Introduction
The new directions we present derive from a relatively mature research program, one that has been in development for some 25 years. The core ideas of this research program have been elaborated in three books and is characterized by three distinctive features (Evers and Lakomski 1991, 1996, 2000). First is an argument that epistemology shapes both the content and the structure of theories in educational administration. Second is a defence of our preferred epistemology which we call naturalistic coherentism. The coherentism part says that theories are justified by a range of considerations, notably, empirical adequacy, consistency, simplicity, comprehensiveness, fecundity, and coherence in the sense that all parts fit together. Taken together, justifying a theory by appeal to both its empirical and so-called super-empirical virtues is known as coherence justification. Naturalism refers to the demand that whatever account is given of the details of this epistemology, it must cohere with natural science, or with that part of science most relevant for the study of knowledge, its dynamics and its justification. And the third feature is to work out in some detail the consequences of the first two features for accounts of administrative theory and its various components such as views of organizations, leadership, decision-making, training, and social science research methodology, to name just a few.
This research program falls within the broad tradition of scientific realism. During the 1950s when attempts to develop a science of administration dominated the scene, the philosophical model of what constitutes science was some form of positivism or, more precisely, logical empiricism. A science of administration was therefore expected to exhibit key features flowing from that model. An administrative theory would be a hypothetico-deductive structure, justified by confirming empirical evidence, with all theoretical terms admitting of operational definitions and reflecting the exclusion of ethics. That is, it should copy the structure and processes of natural science as these are interpreted by logical empiricism.
With the collapse of positivism and its variants as an account of science, the scene shifted to possible post-positivist accounts, of which there were many. Our research program is part of this post-positivist shift, but in a particular direction. For us, a science of administration is simply one that coheres with natural science and is justified by appeals to coherence criteria of justification.
In general terms, the ‘naturalism’ that we refer to in natural science is the doctrine that claims about whatever happens in the world are framed in terms, ultimately, as being the result of interactions between those entities posited by our best natural science. These entities, in whatever form they take, are currently supposed to be instantiations of mass/energy. Over time, this doctrine has had some notable successes in explaining phenomena. We no longer look to the beliefs, desires, or intentions of gods to explain and predict weather, relying instead on complex, computational intensive models of the atmosphere and its dynamics. And we no longer explain the difference between living and non-living things in terms of the former's possession of a posited ‘vital force’.
Whether the explanatory resources of naturalism are sufficient to explain ethics, particularly the giving of ethical justifications, or to explain cognition, particularly the giving of reasons, is more controversial. There is a long tradition that says natural science is an entirely inappropriate model for application to social science (Weber 1947; Winch 1958). Unsurprisingly, some of this controversy has spilled over into discussions of our research program. Sustained critiques of ideas in our three books can be found in Special Issues of the following journals: Educational Management, Administration and Leadership (1993, 21, 3), Educational Administration Quarterly (1996, 32, 3), and Journal of Educational Administration (2001, 39, 6). As a result of these and other critiques, we have been forced to modify our position on some matters, and to strengthen our arguments on others. In what follows, we sketch a number of these new developments.
Particularity versus generality
One feature of coherence justification is that it appears to favour the development of large-scale theories, or at least theories as large as those proposed by earlier logical empiricist attempts to engage in theory building. For example, a preferred theory of leadership might seem to be the outcome of an epistemological argument showing that it is the most coherent theory among alternative rival theories. This is an interpretation that could, with justification, be applied to our earlier work. In a searching critique of this work, however, Donmoyer (2001) raised an important objection. Here is how he puts it:
A commitment to utilize a number of theories rather than an all-encompassing, single theory to symbolically formulate administrative experience would at least minimize the degree of abstraction required and, hence, lessen the distance between experiential understanding and the symbolic formulations we construct to represent it.
(Donmoyer 2001, p. 561)
His argument highlights the following dilemma. Inasmuch as a theory is general, it fails to be sensitive to the particularities of administrative situations. But the particularities of context can be decisive in determining whether the advice a theory gives will work or not. On the other hand, inasmuch as a theory is sensitive to the particularities of context, and hence where it is most relevant or useful, it fails to be comprehensive in the sense required by coherentist justification.
