Abstract
The present paper deals with a research methodology issue. After an introductory literature review, it presents a novel model developed to study the diffusion of educational innovations. This model does not focus on the time course, phases, or characteristics of diffusion, which can be described by various innovation indicators, but on persistent structural elements such as the actors, their relationships, and the territorial, organizational, or other entities that host the actors. The latter, which separate the actors from each other or even constitute a common space for them are called containers. The presented actor-container model (ACM) was developed to help interpret empirical data in the context of a larger research, named Innova project, dealing with the emergence, diffusion, and system-shaping impact of bottom-up innovations initiated by teachers or other local actors in the education sector. In this paper, we demonstrate the application of ACM by analyzing the responses of educational institutions (organizations) at different levels of public and higher education in Hungary, based on the 2018 online questionnaire survey database (N = 2042). The examples presented show that ACM provides a unique perspective for research on innovation diffusion by shedding new light on actors and containers, opening up new possibilities for data analysis and results interpretation. We believe that ACM can be applied not only in the context of educational innovations but also in other innovation fields.
Introduction
This article presents a novel model which can support empirical research on the diffusion of innovations in a wide range of education fields, including public education, vocational education, higher education, market-based education, and adult education. This is a network model describing the relationships of actors, in which actors with the same essential characteristics form well-defined groups. The relationships within and across group boundaries can be explored in this way. The main outlines of the model are described in the final volume of the Innova research available in Hungarian (Fazekas et al., 2021).
The dynamics of innovation diffusion is a key issue in the evolution of any education system. Knowledge about this is not only of academic significance but also of policy relevance. The importance of innovation is underlined by the fact that education can only adapt to the increasingly rapid changes of our times if its actors improve their innovation activities. Innovation diffusion plays a key role in this. Changes are not only challenges but also opportunities. Innovations can also help to make wise use of them. Previous research often focuses on how policies, reforms, and top-down innovations get diffused in an education system (e.g. Sasaki, 2018). Teacher-initiated, bottom-up innovations are rarely put into focus. Our article chose this latter approach, trying to explore the often messy and complex processes of how bottom-up educational innovations emerge and spread across an educational system. The focus of our inquiry is research methodological in nature. We propose a methodology and detail its potential use via secondary data analysis on an existing database, created in a basic research project focusing on the emergence and diffusion of educational innovations in Hungary.
Diffusion of innovations
Rogers (1962) in her seminal book, refers to the diffusion of innovations. Diffusion, from the natural sciences, means mixing substances with different states of matter without compulsion. The classical understanding of the diffusion of innovation as a decision-making process and as an agent-centered model (considering socio-economic and communication aspects) is well-known in the business and management literature regarding commercial innovations in enterprises. In this approach, an agent, after being informed about an innovation, has to decide whether to adopt it or not. It is also a decision if the agent does not obtain information or does not engage with innovations.
Greenhalgh, Robert, Macfarlane, Bate, and Kyriakidou (2004) used a continuum to describe the spread of innovations from diffusion (informal, unplanned) to dissemination (formal, planned). Starting from an unpredictable, uncertain, emergent, self-organizing process (adoption, sensemaking), through a negotiated, influenced, and enabled process (diffusion, negotiation, knowledge transfer) to a scientific, planned, regulated, and managed process (dissemination, reengineering). The continuum shows the complexity of the diffusion of innovations. There are many things to consider as the determinant of this complex process according to the systematic review of Greenhalgh et al. (2004) such as the innovation itself (e.g. relative advantage, compatibility, trialability, complexity, observability, etc.), system antecedents (e.g. structure, absorptive capacity, receptive context), system readiness (e.g. tension, fit, power balances), adopter characteristics (e.g. needs, motivation, skills), the implementation process (e.g. decision making, human resource issues, resources) and the outer context (e.g. sociopolitical climate, incentives and mandates, inter-organizational norm-setting and networks, environmental stability).
Considering the education sector as part of the public service sector, we can identify sector-specific characteristics that could influence how we think about the diffusion of innovations. First, we review some specific aspects of public sector innovation, then we proceed to discuss the specifics of the education sector.
Characteristics of educational innovations in light of public sector and service innovation studies
The intangible and interactive aspect of (public) services directly influences the emergence and diffusion of (public) service innovations (Pelkonen & Valovirta, 2015). Albury (2005) emphasizes the role of innovations in improving the services of the public sector not without its specific barriers, like harmonizing the agenda of a diverse set of stakeholders or privatization, technological shortcomings, inefficiencies in the legal framework, and the lack of market orientation (Velayati, Shabani, & Nazarian, 2020). However, Phusavat, Anussornnitisarn, Comepa, Kess, and Lin (2010) especially identified public-private partnerships as an important driver for public service innovations. In addition, Jäppinen (2015) also emphasizes the role of involving partners in co-designing and co-production service innovations. Bearing these aspects in mind, it is important to consider the specifics of the diffusion of innovations from the literature.
