Use and abuse of the normative in sociological research

There have long been debates about the role of values in sociology; in other words, about whether it should restrict its conclusions to factual matters or be aimed at normative goals defined by practical values. ¹ Normative sociology can take a variety of forms, as well as relying on diverse value commitments (associated with the political Left, Right, or Centre): it may evaluate an institution, policy, or practice in terms of a standard to which the researcher is committed; it can aim specifically to defend or to criticise an institution, policy, or practice; and/or it may engage directly in preserving or changing an institution, policy, or practice, for example via action research, participatory inquiry, or directly through political activism. These are, of course, significantly different enterprises, but they all rely on practical normative commitments on the part of the sociologist.

Of the three ‘founders’ of the discipline typically identified today, both Marx and Durkheim were committed to a normative approach, while Max Weber came to reject this. In the late nineteenth and early twentieth centuries much sociology in the United States was directed at social reform of one kind or another, inspired by religious as well as political ideas. ² However, subsequently, there was increasingly explicit rejection of a normative orientation (see Bannister, 1987; Smith, 1994). This derived, in large part, from attempts to ‘professionalise’ the discipline and promote its claim to scientific status – the latter usually being assumed to rule out normative conclusions. Yet, from the late 1960s onwards, there was a reversion to a normative orientation on the part of many sociologists: ‘value-neutrality’ and even ‘objectivity’ came to be widely rejected. Of particular significance here were ‘critical’ approaches to sociology, whether informed by Marxism, Critical Theory, feminism, queer theory, anti-racism, indigenous politics, postcolonialism/decolonialism, or disability activism. However, generally speaking, the practical values relied on by these approaches were left vague, for instance involving appeals to social justice without much clarification of which concept of justice was being assumed or how it was being interpreted (Hammersley, 2023a; Smith, 2014). In reaction to this, in recent years some proponents of normative social science have emphasised the need for the values involved to be made explicit and justified, for example by drawing on political philosophy (see Abbott, 2018; Modood, 2020, 2022; Sayer, 2011; Sass, 2018).

In my view, sociology should not be normative in the sense of pursuing practical goals (Hammersley, 2024a, 2025b) – it should be solely concerned with the production of factual knowledge rather than with criticising, evaluating, or seeking to defend or remedy the phenomena it investigates, the wider society, or the global system. ³ In making this case, I will start by focusing on the distinction between practical and epistemic values and its implications. I will show that this undercuts various arguments to the effect that sociological research cannot avoid evaluating what is investigated and/or being committed to preserving or changing it, or that it ought to do this. However, towards the end I will address two respects in which practical values do play a key role in sociological inquiry: in the selection of research questions, and in ethical as well as prudential constraints on how it is pursued. Nevertheless, neither involves redefining the goal of research in practical rather than epistemic terms.

So, my argument is not that all normative considerations must be eliminated from social science, so that research becomes ‘value-free’ in a literal sense. This is impossible: for example, research methodology is necessarily normative, albeit relying for the most part on epistemic values (those relating to the pursuit of knowledge), not practical values (Hammersley, 2011: ch1). And I certainly do not believe that normative issues can be treated as an entirely technical matter, or should be regarded as inherently irrational in the manner of emotivist ethical theory (Hammersley, 2024b). My argument is simply that the role of practical normative considerations must be restricted to playing a subsidiary, conditional, or external role, if sociological inquiry is to be pursued well, and in a way that does not exceed its intellectual authority. This is, of course, itself a normative argument, but it is a methodological one and is not at odds with a commitment to restrict the goal of sociological research to the production of factual knowledge.

