University of Groningen Turning the kaleidoscope Oosterhoff, Maria

(1)

University of Groningen

Turning the kaleidoscope

Oosterhoff, Maria

DOI:

10.33612/diss.160821650

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from

it. Please check the document version below.

Document Version

Publisher's PDF, also known as Version of record

Publication date:

2021

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):

Oosterhoff, M. (2021). Turning the kaleidoscope: multiple enactments of professional autonomy in early

childhood education. University of Groningen. https://doi.org/10.33612/diss.160821650

Copyright

Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

(2)

◊

Chapter III

(3)

(4)

Methodologies

As briefly outlined in the introductory chapter, the methodology in this PhD project was built through an iterative process. Three dif-ferent studies were carried out, each drawing on a distinct method-ological approach; each of these aligning with distinct and relevant theoretical frameworks and different research questions (see Figure 3.1). All three of these approaches departed from diverse, mutually contradictory, ontological assumptions.

Firstly, a qualitative study was conducted aimed at exploring the

landscape. This study was initiated within a constructivist paradigm that considers realities to exist in the form of socially and experiential-ly based mental constructions (Lincoln et al., 2018) and emphasizes the importance of multiple subjective realities as an important source of knowledge-building (Hesse-Biber, 2010). In line with this, inductive strategies and qualitative techniques were used to explore the social context in which the research issue was located, to capture the lived experiences of the respondents and to give them a voice. Eight experi-enced ECE teachers were interviewed. The data was analysed themati-cally (Braun & Clarke, 2006) with the help of Atlas.ti (Friese, 2014).

The second, quantitative study, was aimed at generalizing and

theorizing results of the first study. In this study new questions were asked, stemming from a different ontological understanding of social reality. These understandings, generally considered to arise from a (post)positivist paradigm, privilege hypothesis testing and causal-ity as the most important goals of social inquiry (Hesse-Biber, 2010). In this second study, I was interested in generalizations, building on primarily deductive strategies and using quantitative techniques for data-gathering and analysis, with the aim of approximating an objec-tive, ‘real’ reality (Lincoln, et al., 2018, p. 111). Generalizing questions focused on some of the influential actors. This second study also al-lowed the investigation of the relationship between different theoreti-cal concepts of autonomy. Survey data were analysed with Structural Equation Modelling (SEM) using LISREL 8.80 (Jöreskog & Sörbom, 2006) to test a hypothetical model of the assumed relationships.

In the third study, I went on a sociomaterial journey aimed at

speaking with things (Adams & Thompson, 2016). Another methodo-logical shift occurred with this study, with which it became clear that a

(5)

quantitative/qualitative distinction would no longer suffice for making sense of the methodological stance. Actor-Network Theory (ANT) (Latour, 1987, 2005; Law 1987, 2009; Mol, 1999, 2002) seemed to reso-nate with this apparently active role of things that I had seen in the first study. The particular ontology that comes with ANT contradicts both of the worldviews that underpinned the first two studies. As out-lined in Section 2.2, ANT shows how professional practice is produced by a particular gathering of people and things and, as a consequence, is always specific, contradictory and, moreover, in a constant state of flux. The aim of the study was to capture the emergent and co-constitutive character of reality by attending to the everyday socioma-terial work in situated practice (Hultin, 2019). I then realized that the diverse studies were not adding to a more complete picture of reality, but rather to a fragmented picture, and this seemed of a greater value than striving for certainty or completeness. I will come back to this in more depth in the two final chapters (Chapter VIII and IX), where the yields of the project as a whole are harvested.

The connections between the three studies were quite straight-forward: I conducted successive independent studies on the basis of separate research questions that emerged from the first study, which took a mixed method approach that can be labelled a ‘basic explorato-ry sequential design’ (Creswell, 2014, p. 6). Although all studies were connected and contributed to the understanding of the central issue under investigation, each study also stands on its own. Each study has its own specific research questions, and methods chosen in line with these questions. In this project, each study was, therefore, designed as a separate study.

This chapter first provides a methodological justification of the project as a whole (Section 3.1). The following three sections (3.2–3.4) aim to clarify the research methods that were applied in each of the three separate studies.

52 Turning the kaleidoscope ◊ Chapter III Methodologies

Figure 3.1 Overview of the research project: professional autonomy in ECE

Study 1 Exploring the landscape – Inductive approach Data: Open semi-structured interviews

Analysis: Thematic Analysis

Study 2 Generalizing and theorizing results – Quantitative approach Data: Survey

Analysis: Frequency distributions/Structural

Equation Modelling

Study 3 Speaking with things –

Sociomaterial approach

Data: Initial interviews and addititonal object

interviews

Analysis: Posthuman heuristics – Following the

(6)

3.1 Research design and methodological justification

The main feature of the research design as a whole is its emergent character. There was an openness to emerging research questions, to theoretical frameworks and to methodologies. The full research pro-ject unfolded throughout, based on intentional decisions as well as circumstances and opportunities that arose during the process. This stance of openness to this research project enabled the determination of several directions of inquiry, a focus on different actors, the appli-cation of diverse methodological approaches and the use of a variety of theories to answer questions that emerged from the first phase of the study. As a result, a methodologically heterogeneous research pro-ject emerged.

In this section, I will first outline the reasons for choosing an emergent approach. Next, the consequences of this choices are dis-cussed, elaborating on the merits of combining different paradigms in one overarching research project.

Choosing an open and heterogeneous approach

There were several reasons to start with an open research atti-tude and to maintain this stance as the study proceeded. Firstly, there were ethical and political reasons to start the study with an open re-search question and an explorative approach. The experiences of the teachers expressed during the round tables (see Section 1.4), which pointed out how ‘others’ often dictated or attempted to dictate what happens in their classroom, were taken seriously. On this basis, it seemed ethically inappropriate to become yet another outsider im-posing a specific theory on their practices. Instead, it was considered ethically appropriate to begin with the lived experiences of the teach-ers. Furthermore, the study, which arose within the framework of the professorship Early Childhood Studies, had an overt political agenda. The study was initiated to broaden the understanding of contempo-rary concerns in ECE; to add to the critical debate which aimed to bring change; and with the ultimate goal of being able to explore ways to support teachers and prospective teachers to enhance their profes-sional autonomy. In line with this last goal, it seemed appropriate to provide the teachers themselves with an agentic role in the research, highlighting their experiences, the issues they faced and the concerns they had, taking these as a starting point for further research. Ac-cording to Lincoln et al. (2018), inviting research participants to take control, by nominating questions of interest for example, ‘is a means of fostering emancipation, democracy, and community empowerment’ (p. 135). By taking this turn, the project became part of the political agenda of the ECE teachers themselves. In the interviews, it became

(7)

clear that participation in the project was a political choice of the re-spondents. They endorsed the relevance of the study and embraced participation in it as an opportunity to make their voices heard.

