1 I NTRODUCCIÓN
3.6 EJEMPLOS CON LA DISTRIBUCIÓN BETA
3.6.1 Resultados con el Proceso Markoviano
Having identified the common ontological and epistemological basis of Post-Keynesian economics, this poses the question for the appropriate methodology, in the sense of a research strategy, and consequent method(s) for scientific research in a Post Keynesian tradition.
191 In very general terms, the closed system positivist ontology and epistemology of
mainstream economics has been primarily associated with deductive or inductive reasoning, where either pre-formed hypothesis are put against empirical data to test specific propositions (theories) or scientific theories are formed through the
accumulation of verifiable facts (Blaikie 2000; Bryman 2001). Given the underlying objectivism, interested in counting and measuring observable aspects of social life and establishing causal processes, the methods applied in this approach are primarily quantitative with an interest in generalisation, replicability and objectivity (Bryman 2001).134
On the other extreme, open system constructivist approaches have been primarily associated with research strategies which aim to generate scientific knowledge from the research subject‘s account, including abduction, grounded theory etc. Given the
emphasis on social actors‘ meanings and understandings and rich contextual
information, the methods here used are primarily of a qualitative nature (Bryman 2001).
As discussed above, critical realism takes an ontological middle-ground which argues for an existing reality of underlying mechanisms and structures, which, however, only manifest themselves on the empirical level under certain conditions. The
epistemological emphasis in critical realism thus lies in uncovering these underlying real mechanisms and structures which (can) cause empirical surface events. The research strategy suggested to do so is retroduction. Critical realists argue that we can get an insight into mechanisms and deeper structures through beliefs and hypotheses about these underlying factors, which are then investigated against empirical evidence in an iterative and cumulative process (Lawson 1994; Lawson 1997; Walters and Young 1999). However, as Downward and Mearman (2003) argue, little specific practical guidance is offered how this is done in practice. According to Lawson (1997), this will depend on the specific context at hand, although some guidance can be given by realist abstraction and the existence of demi-regularities, which indicate the potential need for retroduction (Downward and Mearman 2003).135
134 Although a strict correspondence between ontology and method is implied here, we will see later that this direct association has been questioned over recent years (Bryman 1984).
135 Realist abstraction stands in contrast to instrumentalism, in which the mechanisms posited can be ideal or fictional, not subject to empirical check, and are assessed only by their ability to yield successful predictions rather than explanations (Downward and Mearman 2003).
192 Given the open, structured and organic nature of reality and the fallibility of knowledge, several authors argue that critical realist ontology and retroduction as its corresponding research strategy, require the use of a mixed-method triangulation (Dow 2001; Olsen 2002; McEvoy and Richards 2006; Downward and Mearman 2007).
Mixed-method research has been suggested for several purposes broadly summarized as triangulation, development and complementarity (Greene, Caracelli et al. 1989; Bryman 2001; Bryman 2006). In this literature, triangulation refers to the simultaneous use of different methods for the same research question with the aim to increasing the validity and reliability of results through counteracting the biases of specific methods.
Development indicates the use of one method to aid or inform research using the other research strategy. Complementarity occurs when two different methods are employed to dovetail different aspects of an investigation (Bryman 2001).
Although critical realists advocate the use of mixed-methods under the heading of triangulation, they ultimately reject the ontological underpinnings of triangulation as defined above, i.e. combining methods with the aim of compensating respective biases and thus increasing the validity of results. Downward and Mearman (2007) argue that this view implicitly assumes that methods are associated with different ontological domains. In contrast, the authors hold that quantitative and qualitative research methods differ more in emphasis than in kind as both of them necessarily invoke a degree of closure (although it could be argued that this closure is more profound in statistical testing). Consequently, they argue for a shift towards seeing methods as re-descriptive devices revealing different aspects of objects of analysis. In a similar vein, Olsen and Morgan (2005) argue that research methods should not be confounded with
methodologies; whereas the latter often have embedded in them an assumption about the nature of reality, methods are primarily to be seen as tools.
Supporting the aim of retroduction, these different tools, or methods, are crucial to reveal different features of the same layered and structured reality. For example, while quantitative methods can uncover demi-regularities in the empirical domain, qualitative methods can uncover the causal mechanisms underlying the observed events. Thus, in an argument coming close to the complementary aim of mixed-method research, the
193 aim is to construct a nexus of mutually supportive claims of reality, without the
presumption of being exhaustive, in which the whole stands distinct from its parts.
These mutually supported propositions are where mixed-method triangulation adds
―validity‖ (Downward and Mearman 2007). The use of different methods is necessary because of the open, organic and structured nature of reality and the fallibility of knowledge (Downward and Mearman 2002). Indeed, referring to Dow‘s
epistemological writings, Downward and Mearman (2002) argue that in a system where knowledge is necessarily incomplete, the use of different methods is not only possible but also necessary.
