Advertisement

A Mixed-Methods Approach

  • Marieke Slootman
Open Access
Chapter
Part of the IMISCOE Research Series book series (IMIS)

Abstract

The study has a mixed-methods research design. It combines data from a structured survey with data obtained from 14 in-depth interviews with university-educated second-generation Moroccan Dutch and Turkish Dutch. The survey was conducted in the context of the TIES research project, prior to this study. The two different methods complement each other in valuable ways and are used in various constellations throughout the study. The chapter explains how results of qualitative interpretivist studies, which are based on small samples, can be generalized in various ways. In addition, it argues how quantitative methods, which are generally used within more objectivist and positivist research traditions, can be used within a more interpretivist perspective and can be even used for the deconstruction of concepts such as ethnic identity, as illustrated in Chap.  5. The chapter furthermore describes the details of the quantitative data collection and the qualitative data collection and analysis.

Keywords

Mixed methods Phenomenology Interpretivism Generalization Ethnic identity TIES survey 

How you collect data affects which phenomena you will see, how, where, and when you will view them, and what sense you will make of them. (Charmaz 2006, p. 15, italics in original)

This chapter describes how I researched the identifications of second-generation Moroccan Dutch and Turkish Dutch and shows the steps that brought me to the claims I make in this book. I first explain the used mixed methods approach and the connection with my ontological perspective (Sect. 3.1). In the subsequent sections, I describe the quantitative data collection (Sect. 3.2) and the qualitative approach (Sect. 3.3). The chapter concludes with a summary (Sect. 3.4).

3.1 A Phenomenological Mixed-Methods Research Design

Different research aims require different methodological approaches. In this case, we want to understand what ethnicity means for minority climbers and understand how, when, and why they identify in certain ways. Such phenomenological and interpretivist study requires an open, ‘qualitative’ method, which is not pre-structured by the researcher and enables free exploration of the phenomenon’s complexities. The aim here is not to draw conclusions that apply to large populations with specified levels of certainty and not to separate the phenomenon from individual experiences and interpretations as in structured or ‘quantitative’ studies. Rather, a phenomenological study ‘describes the meaning for several individuals of their lived experiences of a concept or a phenomenon’ (Creswell 2007, p. 57, italics in original). The purpose is to extract from their individual stories a cohesive description of the essence of the experience for these individuals (Creswell 2007).

The open and interpretive approach that leads to the development of in-depth understanding of the experiences of individuals only allows for data collection from a limited number of individuals. In trying to understand a specific phenomenon or certain human experiences as they make sense to those who live it, between 10 and 20 interviews, or even fewer, might suffice (Dukes 1984, p. 200; Bleijenbergh 2013, pp. 10–11). Polkinghorne recommends a sample size of 5–25 individuals (1989, in Creswell 2007, p. 61). The phenomenological description developed in this book is based on 14 interviewees.

Because of the limited number of cases, a phenomenology does not necessarily present a story that applies beyond the interviewees. Nevertheless, the relevance of a study often does extend beyond the study. Larsson gives an intelligible overview of the various forms of ‘generalization’ of qualitative, interpretivist studies (2009). For example, a study—even a very small-scale study—has broader relevance when findings undermine common assumptions about a phenomenon and nuance an established perspective; Chap.  5 includes an illustration of this kind of generalization.

In addition, for many qualitative studies, the ‘act’ of generalization primarily lies with the audience instead of the author (also see Flyvbjerg 2004). The readers assess the relevance of the study’s findings for situations with which they are familiar. This is why many qualitative studies, including this one, include ‘thick descriptions’. Rich details enable readers to recognize parallels and differences between the study and other situations, and to judge the applicability of the study’s findings for situations they are familiar with. The detailed description of the Dutch context in Chap.  4 is also an example of such ‘thick description’ that helps the reader assess the relevance of this case for other contexts. The researcher can support the audience by articulating aspects of the study’s context that appeared crucial for the studied phenomenon, as well as by suggesting in which situations similar patterns are likely to occur; I provide such suggestions in Chap.  8.

The relevance of qualitative studies furthermore works though ‘theoretical generalization’, or ‘analytical generalization’ (Bryman 2008). Larsson’s reflection on what he calls ‘recognition of patterns’ (2009) helps understand this form of generalization. A qualitative description of a specific phenomenon (as experienced by some individuals) provides others with a lens for looking at the world. The interpretations presented by a study invite readers ‘to notice something they did not see before’ (p. 33); to recognize particular patterns. The descriptive results of qualitative research form ‘interpretational tools for identifying patterns in the everyday world and making better sense of the world around us’ (p. 34). This is what this study offers: a description and interpretation of the experiences and identificational processes of 14 Moroccan-Dutch and Turkish-Dutch climbers that could possibly be found in similar form in other contexts. This is also how I build on other literature: other studies offered perspectives from which I approached my own data and which helped me interpret the data.

Although the contribution of this study primarily lies in the theoretical generalization, at certain places in the book I articulate the broader relevance in more direct ways. Firstly, in all chapters, I mention parallels with the literature. The resemblance of my data—and of the phenomenological description that emerged from it—with other studies, across groups and contexts, indicates that the findings are relevant beyond this study. Secondly, on a few occasions in Chap.  7, I point out similarities between the qualitative stories and the survey data. When a pattern that emerges from the individual stories mirrors a pattern in the answers to the survey questions, this suggests similarity in the underlying experiences or worldviews.