Secondly, apart from an openness to emerging research ques-tions, there were reasons to aim for methodological openness. Basical-ly, this openness was required because emerging research questions might compel the use of specific methods or methodological choices. However, there were more fundamental, reasons for methodological openness and for applying methodological diversity. Chapter II illus-trated how the educational context is characterized by complex dy-namics, in which many stakeholders with differing perspectives are involved. In such an environment, a combination of multiple methods or methodological frameworks can add to a richer understanding of the issues studied than can monodisciplinary approaches (Popa & Guillermin, 2017). Morse and Chung (2003) argue that all methodo-logical frameworks are inherently limited in scope, as they dictate what counts as legitimate data and foreground particular aspects, thus silencing others. Assuming, however, that there are many legitimate types and sources of knowledge, epistemological pluralism and the rejection of ontological monism enables the capture of multiple types of knowledge (Popa & Guillermin, 2017). In addition, undertaking multiple studies also enables a focus on the different actors that might appear on stage, as well as the opportunity to work with different the-oretical perspectives that resonate with the issues that emerge as the research unfolds (Morse, 2003).

Thirdly, there were reasons to strive for methodological open-ness and diversity that had to do with maximizing the opportunities for learning. Popa and Guillermin (2017) argue that combining meth-odological approaches from different paradigms allows for transform-ative reflexivity, which purposefully engages with difference in order to learn from it. Issues of ontological and epistemological differences, as I agree with Lincoln et al. (2018), can create space for dialogue and, in the wake of this, confluence might occur. Over the course of the project, the value of such dialogue became apparent for an under-standing of professional autonomy in ECE practice, in which multiple knowledges seem to compete.

To substantiate the emerging methodologically heterogenous ap-proach, I will sketch the advantages of combining methodologies from different paradigms in this PhD research process, by briefly outlining the transformative reflexivity that was generated by this heterogene-ous approach.

Applying approaches from different paradigms

(8)

as-sumptions about the nature of existence (ontology), leading, in turn, to their perspectives on the nature of knowledge-building (epistemol-ogy) (Hesse-Biber, 2010). These sets of beliefs, interpretative frame-works or paradigms guide the researchers’ action. As Hesse-Biber (2010) puts it: ‘Methodologies lead the researcher to ask certain re-search questions and prioritize what questions and issues are most important to study’ (p. 11). Paradigmatic assumptions also guide the choice of particular methods. However, the choice of methods is not bound to a methodology; rather, methods are shaped by the methodo-logical framework in which they are applied (Hesse-Biber, 2010).

Paradigms are built throughout researchers’ professional biogra-phy, their educational training and their research experience in the scholarly communities with which they engage (Creswell, 2013; Den-zin & Lincoln, 2018). As Law (2004) states: ‘Scientists have a culture. They have beliefs. They have practices’. Somehow, out of this ‘they produce knowledge, scientific knowledge, accounts of reality’ (p. 19). Moreover, research communities differ in their methodological scope. Some communities are narrowly focused on a specific approach, oth-ers are more eclectic and borrow from many disciplines (Creswell, 2013). Some researchers are foundationalists, arguing that there are certain final, ultimate criteria for testing results on their truthfulness; others, the non-foundationalists, argue that such ultimate criteria do not exist, but are negotiated within a certain community at a certain time and under certain circumstances (Lincoln et al., 2018).

The literature that discusses different paradigms often makes a dualistic division between quantitative and qualitative approaches to research (Hesse-Biber, 2010), although it is acknowledged that the two approaches are not mutually exclusive and that the boundaries be-tween qualitative and quantitative research are often blurred (Landrum & Garza, 2015; Lincoln et al., 2018). Scholars also point to the fact that within these broad categories there is a multitude of differ-ences between philosophical assumptions (e.g., Lather, 2006; Tracy, 2010). Furthermore, particular methodologies may prefer particular methods but the choice of methods is not bound to a methodology (Hesse-Biber, 2010). In a broad sense, however, qualitative and quan-titative approaches are understood in terms of different types of data gathered (non-numeric and numeric), knowledge claims (descriptive knowledge about ‘meaning’ and relations of magnitude between vari-ables) (Landrum & Garza, 2015) and methods of data-gathering and analysis (qualitative or quantitative techniques) (Hesse-Biber, 2010). Theory on mixed methods is mainly concerned with how to integrate these two broad categories, qualitative and quantitative techniques, in such a way that the complementary strengths of and differences in both techniques contribute to a more comprehensive understanding

(9)

and more complete description of the often complex reality that is under investigation (Creswell, 2014; Hesse-Biber, 2010; Landrum & Garza, 2015; Morse & Chung, 2003; Plano Clark, 2017). The function of triangulation is a commonly cited reason for incorporating mixed methods into research, ‘to enhance the credibility of research findings’ and also to make a study’s conclusions ‘more acceptable to advocates of both qualitative and quantitative methods’ (Hesse-Biber, 2010, p. 4). Triangulation is a contentious notion, traditionally associated with (post)positivistic notions such as objectivism and validity. However, within the constructivist paradigm it is also denoted as a strategy to allow deeper understanding of an issue under study; thus, as a step towards more knowledge instead of a step towards more objective knowledge (Cho & Trent, 2006; Flick, 2018).

Although the function of making the conclusions of the first qual-itative study more acceptable to advocates of quantqual-itative methods was of relevance in this study, the main value of applying multiple methodologies was – or rather became – different from that promot-ed by both conceptions of triangulation. Starting from the conven-tional understandings of the quantitative/qualitative division, during the course of the study, I adopted a way of thinking about method and methodology that differed from that of many of those traditional understandings. Following Popa and Guillermin (2017), I was primar-ily concerned with the theoretical assumptions and normative com-mitments underlying the diverse research methodologies, including methodologies that extend the boundaries of more traditional con-cepts of qualitative research. Thus, I was concerned with the com-bining of different contradicting paradigms. During the unfolding re-search process questions arose about contradictions inherent to the different approaches and the consequences for the project. In this pro-cess, I moved from the idea that methods should be applied to reveal and represent a pre-existing reality as closely as possible, to the idea that methods should be used to describe the specific, contradictory and fluid nature of reality. When working with ANT, I realized that the diverse studies were not adding to a more complete picture of reality, but rather to a fragmented picture, and this seemed of a greater value than striving for certainty or completeness.

