By Dr Graham Downes and Dr Beth Mitchell
‘There should not be more things dreamt of in my philosophy than there are in heaven or earth’. Nelson Goodman All too often, problems in education fall back onto teachers. The deficit ideas of poor communication, lack of innovation, and lack of impact, often lead to solutions being proposed, such as extra training courses. The growing emphasis on ‘what works’, improvement science, measuring impact, and other discourses associated with performativity, can lead to unintended consequences such as the de-professionalization of teachers. In terms of social justice, teachers are given short shrift, accused of lacking skills that can be measured, overriding experience and professional judgement. Perhaps the solution is less to do with filling in perceived cracks, and more to do with how we conceptualise causality in education. To shift away from current notions of mechanistic, technological cause and effect in education, one might draw from medicine and epidemiology to shift the thinking from ‘impact’ to ‘risk’. Causality has become something of a dirty word in in the philosophy of science. As van Imwagen observed, causation is a 'horrible little word…. a morass in which I for one refuse to set foot. Or not unless I am pushed' (van Inwagen, 1983: 65). Explaining the term has been associated with complications and contradictions for many years, at least as far back as David Hume’s infamous ‘riddle of induction’. Correlations between observed events do not necessarily mean that there is a causal connection between them. Causation, Hume concluded, was ultimately a matter for common sense. In the sciences writ large, the tendency has been to shun the word, leaving it to the domain of folk theory in favour of the pursuit of a more sophisticated scientific language that does not include causation. The term has also all but disappeared from the various methodologies that constitute the social sciences. Often associated with ‘instrumental rationality’ and ‘positivism’ the term is placed in opposition to more normative, ideographic methodologies that foreground understanding and insights into complex worlds. The fields of education and healthcare often borrow ideas from each other. As mentioned before, one of the drivers for this is being public sector bodies. Both fields require a high level of accountability for public spending and reporting to government. Both rely on the justification of budget allocation and the demands for evidence to support decisions that will channel resources into effective strategies. Both fields also rely on similar approaches for evidence-based practice. However, whilst this might be considered appropriate for randomised control trials (RCT) for developing new medicines, this hierarchy falls short when researching more sociocultural interactions that take place in both healthcare and education. In healthcare, causality is traditionally associated with discourses of evidence-based practice, randomised controlled trials and more recently, evaluation through realist approaches, sometimes referred to as ‘what works’ (Pawson and Tilley, Greenhaghl, Dixon-Woods). Evidence hierarchy reinforces the status of more scientific and systematic approaches to healthcare research, with the British Medical Journal (BMJ) recently declaring a moratorium on non-scientific articles. Extending this critique, it has been argued that the positivistic, scientific paradigms are not always the most appropriate ways to research healthcare. A case in point would be the Surgical Safety Checklist (SSC). In a recent review, it was argued that nine years of empirical research has accumulated, but has brought us no closer to answering the question of the efficacy of the SSC in terms of morbidity and mortality. Repeatedly, the studies cited multifactorial effects, some of which are sociocultural in nature. This, along with an open letter to the BMJ protesting against its decision not to publish non-scientific articles, highlights the need for approaches that challenge the input/output, pre/post, before and after, scientific empirical approaches that inform so much research in this area. The conceptualisation of causation in epidemiology has long been countered with more descriptive accounts. Harman makes the case from Latour that causal chains cannot be abstracted from other actors (in epidemiology this might be referred to as ‘context’). There is also the argument that these chains are stuck together by other actants (for example, in Pandora’s Hope, Latour describes how Joliot connected policy and atoms). As a consequence, Harman argues that connections always have something else in-between. This conceptualisation of causality presents causality as ‘vicarious’ (mediated through other actors), ‘buffered’ (not in direct contact) and ‘asymmetrical’ (some actants are considered more important than others). In addition to this, is the idea that ‘cause’ does not directly lead to ‘effect’ and that there are many other possibilities that are not realised. Law refers to this as ‘collateral realities’. These could be abstract or concrete. For example, Nietzche’s writing machine was a typewriter in the shape of a ball, which was designed for its transportability, but looks nothing like later typewriters and keyboards. The big ‘so what’ in all of this is that, although it is recognised that current approaches fall short of what is needed in healthcare and education (and this is a widely held view given the number of signatures on the open letter to the BMJ) it is still difficult to articulate causality in any other way when it comes to empirical research design. One area that has come to the fore more recently, and certainly in the field of epidemiology is counterfactual theory. Initially developed by early theorists such as Nelson Goodman, the approach attempts to work with a subset of distinctive conditionals in which the antecedent is always false (in many cases the consequent can also be false but this is not necessary). Cohnitz and Rossberg (2006) use the following example of a counterfactual: if the match had been struck, it would have lit. Here both the antecedent and consequent are false: neither the match was struck nor did it light. Interestingly, despite both being false, the relationship between antecedent and consequent is inferred as positive (when one occurs, so will the other). Counterfactual theorists have tended to highlight the causal dispositions of objects rather than their causal properties. Dispositions refer to inner states of things and, specifically, about the possible states of any given object (Goodman, 1983). This is similar but not necessarily the same as a property of an object. For example, a disposition of the match is that it is flammable whilst a related property (predicate) of the object could be that it is alight. For Goodman, the problem here is: under which circumstance is it acceptable to extend dispositions as possible properties? For example, under what conditions is it acceptable to extend the disposition of the match ‘flammable’ to the property of the match ‘alight’? Goodman’s solution to this problem is to redefine dispositions as projectable properties of objects. Such accounts of counterfactual causality are predicated on a very different set of assumptions to those that underpin regularity theories of causality. These assumptions emanated from the modal realist (many worlds) theory that was developed by David Lewis. The key axioms of modal realism are thus:
(Lewis, 2001) What is distinctive about Lewis’ model is the equal status is gives to our world (actual) and all possible worlds (real). In other words, all possible worlds are as real as our world. The problem is that, when one considers all possible states of all possible objects, the number of possible worlds is almost infinite. Goodman’s work addresses this problem by utilising the notion of manifest predicates. With reference to Liebnitz, not all such dispositions are projectable into the future: some dispositions are necessary (true in all worlds) whilst others are possible (true in at least one). Thus the match is flammable because it is burning in at least one world, even if it is not burn in our actual world. This leads to issues of projectability i.e. how projectable dispositions are into future events. Many counterfactual methods therefore have a strong link with theories of probability as researchers attempt to establish the various combinations of projectable dispositions and their likelihood of occurring. So how does this deeply theoretical discusion manifest itself within the world of research and educational research in particular? One aspect of research that we have noticed is that different fields of study have different tolerances to counterfactuals. One interesting aspect of this milieu is the way these tolerances shift depending on the discursive framing of the original problem. In particular, we have noticed that approaches that focus on ‘risk’ (amelioration) have a propensity towards counterfactual models whilst those that focus on ‘impact’ (affirmative intervention) tend to shift towards regularity models. If we are to believe Foucault, the reasons for these propensities are deeply rooted in the history of discursive formations. In Birth of the Clinic, Foucault maps out the historical development of French medical practices that evolved around the centralising process of ‘the gaze’ (the shift from religious conceptualisations of truth to the revelatory capacities of the human eye and a process of watching. Foucault goes on to establish the empowering nature of the gaze in Discipline and Punish: the gaze establishes timeless and unbounded truth; to gaze is to assume a position akin to omniscience. As Foucault observes, this discourse is integral to medical practices and is inexorably linked to the emergence of the concept of a state, as medical research extended into the realm of the social through projects to stop the spread of diseases. But the gaze is also timeless: it gives license to see all possibilities devoid of time and hence the propensity towards counterfactuals. A quick look through any introductory text on epidemiology reveals the foregrounding of the concept within the discipline. For those examining the spread of disease, the question of other possibilities is entwined with the various interventions being considered. This opens up all sorts of possibilities in terms of opportunities, causal chains and the causes of non-events (e.g. the white ball caused the red ball not to move because it hit the yellow ball that would have otherwise hit the red ball). Conversely, to be gazed upon is to be the object rather than sense maker. To act is a weaker form of truth than to watch, to ‘do’ requires judgement from the knowing gaze of another. And herein lies the difference: the prima facie unit of difference in education is the teacher/educator and their effect on other minds, whereas the prima facie object of difference in medicine is disease and its effect on the human body. And they are treated in qualitatively different ways- the apparent agency of the teacher converts into an excessive focus on singular casual relationships, which, in turn, generates a qualitatively different type of discourse. A good example of this is the work of John Hattie (2017). Celebrated for his use of ‘big data’, Hattie plots the influences and effect sizes related to student achievement. Top of his league table is collective teacher efficacy-the collective belief of teachers that they can have a positive impact on pupil attainment. This appears to be an affirming message -all teachers can create positive outcomes through their own self belief- and it certainly resonates with wider liberal assumptions about the nature of the individual and their social responsibilities. But this approach also serves to largely omit the possibilities for counterfactual explanations of pupil attainment. Hattie’s model is often interpreted through the lens of regularity theory- a rule that teacher attitudes generate positive outcomes in terms of attainment. A counterfactual approach would suggest this is unlikely given the significant number of additional causal elements, often constituted as causal chains, which can possibly have an impact on the interactions between teacher and pupil. This is as much the product of the filtering of Hattie’s findings as it is of Hattie’s work itself as educators and policy makers look to apply data in ways that have an impact on educational contexts. So maybe we in education can learn a little from epidemiology in particular. Maybe we should focus a little more on ameliorating the multiple risks to education attainment and less on single interventions and impact. It may not have the rhetorical attraction that a focus on individual actions has but it may just produce more effective results. Bibliography Cohnitz, D. and Rossberg, M. (2006) Nelson Goodman. Philosophy now. McGill-Queen’s University Press. Goodman, N. (1983) Fact, Fiction, and Forecast. Harvard University Press. Hattie, J. (2018) Hattie Ranking: 252 Influences And Effect Sizes Related To Student Achievement. [Online] Available from: https://visible-learning.org/hattie-ranking-influences-effect-sizes-learning-achievement/ [Accessed 26 May 2018]. Hume, D. (1752) Political discourses (Google eBook). Lewis, D. (2001) On the Plurality of Worlds. [online]. Wiley. Usher, R. (1996) Understanding Education Research. In: David Scott & R. Usher (eds.). Understanding Educational Research. pp. : 9–32. doi:10.4324/9780203131923.
0 Comments
Leave a Reply. |
Details
Archives
September 2018
Categories |