Practical versus epistemic values

A distinction has sometimes been drawn between ‘theoretical’ and ‘practical’ values; see, for example, Bruun and Whimster's (2012: xxi) discussion of Weber's views about the role of values in science. However, the term ‘theoretical’ is somewhat misleading, given that it has so many different senses (Hammersley, 2023b: ch12). Furthermore, even the meaning that is most directly relevant here, deriving from the ancient Greek distinction between Theoria and Praxis, is far from unequivocal (Lobkowicz, 1967). Therefore, I will use the term ‘epistemic values’, where (as already noted) ‘epistemic’ means ‘to do with knowledge’. By contrast, practical values are concerned with what is good or bad, right or wrong, with what should and should not be done, with which agent or structure is to be praised or blamed for what, and so on. ⁴

The most obvious epistemic value is truth, and this is the guiding principle of research given its aim of producing knowledge. In traditional philosophical terms, I take ‘knowledge’ to mean ‘justifiable true belief’, and this definition points to a second epistemic value: justifiability. We can never know with absolute certainty whether an empirical knowledge claim is true, and in judging the likelihood of its being true we assess the evidence for and against this; in other words, we judge how justifiable it would be to accept it as true.

Some advocates of normative forms of sociology, in taking it to have a practical goal (such as challenging institutions or policies), appear to deny the distinction between epistemic and practical values, insisting that we can have knowledge of what is desirable or undesirable in much the same sense as we can have knowledge of what is true or false, perhaps that the one immediately implies the other. ⁵ One source of this is the work of Hegel, which of course also informed Marx's thinking and much Marxism. Here values are taken to be inherent in the process of historical development, in its teleological trajectory, rather than existing externally and having to be applied to the world, for example in the manner of Kant. Bhaskar (1979: 56–103) provided a more analytic version of this kind of argument, whereby critique is logically implied by description, but in my view it is no more convincing than Hegelian or Marxist versions (see Hammersley, 2014: ch4). This type of argument can also be found in the work of Comte, on which Durkheim draws in adopting a therapeutic approach to society (see Lukes, 1973: 76–77) and was relied on by other early sociologists, such as Hobhouse and Small. Such views came to be challenged by neo-Kantianism and later by ‘analytic philosophy’ in the late nineteenth and early twentieth-centuries; though some kinds of ‘ordinary language philosophy’ insisted that factual description is always evaluative (see Tsilipakos, 2025: ch4). In my judgment, the distinction between factual and value judgements still stands and retains its significance. ⁶

The true and the good have often been assumed to be closely related, or even to form part of a coherent whole. However, post-Enlightenment philosophy challenged both this and the nature of each of these values (notably in the work of Nietzsche and Kierkegaard). ⁷ There are, then, difficult philosophical issues involved here. The best I can do is simply to spell out my view of this distinction. For me, epistemic values relate to logical and factual truths. Logic concerns coherence of argument: the avoidance of internal contradiction and spurious inference. I interpret factual truth in traditional terms as correspondence to reality. This clearly assumes a version of realism; what I have referred to elsewhere as ‘subtle realism’, to distinguish it from the naïve realist assumption that reality simply corresponds to our perceptions, or to what we take to be immediately obvious in common sense terms (Hammersley, 1992: 50–54). ⁸ But subtle realism also contrasts with radical forms of constructionism, or constructivism, in which perception, thought, or discourse is taken to create the character of the very phenomena it is assumed to represent (see Kukla, 2000). There is a three-way relationship between the factual sense we make of the world, the flow of sense impressions (our own and those of others) that are our major source of evidence, and the phenomena that produce those impressions, some of which we are trying to understand. Following pragmatist philosophers, notably Peirce, I do not see this as a passive process but as arising out of our continuing observation of phenomena in the world and of our interactions with them. Of course, there are important distinctions to be drawn between different types of phenomena – the ‘middle-sized, dry goods’ which analytic philosophers often take as their examples (such as tables and chairs) are different in character from social phenomena, but I do not believe that this means the latter are any less real, whether we have in mind families or social classes, organisations or cultures.

While (by definition) factual knowledge represents phenomena accurately, and this is why we aim to achieve it, judgments about what is good or bad, better or worse, right or wrong, are not representational in this sense; hence, the words ‘truth’ and ‘knowledge’ should not be applied to them, if we are to avoid confusion. For me, terms like ‘justice’ do not refer to Platonic forms existing in another world, nor in this one, but rather to evaluative standards human beings employ that arise out of their nature and their conditions of life (Hammersley, 2024b). Moreover, as I noted earlier, there are different notions of justice that can be in conflict: merit-based versus distributional ones, for instance. And there are other values, too, that most of us have at least some commitment to: individual autonomy, our own health and that of others, privacy, the preservation of social order, and so on. But practical value judgments are not just mechanical applications of value principles, they must also take account of contingent circumstances. In the case of the attribution of praise or blame, for instance, they must assess responsibility, awareness, negligence, etc.