Combining qualitative and quantitative methods can be problematic. Generally, research with large samples, which uses structured research methods, is grounded in the perspective that there is a social reality ‘out there’—a reality that exists outside individuals and which is sought to be ‘accurately’ exposed through methods that are ‘completely objective’ (Holstein and Gubrium 1995). In these studies, usually surveys, all included aspects (variables) are predefined by the researcher. Opposite this objectivist and positivist perspective stands the interpretivist perspective, which is the primary perspective of this book. Interpretivism—which is connected with the constructivist view—focuses on the ways in which individuals interpret the world (Bryman 2008). More precisely: interpretivist studies access the world through the interpretations of individuals. Getting to know the world from people’s own perspectives requires an open, unstructured, or qualitative, method. The objectivist perspective is dominant in ideas about ‘good science’ and what is seen as ‘proof’; this is the case in much of the academic world, in everyday reasoning in the media, and in political argumentations.

That qualitative and quantitative methods are generally connected with two different perspectives does not mean that they cannot be combined (Bryman 2008; Johnson and Onwuegbuzie 2004; Niglas 2010). In this study, I combine in-depth, semi-structured interviews with survey data. In the use of the survey data, I take a middle way between the two epistemological perspectives. Although I use the survey data to examine the presence of broader societal patterns, I only use these findings in tandem with the qualitative results, which are crucial for the interpretation of the quantitative figures. Because respondents were offered predefined answering options, I am very cautious in the interpretation of the answers to the survey questions. Particularly for less concrete questions—such as ‘To what extent to you feel Dutch?—I am hesitant to assume that the respondents’ answers reflect static, substantive, ‘factual’ dispositions. I see these answers as nothing more than answers to a question, given at a certain moment in a certain context. In Chap.  5 we will see that the quantitative findings undermine the common ‘thick’ view on identifications; it is a result that asks for a qualitative phenomenological exploration of the phenomenon of ethnic self-identification.
  • The Mixed-Methods Design of this Study

    In this mixed-methods design, semi-structured, in-depth interviews with 14 university-educated second-generation Moroccan Dutch and Turkish Dutch form the main data source. These qualitative data (QL) are supplemented by quantitative (QN) data gathered through a large-scale structured survey conducted among 1500 respondents with a Moroccan-, Turkish- or ethnic-Dutch background in the context of another study. Before I discuss the separate approaches in the next section (Sect. 3.2 and 3.3), this current section contains the overall research design, including a brief description of how the research focus developed. To describe the relation between the QL and QN data, I use the model as explained by Leech and Onwuegbuzie (2009, see also Creswell and Plano Clark 2011; Niglas 2009), who distinguish various main choices in a mixed methods design, which include:

  • Timing of the different methods (parallel, sequential or embedded);

  • Emphasis placed on the different methods (equal emphasis or with one of the approaches considered dominant);

  • Focus (studying the same or different parts of a phenomenon).

I also use the typology of Greene et al. (1989), who identify five purposes of combining different methods.1 These purposes for mixing are:
  • Expansion: increase the scope of the inquiry by using different methods for different themes;

  • Clarification: clarify (or illustrate or interpret) the results from one method with the results of the other method2;

  • Triangulation: seek convergence of results from different methods on the same theme;

  • Development: develop one method based on the results from the other method;

  • Initiation: discover paradoxes, contradictions, and fresh perspectives that (often unexpectedly) emerge from the combination of the methods.

I started my research with the analysis of the quantitative data for a very practical reason: these data were already available when I started my project. In fact, I initially focused on the explanation of social mobility and its relation to the social context and identifications. The data seemed highly useful for this purpose, as they contained many details about educational trajectories, familial backgrounds, social contexts, and identifications of large numbers of second-generation individuals that enabled the exploration of associations between the various factors. Because I was not only interested in mere correlations but also in understanding processes of social mobility as experienced by individuals, I used a less-structured approach that allowed to me learn more about the complexities of people’s experiences and their trajectories. Aiming for triangulation, I conducted 14 in-depth interviews to explore the same theme from a different angle. I tried to understand what made second-generation individuals socially mobile and how this trajectory related to social contexts and identifications. In the analysis phase, I became more and more triggered by the data on identification. In the survey data, I noticed that the respondents’ answers to questions about ethnic identification were not associated with cultural practices in the way I initially expected (see Chap.  5). Likewise, in the in-depth interviews, what fascinated me most were topics related to identifications and the development thereof. When participants reflected on their positions in various social contexts and their ethnic and national identifications, their accounts were full of ambiguities, emotions, and shifting positions (called ‘narrative shifting’, see Holstein and Gubrium 1995, p. 55), which continuously intrigued me (see Chap.  6). That is how my purpose shifted from explaining social mobility to understanding processes of ethnic identification.

The phases of the data collection can be sketched as follows.

This scheme does not reflect the entire setup of the research. In the data analysis phase, the two data sources were used in various compositions, with various aims. For Chap.  5, which explores how strongly the second generation identifies with the ethnic and national labels, as well as the meaning of these identifications, the datasets are used in the following way (the use of upper and lower case reflects the emphasis placed on the different methods).

In Chap.  5, the outcomes of the statistical analyses on ethnic identification ask for clarification. The qualitative data are used to understand the quantitative findings and the two steps focus on the same parts of the phenomenon.