An important consideration in this respect was that, in any re-search, ‘tools, frameworks, and criteria are not value free’ (Tracy, 2010, p. 838). Combining different paradigms within one research project, thus, is a way to create diversity, not only to address the complex character of the research context and the issue at hand, but also to avoid one-sidedness and to allow for transformative reflexivity, which purposefully engages with difference in order to learn from it (Popa & Guillermin, 2017). Furthermore, I felt an explicit sense of the

(10)

rele-vance of this effort to the issue and context under investigation. It resonated with questions that arose as the project unfolded; questions about ‘how knowledge production is shaped by the assumptions and values of participants (such as scientists, practitioners, policymakers) and by the broader sociopolitical context’ (ibid., p. 21). I will come back to these issues in more depth in Chapter VIII, ‘Meeting ontolo-gies’, in which I will pursue deeper methodological reflections and discuss the value of methodological heterogeneity in answering the research questions, and the implications of my approach for ECE prac-tice and policy.

Faithfulness to the different paradigms

Three approaches were applied, all three departing from dis-tinct, mutually contradictory, ontological assumptions. In each study, I was faithful to the philosophical assumptions and traditions of the approach that was applied. I used the different rules, criteria, guide-lines and tools as they were developed over time within the diverse research communities from which they stem. I did this because I agree with Tracy (2010) that ‘guidelines provide a path to expertise’ (p. 838), and also in light of the argument on authority: to connect to the nu-merous allies already accepted within the diverse paradigms (Latour, 1987).

For this reason, working in a research team composed of research-ers with different scientific backgrounds – although with an open para-digmatic attitude – was of great importance for the project. The team consisted of one PhD student and three supervisors, each with differ-ent scidiffer-entific expertise related to differdiffer-ent paradigms. The University of Groningen, under whose responsibility the research project was conducted and supervised, traditionally has its roots in quantitative-oriented research. NHL Stenden University of Applied Scien ces, on be-half of which the study was initiated and co-supervised, has a more qualitative orientation, which is connected to the close relationship with the professional field. Once the ANT approach was incorporated, the research team was expanded, with the addition of a co-supervisor experienced in ANT.

In the following three sections, the methods of each study are ad-dressed in detail. The differences between the studies that stem from their differing philosophical assumptions will become clearly recog-nizable to the reader. Each study gathers different types of data; uses different methods or uses methods differently; aims to meet different quality criteria; uses different languages and terms; and applies differ-ent writing styles. In the subsequdiffer-ent four chapters (IV to VII), in which the results of the three studies are presented, these differences will also become apparent.

(11)

58 Turning the kaleidoscope ◊Chapter III Methodologies

One outstanding example of this is the use of passive and active voice. While the passive voice was used as much as possible for the quantitative chapter (Chapter V), as a means to maintain a distance between the researcher and the data (Lincoln et al., 2018), in the other chapters, the active voice is used as a means to acknowledge the ac-tive interpreac-tive role of the researcher as knowledge-creator (ibid.). This active voice, however, also differs. In the two chapters that stem from the initial study, ‘we’ is used to denote the role of the research team as a whole for the project (Chapter IV and VI); to denote the researcher as a co-constructor of knowledge; and to highlight inter-subjective reasoning and agreement as a quality criteria, enlarging the credibility of the results (ibid.). In the sociomaterial chapters (Chapter VII and VIII), ‘I’ is used to stress the productive role of the individual researcher, a storyteller (ibid.), who, despite utilizing the expertise of knowledgeable, critical friends and colleagues, remains the chief person responsible for the account. In the following three sections, in which the methods of each study are addressed in detail, these differ-ences will also be recognizable.

3.2 Methods Study 1: Exploring the landscape In the initial exploratory study (see Figure 3.2), the following spe-cific research questions were explored:

• What are the experiences of ECE teachers regarding the extent to which they can work according their professional beliefs about good education for children aged 4-6?

• What are the experienced consequences of the situation they face?

• In what way does the workplace environment of the ECE teach-ers support or constrain their professional autonomy?

• How do early childhood educators respond to autonomy-limiting influences?

Figure 3.2 Study 1 as part of the research project: professional autonomy in ECE

Equation Modelling

interviews

(12)

To explore these research questions, a qualitative research de-sign was applied, which aimed to explore the lived realities of the ECE teachers in their daily work, without the imposition of preconceived theoretical notions (on the motives for starting with this approach, see Section 3.1).

Sample

Eight experienced Dutch early childhood teachers (kindergar-ten grades 1-2, ages 4-6) were selected as participants for this study. The size of the sample was deliberately kept small to allow extensive data-gathering over time and in-depth analysis. The sample was pur-posefully stratified (Boeije, 2012). The participants differed equally in two important respects. Firstly, they were distinguished according to whether or not they perceived they had space to act consistent with their own professional beliefs, as it was considered important to be able to contrast the perspectives of teachers who did not feel con-strained in their teaching with those who did. To distinguish teachers in both categories, we asked them to answer the question: Do you

experience pressure stemming from your work environment to work with young children in ways other than those you perceive as desir-able? (yes/no). Secondly, participants were distinguished in terms of their initial teacher training (KLOS or PABO). These differences in ed-ucational background enabled us to explore the probable influence of the initial teacher training programme. In addition, we ensured a regional distribution. Furthermore, we included variation in school characteristics, such as school size, urbanization level, specific school philosophy and religious denomination. See Table 3.1 for an overview of respondent characteristics.

Table 3.1 Respondent characteristics

Pseudonym Initial Experienced Region Age Experience School Denomi- Vision City/ teacher pressure 1st inter- (grade 1-2 size nation rural training (negative) view primary educ.) (pupils)

Annelien3 _KLOS _No _NE ₄₉ _{27 years} ₉₂ _Kath. _Jena- plan Britt PABO Yes N 46 20 years 350 none - rural Charlotte PABO No N 43 11 years 300 Chr. - city Diyt KLOS Yes NW 56 37 years 340 none - city Eef PABO Yes S 48 12 years 400 none - city Fleur KLOS Yes W 59 27 years 200 none - city Geraldien KLAS No E 54 34 years 110 none - rural Hanneke PABO No W 38 12 years 230 Kath. - rural

3 All participant names in this dissertation are pseudonyms.

(13)

The research group was also purposefully kept homogeneous on some characteristics. We opted for experienced teachers, who were at an advanced stage of their career. All respondents had at least ten years of experience in the lowest grades of primary school. These teachers had developed their own teaching style and had made a definite commitment to the profession (Huberman, 1989). Their pro-fessional identity had developed, and their responsibilities were no longer limited to their own classroom but extended to responsibility at the school level (Day & Gu, 2007). All of the respondents were per-manently employed and all were female.