I am not suggesting that value judgments are irrational, but rather that the mode of rational argument they require is different from that involved in factual inquiry, even though all value judgments rely on factual assumptions (see Hammersley, 2024b: 86–87). Moreover, where in the case of factual questions we can usually be confident that there is a single correct answer, even when we find it difficult to know what it is, I do not believe that this is a reasonable assumption in the case of many value questions. Here there will often be more than one reasonable judgment to be made (as well as many unreasonable ones); and these will frequently conflict in their implications. As a result, the scope for disagreement is greater. This is not to deny that there is wisdom (phronesis) that allows us to make better rather than worse value judgments: this is the ability to weigh up various considerations relevant in a particular situation, for example through appropriate comparisons, albeit in a manner that is different from calculation (see Dunne, 1997). However, social scientists (and, for that matter, philosophers) do not have privileged access to such wisdom, in my view. By contrast, they can claim expertise in answering certain types of factual question (social scientists) or conceptual question (philosophers) which may be relevant to practical decisions. ⁹

For these reasons, aiming sociology at practical goals exceeds the intellectual authority of social scientists; and, in the public sphere, this threatens to undermine individual autonomy and democracy, because it involves a false claim to superiority. While it may be true that social scientists have better access to relevant factual information, sociological research is not even the only source of such information relevant to answering practical value questions (this may come from other disciplines, and from practical involvement in relevant activities as well as local knowledge). Moreover, for reasons already explained, while this information may rule out some options, it does not provide strong grounds for unique inference to what is good or bad, right or wrong, etc. Indeed, the same set of factual data can always be the basis for conflicting evaluations, depending on the value assumptions made. The large gap between factual knowledge and evaluations or recommendations often seems to be overlooked by social scientists, though not by them alone – including in notions of evidence-based policymaking and practice (Hammersley, 2002: ch3, 2012, 2023c).

Equally important, directing sociological research at practical goals increases the danger of bias, and therefore of error in the factual conclusions produced. The risk of bias tends to increase where the operational goal is practical rather than epistemic because the research process is prospective-retrospective in character: in making the many decisions required over its course, one is continually looking forward to what one is hoping to produce, and back to what one is currently doing, and has done – to try to ensure that the research is on track to achieve its goals. If we are looking forward towards how well we are answering a value question, or to bringing about some change in the world, then our decisions will be directed in ways that are likely to be less than optimal for, perhaps even obstructive to, the production of factual knowledge. This follows from the fact that there is no intrinsic relationship between the true and the good.

Let me illustrate this with an example of a case where it seems clear that a research project was driven by a desire to bring about a practical change, this leading to spurious results. A team of UK researchers were involved in organising a randomised controlled trial to investigate the effects of implementing ‘best practice’ around ‘setting’ (that is, the allocation of students to teaching groups for particular subjects on the basis of their achievement levels). The researchers’ starting point was the belief that setting in secondary schools increases social inequalities (Francis et al., 2017). However, the trial did not examine the effects of setting directly, as against ‘mixed ability’ allocation to teaching groups – a small-scale pilot had suggested that a trial of this kind was not feasible. Instead, the focus was on the consequences of different ways of implementing setting, those of an intervention devised by the researchers being compared with the effects of various practices of schools in the control group. The results suggested that the intervention made no difference to outcomes, though there had been serious difficulties involved in implementing the trial. Subsequently, members of the team analysed data from the control arm of the trial to determine whether selection for sets involved social inequalities (Connolly et al., 2019). They report a relatively small proportion of what they call ‘misallocations’ of girls and of students from some ethnic minority groups to lower sets than their test results would seem to warrant. However, the correlations involved here are of doubtful validity (Gomm, 2022), and it is unclear why the test results (produced six months previously) should be taken as a more accurate assessment of the capabilities of students at the point of allocation to sets than judgments by secondary school teachers drawing on these results and on other evidence. Despite these problems, and the failure of the trial to support their case, the authors describe their findings as ‘stark’, revealing the ‘exacerbation’ of inequalities, and they insist that their findings ‘require urgent reflection and action on the part of schools’ (Connolly et al., 2019: 895).