This dilemma is serious and requires an answer. The sort of answer that has been developed involves two components. The first is to allow context to define the limits of comprehensiveness. The second is to focus on applying a coherentist epistemology thus modified to the process of building good theories. To see how it works, consider the case of defending a view of leadership that is most effective for promoting student learning. Perhaps the best known meta-analysis of research studies that captures the standard approach is that conducted by Robinson et al. (2008). The basic idea is to calculate from the research studies the effect sizes that different instantiations of leadership – in this case transformational leadership versus instructional leadership – have on measures of student learning. The result in that particular meta-analysis was that instructional leadership had the larger effect size.
But now consider the role of context by imagining two very different schools. The first is in an area of high socio-economic status (SES) and staffed predominantly by very experienced teachers, well trained and well versed in the learning needs of their particular cohorts of students. Under these conditions, the model of school leadership that works best for the promotion of student learning may well be distributed to a large number of teacher-leaders with the principal's role having very little instructional content. On the other hand, another type of school, say one in a low SES area staffed mostly by teachers with relatively modest experience, could well be greatly advantaged by having a principal whose major strength is instructional leadership. So, the question of what type of leadership most promotes student learning can only be answered when the context for exercising leadership is known or understood. What counts as the most coherent approach to leadership is therefore relative to the particularities of the schools.
Specifying the differences in contexts is one fairly broad way of arguing the particularities of leadership options. In addition, a more fine-grained, more nuanced view of leadership can be defended by looking at the details of what is required to improve student learning. In his book Visible Learning (2008), John Hattie summarizes a vast number of research studies – some 200 meta-analyses of 50,000 studies on a total of 200 million students – and gives the effect size of a range of factors in promoting student learning. So, to improve student learning, school leaders would want to target and strengthen a number of practices that have the largest effect sizes. The issues here are, first, not all schools are the same in their current practices, which means that the tasks of leaders, and possibly the types of leadership most appropriate will differ. Second, Donmoyer's dilemma breaks out again at the level of implementation because social science is too general to be relevant for the local particularities. (We take these issues up in more detail in Chapter 4.) Theories need to be developed tailored for particular contexts.
The solution we have adopted to deal with this issue is to focus on an epistemology of knowledge building. Research done by Chitpin and Evers (2005, 2012), Evers and Katyal (2007, 2008), and Evers (2007a, 2007b) supports the notion that theories in applied social science can be built up through a process of guided trial and error of the sort that Karl Popper (1979) proposed to explain the growth of scientific knowledge. We explain what we call ‘Popper Cycles’ in detail in Chapters 2 and 6, but in general terms, this problem-solving model begins with an initial problem to be solved, where ‘problem’ is to be understood as a set of constraints plus the demand that something be done. We start with an initial tentative theory for dealing with the problem and proceed with the process of error elimination that involves testing the theory to see if it meets the demands of problem-solving. The next problem to be solved arises out of the error elimination process, and the process continues over numerous cycles where each new tentative theory is the result of making coherent adjustments to earlier theories in the light of feedback evidence. Here is a simple example of how the model operates. At a school in a relatively low SES area with a large multicultural population, one factor that is thought to diminish student learning is a 20% student absentee rate on Fridays and Mondays. However, for the school's leaders it is not obvious what is to be done, especially in a time before text messaging was common as is the case in this example. There's no textbook answer. Instead, staff try out different ideas, measuring the effects of each. Table 1.1 captures this process.
TABLE 1.1 Decision-Making to Solve Absenteeism Popper cycle 1 | Popper cycle 2 | Popper cycle 3 | Popper cycle 4 |
P1: Absenteeism among students peaks at 20% on Fridays and Mondays disrupting the pacing of classroom teaching. | P2: How can we communicate with parents of students who are absent on Fridays and Mondays? | P3: How can we arrange for letters to reach parents without being intercepted by students? | P4: How can the school communicate directly with parents of chronically absent students? |
TT1: Mark the roll when school starts and then utilize all general staff to phone the home of all absent students. | TT2: Send a letter from the school to the address where students live, asking parents to contact the s... |