Procedure and design

The respondents were recruited through the researchers’ profes-sional network, by asking the question: Do you know early childhood

teachers who might be willing to participate as respondents in research about professional autonomy? We are looking for teachers who feel con-strained as well as teachers who do not feel concon-strained in carrying out their work according to their professional beliefs about teaching young children. Recruitment through the general network yielded only one respondent (Hanneke). The other respondents (seven) were found through the network of the professorship (lectoraat) Early Childhood Studies, on behalf of which the research was conducted. Some of them were involved in the network itself: Charlotte joined the broad ‘knowledge circle’ of the professorship; Dyt attended a round table conference; and Eef’s school was making use of the support activities of the professorship. The other respondents were suggested by a con-tact of the professorship. We consciously considered the implications of this, because of the politically laden task of the professorship. We understood that recruitment through this network might have been biased due to the political interests of the candidates (Creswell, 2013). However, by selecting teachers who were either constrained or non-constrained to the same extent, we did the upmost to guarantee a bal-anced influence with respect to our research design.

All potential candidates received an email describing the aims of the project and the character and expected duration of the inter-views. In a telephone follow-up, the selection criteria were checked. As there were more than eight suitable candidates who were willing to participate, we could add variation based on school characteristics (see Table 3.1)

The interviewees were free to choose the location where the in-terviews took place. Most inin-terviews took place at the schools of the respondents, although some (Dyt and Fleur) chose their homes as in-terview locations to protect their privacy. The participants were inter-viewed three times, at yearly intervals, in 2014, 2015 and 2016. The

(14)

duration of the interviews was 90, 60 and 30 minutes, respectively. All interviews were audiotaped. The initial eight interviews were tran-scribed and uploaded as primary documents into Atlas.ti for analy-sis (see below, in the analyanaly-sis section). Follow-up interviews (the first face-to-face, the second by telephone), enabled us to verify the syn-thesis texts, to clear up ambiguities and to track developments over time. Guidelines were designed for these interviews, partly tailored to the individual respondents. Relevant parts of the follow-up interviews were transcribed; that is, developments and changes in the work con-text that contradicted, illustrated and/or deepened the insights from the first analysis.

Instrument

The participants were interviewed in open one-on-one interviews that focused on the teachers’ perceptions of professional autonomy and of the presence or absence of tensions due to the balancing of pressures related to autonomy and accountability. An exploration of the relevant literature led to the identification of some broad sensitiz-ing concepts (Boeije, 2012) that functioned as lead topics in the inter-views: professional beliefs, work environment (context), consequences for practice (effects) and responses to autonomy-limiting influences (strategies). The leading questions in the interviews, derived from these topics, were: What, in your opinion, constitutes good education

for young children aged 4-6? Who or what in your work environment enables you to work or constrains you from working in that way? What experiences have you had as a consequence of the situation described? In what way have you responded to autonomy-limiting influences and what were the effects of this?

An interview guideline4_{was designed and tested once. This}

test-ing was aimed at verifytest-ing the duration of the interview (maximum 90 minutes), practising the intended interview performance, seeking a balance between recognition and distance, and testing the usability of the guideline. No adjustments appeared to be required. The guideline was used as checklist during the interview. All interviews opened with the same question: What, in your opinion, constitutes good education

for young children aged 4-6? The subsequent order in which the topics were addressed was mainly determined by the respondents. The ques-tions were open and there was ample opportunity for the respondents to develop their own story line, putting their own accents and bringing their own issues to the fore.

During the interviews, the researcher was aware of maintain-ing a balance between involvement and distance. A relationship of trust was needed to make the respondents feel at ease and able to talk freely about, for some of them, emotionally laden subjects. Distance

4 All research documents referred to in this and the following subsections of Chapter III, such as interview guidelines, procedures, code tree, survey, analysis logs, codebooks, are stored in a publication package, in accordance with the principles of the Research Data Management Protocol for Nieuwenhuis Institute Researchers (2017).

(15)

was needed in order to prevent undesired influence. In seeking and safeguarding this balance, the involvement of the researcher was an important issue. An open, interested and encouraging but non-judge-mental listening attitude was sought. In this respect, see the measures used to enhance trustworthiness discussed below (Trustworthiness).

Analysis

Verbatim transcripts of the initial interviews were thematically analysed. Such analysis ‘provides a flexible and useful research tool, which can potentially provide a rich and detailed, yet complex account of data’ (Braun & Clarke, 2006, p. 5). In this dissertation, we report on the analysis of the dataset that was drawn from the entire data cor-pus (all interviews and follow-up interviews). The dataset comprised all moments in the interviews that referred to the topics of interest: beliefs, context, effects and strategies. By listening to the audiotaped interviews and reading through the transcripts, primary issues, state-ments or incidents of interest that were related to the topics were not-ed and collectnot-ed in an initial first-stage coding process, with the use of Atlas.ti and a ‘Computer assisted NCT-analysis’ (Friese, 2014). The three basic components of this model, Noticing, Collecting and Think-ing about the data, were applied in an iterative process, in which open and axial coding alternated (Boeije, 2012). Initial codes were combined to form broader themes of interest.

In the subsequent phase, these themes were checked for patterns and for variability and consistency across the entire dataset, querying the data with the help of Atlas.ti software (Friese, 2014; Mortelmans, 2001). An important aim of this horizontal analysis was to explore if and how differences in respondents’ experiences regarding profes-sional autonomy related to differences in contexts. While the research questions and related topics led the analytic process, these broad, ex-ploratory questions evolved and were refined, leading to the formula-tion of more specific quesformula-tions in the process of querying the data and looking for relationships and patterns. This process was supported by the accurate writing research question memos (Friese, 2014). Further interpretation of the themes was carried out by a process of reading and re-reading, with reference to the relevant literature (Boeije, 2012; Creswell, 2013).

In the next step of the analysis, the results were summarized and related to the literature in an extended internal research report. Subse-quently, the findings were summarized and expanded in a comprehen-sive synthesis text for each respondent (Kelchtermans, 1994). All cita-tions and interpretacita-tions from the horizontal analysis, as described in the general result report, were included in the synthesis texts to complete each of the stories, which represented the interview of each

(16)

respondent as closely as possible. All synthesis texts were arranged into the same format under headings that reflected the main topics of the interview. This step in the analysis procedure also included an ‘interview check’. All of the questions and notes in the results report were checked by re-reading the original transcripts. Any remaining unclarity was translated into interview questions for the follow-up in-terview.

The analysis of the data derived from the second and third in-terview rounds was done without Atlas.ti, because the inin-terviews had a more structured and targeted character. While part of the interview was open, comparing the new data with earlier results did not give rise to new, open analysis strategies. The second and third interviews de-livered new illustrative material, which was processed in the result re-port. Finally, the writing of articles on the basis of the results formed an integral part of the analysis (Braun & Clarke, 2006).

For an overview of the entire analysis procedure, see Table 3.2. Although this overview might create an impression of linearity, we stress that the analysis process in practice was not a linear but a re-cursive process. More detailed clarification of the separate steps can be found in the research log.