It must be emphasized that the diffusion of innovations in the education sector should be treated differently than the processes of for-profit organizations. Although the general model of diffusion was successfully applied in education (see for example Roger's diffusion of innovation theory applied to Czech teachers in the paper of Cirus and Simonova (2020)), the specific characteristics of the sector must be taken into consideration. Educational institutions operating in the public sphere have a more regulated environment and very different stakeholder relationships than organizations in the private sector. Schools can be considered complex adaptive systems (Keshavarz, Nutbeam, Rowling, & Khavarpour, 2010) characterized by distributed controls, emergent changes, and a nested system. As Dede (2006) puts it, it is very rare for an innovation to get diffused within a school and it is much rarer than it gets diffused between schools. While innovations in a fast-food restaurant can be easily adopted by other restaurants in the franchise, it is that much harder to get a new teaching strategy to work even for a different class in the same school. Sargent (2015) emphasized that for innovations to spread in teacher networks (especially in remote rural schools), channels of communication linked to external networks are very important. As it was emphasized by Huang (2019) as well, it is important to supplement top-down approaches with bottom-up perspectives, creating an integrated approach for the diffusion of educational innovations, taking into consideration the process-like nature of the phenomenon.
One common theme that is important for the private and public sectors as well in terms of the diffusion of innovations is the spatial aspect (Asheim & Gertler, 2009; Baptista, 2001). Taking the geographical aspects into consideration is well-aligned with contemporary views of innovation diffusion focusing on an open model of innovation ecosystem (Xiong, Kuan Lim, Tan, Zhao, & Yu, 2022). Geographical aspects also raise issues of different cultural approaches. As Zhang, Tian, and Hung (2020) found, the diffusion of innovation is deeply influenced by cultural distance as well. The proximity of stakeholders and regionality is an important aspect of diffusion, especially if we think about the role of public-private partnerships and the user-driven aspect of public sector innovations. Considering the geographical diffusion of educational innovations, the processes can be described as knowledge spillovers from boundary-crossing linkages and multiplier effects (Toh, Hung, Chua, He, & Jamaludin, 2016).
Taking into consideration that Rogers' diffusion of innovation theory also emphasizes the role of actors in the process of adaptation categorizing them from innovators to laggards, we propose an actor-container model (ACM) considering these aspects for explaining the diffusion of educational innovations. The container is a new conceptual abstraction in this model, reflecting the different spatial (e.g. geographical, territorial, or organizational) positions of the actors as described below.
Research question
The results presented in this study come from a larger research, named Innova project, dealing with the emergence, diffusion, and system-shaping impact of local innovations in the education sector. The project website in Hungarian: https://nevtud.ppk.elte.hu/content/innova-kutatas.t.6078?m=2637.
The main question of the innovation diffusion study, which was part of this large project, was which patterns and characteristics of diffusion can be observed and how these depend on the actors, the innovation, the organization, the geographical location and other characteristics of the context. A related question was whether a formal model could be constructed that could effectively support the analysis of the data obtained. Based on the following, we believe that the second question can be answered in the affirmative.
Methods
The Innova project was carried out between 2016 and 2020 in Hungary. The research methods included (1) a review and synthesis of previous relevant research, (2) a secondary analysis of some highly relevant existing databases, and (3) own empirical data collection in public and higher education and vocational training. The latter included data collection using self-developed questionnaires, interviews, and case study analyses (covering some longitudinal aspects as well).
The secondary analyses carried out in the Innova research affected the following databases: the regular national pupil achievement survey (OKM), the ImpAla research on centrally initiated development interventions, the KÖVI database on school management and organizational processes (University of Szeged), the OFI research on professional learning communities (Institute for Educational Research and Development), and some platforms for sharing good practices. For the specific research field addressed in this study, we have used the regional databases of the Central Statistical Office (KSH) and the Institute for Economic and Enterprise Research (GVI).
In the Innova project 14 complex case studies were carried out covering all major types of public and higher education institutions, from kindergarten to higher education. The protocol covered several elements of the fieldwork, such as the duration and method of data collection, the proposed interview questions, and the documents recommended for analysis. Accordingly, a minimum of two researchers spent five days in the institutions, conducting semi-structured interviews with pre-defined individuals, informal discussions, observations of lessons/activities and other areas, and analysis of available data in databases and institutional documents.
For the research, innovation is defined as a change in the practices of individuals or organizations that results in a significant departure from their previous routines or practices and increases the potential for them to become more effective or solve a problem. This interpretation is very close to the innovation definition in the Oslo Manual (OECD/Eurostat, 2018).
The surveys (2016 and 2018) focused in two cases on the innovation characteristics of the organizations and in one case on the innovation activity of the responding individuals (educators or teachers). A specific feature of this research was that the surveys always covered the entire education vertical, with all types of institutions from kindergartens to doctoral schools of universities, all completing the same questionnaire. The analysis of the diffusion of innovations is based on the organization-level survey carried out in 2018. The online questionnaire was completed in this survey by the head or a representative of the educational institution (or a separate larger organizational unit). A total of 13,809 questionnaire links were sent out to all known institutions of public and higher education, as well as to education market players, of which 2042 usable voluntary responses were received, and processed by the IBM SPSS Statistics 25 program.
The variable structure covered a wide spectrum of innovation-relevant characteristics of the institutions, some examples of which will be presented in the results section below. Such variables were, among others: the maintainer of the institution, its geographical location, its operational profile, its participation in different development programs, its learning organization characteristics, its self-perceived effectiveness, its innovation activity, the areas of professional activity affected by innovations, the drivers of innovations, the adoption and transfer of innovations, and the characteristics of a selected important own innovation.
The most important question for the diffusion analysis reads: “Can you name another institution/organization whose good practice or innovation you have adopted, learned from, and whose practice has had an impact on the functioning of your institution/organization?” For the analysis, we have also taken into account the different ways in which named institutions were designated, and their geographical location (e.g. which county they are in).