Chap.  6, which explores the contextual character of the participants’ self-identifications, is entirely based on the data of the in-depth interviews.

Chap.  7, which deals with the temporal aspect of social contexts and identifications, relies primarily on the interview data. Some of the findings are backed up with findings from the survey data to indicate the generalizability of certain results. As the quantitative data are used to understand the breadth of the qualitative findings, here the purpose of mixing is again for clarification.

3.2 Quantitative Approach. Use of the TIES Survey Data

The survey data were collected in 2006 and 2007 in the context of the international TIES project. This project focused on The Integration of the European Second Generation (TIES) and was coordinated by the Institute of Migration and Ethnic Studies (IMES) of the University of Amsterdam and the Dutch Interdisciplinary Demographic Institute (NIDI). The project studied the incorporation of children of immigrants who were born and educated in their countries of residence, in 15 cities across eight European countries. For the Netherlands, the TIES project is the first large-scale study focusing specifically on second-generation youths (Crul and Heering 2008). The description of the data collection is based on Groenewold (2008) and Groenewold and Lessard-Phillips (2012). The Dutch segment of the survey was conducted face-to-face among 1505 respondents aged between 18 and 35 years in the cities of Amsterdam and Rotterdam. The respondents were equally spread over three ethnic categories: second-generation Moroccan Dutch, second-generation Turkish Dutch (at least one parent born in Morocco or Turkey) and a control group of ethnic Dutch (both parents born in the Netherlands). The questionnaire contained detailed questions about a range of themes, including educational trajectory, employment, household, neighbourhood, parental background (education, employment, and migration history), language use, family relations, identifications, sociocultural practices, attitudes, religiosity and discrimination. Data collection and processing were carried out by the survey organisation Bureau Veldkamp.
  • Sampling Procedure

    The aim of the Dutch survey was to obtain statistically-representative information on second-generation Turkish and Moroccan Dutch in Amsterdam and Rotterdam (Groenewold 2008). The sampling was carried out in various steps. First, neighborhoods were sampled. In the two cities, 47 of the 167 neighborhoods were sampled. This number was based on a cluster size of 30 (ten respondents per ethnic category) and a compromise between having enough respondents per cluster and having enough clusters. To get an optimal spread of the respondents over neighborhoods with different concentrations of second-generation Moroccan Dutch and Turkish Dutch, the selection-probability for each neighborhood was proportional to the number of residents with Moroccan and Turkish backgrounds. The sampling frame only included neighborhoods with residents from all three ethnic categories; a few small neighborhoods were excluded. Subsequently, individual respondents were selected from the sampled neighborhoods. Because of the expected non-response, initially 6000 addresses were sampled from the municipal population registers (GBA)—four times the minimum effective sample size of 1500—of which 4999 addresses were valid. The GBA (Gemeentelijke Basis Administratie) contains information about all legal residents in the municipality, including address, gender, date of birth, country of birth parents, and nationality. Later, another 271 additional addresses were sampled to increase the numbers of respondents. Ethnic-Dutch respondents were sampled from the same neighborhoods in similar numbers as the second-generation respondents. Eventually, 1505 individuals were interviewed (see the size of the three ethnic categories of respondents in the two cities in Table 3.1). The overall response rate was 30%; it was slightly higher for the ethnic Dutch than for the Moroccan Dutch and Turkish Dutch. Low response rates are common for young respondents with immigrant backgrounds. A comparison of the population and the sample suggests that the non-response bias is only small (For further information on the methodology and the broader project see: Crul and Heering 2008; Crul et al. 2012; Groenewold 2008; Groenewold and Lessard-Phillips 2012, and the webpage of the TIES project: www.tiesproject.eu).
    Table 3.1

    TIES respondents (size of ethnic groups per city)

     

    Amsterdam

    Rotterdam

    Total

    Turkish background

    237

    263

    500

    Moroccan background

    242

    251

    493

    Ethnic-Dutch control group

    259

    253

    512

    Total

    738

    767

    1505

    Data TIES survey for the Netherlands, 2007, NIDI and IMES

  • Data Collection

    After a pilot phase during which the questions were tested and adjusted, the surveys were conducted between May 2006 and July 2007 by 83 experienced and trained interviewers. Most of them had an ethnic-Dutch background. Invitation letters were sent to explain the study’s objectives and announce the visit of the interviewers. Participants received ten euros for their participation, which was also mentioned in the letter.

The interviewers encountered various problems, such as: selected individuals who did not live at the registered address, inaccessibility of apartment buildings, and suspicious or hostile individuals. Also the duration of the interview—which took one hour and fifteen minutes on average—was sometimes experienced as problematic, as was also the sensitive nature of some questions. Interviewers sometimes skipped questions or conducted the second part of the survey on paper that could be filled out at later time. To reduce non-response, reminder letters were sent, the participation fee was increased, and the interviewers were trained in persuasion techniques.
  • Reflection on the Use of the Data

    The use of the statistical data in this book illustrates that statistical analysis of structured data is not necessarily based on an objectivist and positivist perspective, nor does it necessarily focus on testing strictly-defined hypotheses. From an interpretivist perspective, the use of concepts that are predefined by the researcher (variables) is somewhat problematic as this ignores, or even overrules, understandings of the people themselves. This study illustrates how quantitative data can be used within a study that primarily holds an interpretivist view. In Chap.  5, statistical analyses are used to deconstruct an objectivist and groupist conception of identification by showing that identification with an ethnic label does not necessarily reflect a specific coherent cultural content. In Chap.  7, quantitative data help us reveal the intersectional character of education level and ethnic background, nuancing the groupist idea that Moroccan Dutch and Turkish Dutch are more conservative than ethnic Dutch. However, the data are also used in more objectivist ways, for example in Chap.  4, where I present a descriptive comparison of Moroccan Dutch and Turkish Dutch sociocultural practices, and in Chap.  7, where I present the demographic characteristics of the respondents’ social networks.