Trustworthiness

Like all of the choices made in setting up and conducting this initial study, the rules of rigour in this particular study were derived from the constructivist paradigm, within which the study approach is situated. In this section, I present how this study strives to improve trustworthiness, seeking ‘to insure an accurate reflection of reality (or at least, participants’ constructions of reality)’ (Cho & Trent, 2006). The research was conducted in the research team, which at that mo-ment consisted of three researchers: Alexander Minnaert (supervisor), Ineke Oemema-Mostert (co-supervisor), and me. When I use ‘we’ in this section, I am referring to these three people.

The interviews and analyses were carried out by me. Here, it is important to emphasize the active role of the researcher in the pro-cess of collecting, selecting and interpreting the data (Braun & Clarke, 2006), and thus, the influence of the researchers’ goals, assumptions and values. Although the coding process was data-driven, ‘researchers cannot free themselves of their theoretical and epistemological com-mitments’ (Braun & Clarke, 2006, p. 12). To enhance credibility, several strategies were applied, noting that in qualitative research ‘trustwor-thiness is never absolute proof, but by persuasion, one is compelled to accept the findings’ (Morse, 2018, p. 801). Firstly, intersubjective agreement and reasoning was sought (Lincoln et al., 2018). The inter-view guideline, the developing code tree and the research question

(17)

memos were discussed by the research team on a regular basis dur-ing the process of analysis. Additional peer debriefdur-ing was applied to enhance reflexivity during the entire research process, by presenting and discussing the quality of the research at different stages of the process, both internally, with colleagues at the University of Gronin-gen and NHL Stenden University of Applied Sciences, and externally at national and international scientific conferences as well as at sympo-sia in the Dutch educational field.

Furthermore, the synthesis texts were sent to each respondent and member checked in a structured one-on-one follow-up interview to verify interpretations and clarify ambiguities. These synthesis texts were based entirely on the interview transcript (with the exception of information about the board, which was derived from the school web-site). Interpretations were illustrated with quotations of the respond-ents, which were presented in literal, colloquial language, although in some instances they have been corrected and repetitions omitted to promote readability. All participants agreed on the interpretations and considered the synthesis texts to be a good representation of their interview. The data from the follow-up interviews was compared with the results of the initial interviews. Minor changes were made and marked in a second version of the synthesis texts and supplemented with additional interpretations. Again, findings were summarized and sent to the participants for confirmation. See Table 3.2 for an over-view of the entire analysis procedure, including strategies to enhance trustworthiness.

Moreover, the trustworthiness of this qualitative study, derived from a constructivist paradigm, is determined by the extent to which it is possible to inspect how the researchers have reached their con-clusions. Clarity about the methodological process allows readers to evaluate the research and to compare it with other studies on the same topic (Braun & Clarke, 2006). With this purpose, much attention was paid to accurate description and documentation of the research pro-cess, in the form of research question memos. Participants agreed to save the audio-recordings of the interviews for up to five years after finishing the research. In the presentation of the results, thick descrip-tions were used to enable the reader to judge the transferability of the results to other situations (Boeije, 2012). The use of the active voice ‘we’ in the presentation of the results (Chapter IV), ‘reminds the readers of the researchers’ presence and influence in participating and interpreting the scene’ (Tracy, 2010). In addition, the use of ‘we’ re-minds the reader of the teamwork-setting in which the research was conducted.

Finally, theoretical triangulation took place by relating the quali-tative findings to the literature at several moments in the research

(18)

process (Boeije, 2012). This occurred, fi rstly, in advance, by reading about and defi ning sensitizing concepts; secondly, in Stage 6, when summarizing results and relating the results to the literature; and, thirdly, in Stage 11, during the writing for publication.

Ethical aspects

The present research project complies with the EECERA Ethi-cal Code for Early Childhood Researchers (Bertram, 2015), with the Ethics Regulations of the Ethical Committee for pedagogy and edu-cational science of the University of Groningen (University of Gronin-gen, 2018a) and with additional ethical codes designed by the Ethical Committee for Psychology of the University of Groningen (University of Groningen, 2018b). In general, this kind of ‘procedural ethics of-fers researchers an ethics “checklist”’, which reminds the researcher

Table 3.2 Overview of the main steps of the analysis process

Stages Product

1 Preparation: transcription Primary documents/summaries audiotapes/summarizing headlines

2 Open and axial coding of the fi rst four Provisional code system primary documents (balanced distribution

of subpopulations)

3 Horizontal analysis of the fi rst four Provisional results, documented in research primary documents (querying the data question memos

with the use of Atlas.ti)

Discussion of the developing code tree and RQ memos in the research team

4 Coding second series of interviews Confi rmed and further developed code tree 5 Second horizontal analysis Confi rmed and further developed provisional

results

Discussion of the developing code tree and RQ memos in the research team

6 Relating results to the literature Confi rmed and developed code tree Research report

7 Summarizing results Internal research report

Discussion of report in the research team

8 Vertical analysis Synthesis texts

Discussion of synthesis texts in research team/member validation (follow-up interview)

9 Processing new data and comparing with results Confi rmed and nuanced results

Member validation (telephone follow-up)

10 Processing new data and comparing with results Confi rmed and nuanced results 11 Writing for publication Two articles (Chapters IV and VII)

Additional paper presentation (Oosterhoff et al., 2016)

Discussion of evolving articles in the research team/ on conferences

(19)

66 Turning the kaleidoscope ◊ Chapter II Theoretical frameworks

to consider ethical issues, such as potential risks for the participants, balancing the benefits and the costs of the research and ensuring the confidentiality of data (Guillemin & Gillam, 2004, p. 268). With respect to the use of these ethical codes, two main procedural issues, as indi-cated by Guillemin and Gillam (2004, p. 275), were addressed: respect for the autonomy of the participants, achieved by the mechanism of informed consent, and respect for privacy and confidentiality.

From the very first contact with the respondents, confidential-ity was guaranteed by the implementation of several procedural de-cisions. Firstly, the managers of the respondents were not informed about the nature of the visits, and some respondents deliberately chose their homes as the interview location. Secondly, all real names (such as the names of colleagues, or the school head) were omitted from the transcriptions and replaced by general labels, and all of the respondent’s names were replaced by pseudonyms. When present-ing the results, the use of thick descriptions of the situation meant the respondents could be easily recognized in some cases. This issue was discussed during the follow-up interview. Some respondents were concerned about this and it resulted in some modifications to the syn-thesis texts, such as changing the name of a location or using more general labels.