An important opportunity was to combine individual and organizational level datasets from our different surveys, linking individuals to their organizations, and including additional data from other relevant external databases. In terms of the new approach (see below), the former helps to associate actors with containers, and the latter to capture the characteristics of actors or containers. To better analyze and understand the diffusion phenomenon, we have developed a novel model. On this basis, as described later in this paper, it was possible to create several new variables from the basic data, the statistical analysis of which sheds new light on the innovation actors under study.
Results
This section gives examples of how to use the model. As the developed novel model is considered a significant result of the research, its presentation is included also in this section.
Introducing the concept of container
The territorial analysis of the diffusion of educational innovations was based on a specific logical framework. Two concepts play a fundamental role here: the notions of actor and container. The former is widely used, while the latter is novel in this context. A general model describing certain aspects of diffusion processes can be built on these two concepts as follows.
An actor in our interpretation is an entity who can be the subject of an activity (in this case, the transfer or adoption of innovations), such as an individual, a group of individuals, an organizational unit, or an organization, which is a well-identifiable unit in a given process. Research questions on actors may include how to characterize actors who excel or lag in the transfer or adoption of innovations in particular about the effectiveness and importance, size, value, or domain of innovations, and to its organizational and individual characteristics; and what about other characteristics of their innovation activity, barriers encountered and territorial aspects.
A container is a well-identifiable entity that includes a group of actors. Examples: a subsystem (sector) of education, a territorial unit, an organization, a clearly defined natural or artificial group of organizations or individuals (e.g. the group of Waldorf institutions in Hungary, or a permanent collaborating professional community of mathematics teachers in a school). Worth noting, that the word "container" may not be entirely appropriate, as it can have misleading associations in colloquial terms. The term is used in an abstract sense. Here we are talking about an entity that contains others and makes it possible to interpret boundary-crossing relations. It is a logical abstraction, which helps analysis by highlighting the relation of containment.
Since it has a well-defined boundary, internal and external connections become meaningful for each container. In the case of an internal connection, the source and the user of the innovation are located in the same container. An external connection, on the other hand, crosses the boundary of the container, as the source and the user of the innovation are located inside and outside the container. The application of the container approach is useful where boundary-crossings may occur, as it helps to examine them.
It is worth noting that in network science, the concepts that refer to different clusters, blocks or sub-networks of actors (nodes) within a network are well known (Barabási, 2002). These groups can be identified by inductive analysis of the relationships between nodes of the same type within a network, or by a deductive classification approach based on the intrinsic attributes of the nodes. ACM falls into the latter category since the notion of a container used in our model is based on the fact that actors belonging to a given container (group) are placed there by definition according to some essential property. Such a property could be, for example, the geographical region in which they live. Thus, in the ACM, actors are not of the same type but initially differ according to the container to which they belong.
A set of containers of the same kind creates the container space in which the diffusion across borders can be studied (for example, within and between counties in Hungary). A special case is when a single container and its environment are studied (for example, the capital and its environment). Of these, the organization and the group of individuals acting as a unit are dual in nature for our model: they can be both actor and container; while the other entities listed above can be only one or the other. Some containers are scalable in the ‘smaller-larger' dimension and can usually be nested along the containment relation, embedding into each other (such as districts in counties, or counties in regions). Such multi-level hierarchical structures allow for more complex analysis. For example, if we look at a large university and its environment, the internal structure of the university appears as a multi-level container system, where both internal and external relationships can be interpreted. Those relations that remain external relations even using a larger scale container space suggest more distant effects. Through these, the impact distance of each source or group of sources can be examined.
Research questions related to containers may include: how to characterize the spatial dimension of diffusion in terms of containers and what container properties play a role in this, in particular, concerning boundary-crossings; and how boundary-crossings are related to the effectiveness, seriousness, value, and other properties of innovations in different container spaces. One question linking actors to containers is how to characterize the actors in different containers that play a prominent or lagging role in the diffusion of innovations, with attention also to border crossings.
The above research questions on actors or containers as specific entities will not be addressed in this paper, as we will now focus specifically on diffusion itself. The innovation diffusion is represented in our model as a relationship between two actors (a transferor/source and a recipient/user), which either crosses the boundary of the transferor's container or not. Arrows in the illustrative figures indicate this directed connection. Figure 1 illustrates some typical examples of the actor-driven approach, showing different types of actors and containers. For some specific empirical research results on actors and containers, we refer in particular to the final volume of the Innova research (Fazekas et al., 2021).
Illustrative examples of the application of the actor-container approach for innovation diffusion (Pálvölgyi, 2020)
Note. This principle diagram shows only a few possible typical relationships. The sources or containers of sources involved in the innovation transfer are highlighted in yellowish shades, while the recipients (adopters) are highlighted in bluish shades.
Citation: Journal of Adult Learning, Knowledge and Innovation 6, 2; 10.1556/2059.2023.00084
The actor-container model
A more specific actor-container model (ACM) for the analysis of locally driven spatial diffusion of education innovation was developed using the above-outlined general approach. This can be seen as a further concretization and territorial application of the actor-container approach of innovation diffusion, presented in Fig. 2. In this special case, the respondent organizations are the actors and the containers are the territorial units (the latter considered at the same level). This construction can serve as a basis for defining several additional concepts and indicators that may be useful in the analysis of diffusion. It is worth noting that the concepts introduced below can be interpreted similarly in a more general actor-container model as well, allowing for arbitrary actors and arbitrary containers depending on the research question.