The survey was carried out in 2006 among respondents between 18 and 35 years old, whereas most of the in-depth interviews were conducted in 2011 with participants who were over 40 years old. Nevertheless, the strong parallels between the findings imply that these differences are not problematic and that the qualitative data still enhance our understanding of the patterns in the quantitative data. This is aligned with the description in Chap.  4, which shows that Dutch context did not abruptly change in this period.

3.3 Qualitative Approach. In-depth Interviews

This section on qualitative data collection and analysis is relatively detailed. The reason is that qualitative data collection and analysis are much less straightforward than quantitative approaches. Not only are qualitative approaches less structured, there is also a lack of standard guidelines for reporting about qualitative approaches (Guba and Lincoln 2005). This requires a relatively detailed justification of the approach. In addition, I find it important to open the black box and exemplify a possible approach for qualitative data analysis.

In the qualitative tradition, there is not even full agreement on the criteria for evaluating research that could provide guidance for writing a sound methodological justification (Bryman 2001, p. 270; Guba and Lincoln 2005; Silverman 2006). I agree with the view that producing valid knowledge is not about uncovering ‘the truth’, but obtaining and presenting findings that are credible (Silverman 2006, p. 281). According to Riessman, it comes down to the question: why should we believe it? (2008, p. 184). I agree with Silverman (2006) that we can evaluate the credibility of qualitative research using the same core criteria as in quantitative research: validity and reliability. I would say that research findings are credible when they are likely to accurately represent the social phenomena to which they refer; in other words: when they are valid (see Hammersley 1990 in Silverman 2006, p. 289). Therefore, it is important to show that the findings are not accidental results shaped solely by the circumstances of the research. In other words: the findings need to be reliable (see Kirk and Miller 1986 in Silverman 2006, p. 282). In order to judge the reliability of the findings, it is crucial that the research process is transparent; that it is clear how the data were obtained, what the influence of the research setting was, and how the conclusions were developed from the data through processes of interpretation (Silverman 2006, p. 282). This means that bias, which cannot be avoided in any study, needs to be understood and explained (Small 2009, p. 14). As Riessman argues: good research is credible or persuasive when the researcher demonstrates that ‘the data are genuine, and analytical interpretations of them are plausible, reasonable, and convincing’ and when the researcher’s theoretical claims ‘are supported with evidence from informants’ accounts, negative cases are included, and alternative interpretations are considered’ (2008, p. 191). The report of a scientific study should be transparent in how the final claims are developed, based on a ‘trail of evidence’, consisting of data, analyses, and interpretations (Riessman 2008, p. 188). This transparency is particularly important for less-structured approaches, in which findings are more strongly shaped by circumstances and by the decisions and the personality of the researcher, hence this relatively extensive section.

Following Holstein and Gubrium (1995), I see an interview as an ‘active interview’, as something that is created in a particular setting and is the result of a situated interaction between the interviewer and the research participant. The situated character of the narrative does not mean, however, that the interview is created from scratch during the interview or that the respondent is making things up (p. 28). Instead, a story is created that is ‘true to life’—faithful to subjectively meaningful experience (p. 28). To assess what the participants’ words mean, we should consider the context of how the narrative came into being. I do this by carefully describing the interviews and the analytical steps, and by reflecting on my personal role as interviewer and researcher during the interviews and the interpretive process. In the empirical Chap.  5 7, I show how the conclusions of this research are tied to the empirical data. In this section, I further discuss how I approached the qualitative data collection and analysis.
  • Data Collection

    I describe successively the selection of the participants, the interview, and the processing of the interviews.

  • Selection of the Participants

    I conducted 14 interviews with socially-mobile second-generation Moroccan-Dutch and Turkish-Dutch men and women. The criteria for selection were that they were born in the Netherlands from parents (at least one) who migrated from Morocco or Turkey to the Netherlands, or that they arrived here with their parents at a very young age, that is, before entering the educational system. In addition, they had to have graduated from university and hold jobs matching their education level at the time of the interview. As I intended for them to reflect on their trajectory of mobility, I selected people who were not at the very beginning of their professional careers and were over 30 years old. In the end, two male participants with Moroccan backgrounds did not fit these criteria, as one had come to the Netherlands at an older age and one had not attended university but had graduated from higher vocational training (HBO); however, as they nevertheless contributed to my findings and their stories did not substantially deviate from the other stories, I did not exclude them. As my final focus excluded individuals with a mixed ethnic background, I did exclude the fifteenth interview, with Nathalie, a participant with a Moroccan-Polish background.

