(2), 281-317.

Folgheraiter, F. (2004). Relational Social Work. Toward Networking and Societal Practices. London: J. Kingsley.

Folgheraiter, F., & Raineri, M. L. (2017). The principles and key ideas of Relational Social Work. Relational Social Work, 1(1), 12-18.

Guadarrama Sánchez, G. J. (2018). Las asociaciones y los bienes que generan. El debate conceptual y metodológico. Espiral. Estudios sobre Estado y Sociedad, 25(71), 39-70.

Guardini, R. (1925). Der Gegensatz: Versuche zu einer Philosophie des Lebendig. Mainz: Matthias-Grunewald.

Hałas, E. (2016). Through Social Values to the Reinterpretation of Sociology’s Ethical Neutrality. Sociologia e Politiche Sociali, 19(1), 67-79.

Holmwood, J. (2002). La sociologia dopo l’epoca delle ideologie. Talcott Parsons e la sociologia come professione. Quaderni di teoria sociale, 2, 61-84.

Lidz V. (1981). Conceptions of Value-Relevance and the Theory of Action. Sociological Inquiry, 51(3-4), 371-408.

Luhmann, N. (1995). Social Systems. Palo Alto, CA: Stanford University Press.

Manterys, A. (2021). Social Bonds and the Relational Nature of Embeddedness. In E. Hałas & A. Manterys (Eds.), Relational Reason, Morals and Sociality (pp. 95-128). Berlin: Peter Lang.

Marsden, P. & Lin, N. (1982). Social Structure and Network Analysis. Beverley Hills: Sage.

Mills, C. W. (1959). Sociological Imagination. New York: Oxford University Press.

Olsen, W. (2007). Critical Realist Explorations in Methodology. Methodological Innovations Online, 2(2), 1-5.

Ringer, F. (1997). Max Weber’s Methodology. The Unification of the Cultural and Social Sciences. Cambridge Mass.: Harvard University Press.

Ritz, B. (2020). Comparing abduction and retroduction in Peircean pragmatism and critical realism. Journal of Critical Realism, 19(5), 456-465.

Selg, P. (2022). What is relational explanation. Paper presented at the conference «Methodology of relational sociology. Approaches and analyses», University of Warsaw, March 14-15.

Shaikh, A. (1990). Capital as a social relation. In J. Eatwell, M. Milgate, & P. Newman (Eds.). The New Palgrave: A Dictionary of Economic Theory and Doctrine. London: The Macmillan Press.

Teubner, G. (2001). The Economics of Gift – Positivity of Justice: The Mutual Paranoia of Jacques Derrida and Niklas Luhmann. Theory, Culture & Society, 18(1), 29-47.

Vargas, F. J., Ibáñez, R. E. J., Soto, A. C. E., & Javier C. K. (2014). La relación entre la diferenciación y la calidad de red. Revista Electrónica de Psicología Iztacala, 17(3), 1308-1332

Wagner, A. (1991). On Sharing. Preface to an Economic Theory of Voluntary Action. Nonprofit and Voluntary Sector Quarterly, 20(4), 359-370.

Weaver, B. (2016). Offending and Desistance. The Importance of Social Relations. London: Routledge.

Weber, M. (1958). Il metodo delle scienze storico-sociali. Turin: Einaudi.

Appendix

Fig. 1 The sequence followed by relational sociological research.

Fig. 2 The research design according to Donati’s relational sociology.

Fig. 3 Relational analysis seeks to link the micro level (qualitative) and the macro level (quantitative) through the meso level.

Fig. 4 The molecular structure of the social fact as a relational AGIL.

Fig. 5 The space-time structure of the black box as AGIL.

Fig. 6 The morphostatic/morphogenetic dynamics in the black box.

Fig. 7 An example of an action research inspired by relational sociology on youth gangs.

---