The actor-container model (ACM) for the analysis of the territorial diffusion of local education innovations as an application of the actor-container approach (Pálvölgyi, 2020)
Note. In this conceptual diagram, sources are marked in red, and users in blue.
Citation: Journal of Adult Learning, Knowledge and Innovation 6, 2; 10.1556/2059.2023.00084
The definitions of the concepts shown in Fig. 2 and other related indicators are summarized below. Further ACM concepts and indicators can be constructed following a similar logic.
User: responding organization that has named at least one inspiring organization in the survey. Indicated by a blue dot in the figure.
User density: the number of users within a given territorial unit divided by the number of responding organizations from that territorial unit. This refers to the proportion of users of the given territorial unit who named inspiring organization(s) in any territorial unit
Source: an organization that was named as an inspiring organization by at least one respondent. Indicated by a red dot in the figure.
Source density: the number of sources operating within a given territorial unit divided by the number of responding organizations from that territorial unit.
Penetration: the user density for a given source or innovation.
Transfer connection: a connection from the source that connects the source to its user (that named it in the survey). Indicated by an arrow in the figure, the arrow points to the user.
Transfer density: the number of transfer connections from named sources operating within a given territorial unit divided by the number of responding organizations from that territorial unit.
Adoption connection: the relationship between a user and the named source. This is the inverse view of the transfer connection.
Adoption density: the number of adoption connections of users within a given territorial unit divided by the number of responding organizations from that territorial unit.
Export connection: a transfer connection from the source to a user located in a different territorial unit from the source.
Export density: the number of export connections of sources in a given territorial unit divided by the number of responding organizations from that territorial unit.
Import connection: an adoption connection between a user and a source located in a different territorial unit from the user. This is the inverse view of the export connection.
Import density: the number of import connections of users in a given territorial unit divided by the number of responding organizations from that territorial unit.
Export orientation (or dominance): the number of export connections in the given territorial unit divided by the number of import connections (provided the latter is not zero). Its value is numerically equal to the ratio of export density to import density. A value of less than one indicates import orientation.
Import orientation: reciprocal of export orientation. A value of less than one indicates export orientation.
Internal connection: a relationship that connects a source and its user located in the same territorial unit. This connection does not cross the territorial unit boundary (i.e. it does not appear in the export/import balance).
Internal connection density: the number of internal connections within a given territorial unit divided by the number of responding organizations from that territorial unit. (Another theoretical possibility would be to relate the actual number of internal connections to the mathematically possible maximum number of internal links.)
External connection: a relationship that connects a source and its user located in a different territorial unit. This connection crosses the boundary of the territorial unit (is either an export or an import connection).
External connection density (or integration): the number of external connections of a given territorial unit divided by the number of responding organizations from that territorial unit. It refers to the extent to which a given territorial unit is integrated into its environment through its connections.
Closeness: the number of internal connections of the territorial unit divided by the number of external connections (provided the latter is not zero). This indicator usually gives a value greater than one, indicating a higher number of internal connections.
Openness: reciprocal of closeness.
While the concepts of export and import refer to more distant connections across territorial unit borders, the concepts of transfer and adoption differ in that they include connections within territorial units as well. How many connections of each type the actor has is a good description of the given actor, while the density indicators describe the containers. A high transfer density often (but not always) indicates a high density of sources, while a high adoption density usually indicates a high density of users. These are therefore primary indicators of inter-institutional connections in terms of innovation diffusion.
The ACM description includes some concepts that are very similar to some of the network science concepts, but not identical with them. One example is density, which is understood in network science as the ratio of actual to potential connections. This term is used in our model with different labels and characterizes the activity of actors in a given container in different contexts, as described above. It is worth noting that the theoretical definitions of the density indicators in our model can always be based on the ratio of all the connections of a given type (e.g. transfer connections) to all the actors in a given territorial unit. In contrast, however, the number of organizations responding from a given territorial unit in the survey has been included as a divisor in the above definitions, because we have now looked directly at the available sample rather than the whole population, and only those relations that were indicated by the users in the sample have been included. Due to the limitations of the given sample, these densities can only be considered as approximate estimates of the actual real densities in the population, in the sense that we are used to in statistical analyses.
The above model and related concepts provide a novel conceptual-logical framework for analyzing the diffusion of local innovations from a territorial perspective. It is important to underline that the concepts can be interpreted similarly for other types of (non-territorial) containers as well. With the concepts just presented, the diffusion of innovation can be characterized not only at different levels of territorial units (regional, county, district) but also, for example, at different levels of organizational units. These units can be described in terms of their internal and external connections. The additional indicators derived from these allow the application of a variety of different analytical perspectives. The impact of innovation sources is indicated by the penetration pattern. The application of the indicators introduced is illustrated by the following examples.
Examples of the use of the model
Before presenting examples of territorial analyses, it is worth reiterating that many types of containers are possible. For example, if individuals are the actors, then containers can be well-defined permanent groups of these actors or different levels of organizations hosting the actors. If the actors are organizations, then containers can also be different subsystems (sectors) of education. The first example applies to the latter case.