The authors report that the original trial was motivated by their belief that policymakers and teachers had failed to act on the results of previous research about setting, which these researchers took to have normative conclusions (notably, that setting worsens social inequities and should be abandoned). They report that: ‘the project's experimental design reflects powerful (positivist) enlightenment discourses of “science” and the search for “truth”, which we hope may support the discursive traction of our findings’ (Francis et al., 2017: 11). It looks as if, when the trial failed to deliver results supporting the change they believed to be desirable, members of the research team set out to find evidence that would support it, and did this in a manner that involved highly questionable inferences. In short, they seem to have been hunting for evidence of inequalities, which they automatically treated as inequitable, in order to show that setting is detrimental. ¹⁰

I am not suggesting that a normative orientation aimed at practical goals always causes bias, or generates false conclusions, even less that research aimed solely at epistemic goals always produces the truth; only that a normative orientation increases the risk of error considerably; and that producing factual knowledge about the social world is, in itself, very difficult (more so than some researchers apparently believe). Therefore, we cannot afford to increase the difficulties still further, given that producing such knowledge is our main responsibility as researchers.

The claim that normative assumptions are unavoidable

It is sometimes argued that social research is always normative (or ‘political’) in the sense outlined above because it necessarily depends upon basic assumptions about the nature of the social world, and how it can be understood, and that these cannot be empirically assessed and therefore arise from value commitments. Such assumptions, it is insisted, are built into the theoretical frameworks and methodological paradigms within which researchers operate, and they convey determinate evaluative implications, which can be viewed as either progressive or regressive. For example, positivism is said to be committed to an instrumental attitude towards the world, so that the resulting knowledge is used to exercise social control (this usually being viewed as undesirable). By contrast, interpretivism assumes that communication and cooperation are central to social life and is aimed at facilitating these, for example by enabling people, especially those in marginalised or subordinated positions, to voice their own experiences and perspectives. This may or may not be viewed positively. ¹¹ In short, it is suggested that there are different views about human nature – as inherently bad or good – or about human society – as consensual or inherently conflictual – built into different approaches, and these are taken to be necessarily linked to opposing political stances. While there may be some truth in this, the links here are mediated and weak, and therefore allow multiple possibilities. And, if we look at the history of many social science approaches, we find significant discrepancies. To take just one example, many of the logical positivists of the Vienna School were committed to Leftist politics, while Heidegger, often appealed to for inspiration by interpretivists, joined the National Socialists.

A currently influential version of the claim that diverse ontological and epistemological assumptions are to be found in the work of social scientists, and that these are adopted on the basis of practical value commitments, is to be found in criticism of Western social science as Eurocentric or even colonialist in character (Bhambra and Holmwood, 2021; Meghji, 2020; Scheurich and Young, 1997). But there has been a long line of arguments of this kind, treating intellectual products as reflecting and serving the interests of particular social categories: the original source was Marxism and Critical theory, denouncing the bourgeois character of social science (Blackburn, 1972; Horkheimer, 1972; Shaw, 1975); and, subsequently, there were feminist critiques of masculinist bias in prevailing forms of scientific inquiry (see, for instance, Delamont, 2003; Harding, 1991). All of these critics tend simply to assume that practical normative commitments inevitably direct social science, one way or the other, but they are mistaken about this.

While Marxists and feminists have usually sought to document the falsity of the assumptions on which conventional social science relies, whether through philosophical argument or through revealing the operation of bias in particular bodies of work, other proponents of the idea that theoretical frameworks or methodological paradigms are inevitably normative adopt a more relativistic position. One persisting influence here is Thomas Kuhn's (1962) revisionist account of the development of natural science. He argued that science relied upon paradigmatic assumptions built into exemplary pieces of research, these paradigms being ‘incommensurable’ and subject to revolutionary change when they could no longer solve the problems that investigation under their auspices was generating. This concept of paradigm was widely adopted by social scientists, although it was distorted in the process. It came to be argued that sociology is a multi-paradigm science in which there are diverse approaches that rely on fundamentally different ontological, epistemological, and/or axiological assumptions: in other words, they share no common ground, simply constituting the social world in divergent ways (see, for instance, Smith, 1989). This more relativist position was subsequently reinforced by the influence of post-structuralist philosophy. ¹²

