Chapter 6. Methods and material
1. Interviewing economists
The decision to use an interview study as one of the two empirical studies was driven by several factors. The overarching problem this study attempts to understand, the dynamics of the styles of reasoning of the economics discipline, could have been studied using a variety of approaches. As covered in the literature reviews in chapters 2 and 3, as well as the historical overview in chapter 5, there has been quite a lot written on similar topics. A lot of this literature exists within the history of economic thought, especially by those authors inspired by various STS approaches. However, most of those studies examine scientific writings, that is, the finished products of scientific knowledge production. Following the general thrust both of the STS field, and of the more recent turn towards the sociology of social knowledge (Camic et al. 2011), my aim is to reach closer to actual knowledge-producing practices and the knowledge producers themselves. The method of choice in science studies has long been ethnography. While this could have been an interesting and viable option, there is an obvious problem with ethnographic fieldwork on something that, after all, takes place to such a large extent “in the head”. Of course, scientific seminars, conferences, and in this case
perhaps also doctoral coursework, should be interesting sites to study ethnographically. On a yet smaller scale, collaborative research work, and even shadowing single researchers, could have been good methodological options.
However, since my core interest is also in “sociologizing” the concept of styles of reasoning, and exploring how it can be understood on the level of individual actors and a scientific habitus, I also needed to get closer to the actors and be able to elicit ideas, conceptions and dispositions that may not naturally emerge. For, as I will argue, interviewing can fruitfully be understood in a realist sense as similar to experimentation, in that it may bring out and produce phenomena that do not (often) occur spontaneously in nature or society. In the following sections, I will start with a very general discussion about the epistemology of interviewing, and move towards the more concrete questions of interview techniques, selection of informants, and the handling and analysis of the material.
The epistemology of interviewing—three views and their problems The approach to interviewing employed here is close to that in the well-known handbook Doing Interviews by Norwegian psychologist Steinar Kvale (Kvale 2007), and, like Kvale, I am indebted to Bourdieu’s fine methodological piece
“Understanding” (Bourdieu 1996). The latter combines lessons for the practical craft of interviewing with a sound epistemological framework for thinking about sociological interview research. Kvale is probably one of the foremost authorities on qualitative research interviews in the social sciences, with widely-read handbooks on the topic (see also Kvale and Brinkmann 2009, 2014). To briefly explain this approach to interviewing, it can usefully be contrasted to, first, an older view of interview methodology inspired by positivism, second, a hermeneutical view, and third, the currently more widespread constructivist approach.
The positivist approach treats the interviewee as a “vessel-of-answers”
(Gubrium and Holstein 1999, 2001b; Marvasti, Holstein, and Gubrium 2012) and the interviewer as a neutral “miner” of information (Kvale 2007:19). If the informant is a vessel containing information to be mined by the miner-interviewer, one of the central methodological problems becomes how to retrieve the valuable information-ore without the miner contaminating it. The interviewer must be neutral in order not to affect the interviewee. Such neutrality is often framed in terms of practical advice like dressing properly and avoiding leading questions at all costs. Such a “vessel view” is based upon the ontological assumption that there already exists something (beliefs, ideas, opinions, etc.) out there in the informants for the researcher to collect as-is. Furthermore, the
empiricist epistemology emphasises the neutral observer who neutrally perceives empirical reality as the prime road to knowledge. The goal is exact and generalisable knowledge, framed in terms of validity and reliability. If you ask a range of informants the right questions, they will tell it like it is. However, you must not influence the informant by formulating questions that are suggestive or introduce bias of any sort, which will ruin the validity of the study. There is a whole literature in witness psychology that experimentally shows the large extent to which the formulation of questions may shape answers.
A major problem with the positivist conception is that it grossly underestimates the role of the interviewer. For example, Kvale emphasises that interviewing is a craft rather than the neutral application of methodological rules. Instead, the interviewer’s person is the research instrument. Against what he calls the
“bureaucratic conception of method”, where “the ideal interview would be an interviewer-free method”, he posits the highly trained and skilled interviewer exercising judgement rather than context-free rules of method as a prerequisite for high quality interview data (Kvale 2007:48). The insight that the understanding of human existence requires another human has always existed as a parallel to positivism in hermeneutics and in anthropology fieldwork, which has been a main source of influence for contemporary qualitative methods. According to this conception, qualitative interview research is all about understanding the rich particularities of actors’ life-worlds. Questions of validity and objectivity are replaced by the search for rich authentic descriptions of particular local settings.
If the interviewer can get access to and listen carefully and attentively to the informant, the researcher may enter into the meaningful life-world of the informant. Ontologically, according to the hermeneutic conception, there are universes of meaning out there to discover (just like in the positivist conception), but epistemologically, there can be no separation of knower and known, no neutral outside observer. There is no escape from the researcher using his or her self as the only viable “instrument” of knowing.