Looking at the two main subsystems (sectors) of education, public, and higher education, as two different containers, we see in our sample, that public education institutions (N = 1,668) predominantly use public education sources, while higher education institutions (N = 285) predominantly use higher education sources to receive innovations or inspirations. In this sense, therefore, these sub-systems as containers can be considered strikingly closed, as shown in Fig. 3 based on the innovation diffusion question cited above (crosstab: χ2(3) = 412.28, Cramer's V = 0.459, p < 0.001). Note: the Hungarian public education system includes kindergartens, primary schools, and various secondary education institutions. Students attend these institutions until they are about 18 years old.
The proportion of institutions using public education or higher education sources of innovation in Hungary – an example of the relative closeness of these sectors as containers
Source: Innova research 2018 survey, organizational database (N = 1953) (Fazekas et al., 2021).
The public education institutions on the left side of the diagram have a strong tendency to use public education sources, while the higher education institutions on the right side have a strong tendency to use higher education sources.
Asked question: “Can you name another institution/organization whose good practice or innovation you have adopted, learned from, and whose practice has had an impact on the functioning of your institution/organization?”
Citation: Journal of Adult Learning, Knowledge and Innovation 6, 2; 10.1556/2059.2023.00084
A very similar trend was observed when we looked at another item of the questionnaire to see where the answering institution took ideas and inspiration from when developing a more significant specific innovation of their own that they freely identified. Whereas the previous item asked about diffusion in general, this one asked about the same for the specific innovation selected. 30.7% of respondents in public education institutions who named a specific innovation received inspiration from another Hungarian public education institution and only 2.3% from a domestic higher education institution. For higher education institutions, 25.9% received inspiration from another domestic higher education institution (while 22.9% from a foreign higher education institution) and only 4.4% from a Hungarian public education institution (crosstab: χ2(3) = 222.38, Cramer's V = 0.433, p < 0.001). A foreign public education institution very rarely inspired the named own innovation: only 3.1% of the sampled domestic public education institutions and only 1% of higher education institutions cited such a source.
Before presenting ACM-based cases of territorial analyses, it is worthwhile to briefly outline the national context in which the research took place. Between the early 1990s and the early 2010s, Hungary's decentralized education system strongly supported innovation. More than a third of institutions participated in EU-funded programs to modernize classroom processes (Fazekas, 2018). Knowledge-sharing networks were created and a market for best practice was established. Hungary was ranked sixth in the OECD ranking of countries between 2000 and 2011 significantly ahead of the OECD average (OECD, 2014). After 2010, a strong process of centralization emerged. The state took over the management, control, and financing of schools from local governments, and the autonomy of school principals and teachers decreased (Semjén, Le, & Hermann, 2018). Knowledge level test results show that the performance of the Hungarian education system is low by international standards and is steadily deteriorating; the reasons include low levels of funding, low teacher pay, and a shortage of teachers (Csapó, 2015).
Research in recent years shows that even under these conditions, considerable innovation potential and activity can be identified for locally initiated small-scale innovations, with significant differences between various schools depending e.g. on organizational capacities or participating in top-down development interventions (Halász, 2018). Among the more significant innovations are the practices of foundation and private schools with various alternative programs, which have been reduced in their degree of freedom by new regulations (Langerné Buchwald, 2020), and the centrally funded Complex Basic Program (Révész et al., 2018), which is being widely distributed, and will be returned to later. Although the data was collected before the Covid epidemic, it is worth noting that the introduction of distance learning in 2020 due to the epidemic has led to a dramatic increase in the use of digital solutions and a series of innovations, as in other countries.
ACM was created in this national context, but we believe that, due to its abstract and general nature, it can also serve a useful purpose in similar innovation research in other countries as can be seen from the following. For the next application examples, we will now focus only on public education organizations as actors; and we will consider counties and the capital Budapest as territorial units as containers. In the scatterplot in Fig. 4, red dashed lines represent the mean of the export density (0.105) and the mean of the import density (0.147). These divide the whole diagram into four areas. Each field is interpreted as follows: highly integrated counties are in the top right field, more export-oriented in the top left field, more import-oriented in the bottom right field, and rather insulated in the bottom left field. While both exports and imports of innovations are above average for the highly integrated, the rather insulated counties do not perform well in either area. Near the intersection of the two red dashed lines are the counties that show a relatively balanced picture on the average level in terms of export-import compared to the others, and further away are those that display more pronounced characteristics of the respective field.
Level of export and import density of innovations in the public education sector in Budapest and by county in Hungary, based on the responses (explanation in the text)
Source: Innova research 2018 survey, organizational database (N = 626) (Fazekas et al., 2021).
Question asked: see Fig. 3.
Citation: Journal of Adult Learning, Knowledge and Innovation 6, 2; 10.1556/2059.2023.00084
In Fig. 4, the export-density position of No. 11 (Jász-Nagykun-Szolnok) and No. 4 (Borsod-Abaúj-Zemplén) counties is particularly strong. The most numerous group is the group of import-oriented counties. No. 8 (Győr-Moson-Sopron) and No. 3 (Békés) counties are the most insulated with their low values. Interestingly No. 5 Budapest's very high export density is coupled with the lowest import density.