This case for mixed-method research holds also true for the Babylonian method. As discussed above, Dow highlights that true knowledge can never be attained. Given the complexity of the economic system, Dow advocates using a range of methods to gain - at least partial - knowledge of the whole. According to Dow, ―diversity of method, indeed, is the inevitable outcome of an epistemology which focuses on grounds for rational belief when knowledge is imperfect‖ (Dow 1998: 379). All that is ruled out on epistemological grounds is the assertion that a theory is anything other than partial.136
By advocating mixed-methods studies to support their ontological and epistemological background, Post Keynesian and critical realist methodology break down the dichotomy between those who advocate mixed-methods for pragmatic reasons and the
―methodological purist‖ camp, which argues that quantitative and qualitative methods constitute irreconcilable views about how social reality is constituted. In this latter view, mixed-method research would not be possible from a consistent ontological,
methodological perspective (Bryman 1984; Bryman 2001; Plano Clark and Creswell 2008). By viewing methods as mere instruments to illuminate certain aspects of a phenomenon, critical realists/Post Keynesians adopt a clear pragmatic approach, which is nonetheless entirely consistent with their ontological and epistemological view of a structured reality and fallible knowledge.
136 A third research strategy that has been suggested for Post Keynesian economics is grounded theory, where theory is developed directly from the data (which are not necessarily objective data as in the inductive method) (Lee 2002). Indeed, in its iterative process where data collection, theoretical analysis and theory building occur simultaneously, this method could come close to the retroductive method.
Danermark, Ekstrom et al. (2002), however, argue that grounded theory is ultimately too inductive.
Moreover, Downward and Mearman (2007) opine that grounded theory does not provide a strong ontological base for mixed method research.
194 Thus, the critical realist/Post Keynesian pragmatic view of methods is only appropriate if a common ontological and epistemological position is sustained (Dow 1998; McEvoy and Richards 2006). This consistency has been particularly controversial when it comes to the role of econometrics in an open system ontology (Sayer 1992; Lawson 1997).
This is so, because econometrics requires both intrinsic and extrinsic closure (Lawson 1989).
In line with the arguments for methodological triangulation, several authors have argued that the use of closed methods, such as analytical statistics or econometrics, does not necessarily violate the open system ontology. Methods are just practical devices and it is the methodology, the way research is conducted, which matters. In this vein,
Downward and Mearman (2002) argue that most forms of empirical research which want to establish regularity, including qualitative methods, imply some form of
closure.137 From an open system ontology, however, it is important to acknowledge that these are temporary, local closures in an inherently open system (Lawson 1997). This line of reasoning is also echoed by Olsen and Morgan (2005) who hold that
methodological closure does not necessarily imply realist closure.
This ―critical‖ view of econometrics also implies that certain methods are more
consistent with a critical realist methodology than others. For example, time-varying or non-parametric methods which analyse one case study are preferable to panel or cross sectional studies. In addition, this means that econometrics should be conducted
primarily for explanatory rather than predictive reasons. Finally, this view also puts the emphasis on the interpretation of economic results. For example, Olsen and Morgan (2005) write:
― ..the validity of interpretation..is what social scientists should argue about. Therefore what matters is how warranted arguments are built by the researcher using statistics.
137Indeed, even in an open system there are underlying forces which maintain or restore order (if in an indeterminate way) (Dow 1996). This is the result of (a) the existence of relatively enduring underlying mechanisms and processes and (b) economic agents seeking stability in their decision making (Keynes‘
conventions are an excellent case in point). If these two coincide, underlying mechanisms and structures might be reflected in observable, regular events on the empirical level (so-called demi-regularities in critical realist terminology) (Downward and Mearman 2003; Setterfield 2003; Mearman 2004) .
195 Our argument supports seeking surprising findings; being aware of the caveat that demi-regularities do not necessarily reveal laws; and otherwise following advice given from the ―sceptical‖ school‖ (p. 280).
However, both Olsen and Morgan (2005) and Downward and Mearman (2002) do not only build a justification, but also a positive case for the use of econometrics as part of a retroductive mixed-method triangulation. In addition to pointing to demi-regularities on the empirical level, which can then be explored for causal mechanisms, econometric results can add important insights into the causal mechanisms and deeper structures in a world of fallible knowledge. For example, according to Downward and Mearman (2002) : ―...while descriptive and historical analysis might be suggestive of the causal mechanisms themselves, the effect of their action can be assessed, and hence the
purported causal mechanism supported, with reference to more quantitative analysis‖ (p.
15). In a similar vein, Olsen and Morgan (2005) argue : ―...some aspects of relationships – e.g. liabilities, capacities, and generative mechanisms – can be revealed to observers through the use of analytical statistics...‖ (p. 266). In sum, econometrics can play an important role in adding knowledge about a multi-layered, complex, open and dynamic reality. At the same time, however, the researcher needs to be constantly aware that she is moving in an open system reality, where closure will be temporary and context-specific.