The caricatured conception of interviewing I label “constructivism” is also formed as a critique of positivism, but in a more radical sense than hermeneutics.
Paraphrasing Marx on commodity fetishism in Capital (1976), Jaber Gubrium and James Holstein claim that if “[a]t first glance, the interview seems simple and self-evident” (Gubrium and Holstein 2001a:1), a closer examination shows that it is more intriguing than that. Gubrium and Holstein (2001a:13) turn against the vessel-of-answers view of interviews, where “the subjects behind respondents are basically conceived as passive vessels of answers for experiential questions put to them by interviewers. Subjects are repositories of facts, feelings, and the related particulars of experience” (emphasis in original). Instead, they claim that the
discourse produced in an interview is constructed in the act of interviewing. It is not a representation of something that already existed out there (i.e. in the vessels).
Taken in its pure form, such constructivism rests on an ontology of radical becoming, where persons do not hold any opinions or have any personalities, cultures, habitus or what have you, prior to the act of performing/ constructing belief etc. in the act of speech. Here, the interviewer is active in a much stronger sense (compared to the hermeneutic position), in that the interview discourse is a co-construction by interviewer and interviewee. The critique of the vessels-of-answers view has its merits. It leads us to consider the extent to which narratives are actively constructed by the agencies of both interviewer and interviewee as participants in the interview situation, and to the way that the subjectivity expressed by interviewees may in fact belong to different subject positions and voices (Gubrium and Holstein 2001a:22). For example, the respondent may speak from the position of individual experience, or as a representative of the profession, or perhaps as a citizen, shifting subjectivities during a single interview. It also leads us to think beyond individual subjects, instead focusing on the institutional discursive environment of subjects (Gubrium and Holstein 2001a:26). This means that the origin of beliefs and opinions must be sought beyond the individual, in the institutions that provide distinctive ways of speaking and interpreting everyday life. A similar point is made by Bourdieu, who turns against what he calls the naïve personalism in some interview research that doesn’t understand how persons are not as unique as they may appear, but instead always products of social structures (Bourdieu 1996:27).
Interviewing as Socratic sociological realism
Let me now try to elaborate the conception of interviewing that informs this study, based primarily on Kvale (Kvale 2007) and Bourdieu (Bourdieu 1996), and filtered and refined through a largely critical realist understanding of social scientific research (see Bhaskar 1998; Sayer 2000). This conception draws heavily on insights from the other three positions, but also tries to remedy their respective weak spots. From positivism, we learn to avoid leading questions and adverse effects of the interviewer. From hermeneutics, we learn the necessity of understanding through the active and personal engagement of the interviewer.
From constructivism, we learn that the interview is a setup, a constructed situation with a constructed outcome. Added to that, the conception I am proposing introduces a number of ideas.
First, the “sociological” aspect of interviewing is one of Bourdieu’s central claims. Bourdieu, like Kvale, argues that interviewing must be understood as a
craft that requires experienced and active interviewers. But Bourdieu is not a constructivist emphasising the local construction act of the interview, like Gubrium and Holstein. He positions himself against two common conceptions:
the scientistic, rigorously methodological, stance of the positivist tradition, which seeks to be free of any influence of the interviewer. On the other hand, he is equally critical of what he terms the “antiscientific” advocates of a hermeneutical approach that seeks the “mystic union” of interviewer and interviewee in a supposedly distortion-free melting together of understandings (Bourdieu 1996:18). The main obstacle to understanding subjects in Bourdieu’s view is the objective social distance between the interviewer and the interviewed, which inevitably influences and distorts the social exchange.
However, the only way to counter the distortion caused by social distance is through the interviewer’s sociological grasp of the social conditions that structure the subject’s life through giving oneself
a general and genetic comprehension of who the person is, based on the (theoretical or practical) command of the social conditions of which she is the product: a command of the conditions of existence and the social mechanisms which exert their effects on the whole ensemble of the category to which such a person belongs [. . .] and a command of the conditions, psychological and social, both associated with a particular position and a particular trajectory in social space. Against the old Diltheyan distinction, it must he accepted that understanding and explaining are one. (Bourdieu 1996:22–23, emphasis in original)
Contra the positivist non-interference view, Bourdieu emphasises the craftsmanship of interviewing centred on calibrating the effects of social distance and countering them continuously during the interview situation. This goes beyond the mechanic implementation of a methodology, it requires what he calls a “reflex reflexivity” that “enables one to perceive and monitor on the spot, as the interview is actually being carried out, the effects of the social structure within which it is taking place” (Bourdieu 1996:18). Such a conception is founded upon a realist social ontology of pre-existing social structures (contra radical constructivism).