The ‘principle of beneficence’, translated into balancing the bene-fits and the costs, is also addressed in the procedural ethics guidelines of the University of Groningen. According to Guillemin and Gillam (2004), in qualitative research, these principles are ‘quite subtle and stem from the nature of the interaction between researcher and par-ticipant. As such they are hard to specify, predict and describe in ways that ethics application forms ask for’ (p. 272). Moreover, ethical issues not only occur prior to conducting a study, but also during the entire research process, including the publication of the results. Thus, the researcher should specifically address ‘ethics in practice’ (Guillemin & Gillam, 2004, p. 269), that is, the ethical dimensions that occur in the complex dynamics between researcher and participants in the daily practice of research. In this study, it was especially important to pro-tect individual autonomy, emphasizing ‘that it is an individual’s right to act according to his or her own purposes rather than the purposes of others’ (Guillemin & Gillam, 2004, p. 271). In this respect, it is im-portant to note that all eight respondents explicitly expressed and underscored the political importance of the research project. Their motives for taking part sprang from this interest, although their mo-tives differed in the details, and their stories, which sprang from these motives, also differed.

The investment of time by the respondents was substantial (three hours of interview time and also reading through several drafts of

(20)

synthesis texts). On the one hand, the inductive, qualitative start of the project was deliberately chosen to facilitate the voices of the ECE teachers, allowing them to ‘be heard and honoured, so that others can learn from them’ (Denzin & Lincoln, 2018, p. 33). In this research pro-ject, the respondents were considered participants rather than sub-jects (Guillemin & Gillam, 2004). However, we, as researchers, could not guarantee in advance that the results would accord with the ex-pectations of the participants, nor were we able to acknowledge all their individual concerns in the final report. Ultimately, the analysis is undertaken by the researchers, and, as Braun and Clarke (2006) em-phasize, the ‘giving a voice approach’ should not be understood in a simple realist sense. Instead, the researcher plays an active role, identifying patterns of interest and selecting pieces of evidence that are edited and deployed to build up arguments that are presented to an academic audience.

In contact with the participants, the researcher attempted to be as clear as possible about the opportunities as well as the limitations of the project. This occurred beforehand, by informing the respond-ents about the research aims and the research questions, and the es-timated investment of time. It occurred during the research process, by informing them about the emerging findings, discussing them and deliberately attempting to keep a balance between respecting the au-tonomy of the respondents, honouring their individual experiences and perspectives on the one hand, and being mindful of one’s role as a researcher, keeping a balance between involvement and distance, on the other hand.

Furthermore, in several cases, the participants reported an ad-ditional, unforeseen benefit of participating: they considered the in-terview to have been a pleasant and reflective experience. The fol-lowing quote from the interview with Charlotte illustrates both these personal benefits and the broader expectations.

Interviewer: How was your experience of this interview?

Charlotte: Yes, pleasant. Very reflective, actually. [...] When I hear

myself talking, I think: ‘Wow, I do a ridiculous lot’. [...] And this is actually why it was nice for me to have this talk, I realized I am proud of what we have achieved and I like sharing that with you, and hopefully also with others.

Moreover, all of the respondents appreciated being kept informed about the progress of the research. This was carefully taken care of, by sending the respondents all of the results as they emerged and were published in one form or another. In addition, all of the respondents were informed on a regular basis about the general progress of the

(21)

project by email. These were personalized, with the exception of gen-eral information, thus also thoughtfully acknowledging their specific context or situation.

3.3 Methods Study 2: Generalizing and theorizing results

A quantitative study aimed to generalize and theorize results from the first study (see Figure 3.3). Two findings were thought to be important for generalizing to the population of Dutch ECE teachers as a whole. Firstly, the apparently important role of the manager in enhancing the professional autonomy of the ECE teacher (see Chapter IV). The decision to conduct a quantitative study to generalize this finding primarily lay in the desire of external participants involved in the professorship Early Childhood Studies, on behalf of which this research was initiated. The professorship deliberately wanted to team up with ECE professionals in the field and to that end established a network group (kenniskring) that met on a regular basis. The findings of the first study were presented at one of these meetings. Conse-quently, all of the members, of whom a substantial part were manag-ers of primary schools themselves, recognized the significance of the findings on the role of managers, and firmly supported the importance of this finding. However, they also expressed the fear that the qualita-tive study would not be sufficient to convince their colleagues in the field, due to the relatively small number of participants.

Secondly, it was also deemed important to generalize the find-ings on the apparently constraining role of the initial teacher train-ing programme with respect to enhanctrain-ing professional autonomy, for the same political reasons: to be able to articulate assumptions about the Dutch ECE population as a whole. The participants themselves brought this issue to the fore in the first interview round (see Chapter VI). All participants signalled a lack of attention to specific ECE

exper-Figure 3.3 Study 2 as part of the research project: professional autonomy in ECE

Equation Modelling

interviews

(22)

tise in current teacher training programmes and related this to a lack of agency in prospective teachers. They emphasized the present ur-gency of this issue, due to the prospective retirement of colleagues in the field who were educated in former teacher education programmes that focused on ECE.

Both issues led to two broad research questions that guided the second study: ‘Generalizing and theorizing results’.

• What is the relationship between management style, profes-sional autonomy and job perception?

• To what extent do initial teacher training programmes contribute to the development of expertise?

The results of the first study that were related to these two ques-tions were translated into an extended conceptual model (Figure 3.4). The model was divided in half (see the dotted ‘waterline’), with the relationships between the variables above the waterline tested using Structural Equation Modelling (SEM). Not all management variables were included in this hypothetical model because this would make the model too complicated for testing with SEM and thereby violating the assumptions for model testing. The ratio between the number of variables and respondents would not allow testing with SEM (Jöreskog & Sörbom, 2006). The variables and relationships under the waterline were analysed in a more explorative way, using frequency

distribu-Figure 3.4 Conceptual model for Study 2: Generalizing and theorizing results. Fields with grey text have not yet been processed in the result chapters.

(23)

tions on a descriptive level and a Mann-Whitney U-test to investigate differences. The fields with grey text have not yet been processed in the result chapters. Below, the design of the study will be outlined. Measures to enhance the validity and reliability of the study are ad-dressed throughout this section. Additional measures are presented in Chapter V.

Sample

The study sample consisted of Dutch primary school teachers working in grades 1 and 2 of primary education. The operational pop-ulation comprised all Dutch teachers working with these grades at the time of the study (April 2016).

A random sample was drawn which was of a size in line with rec-ommendations for model testing (at least ten times as many respond-ents as variables and at least 200 respondrespond-ents) (Bentler & Chou, 1987). Multistage sampling was used to ensure a balanced distribution across the Netherlands, with random selection carried out in each province. In order to avoid a nested data structure, only one respondent from each school was admitted to the sample. School characteristics, which were not measured, could play a background role.

Phone calls were made to a total of 349 schools in accordance with the procedure outlined. At 41 schools, contact could not be estab-lished with a grade 1 or 2 teacher because the teacher was not avail-able or the gatekeeper refused participation (in 10 cases). In all, 308 teachers were approached. Seven did not wish to complete the ques-tionnaire. In most cases, work pressure was cited as the reason for not taking part. In total, 301 teachers agreed to participate and were sent a questionnaire, with 250 questionnaires returned, five of which were not filled in completely and were therefore eliminated from the database. Thus, we ultimately had a sample of 245 teachers who filled in the entire questionnaire (i.e. a high response rate of 80% of the 308 teachers approached).