Ten of the interviews were conducted with Moroccan Dutch (of which three were female) and four with Turkish Dutch (two female and two male participants) (Table 3.2). I conducted four of the interviews in 2006, for a previous project on ethnic identification, while the rest were conducted in 2011. That the context did not change that much between these years was reflected in the interviews, which did not radiate a different Zeitgeist.3 All participants were in their thirties or early forties at the time of the interview. This meant that they were born shortly after (or before) their families migrated to the Netherlands, which makes them what I call members of the ‘early’ second generation. Some were in relationships (mostly married), and others were single. Some had children. They lived all over the Netherlands and grew up all over the Netherlands, in cities as well as in villages. Several worked as consultants in various sectors, some ran companies they (co-) owned, one worked in the medical field, and others worked as researchers, technical engineers, and teachers. All participants spoke Dutch fluently. Most of the participants did not have any accent that revealed their immigrant backgrounds. Nearly all participants had—in my view—a ‘professional’ appearance. They were dressed according to standard business codes, radiated confidence and reflexivity and formulated their thoughts with a certain ease and determination. Although nearly all participants call themselves Muslim, their level of religiosity varied. It seemed to me that for three of them, their religiosity was more important emotionally and for providing practical guidelines than for the rest. To protect the anonymity of participants, I do not connect the various personal characteristics with each other and do not create detailed profiles of the individual participants. I furthermore use pseudonyms and altered some factual details.
Table 3.2

Interview participants (pseudonyms; ethnic backgrounds and gender)

Participants (pseudonyms)

Mor/Tur Dutch

Gender (m/f)

Participants (pseudonyms)

Mor/Tur Dutch

Gender (m/f)

1. Bouchra

M

f

8. Karim

M

m

2. Hicham

M

m

9. Yunus

M

m

3. Mustapha

M

m

10. Imane

M

f

4. Ahmed

M

m

11. Berkant

T

m

5. Said

M

m

12. Aysel

T

f

6. Masud

M

m

13. Adem

T

m

7. Hind

M

f

14. Esra

T

f

In chronological order of the interviews

To avoid selecting participants based on their ethnic identifications and thus selecting on the dependent variable, I did not use organizations with ethnic signatures as starting points for recruiting. I recruited most participants via my own (primarily ethnic-Dutch) private network, covering various professional branches and various parts of the Netherlands. I recruited a few participants via my professional academic network. Furthermore, I avoided an emphasis on ethnicity in the announcement of the interview topic, which I formulated as ‘the social mobility of children of immigrants’. Nonetheless, all participants (partially) identified in ethnic terms. As participation was voluntary, a certain bias could not be completely avoided. In explaining their willingness to participate, most participants mentioned the importance of contributing to the Dutch debate, to have their voices heard and challenge negative stereotypes. This implies that the participants have a relatively strong social involvement.
  • The Interview

    The interviews were semi-structured, lasted between one and four hours, and were all conducted in Dutch. All interviews were recorded on audiotape, except for one, in which the participant objected to the recording. A translated, English version of the topic list is included in Appendix A.

The first part of the interviews did not explicitly focus on ethnic identity. I started by asking the participants to describe their educational trajectory chronologically—including familial background and educational trajectories of siblings—focusing on social environments and the role of social others. This provided a detailed picture of the composition of the various social contexts they moved in (in characteristics of gender, class, and ethnic background) and how they experienced their social relations and positions in these various contexts without the participants interpreting these situations through the lens of ethnic identification. By focusing on the process of social mobility instead of ethnic and national identifications in the initial stage of the interview, I followed one of Fox and Jones’ suggestions (2013) to avoid the trap of unwillingly applying an ethnic lens (see Sect.  2.2) by focusing on the ‘everyday’ as a means to explore practices beyond ethnic practices. The focus on trajectories of social mobility had a similar effect. When we discussed the theme of feeling ‘Moroccan’, ‘Turkish’ and ‘Dutch’ later in the interview, many details had already been discussed, which we could then use to reflect on expressions of ethnic identification. (Later, in the analysis, experiences that had been formulated in terms of ‘feeling different’ and ‘feeling similar’ and ‘feeling normal’ helped me understand the role of ethnic, Dutch, and other identifications.)

Throughout all of the interviews, I was uneasy asking about them feeling Moroccan, Turkish, and Dutch, about ethnic backgrounds and the role of ethnicity. I feared that this focus made me contribute to a discourse that presupposed the relevance of ethnicity for individuals with an ethnic-minority background, and I therefore wanted to avoid the impression that I myself assumed that ethnicity is always greatly relevant. However, the participants’ responses to these questions were insightful. As I will show in the coming chapters, in some responses participants did not problematize these questions at all, whereas in other responses they challenged the underlying views.

In many of the interviews, I felt that my own educational and professional background contributed to the mutual rapport. In only a few cases, I felt that my gender played a role and enhanced the rapport with other female participants, when we discussed the theme of being a gender minority in educational or professional settings. I did not feel that my gender influenced the interaction with other (male or female) participants. I do not know how the fact that I did not have the same ethnic background affected the situation. I can imagine that this made the participants hold back in relating negative experiences, as they might have wanted to portray an extra-positive image to challenge stereotypes that are related to their ethnic category. Nevertheless, this effect seemed limited, as participants often did reflect on relationships with coethnics, and also mentioned disagreements and struggles in what seemed to be quite an honest way.