The distribution of transfer densities and adoption densities based on public education sources and users was analyzed in a similar way to the previous one for the counties and Budapest. In Fig. 5, we have plotted the red dashed lines indicating the averages (averages: transfer density: 0.441; and adoption density: 0.411.) The four resulting fields are interpreted as follows: the top right field indicates those counties whose institutions are more integrated than those of the other counties because they are more often listed as sources and/or users. The top left field includes counties with relatively more transfer-oriented institutions, while the bottom right field includes counties with relatively more adoption-oriented institutions. Finally, the bottom left area contains those counties that are more likely to have a higher prevalence of neither behavior, i.e. schools are more insulated here in this sense.
Level of transfer and adoption density of innovations in the public education sector in Budapest and by county in Hungary, based on the responses (explanation in the text)
Source: Innova research 2018 survey, organizational database (N = 626) (Fazekas et al., 2021).
Question asked: see Fig. 3.
Citation: Journal of Adult Learning, Knowledge and Innovation 6, 2; 10.1556/2059.2023.00084
As before, in Fig. 5, the counties whose institutions show a more balanced picture are marked near the intersection of the two red dashed lines. Further away are those that display more pronounced characteristics of the respective field. The coverage of each field is similar to the previous scatter plot, but partly with the presence of other specific counties.
We cannot go into a detailed analysis of the territorial units here. Their spatial distribution is shown in the next figure. It can be stated that the pattern of the average level of innovation activity of the territorial units is quite similar to the pattern of their transfer density, as indicated by the very high correlation between these two variables (r = 0.742, p < 0.001). The innovations themselves most often focused on competence development, measurement & evaluation, environmental pedagogy, digitalization, gamification, group work, project pedagogy, differentiation, talent management, and addressing disadvantages.
The geographical interpretation of the results can be facilitated by the map below, which color-coded shows the transfer density of innovations in the public education sector in Budapest and by county (see Fig. 6), using the code numbers applied in the previous two figures. Transfer connections are particularly worth paying attention to because a higher level of innovation transfer is a relatively sensitive indicator of higher innovation activity in general (Fazekas et al., 2021).
Level of innovation transfer density in the public education sector in Budapest and by county in Hungary, based on the responses, map visualization (explanation in the text)
Source: Innova research 2018 survey, organizational database (N = 626) (Fazekas et al., 2021).
Question asked: see Fig. 3.
Citation: Journal of Adult Learning, Knowledge and Innovation 6, 2; 10.1556/2059.2023.00084
As shown in Fig. 5, No. 4 Borsod-Abaúj-Zemplén (BAZ) county ranks second to the capital in terms of transfer density. The good result of this north-eastern county (marked on the map with BAZ) is also because it is home to the Béla IV Primary School in Hejőkeresztúr, the base institution of the Complex Basic Program, which is a state-supported innovative initiative of national importance to reduce early school leaving (drop-out). The Innova research data show that the school is a very active source with a high number of transfer connections, and its inspiring effect can be detected in many counties. It achieved the highest penetration in its county (9.77%), as well as in the counties of Nógrád (4.55%), Csongrád (3.85), and Fejér (3.39%). Incidentally, Fejér and Nógrád counties are generally at the top of the import density ranking, along with a few other counties, but Csongrád county's import density is only average (see Fig. 4). It is worth pointing out that the cited innovation had a very small penetration in Budapest (0.35%).
Discussion
As seen in the previous chapter, the application of the proposed actor-container model (ACM) has led to interesting results. We now show that the model can also help interpret these results adding a novel perspective to the research of educational innovations by shedding new light on actors and containers, opening up previously unused possibilities for analyzing diffusion.
Based on the data summarized in Fig. 4 there is no county truly integrated in the sense considered here: it is striking that no county, including the capital, has simultaneously both export and import connection densities above the average of all counties. In other words, the territorial units under study are either export-oriented or import-oriented, or (relatively) insulated, the latter meaning that both indicators are low compared to the average. It is worth pointing out that we are talking here about territorial units. The chart allows plenty of scopes for some specific schools to stand out and be truly integrated in the sense that they have much richer than average export and import connections. This diagram shows the connections aggregated at the level of territorial units, positioning them among each other. The study of indicators for individual institutions would be a different level of research, where, for instance, the actor could be the teacher and the container could be the school.
Striking that the capital's very high export density is coupled with the lowest import density according to Fig. 4, which indicates a specific communication situation in boundary-crossing educational innovation diffusion. The capital's position here indicates the different conditions and the specific leading position of the capital. It is no coincidence that it is less able to adopt innovations from rural institutions, while the latter often adopt innovations from schools in the capital.
It is worth comparing the position of the capital Budapest in Figs 4 and 5. In terms of transfer density, it has the highest value, which is in line with the very high export density level shown in the previous figure. In terms of adoption density, we see a good medium value. The very low import density in Fig. 4 is therefore not because schools in the capital have fewer adoption connections, but rather because their sources are large schools in Budapest, i.e. within their own container. Similar interesting conclusions can be drawn if a larger difference in the location of a particular county is shown in these two figures (see No. 19 Veszprém and No. 2 Baranya counties).
Based on ACM we have defined the penetration numbers in different territorial units of a major government-supported innovative program, which has been rolled out from a rural base institution. In this case, it also showed that the capital is less open to innovations from outside. This may be again because the capital city enjoys a quite special position in the country, both in terms of its mass and its development, with many institutions and resources concentrated here. Furthermore, it is much more difficult to achieve greater penetration in a large city of 1.7 million inhabitants than in much smaller territorial units. In addition, the specific nature of the Complex Basic Program in the discussion, which tended to target rather rural schools, may also have played a role in this observed specific penetration pattern.