Second, interviewing involves active and methodological listening, which goes beyond hermeneutic understanding or constructivist co-construction. In this conception, the active interviewer acts as a Socratic midwife who uses encouragement, attentiveness and follow-up propositions, helping to deliver the subject’s “truth” which was already out there (contra Gubrium and Holstein), but which required an ideal and constructed situation (contra the positivist emphasis
on interviewer neutrality) for the informant to deliver ideas and conceptions that are deeply held, but that would not be shared under most social circumstances.
This is in line with Kvale(2006), who claims that there is a common but problematic conception of interviews as dialogues. The interview method has sometimes been depicted as more egalitarian than “the objectifying positivist quantification of questionnaires”, giving voice to common people in a gentle, mutual and caring way (Kvale 2006:481). However, such a conception of interview research is as false as notions of dialogue in contemporary management, politics or education, because it is blind to inbuilt power asymmetries. This he asserts, is a false view:
In contrast to the mutuality of the dialogue, in an interview, one part seeks understanding and the other part serves as a means for the interviewer’s knowledge interest. The term interview dialogue is therefore a misnomer. It gives an illusion of mutual interests in a conversation, which in actuality takes place for the purpose of just the one part—the interviewer. (Kvale 2006:483)
As an alternative to the conception of interview-as-dialogue, Kvale proposes a range of interview practices that acknowledge this fundamental power asymmetry.
These alternative conceptions all share an agonistic component, which means that the interviewer should actively follow up on answers and provide some form of resistance to the interviewee. The level of conflict ranges from what Kvale (2006:486) calls the “Platonic dialogue”, mentioned above. This may be suitable for expert interviewing, “where the interviewer confronts and contributes with his or her conceptions of the interview theme”. The interview then becomes a conversation that stimulates both parties to formulate and sharpen ideas that were perhaps not formulated previously. A more agonistic interview style would be what Kvale (2006:487) calls the “actively confronting interview”, where the goal is not consensus, but where the interviewer confronts the informant with critical questioning if, for example, the informant contradicts herself. The goal is not to impose the interviewer’s ideas on the informant, but to uncover and make explicit the informants’ hidden assumptions. This is the Socratic interview style employed by Bourdieu in The Weight of the World, held forth as a prime example of interview craft by Kvale (Bourdieu 1996; Kvale 2007). The practical implication of this is the imperative to actively and critically follow up on the interviewee’s answers, to probe for assumptions and to take a maieutic approach to conversationally formulating conceptions that were perhaps not already consciously formulated by the informants.
Third, high quality “spontaneous” accounts of informants’ life-worlds are produced not through the passivity of the interviewer, but in a carefully constructed
interview situation with an active “Socratic” interviewer. This conception of interview epistemology is in turn based on a realist ontology:
Thus, against the illusion which consists in searching for neutrality through the elimination of the observer, it must he admitted that, paradoxically, the only
“spontaneous” process is one that is constructed, but it is a realist construction.
[. . .] It is only when it rests on prior knowledge of realities that research can bring the realities it wishes to record to the surface. (Bourdieu 1996:28)
The realities Bourdieu speaks of here comprise the inherently structured social world that invisibly shapes us as social beings. In the present study this is, for example, the contours of a scientific habitus shaped through socialisation into the economics discipline, pointing to structures beyond individuals as their carriers.
Against Gubrium and Holstein’s constructivist view, we see here a robust formulation of a non-naïve realist conception of the ontology and epistemology of interview research. One can compare this to a caricatured empiricist/ positivist versus a scientific realist understanding of experimentation in the natural sciences.
Whereas the empiricist would claim that experiments work through observing event regularities in a controlled setting (without researcher bias/ interference) to generalise about causality, the realist would point to the necessarily highly constructed nature of experiment. To perform even the simplest experiments in classical physics, one needs to carefully construct the experimental situation. We cannot usefully study gravity in spontaneous events in nature (since very few things are preoccupied with just constantly falling before us), so we need to carefully construct inclined planes and perfectly round balls to roll down them to observe how the real but unobservable law of gravity produces effects that can be empirically observed. Furthermore, what we want to understand are the properties of the general underlying law of gravity, not what actually happened to a particular pile of balls.52
To sum up this “Socratic sociological realism”, the purpose of sociological interview research is to uncover real pre-existing systems of belief and social structures beyond individuals’ particular conceptions. However, interviews require interviewers who actively listen and ask critically “Socratic” questions, and who can help informants “deliver their truth” through sensitive follow-up questions. However, one must simultaneously be aware that this is not the same things as posing “leading questions”, and recognise the obstacles that social distances may create. Interviewers should strive to reduce social distance through
52 This account of experimentation and the contrast of empiricism to realism draws on Bhaskar’s work (Bhaskar 1975a; Collier 1994).
being consciously reflexive about the social interaction, through all aspects of self-presentation in the interview situation.