  • Data Analysis

    As I argued before, in order to enhance the credibility of the research findings, it is important to show how the claims I make in this book relate to the empirical data. In this section I describe the analytical steps I took to develop the themes as discussed in this book. I believe that it is important to also include the more initial, explorative analytical phases, as these are crucial steps in the process of meaning-making, in the interpretation of the data. The research log, in which I kept track of my analytical steps as well as my considerations and confusions, not only helped me retrace my analytical steps, but—like Riessman suggested (2008, p. 191)—also fostered my reflexivity and awareness. The challenge here is to offer an overview that elucidates the process but is also concise. I start with the transcription phase, proceed to the explorative stage of open coding and memo writing, and conclude with a description of the main analyses.

  • Transcriptions

    However straightforward it sounds to make a transcription, ‘the “same” stretch of talk can be transcribed very differently’ (Riessman 2008, p. 29). As I increasingly wanted to attend not only to the ‘what’ of the interview, but also to the ‘how’ (Holstein and Gubrium 1995), I improved the first transcriptions several times, every time including more details on the ‘how’ of the interviews. Following Gillham’s suggestion (2005), I included my own speech (including my questions, probes, and audible reactions) and ‘paralinguistic’ features (such as hesitations or emphases) when they seemed important for the interpretation. I also included speech repetitions, such as ‘you know’, because these often appear to express emotions such as unease or agitation. I transcribed all the interviews myself, as I agree with the view that transcription is an interpretive practice (Gillham 2005; Riessman 2008). After the transcriptions, I listened to the interview again and added interpretative notes, using the qualitative data analysis software MaxQDA. An example of a brief interpretive note, called a ‘memo’ in MaxQDA, is the following memo I attached to Karim’s words about his disappointingly low secondary school advice:

Memo: These sentences already radiate frustration. (He returns to this theme later in the interview). And that he mentions that in the end his graduation was ‘with honors’ sounds like a redress, illustrating how ridiculous the previous advice was. It sounds like ‘I told you so!’ (Memo dd. 13 August 2012, translation MS)

  • Exploring the Data: Open Coding and Memo Writing

    As I did not want to force any structure upon the data by using preconceived categories, I started with a bottom-up coding approach, conforming with the principles of Grounded Theory (Corbin and Strauss 2008). I started with the process of open coding and assigned codes to text segments that reflected the theme, meaning, or emotions of the participant’s words or more processual aspects, such as instances of reflexivity. I created memos about the content of the specific codes. The coding resulted in 120 codes and nearly 1800 coded segments.

To make sense of these codes, I divided the codes into four categories: ‘arrear and success,’ ‘identification, ethnicity and social relations’, ‘life phases’, and ‘other’. I then explored the relationships by grouping the subcodes that were similar in meaning or theme and explored how the various themes connected to each other, trying to piece together a diagram that reflected a coherent argument. This sorting exercise invited me to play with the data but did not lead to an unambiguous, coherent, innovative diagram and argument. While searching for coherent arguments and trends in the data, I tried to be perceptive of variations and negative examples, as suggested by Charmaz (2006, p. 102) and Corbin and Strauss (2008, p. 84).

The numerous memos I wrote, following the approach of Corbin (Corbin and Strauss 2008, see also Charmaz 2006, Chap. 4) turned out to be most useful in furthering the meaning-making process. In the entire project, I wrote 521 memos, which were attached to codes or text segments. I found that extensive memo-writing enhanced my insight by helping me disentangle complexities in the data and further my thinking on issues I did not understand right away. I used the memos both to describe the ideas behind the developed codes and also to explain why I found certain expressions intriguing; what I found surprising or confusing, and what confirmed my hunches; and how participants’ experiences or interpretations paralleled or contradicted each other. The following memo, assigned to a specific interview segment and connected to the codes ‘reluctant to use ethnicity/ethnic explanations’ and ‘being Dutch, Moroccan, Turkish’, illustrates how I used the writing of a memo when I was confused:

Memo: Suddenly, here she seems very resistant to categorization in ethnic categories. Why? I feel it fits her cynical outlook on the world. Why then does her resistance surprise me? That is because earlier in the interview she did talk about not-being-Dutch, and being-Turkish herself. So, she does employ such categorizing language herself. But now it suddenly frustrates her. I think she might be afraid that such approaches are not constructive – that they too strongly reflect the exclusivist thinking of the dominant discourse. Either way, she is critical every time – in reaction to nearly everything happening in the Netherlands, and to nearly everything I say. (Memo dd. 28 September 2012, translation MS)

  • Advancing the Analysis

    Three analytical steps furthered my thinking on the themes and arguments in the data. The first was to combine all memos that were connected to the codes within the main theme ‘identification, ethnicity and social relations’: 70 at that stage. This collection of reflections formed the basis for a document in which I described various mechanisms and concepts that emerged from the interviews (using labels such as ‘practices of in- and exclusion’, ‘process of developing pride’, ‘the role of social others’ and ‘categorization resistance’), which I discussed in various versions with various colleagues.

The second step was an analysis of the social contexts, inspired by Corbin and Strauss (2008, Chap. 10). For each interview, I created an overview of the various contexts that were mentioned (such as family (parents and siblings), neighborhood, local coethnic community, primary school, secondary school, university, work, partner, peers), including the participants’ evaluations of these contexts. Obviously, social contexts differed per life phase, but how participants positioned themselves also showed development.