It is worth pointing out that, according to the Innova survey, the level of innovation adoption is significantly higher than the level of innovation transfer, for all types of schools. This is not a problem for the diffusion of innovations, since an education system can become innovative in such a way that the vast majority of its organizations regularly adopt innovations, while only a small proportion of them create and disseminate them. (Innovations can be collected on a central platform and communicated in various ways.) A bigger problem would be rather the reverse, i.e. the case of few adopters.
The willingness and openness of organizations to adopt innovations is, therefore, a key strategic issue for the diffusion of innovations (and thus for the development of education), and to some extent more important than the frequency of transfer behavior. The emergence and rapid spread of a large number of innovative solutions are better facilitated by mass adoption activity than by the repeated creative re-discovery of identical or similar innovations in different organizations (without being informed in any meaningful way about the good practices of others). Of course, selecting the right innovation and, in particular, adopting it competently, to one's own circumstances requires a significant amount of expertise and creativity. Actors must be given here some degree of freedom, and professional room for maneuvers to be successful, and they can be appropriately supported and motivated to do this. As the Innova case studies have shown, the vast majority of innovation takeovers involve some degree of creative adaptation. The successful professional adaptation of any innovation can be considered a full-fledged innovation for the organization, regardless of the extent to which the original solution is modified. It is no coincidence that both the Oslo Manual (OECD/Eurostat, 2018) and our definition has been interpreted innovation as a novelty in the own practices of a given organization.
Conclusion
Based on the experience of using ACM in our research, we conclude that this novel model describes some important aspects of innovation diffusion and provides valuable new possibilities for data analysis and results interpretation in related empirical studies. Moreover, we believe that ACM can be applied not only in the context of educational innovations but also in other innovation fields.
In terms of practical application, the research carried out concludes that a web-based repository of educational innovations and innovative good practices can be particularly effective in supporting the dissemination and diffusion of innovation. Such a repository can provide a very useful service in any country, or even internationally. It can help a relatively small number of resources to serve a very large number of users. The technical conditions for this are relatively easy to provide. The web-based repository could also be linked to a database in which diffusion data of innovations automatically generated by use would be ongoing collected and stored for further analysis to support education research and evidence-based policymaking. With GDPR compliance (see: General Data Protection Regulation https://gdprinfo.eu/), the emergence, characteristics, and diffusion of innovations can be monitored and analyzed in this way (See Fig. 7). The resulting database can be compared with other relevant databases to explore, among other things, the relationship between the characteristics and diffusion of innovations and other parameters of the education system. The ACM presented in this paper is well-suited to support all of this.
The principle of a web-based repository of educational innovations to support innovation diffusion, building database for research, and founding evidence-based policymaking (own edited figure)
Citation: Journal of Adult Learning, Knowledge and Innovation 6, 2; 10.1556/2059.2023.00084
As regards the limitations of current research, the following should be highlighted. Although ACM is a rather abstract model, and therefore widely applicable, the specific empirical research in which it has been developed and tested has, of course, several limitations. These include the fact that, although the sample is large, it cannot be considered nationally representative as it is based on a voluntary response, and that relatively few questions focused on the innovation diffusion itself. However, these are not relevant limitations for the demonstration of the applicability of the model.
More importantly for the use of the ACM, the current research has focused on single-step diffusion connections (and has not identified, for example, chains or typical groups of closely interacting actors). However, at sufficiently high connection intensities, more complex patterns of connections between agents can arise, such as those known from sociometry. In the present case, few such relationships between entities were detected, but none were explicitly explored. The number of responses related to the diffusion of innovation was not high enough for such more complex relationship patterns to emerge frequently. Under a different research set-up, for example, considering teachers as actors, there would be a greater chance of discovering such a dense connection network reflecting innovation diffusion between and within more closely collaborating schools. To describe this, the ACM could easily be further developed by adopting concepts known from sociometry. This could even be a promising direction for further research, allowing the later application of network science methods.
Acknowledgement
Project no. 115857 has been implemented with the support provided by the National Research, Development, and Innovation Fund of Hungary, financed under the OTKA-K_16 funding scheme.
References
Albury, D. (2005). Fostering innovation in public services. https://ssrn.com/abstract=650420.
Asheim, B. T., & Gertler, M. S. (2009). The geography of innovation: Regional innovation systems. In J. Fagerberg, & D. C. Mowery (Eds.), The oxford handbook of innovation (pp. 291–317). Oxford. https://doi.org/10.1093/oxfordhb/9780199286805.003.0011.
Baptista, R. (2001). Geographical clusters and innovation diffusion. Technological Forecasting and Social Change, 66(1), 31–46. https://doi.org/10.1016/S0040-1625(99)00057-8.
Barabási, A. L. (2002). Linked. The new science of networks. Cambridge MA: Perseus Publishing.
Cirus, L., & Simonova, I. (2020). Rogers’ diffusion of innovation theory applied on primary education: Case study of Czech teachers. 2020 International Symposium on Educational Technology (ISET). https://doi.org/10.1109/ISET49818.2020.00017.
Csapó, B. (2015). A magyar közoktatás problémái az adatok tükrében : értékek és viszonyítási keretek. Iskolakultúra, 25(7–8), 4–17. https://doi.org/10.17543/ISKKULT.2015.7-8.4.