As I noted, the key assumption here is that research relies on fundamental philosophical assumptions that are not open to evaluation in epistemic terms, so that they must be adopted on practical normative grounds. There are at least two problems with this argument. The first is reliance on a hierarchical or foundationalist model in which a set of basic assumptions determines the whole character of a social scientific approach, different assumptions necessarily leading to the adoption of different methods and producing different results. But a better metaphor in seeking to understand such assumptions is that they form a web, in which there are relationships of varying strength among the components of an approach, with none of these being fixed; and though some will be more central and others more peripheral, this can change over time (Quine, 1980). The second problem is that, while ontological and epistemological assumptions are not open to direct empirical test, we can judge their coherence and plausibility, and we can explore the consequences of adopting them as a basis for research investigations: for example by asking whether they produce what we can reasonably take to be knowledge. While it may be claimed that what counts as knowledge is itself paradigm-dependent, this is not true in the case of practical activities, where we all rely on more or less the same means of judging empirical claims (at least when we are solely concerned with whether they are true). It is hard to see why sociological research should be any different.

Another argument insisting that social science is unavoidably normative is that the language we use to talk about social phenomena, unlike that applied to physical phenomena, is inherently value-laden (see Alexander, 1989: 20; Louch, 1966; Raza and Watts, 2025; Strauss, 1953: chII). However, this is false. Even when words are evaluative in conventional usage, we can distinguish factual from genuinely evaluative components. For instance, though the term ‘inequalities’ is frequently used to refer to inequities, we could (and should) restrict its reference to factual differences between two or more things in some relevant respect, irrespective of how those differences are evaluated. ‘Inequity’ is an evaluative term whose factual component refers to an inequality of some kind but which also contends that this inequality is unjust. There is no good reason to treat these two words as having the same meaning; inequalities can be good as well as bad – for example unequal treatment of those in greater and lesser need.

Similarly, Strauss (1953: 52) is mistaken to claim that describing the behaviour of concentration camp guards towards inmates as anything other than cruel is to misdescribe that behaviour. There may, of course, be a Gricean maxim to the effect that, if we describe the behaviour of these guards solely in factual terms, we will be taken to be implying that their behaviour is not cruel, perhaps even that it is legitimate (though who in their right mind would think that?). But this is a convention that can and should be suspended for sociological purposes: what it points to is a need for disciplined reading and interpretation within research communities, as well as for the education of relevant publics who draw on their work. The distinction between factual and evaluative statements needs to be respected, and it must be recognised that use of factual statements does not automatically carry evaluative implications within the context of social research. Language practices are not immutable, they are designed to serve particular purposes, but effort is needed to ensure that they do this well. ¹³

In summary, then, sociology is not inevitably normative in terms of practical values, and there are good reasons for it not to be, for it to restrict its aim to the production of factual conclusions. But this does not mean that such values play no legitimate role within it.

Two ways in which practical values play a role in social science

While I insist that there are no good reasons to gear academic social research to practical goals, and that it does not inevitably serve such goals, there are two respects in which practical values do and should shape it: in the selection of research questions, and through the role of research ethics and prudential considerations. However, neither of these involve using practical normative ideas to define the goal of sociological inquiry.

Some final complications

I have argued that the intellectual authority of sociologists is limited to factual conclusions coming from investigations whose results have met the necessary threshold of likely validity (Hammersley, 2011: ch5). I will call this ‘academic authority’. And I suggested that exceeding this authority by pursuing normative conclusions threatens to increase bias and error in factual findings; moreover, presenting such conclusions as if they have been validated by research distorts democratic discussion and could undermine lay people's freedom to think for themselves. However, there is an additional kind of intellectual expertise that academics can legitimately claim: about the conditions required for carrying out academic work well, this expertise relying on the practical experience of doing such work. There is a parallel here with similar claims by those performing other types of social role, so I will give it the label ‘role expertise’. Democracy does not require, and should not be taken to imply (as often seems to be the case), that participants in the public sphere must be treated as all equal in expertise and therefore in intellectual authority. At the same time, the limits to these various kinds of authority, underpinned by diverse forms of expertise, must be respected and not exceeded. Furthermore, the distinctive risks of bias associated with them should be recognised.