This development explains why I got stuck when using the grounded theory approach the way that I did. As grounded theory approaches invite the use of text segments in fractured, decontextualized ways, it is easy to lose narrative aspects of the interview (Mishler 1999, p. 23). Instead, ‘narrative analysis’ attempts to keep the ‘story’ intact, and attends to sequences and the personal interaction in the interview setting (Riessman 2008). The idea of narrative analysis led me to pay more attention to developments and mechanisms, as well as arguments constructed by the participant. I looked for words that indicated a specific relation between two parts of a narrative (since, due, when, because, results in) and words that were indicative of temporality and change (initially, gradually, current, ‘now I feel…’, ‘this has become…’, ‘I have learnt’) (see Corbin and Strauss 2008, p. 83). This way of looking also enabled me to notice ambiguities within interviews, as participants at times seemed to contradict themselves. As Chaps.  6 and  7 show, this attention to ambiguities and temporality appeared very valuable for the further crystallization of my findings.

The third step, in which I employed the idea of narrative analysis, focused on ‘processes’. Again, I was particularly inspired by the ideas of Corbin and Strauss. They developed a perspective to help the researcher identify the role of context and link context to process and outcome (2008, p. 89). In the transcripts, I searched for narrative chains consisting of (a) conditions, (b) interactions and emotions, and (c) consequences. I wrote a memo on every such process/chain, outlining the specific conditions (triggers, context, and causes) and responses (emotions, actions, reactions, results, or aimed results). Per interview, I coded between 12 and 51 text segments, which I finally classified (and re-classified) into three categories that emerged from the data:
  1. 1.

    ‘Netherlands’: 181 segments, relating to interactions with ethnic Dutch and to the Dutch discourses,

     
  2. 2.

    ‘Coethnics’: 102 segments, relating to parents, the local coethnic community, the abstract coethnic community,

     
  3. 3.

    ‘Friends’: 29 segments that I identified as processes relating to people who are considered friends, regardless of their ethnic background, and to partners.

     
Per category I considered the various relevant actors, triggers, effects, and reactions. This led to detailed descriptions about the range of interactions and responses in different interactional contexts, which formed the basis for Chap.  6.
  • A More Structured Approach for Chap.   5

    The analysis of the qualitative data for Chap.  5 was more straightforward. Based on the outcomes of the statistical analyses, I analyzed how the participants described their identification as Moroccan or Turkish or Dutch. I retrieved the 48 text segments that were coded ‘being Dutch, Moroccan, Turkish’. I developed thematic subcodes, such as ‘language’, ‘attitudes’ or ‘bond with the country’. I also looked into the combinations of these themes per participant, and I considered if the use of these themes noticeably varied between the Moroccan and Turkish Dutch and between men and women.

3.4 Summary

This phenomenological mixed methods study of the identifications of Dutch social climbers of Moroccan and Turkish descent is based on in-depth interviews, which are combined with survey data collected in the context of another study. Both the qualitative and quantitative data were used within an interpretivist perspective. Nevertheless, I described the quantitative data collection according to the standards that are common in positivist research traditions. For example, I did not reflect on how the context of the data collection might influence the findings. The practical reason is that I was not involved in the collection of the survey data and that I base my description on the reports of others.

The description of the qualitative approach is more elaborate than that of the quantitative approach and includes discussions of my role as an interviewer, my perspective on the interview data as empirical evidence, and the processes of making sense of the data. The reason is that less-structured approaches by definition lack high levels of standardization. For reasons of credibility and transparency, qualitative methods require a detailed presentation of the ‘chain of evidence’, showing how the data were gathered and how the findings follow from the data.

Footnotes

  1. 1.

    Various typologies are mentioned in the literature. Some authors present typologies of mixed methods studies based on technical characteristics, such as the sequence of the methods and the emphasis placed (e.g. Caracelli and Greene 1997; Leech and Onwuegbuzie 2009; Teddlie and Tashakkori 2009). Personally, I do not find these very useful in designing a study because the technical design is not a consideration in itself, but should arise from the purpose of the study. Other authors, like Creswell and Plano Clark (2011), give a typology that mix purposes and designs. I find their typology confusing because the options they sketch are too rigid and limited.

  2. 2.

    Greene et al. (1989) refer to this purpose as ‘complementarity’, but I prefer the label clarification to clearly distinguish this purpose from the purpose of expansion.

  3. 3.

    I asked the 2006 participants for permission to use their interviews in this project as well. One I was unable to reach again, but decided to use the interview for the analysis, but include only a few quotes in the book in a decontextualized way, without any other personal information.