There was an even national distribution within the sample. Of the 245 teachers in the sample, 98% were female. The respondents’ ages ranged from 20 to 64 years, with an average age of 46; 52% were older than 49. In terms of qualifications, the majority of respondents had primary school teacher training (4-year PABO) (51.4%), while 36.7% had preschool teacher training (KLOS), 9.4% had former primary teacher training (PA) and 2.4% had another qualification. The schools where the respondents worked were 33.5% non-denominational, 26.9% Prot-estant, 28.9% Roman Catholic and 10.9% another denomination. Of the schools, 12% were of a specific school type (Jenaplan, Montessori, Steiner, Dalton, experience-based education), of which over half (6.5%) were Dalton.

(24)

It is also important to bear in mind systematic, random non-response, which model testing could combat. The questionnaire might not have reached a relevant section of the study population, namely teachers who may have left the profession because of a lack of profes-sional autonomy or who were on sick leave for the same reasons. This could distort the results, because participants with a perceived man-agement style that was relatively highly restrictive and a perceived autonomy that was relatively limited would be missing from the study sample. The size of this potential group of respondents who could not be reached is unknown.

Procedure and design

The DUO list (2016) of main locations of primary schools, which is organized by province, was used as a sampling frame, with every twentieth school on the list selected. The grade 1 and 2 teachers were approached by phone at random for each school. In some cases, there was a ‘gatekeeper’ who requested information about the study before passing on the teacher’s details. In these instances, the gatekeeper was only given general information about the study, and not the question-naire itself. They were told that the work environment was one of the items being asked about. No explicit mention was made of the fact that management style was one of the aspects of the work environment. This was done to reduce any possible influence of management on participation.

Each participating respondent received an individual link to the questionnaire, which could only be used once. The accompanying email contained a short description of the incentives and aim of the study and a notification that respondents would receive a summary of the research findings as soon as they were available. It also stated that data would be processed anonymously. Teachers or gatekeepers who refrained from participation were asked about the reasons behind this. Respondents who had agreed to take part but had not responded were sent an email reminder after 14 days.

Instrument

A digital questionnaire was compiled in Qualtrics. The question-naire opened with a number of general questions about respondent characteristics (age, gender, education, experience, extent of appoint-ment) and school characteristics (school size, denomination, school type). Further variables were measured using statements with re-sponse options on a 6-point scale. Uniform scales made it possible to make comparisons. Providing six response options meant respond-ents could give a nuanced response, and using an even number of response options, with no neutral mid-point, forced respondents to adopt a position.

(25)

For the purposes of model testing, the survey questions were formulated in three categories: professional autonomy experienced; management style of the school leader; and job perception. This was done to investigate the question: What is the relationship between

management style, professional autonomy and job perception? The operationalization of the variables in each category is explained in detail in Chapter V, ‘Room for autonomy’ (Section 5.3, ‘Instruments’). In part, standardized instruments were used and adapted. Information about the validity of these instruments can be found in Chapter V. The instruments were adjusted to the needs of the current study: some had to be translated, some had to be shortened, and scales had to be brought into line. Each variable was measured by different statements, which were combined into a total score. For each adjusted composite scale, Cronbach’s

α

was calculated to check the reliability of the new scale.

To answer the question: To what extent do initial teacher training

programmes contribute to the development of expertise?, the survey questions were formulated under the heading ‘Working with young children’. These questions aimed to explore two topics: the role of specific ECE expertise in conducting daily work with young children (three statements); and, subsequently, the role of the initial teacher training programme in developing this expertise (three statements). Another statement was added that was related to autonomy: I feel

that I must actively protect young children from outside influences on the educational approach in grades 1 and 2. All items related to this second research question were based on statements by respondents derived from the findings of initial study, ‘Exploring the landscape’. The quantitative data with respect to the initial teacher training is pro-cessed in Chapter VI, ‘Aiming for agency’ (for the operationalization of this item, see section 6.2, ‘Methods’).

Additionally, questions were asked about the way in which the respondents experience their collaboration with colleagues (eight statements), and about the way the management was facilitating team collaboration and professional development (three statements) dur-ing the penddur-ing school year. The data on these statements has not yet been analysed. Finally, the engagement of management with the young child was investigated using three statements. One of these:

The principal actively safeguards the quality of teaching for young chil-dren, is incorporated in the results in Chapter V, ‘Room for autonomy’. The questionnaire closed with an open question (Finally, do you

have any comments, additions, or would you like to explain your an-swers in this questionnaire?) and respondents were asked to rate the autonomy experienced on a scale from 0 to 100 (To what extent can

you currently design your education according to your own vision of education for young children?

(26)

The questionnaire was tested before distribution by a group of 10 respondents approached through the network of the professorship Early Childhood Studies. The aim was to check the time investment re-quired, the comprehensibility of the questions and technical fluidity. Minor modifications were made on the basis of this test. Subsequently, the digital questionnaire was distributed with the help of Qualtrics.

Analysis

Firstly, the data was transferred from Qualtrix to SPSS. The data-set was anonymized, with the names of the respondents replaced by numbers. Incomplete questionnaires were deleted (five). The remain-ing data was prepared for analysis. Questions were replaced by rec-ognizable codes. Composite scale scores were combined into a single score and the homogeneity of each variable was checked by means of item analysis. The Cronbach’s alphas of all composite scores were noted in the codebook. The level of measurement was nominal for some general questions (gender, education, denomination, school type), while other general questions were measured at the ratio level (age, extent of appointment, experience, school size). The scale ques-tions were measured at interval measurement level. SPSS summarizes the interval and ratio levels under ‘Scale’ (Baarda et al., 2014).

Investigating frequency distributions and differences

Secondly, the study sample was described in terms of the regional distribution, the general characteristics of the sample and responses to the general statements about working with young children in the current social context, based on frequency distributions at a descrip-tive level. Furthermore, the questions aimed at quantifying the extent to which Dutch initial teacher training programmes contribute to the development of ECE teacher expertise were also analysed based on frequency distributions at a descriptive level. Additionally, a Mann-Whitney U-test was performed to investigate differences between re-spondents who received their teacher training in the PABO and KLOS programmes, respectively (Field, 2013). The Mann-Whitney U-test is suited to determine whether one group of respondents within a sam-ple scores significantly higher more often than the other.