Two further complications arise for sociology here. The first is that, while for most sociologists what I have called academic and role expertise will be quite separate from one another, this is not true of those who specialise in the field of higher education. Second, some sociologists working in applied areas can also claim occupational expertise in those areas: for example, ex-schoolteachers who have become sociologists of education, or ex-nurses who have studied sociology and who work in the sociology of health. Even here, though, I believe that the two sorts of expertise should be analytically distinguished, and their distinctive features recognised. Indeed, there may well be respects in which they conflict with one another.

Conclusion

I have argued that it is false to claim that sociology is necessarily normative in character, in the sense of being directed towards goals defined by practical value commitments or serving political interests. And I insisted that its task should be limited to the production of value-relevant factual knowledge; though I recognised that this does not rule out conditional evaluations and recommendations – these indicating what normative conclusions might be reached if some set of value priorities were to be adopted. At the same time, I outlined ways in which practical values nevertheless do play an essential role in academic sociological work even though they do not define the research task.

I began by clarifying, and emphasising the importance of, the distinction between epistemic and practical values. The former properly define the goal of sociological inquiry: as directed towards producing knowledge. And I emphasised that there is no strong correspondence between the two types of value: pursuit of what is true does not necessarily lead to good consequences, even though knowledge can be important for pursuing the good (as well as the bad). Indeed, these values, like others, can be in conflict. I outlined how seeking to redefine the goal of sociology in practical terms increases the dangers of bias in factual conclusions. I also argued that to put forward practical value conclusions on the basis of sociological research threatens to distort democratic processes, and to undermine people's autonomy: it amounts to a form of scientism. In addition, I challenged the idea that practical values are built into theoretical frameworks and methodological paradigms. I suggested that ontological and epistemological assumptions can be rationally assessed, especially in light of the findings produced by research relying on them.

It is true that sociology, and indeed all academic inquiry, relies on a set of practical normative presuppositions: it assumes that knowledge is more desirable than ignorance; that knowledge is produced through rational inquiry involving a critical orientation which refuses to accept what is not reasonably certain; and that this orientation should be tolerant of competing views, so long as there is a commitment to discovering the truth on the part of those putting them forward. Such a commitment is essential to the process of checking what assumptions can be relied on and what conclusions seem likely to be true. However, these assumptions do not define the goal of research, they are the rationale for its goal and for how it is to be pursued.

I also recognised that practical values play a subsidiary, albeit important, role in sociological research in two other respects: in the selection of research questions and through ethical and prudential considerations about how sociological work should be done. However, neither involves redefining the research task away from the exclusive operational goal of producing factual knowledge. The first relies on practical values in determining what topics are worth investigating, but research questions can (and should) still be factual in character, rather than normative. Similarly, contrary to some current views, ethical considerations operate as an external constraint on the means employed in inquiry, they should not make up the research goal. In other words, they perform a regulative, not a constitutive function. Finally, I discussed the complicating factor of role-based intellectual authority.

It is sometimes insisted that focusing solely on producing and disseminating factual knowledge is too paltry a contribution to justify the existence of sociology. Along with this, it often seems to be assumed that producing such knowledge is relatively unproblematic (for what is perhaps an example of both these misconceptions, see Simey, 1968: x). Yet in a ‘post-truth’ age, when many people appear to be gullible in their acceptance of ‘false news’ and ‘conspiracy theories’, the production of factual knowledge is more important than ever. And the idea that producing reliable factual knowledge is easy lacks credibility (Hammersley, 2025a). Equally important, any practical evaluations and recommendations presented by researchers are likely to be as questionable as those produced by others. And presenting them as the outcome of inquiry erodes the intellectual authority of academics, as well as distorting democratic discussion, and individuals’ autonomy, through spurious appeal to social scientific expertise.