References

  1. Bleijenbergh, I. (2013). Kwalitatief onderzoek in organisaties. Den Haag: Boom Lemma.Google Scholar
  2. Bryman, A. (2001). Social research methods (1st ed.). Oxford: Oxford University Press.Google Scholar
  3. Bryman, A. (2008). Social research methods (3rd ed.). Oxford: Oxford University Press.Google Scholar
  4. Caracelli, V., & Greene, J. (1997). Crafting mixed-method evaluation designs. In J. Greene & J. Caracelli (Eds.), Advances in mixed-method evaluation: The challenges and benefits of integrating diverse paradigms (pp. 19–32). San Francisco: Jossey-Bass.Google Scholar
  5. Charmaz, K. (2006). Constructing grounded theory: A practical guide through qualitative analysis. Londen: SAGE.Google Scholar
  6. Corbin, J., & Strauss, A. (2008). Basics of qualitative research: Techniques and procedures for developing grounded theory (3rd ed.). Thousand Oaks, CA: Sage Publications.Google Scholar
  7. Creswell, J. (2007). Qualitative inquiry and research design. Choosing among five approaches. Thousand Oaks, CA: Sage Publications.Google Scholar
  8. Creswell, J., & Plano Clark, V. (2011). Designing and conducting mixed methods research (2nd ed.). Thousand Oaks, CA: Sage Publications.Google Scholar
  9. Crul, M., & Heering, L. (2008). The position of the Turkish and Morocan second generation in Amsterdam and Rotterdam. The TIES study in the Netherlands. Amsterdam: Amsterdam University Press.Google Scholar
  10. Crul, M., Schneider, J., & Lelie, F. (2012). The European second generation compared. Does the integration context matter?. Amsterdam: Amsterdam University Press.Google Scholar
  11. Dukes, S. (1984). Phenomenological methodology in the human sciences. Journal of Religion and Health, 23(3), 197–203.CrossRefGoogle Scholar
  12. Flyvbjerg, B. (2004). Five misunderstandings about case-study research. In C. Seale, G. Gobo, J. Gubrium, & D. Silverman (Eds.), Qualitative research practice (pp. 420–434). London: Sage Publications.Google Scholar
  13. Fox, J., & Jones, D. (2013). Migration, everyday life and the ethnicity bias. Ethnicities, 13(4), 385–400.CrossRefGoogle Scholar
  14. Gillham, B. (2005). Research interviewing: The range of techniques. New York: McGraw-Hill International.Google Scholar
  15. Greene, J. C., Caracelli, V. J., & Graham, W. F. (1989). Toward a conceptual framework for mixed-method evaluation designs. Educational Evaluation and Policy Analysis, 11(3), 255–274.CrossRefGoogle Scholar
  16. Groenewold, G. (2008). Annex 1 sample design, survey implementation and evaluation. In M. Crul & L. Heering (Eds.), The position of the Turkish and Moroccan second generation in Amsterdam and Rotterdam. The TIES study in the Netherlands (pp. 169–180). Amsterdam: Amsterdam University Press.Google Scholar
  17. Groenewold, G., & Lessard-Phillips, L. (2012). Research methodology. In M. Crul, J. Schneider, & F. Lelie (Eds.), The European second generation compared: Does the integration context matter? (pp. 39–56). Amsterdam: Amsterdam University Press.Google Scholar
  18. Guba, E., & Lincoln, Y. (2005). Paradigmatic controversies, contradictions, and emerging confluence. In N. Denzin & Y. Lincoln (Eds.), The Sage handbook of qualitative research (pp. 191–215). Thousand Oaks, CA: Sage Publications.Google Scholar
  19. Hammersley, M. (1990). Reading ethnographic research: A critical guide. London: Longmans.Google Scholar
  20. Holstein, J., & Gubrium, J. (1995). The active interview. Thousand Oaks, CA: Sage Publications.CrossRefGoogle Scholar
  21. Johnson, R., & Onwuegbuzie, A. (2004). Mixed methods research: A research paradigm whose time has come. Educational Researcher, 33(7), 14–26.CrossRefGoogle Scholar
  22. Kirk, J., & Miller, M. (1986). Reliability and validity in qualitative research. London: Sage Publications.CrossRefGoogle Scholar
  23. Larsson, S. (2009). A pluralist view of generalization in qualitative research. International Journal of Research & Method in Education, 32(1), 25–38.CrossRefGoogle Scholar
  24. Leech, N., & Onwuegbuzie, A. (2009). A typology of mixed methods research designs. Quality & Quantity, 43(2), 265–275.CrossRefGoogle Scholar
  25. Mishler, E. (1999). Storylines: Craftartists’ narratives of identity. Cambridge, MA, London: Harvard University Press.Google Scholar
  26. Niglas, K. (2009). How the novice researcher can make sense of mixed methods designs. International Journal of Multiple Research Approaches, 3(1), 34–46.CrossRefGoogle Scholar
  27. Niglas, K. (2010). The multidimensional model of research methodology. An integrated set of continua. In A. Tashakkori & C. Teddlie (Eds.), Handbook of mixed methods research (pp. 215–236). Thousand Oaks, CA: Sage Publications.Google Scholar
  28. Polkinghorne, D. E. (1989). Phenomenological research methods. In R. Valle & S. Halling (Eds.), Existential-phenomenological perspectives in psychology (pp. 41–60). Boston, MA: Springer.CrossRefGoogle Scholar
  29. Riessman, C. (2008). Narrative methods for the human sciences. Thousand Oaks, CA: Sage Publications.Google Scholar
  30. Silverman, D. (2006). Interpreting qualitative data (3rd ed.). Londen: Sage Publications.Google Scholar
  31. Small, M. (2009). “How many cases do I need?”: On science and the logic of case selection in field-based research. Ethnography, 10(1), 5–38.CrossRefGoogle Scholar
  32. Teddlie, C., & Tashakkori, A. (2009). Foundations of mixed methods research: Integrating quantitative and qualitative approaches in the social and behavioral sciences. Thousand Oaks, CA: Sage.Google Scholar

Copyright information

© The Author(s) 2018

Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

The images or other third party material in this chapter are included in the chapter's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

Authors and Affiliations

  1. 1.Vrije Universiteit AmsterdamAmsterdamThe Netherlands

Personalised recommendations