Such an articulation of a mixed-methods approach requires a simultaneous expression of the meta- theoretical underpinnings which inform it. As Maxwell and Mittapalli (2010) note, there is a ‘widespread view…that the appropriate philosophical “partner” for qualitative research is constructivism, and that for quantitative…post-positivist empiricism’ (p146). However, as Scott (2007) contends, drawing upon Pring (2000), this represents a ‘false duality argument’ (p5), which has come to represent an outmoded and restrictive dogmatism, resulting in the exclusion of methodological practices which may be of real value to a particular research question. Johnson and Onwuegbuzie (2004) argue that ‘this linkage between research paradigm and research methods is neither sacrosanct nor necessary’ (p15),
articulating instead a vision of ‘a non-purist or compatibilist…position [which] allows researchers to mix and match design components that offer the best chance of answering…question[s] (Ibid). However, it is worth noting that their perspective is one which is strongly informed by the pragmatist perspective, drawing on the tradition of Peirce (1878) amongst others. Yet it could be argued strongly that pragmatism seeks to elide questions about the ontic and epistemic, rather than engage with them.
39 By contrast, critical realism offers a fully engaged ontological and epistemological stance, but with many of the methodological virtues of pragmatism. It is for this reason that Scott (2005) argues, ‘any
reconciliation between qualitative and quantitative…methodologies…has to take account of the principles…[of] critical realist meta-theory’ (p633). It is therefore important that critical realism is defined. Bhaskar (1975) delineates basic realism as the idea that ‘perception gives us access to things and experimental activity access to structures that exist independently [my italics] of us’ (p30). However, as McGrath (2004) puts it, crucially, for critical realists: ‘ontology (the way things are) determines epistemology (the way things are known)’ (p107). This departs from Kantian transcendental idealism, whereby knowledge comes from experience, but where our minds impose what that experience is (ie. the real does not exist independently of the mind, or, in some of Kant’s (1781) more ambiguous moments in respect to realism, that it doesn’t matter if they do, to our perception of them. Rather, for critical realism, the real nature of things (though ultimately unknowable) exerts an influence how we perceive them.
A key concept that then arises here is that of stratification or ‘ontic depth’ (Olsen, 2009). Lipscomb (2008) defines this well as a researchers having an awareness of the ‘stratified and complex nature of the reality they are investigating’ (p42), that there are many and layered ways of attempting to know a thing: in other words, this leads inevitably to epistemic relativism. From this, a logical consequence is methodological pluralism; if there are many and sometimes changing ways in which an object (including social objects) manifests itself, this implies that it must be viewed in multiple ways methodologically. In essence, as Scott (2007) argues: ‘critical realism is realist and critical for two reasons: objects in the world, and in particular social objects, exist whether the observer or researcher is able to know them or not [hence realist]; and secondly, knowledge of these objects is always fallible [hence critical]’ (p14). It is perhaps therefore unsurprising that the likes of McEvoy and Richards ‘emphasize critical realism’s theoretical ability to ‘circumvent many of the problems associated with paradigm “switching” (2006, p76), a view shared not least by Lipscomb (2008).
Yet in order to explore the implications of this fully, it is worth noting that early critical realists such as Bhaskar (1979) have argued (based on what is arguably a qualitative bias) that quantitative work is innately incompatible with the critical realism perspective. This has perhaps best been taken to task by Pratschke (2003) who argues that modern statistical models in the social sciences do not assume the positivist rigidity which early critical realists have perhaps naively accorded to them. For example,
40 Pratschke (2003) notes that an ‘objection that critical realists have made to statistical models concerns the validity of the assumptions implied by these models [like] homoscedasticity and multivariate normality’, but goes on to argue that this is less relevant due to ‘recent developments in statistical theory and…robustness [in response to] deviations from normality’ (p22). This argument is perhaps even more true as a consequence of innovations subsequent to Pratschke’s (2003) time of writing, with the emergence of work by the likes of Basto and Pereira (2012) and Lorenzo-Seva and Ferrando (2013) in respect to ordinal factor analysis being an example of this. Secondly, Pratschke also notes the tendency of early critical realists to regard quantitative work as occurring within a reductively ‘closed system’ whereby analytical closure can be achieved (ie. that a ‘question’ can be definitively and conclusively answered). By contrast, Bhaskar (1975) would contend that social enquiry operates within an ‘open system’, one which is fluid, irreducible and inconclusive. However, there is no such assumption with quantitative work that such enquiries are conclusive. Similarly, they (‘classic’ critical realists) also believe that in an open system it is possible for ‘demi regularities’ to occur, which are defined as ‘partial event regularit[ies] which prima facie indicates the occasional, but less than universal, actualisation ofa mechanism or tendency, over a definite region of space-time’ (Lawson, 1998, p149). Pratsche (2003) argues persuasively ‘that associations between events, attributes, actions andbeliefs that are recorded by the covariances between variables can beinterpreted as equivalent to 'demi -regularities'’ (p25); in other words, that statistical analyses such as factor analyses readily align with critical realist thinking when properly understood. Further critical realist concepts of relevance here in respect to quantitative work are abduction and retroduction, which are natural companions of ontic depth and epistemic relativism. Meyer and Lunnay (2012, online), drawing on the ideas of Danermark et al. (2002), define these succinctly: ‘abduction involves analysing data that fall outside of an initial theoretical
frame…retroduction is a method of [re]conceptualising which…identif[ies] the circumstances without which something (the concept) cannot exist…in conjunction, these forms of inference can lead to the formation of a new…theory’. Importantly here, this implies that a theoretical framework being used in quantitative research is not simply proved or disproved in the sense of deductive logic, but rather is evaluated, revised and reconceptualised on the basis of the findings, as is the case in the present study.
Drawing on the important work of Margaret Archer (1990, 1995), Scott (2005) identifies a key tenet: ‘For critical realists…the central relation of social reality is between agency and structure’ arguing that ‘social structures pre-exist agential operations, and in turn human beings reflexively monitor the social
41 been (and are being) ‘created’ in the broadest sense of the term, but these then assume an independent reality (hence critical realist) which precedes and impacts upon the agency of their original human progenitors. This leads to the conclusion, as Scott (2007) argues in a later work, that
‘methodologically…investigation can only take place at the intersection or vertex of agential and structural objects, and thus indicators that researchers use have to reflect this close relationship between the two’ (p14-15). In other words, the criterion for choosing methods must be their pragmatic ability to ‘get at’ the interplay between structure and agency, rather than any outmoded notion as a paradigmatic bias towards quantitative or qualitative. As the present study looks at the impact of social structures upon motivation (and by extension agency), this therefore necessitates a mixed methodology that can explore such.