Dede, C. (2006). Scaling up: Evolving innovations beyond ideal settings to challenging contexts of practice. In R. K. Sawyer (Ed.), Cambridge handbook of the learning sciences (pp. 551–566). Cambridge, England: Cambridge University Press.
Fazekas, A. (2018). The impact of EU-funded development interventions on teaching practices in Hungarian schools. European Journal of Education, 53(3), 377–392.
Fazekas, Á., Halász, G., Horváth, L., Pálvölgyi, L., Balázs, É., & Antoni-Alt, P. (2021). Innováció az oktatásban. Budapest: Akadémiai Kiadó. https://doi.org/10.1556/9789634547143.
Greenhalgh, T., Robert, G., Macfarlane, F., Bate, P., & Kyriakidou, O. (2004). Diffusion of innovations in service organizations: Systematic review and recommendations. The Milbank Quarterly, 82(4), 581–629. https://doi.org/10.1111/j.0887-378X.2004.00325.x.
Halász, G. (2018). Measuring innovation in education: The outcomes of a national education sector innovation survey. European Journal of Education, 53(4), 557–573.
Huang, J. S. (2019). Diffusing innovative pedagogies in schools in Singapore: case studies on school leaders’ diffusion approaches and their rationalisations. Innovations in Educational Change, 205–222. https://doi.org/10.1007/978-981-13-6330-6_10.
Jäppinen, T. (2015). How to manage a service innovation process in the public sector: From Co-design to Co-production. In R. Agarwal, W. Selen, G. Roos, & R. Green (Eds.), The handbook of service innovation. Springer. https://doi.org/10.1007/978-1-4471-6590-3_32.
Keshavarz, N., Nutbeam, D., Rowling, L., & Khavarpour, F. (2010). Schools as social complex adaptive systems: A new way to understand the challenges of introducing the health promoting schools concept. Social Science & Medicine, 70, 1467–1474. https://doi.org/10.1016/j.socscimed.2010.01.034.
Langerné Buchwald, J. (2020). Az alternatív iskolák helyzetének változása a rendszerváltástól napjainkig. Iskolakultúra, 30(1–2), 70–88.
OECD (2014). Measuring innovation in education: A new perspective, educational research, and innovation. Paris: OECD Publishing.
OECD/Eurostat (2018). Oslo manual 2018: Guidelines for collecting, reporting and using data on innovation. In The measurement of scientific, technological and innovation activities (4th ed.). Paris/Eurostat, Luxembourg: OECD Publishing. https://www.oecd.org/science/oslo-manual-2018-9789264304604-en.htm.
Pálvölgyi, L. (2020). Az oktatási innováció terjedésének értelmezési modellje .Unpublished manuscript. Budapest: ELTE PPK Innova Project.
Pelkonen, A., & Valovirta, V. (2015). Can service innovations be procured? An analysis of impacts and challenges in the procurement of innovation in social services. Innovation: The European Journal of Social Science Research, 28(3), 384–402. https://doi.org/10.1080/13511610.2014.999026.
Phusavat, K., Anussornnitisarn, P., Comepa, N., Kess, P., & Lin, B. (2010). Service innovation and organizational development through public-private partnership. International Journal of Management and Enterprise Development, 8(4), 383–397. https://doi.org/10.1504/IJMED.2010.033621.
Révész, L., Nagy, K. E., Szivák, J., Rapos, N., Török, B., & Mrázik, J. (2018). A Komplex Alapprogram Koncepciója. Eger: Líceum Kiadó.
Rogers, E. (1962). Diffusion of innovations. New York: The Free Press.
Sargent, T. C. (2015). Professional learning communities and the diffusion of pedagogical innovation in the Chinese education system. Comparative Education Review, 59(1), 102–132. https://doi.org/10.1086/678358.
Sasaki, M. (2018). Application of diffusion of innovation theory to educational accountability: The case of EFL education in Japan. Language Testing in Asia, 8(1). https://doi.org/10.1186/s40468-017-0052-1.
Semjén, A., Le, M., & Hermann, Z. (2018). The goals and consequences of the centralization of public education in Hungary. Acta Educationis Generalis, 8(3), 9–34.
Toh, Y., Hung, W. L. D., Chua, P. M.-H., He, S., & Jamaludin, A. (2016). Pedagogical reforms within a centralised-decentralised system: A Singapore’s perspective to diffuse 21st century learning innovations. International Journal of Educational Management, 30(7), 1247–1267. https://doi.org/10.1108/IJEM-10-2015-0147.
Velayati, R., Shabani, E., & Nazarian, A. (2020). Hope for the best, prepare for the worst: Barriers to service innovation. Qualitative Market Research, 23(4), 843–863. https://doi.org/10.1108/QMR-05-2020-0062.
Xiong, B., Kuan Lim, E. T., Tan, C.-W., Zhao, Z., & Yu, Y. (2022). Towards an evolutionary view of innovation diffusion in open innovation ecosystems. Industrial Management & Data Systems, 122(8), 1757–1786. https://doi.org/10.1108/IMDS-11-2021-0686.
Zhang, H., Tian, M., & Hung, T. K. (2020). Cultural distance and cross-border diffusion of innovation: A literature review. Academia Revista Latinoamericana de Administración, 33(2), 241–260. https://doi.org/10.1108/ARLA-10-2018-0239.