Model testing

The anticipated relationships between the variables of manage-ment style, professional autonomy and job perception were described within a hypothetical structural model, developed on the basis of sub-stantive theory (Bentler & Chou, 1987). Structural Equation Modelling (SEM), with the aid of the LISREL 8.80 analysis program, was used to test the relationships within the hypothetical structural model

(27)

74 Turning the kaleidoscope ◊Chapter III Methodologies

(Jöreskog & Sörbom, 1996). SEM enables the researcher to create and test complex path models with more than one dependent variable and an intervening variable between the predictor(s) and dependent variable(s) (Hox, 1999). In preparation for the model testing, Pearson’s correlations were calculated for all individual variables.

Several goodness-of-fit measures were determined (Byrne et al., 1989; Hox, 1999), including: the χ²/df ratio (i.e. the ratio between χ² and the degrees of freedom); the P-Value for the closeness of fit; the Root Mean Square Error of Approximation (RMSEA); the Goodness of Fit Index (GFI); the Incremental Fit Index (IFI); and the Adjusted Goodness of Fit Index (AGFI). Because χ², the P-statistic and the (A)GFI are sensi-tive to the size of the sample and deviations from the assumed normal distribution, we also report the χ²/df ratio, the RMSEA and the IFI. Fit indices above .95, an RMSEA below .05 and an χ²/df ratio below 2 are indicative of a good fit (Marsch et al., 2004; Byrne, 2010).

Furthermore, the correlations of diverse background variables were also checked. They were: teacher training, gender, experience, denomination and school type. None of these checks gave rise to sepa-rate model testing.

3.4 Methods Study 3: Speaking with things

The ideas of ANT offered a theoretical and methodological frame-work to explore the complexity of everyday life in education (see Fig-ure 3.5). For ANT, practices are the key objects of research (Latour, 1999). Practices, as outlined in Chapter II, are unpredictable, fluid and emergent, as they are understood as constitutive entanglements of humans and things (Reich & Hager, 2014). The aim of ANT as a prac-tice-based approach is to describe micro-practices in an investigation of how humans and nonhumans constantly co-constitute mundane, emergent, local practice. This practice-based approach invited to ex-plore the question about the role of things that appeared in the stories

Figure 3.5 Study 3 as part of the research project: professional autonomy in ECE

Equation Modelling

interviews

Analysis Posthuman heuristics – Following the

(28)

of the eight respondents in the initial study: How do people and things

together enact everyday Dutch ECE practices?

The main theoretical and philosophical positionings and sensi-bilities of ANT were outlined in Chapter II, but how is it possible to ac-tually ‘use’ this framework? ANT does not consider itself to be a single method. There is not one way of ‘doing’ ANT. There are no handbooks, systems, grids or rules to follow step-by-step. It is inherent to ANT philosophy, which argues against fixity and singularity, that ANT re-searchers take up ANT in many different ways. As Mol (2010) warns: ‘It is impossible to “use ANT” as if it were a microscope’ (p. 261). Law and Singleton (2013) further explain: ‘ANT is probably best understood as a sensibility, a set of empirical interferences in the world, a worldly practice, or a lively craft that cherishes the slow processes of knowing rather than immediately seeking results or closure’ (p. 485). This open approach does not make it easy to take up ANT, especially for novice researchers. However, experienced ANT scholars have provided many examples through their work (e.g., Röhl, 2015; Fenwick, 2014; Mol, 2002; Plum, 2016; Thompson, 2018) and recent doctoral work based on ANT also encourages novice researchers to pursue this direction (e.g., Ceulemans, 2015; Hultin, 2019; Scoles, 2018). Moreover, in this specific study, heuristics developed by Adams and Thompson (2016) were of great help in finding a way of working with ANT.

Based on their own more-than-human research experiences, Adams and Thompson (2016) provided a series of heuristics for interviewing

objects to support sociomaterial research. Such heuristics, Adams and Thompson (2016) state, are not step-by-step procedures, but possible

ways of inquiry; a series of questions drawn from theoretical sources such as ANT, (post)phenomenology and critical media studies, which help to become attuned to important nonhumans and untangle the way people and things co-constitute practices. These questions might help to find fresh answers concerning the way people and things are entangled in practice. ‘The researcher or practitioner starts out on a footpath taken by others before, but with an eye for surprises and a willingness to make unexpected detours’ (p. 20). Adams and Thompson (2016) present eight heuristics which provide ways of ‘speaking with things, i.e. making visible and questioning relevant nonhumans found at one’s research site’ (p. 19).

These heuristics helped me to start on the journey, to explore possible ways to travel, and above all to keep me on the right track, especially to avoid the pitfall of the anthropocentric disposition to which I was repeatedly tempted. Adams and Thompson (2016) pre-sent their heuristics in two sets that are ‘loosely aligned’ (p. 20) to the main methodological acts of qualitative research – data collection and data analysis. They say ‘loosely’, because the two gestures of data

(29)

collection and data analysis, are closely intertwined in a sociomaterial research process. Below, I will first briefly describe the starting points of the approach that I took – the data and questions about it – and, subsequently, I will explain the way I analysed this data through an ANT lens. Finally, some issues of scientific rigour will be addressed.

Data

The starting point of the ANT analysis was the stories of eight experienced Dutch early childhood teachers, which they recounted during the first two in-depth, semi-structured one-to-one interviews conducted as part of the initial qualitative study, ‘Exploring the land-scape’. For details about this study, respondents and interviews, see Section 3.2. These stories were about the everyday work of the teach-ers of kindergarten grades 1-2, working with children aged 4-6. The teachers told about their drives and their frustrations, about events that occurred in their practices, about happy and sad moments, about tensions and dealing with them, and about the many people and things that were involved.

Mol (2002) has shown that the stories people tell in interviews do not merely present meanings or perspectives, but also tell about events they have lived through; and these stories are animated by the things that made these events happen. While the initial inter-views focused on the daily practice of the teachers, they were not conducted from a posthuman perspective, that is, they did not intend to trace the human-nonhuman actors and actions. However, a variety of nonhumans appeared in the stories of the eight participants. The transcripts, so to speak, invited me to zoom in on the data with a so-ciomaterial lens. Accepting this invitation wholeheartedly, I began to explore the data anew with an ANT sensitivity. To begin with, I did so in a broad sense, by re-reading the interview transcripts, attuned spe-cifically to the variety of objects and things that animated the stories. In doing so, I became interested in the role that standards seemed to play in creating tensions in ECE. More specifically, a question began to emerge: How do the standards work as part of extended sociomaterial

networks of accountability policies? Recalling Mol (2002), who stressed that when investigating objects they should never be isolated from the practices in which they are enacted, Britt’s story came to the fore. It exposed how things, such as standards, take on their particular shape, are enacted, in and through daily educational activities.

To explore ‘the doings of standards as networks in Dutch ECE practice’ through an ANT lens, I thus selected one of the eight cases of the initial study, Britt’s story, to analyse situated sociomaterial prac-tice in detail. Why did I choose to concentrate on this particular case? Britt had told me about a clash that had occurred during a meeting