On the ground: Guidelines for interviewing
Kvale (2007:80) lists six qualities of a good interview, that I have taken as useful ideals. A high-quality interview should: i) provide spontaneous answers that are qualitatively rich and relevant; ii) have short questions that result in long answers;
iii) have a high level of follow-up questions and clarification of informants’
answers where needed; iv) be characterised by “on-the-fly interpretation”, where the researcher immediately interprets what is said; which ideally leads to v) a high level of verification of interpretations, as informants agree (or by disagreement point to other interpretations) on and so verify the interviewer’s interpretations;
and, finally, vi) to a large extent be “self-reported”, that is, form a story that could be quoted and presented as-is without the need for further interpretation.
Good interviews require skilled craftsperson-interviewers. According to Kvale (2007:81) the qualities of such an interviewer include being: i) knowledgeable about the topic; ii) structuring in relation to the interview situation; iii) produce questions with clarity; and iv) be gentle, letting informants go on, pause, etc.
Furthermore, the interviewer should be v) sensitive and attentively listening to what is being said; vi) open to unexpected turns with a “hovering attention”; but vii) steering when interviewees slide off topic. Finally, a good interviewer is also viii) critical and does not automatically accept information at face value; ix) remembering, to avoid repetition and connect to what has been covered earlier;
and x) interpreting, actively trying to clarify the meaning of what is being said for informants to confirm or disconfirm on the spot. These are all notions that should be treated as ideals to strive for and develop as part of the craftsmanship of interviewing. I do not claim to have been an outstanding follower of these steps, but have consciously strived to continuously improve my interviewing practice following these guidelines.
Selection of informants
The informants selected for interviewing were all active Swedish researchers (including doctoral students) in economics or in two cases, had been researching, written and engaged in some form of heterodox economics, or had an economics educational background. “Swedish” means active at a Swedish institution and has nothing to do with ethnicity, although all interviews were conducted in Swedish.
The selection of informants did not follow the principles of random sampling
since the qualitative interview data are not going to be used for a statistically representative analysis. However, I do not think it is warranted to drop the issue of representativeness altogether just because one works with qualitative methods.
Therefore, care has been taken to achieve a sample that is not obviously biased in important dimensions.
A wish list of informants was constructed based on a number of criteria. Since the aim of this study is to investigate mainstream economists, as well as the nature of Swedish heterodoxy and the relationship between these two supposed groupings, this led to a list of criteria for selecting respondents. First, the majority of informants were selected as representatives of a broad mainstream. For this group of informants, I used seven criteria for selection. A wish list was constructed so that all criteria were satisfied by at minimum of one informant. Only the first criterion, i.e. belonging to one of the top five universities, was used as a necessary condition for inclusion in the mainstream group. The selection criteria are presented in table 1.
The interviewed informants in this group: i) all belonged to one of the top six economics departments.53 In a hierarchical and strongly top-down discipline, this ensures that informants are recruited from economics departments held in high esteem by economists themselves (see Fourcade et al. 2015). Some were ii) selected from among the top-ranked researchers according to bibliometric measures,54 also a measure to include some of the most influential persons as measured by scientific output. Some were, furthermore, iii) authors of widely-used textbooks, which is another very important channel for exercising influence on the reproduction of the discipline. Others held iv) influential positions in doctoral programmes (director of studies or similar), and were thus authorities on the structure and content of these programmes, arguably a very important determinant of disciplinary reproduction (see Colander 2005). Even if doctoral programmes are never the domain of one or even a few persons, but rather the collective responsibility of senior faculty, such persons are well-informed about the doctoral programmes and any discussions about them.
53 These are the economics departments at the universities of Lund, Uppsala, Gothenburg and Stockholm, and Stockholm School of Economics. However, Stockholm University hosts both a department of economics and a research institution, the Institute of International Economic Studies (IIES). See chapter 5 for metrics. While administratively separate units, they are located just one floor apart and there is naturally a not insignificant degree of interaction between the two departments. However, I made sure to include informants from both departments.
54 Rankings of authors from recent literature were used. However, to protect the anonymity of informants, these rankings are not referenced here.
Another selection criteria was v) a position as public intellectuals, being known in public debate. This is arguably also a way of exercising influence on the academic profession, as demonstrated by the example of Paul Krugman, who is widely read not only by the general public, but importantly also by his fellow economists. The next basis of selection was vi) membership in the Nobel Committee. The Bank of Sweden Prize in Economic Sciences is an internationally unique mechanism of creating disciplinary hierarchy in the social sciences. The winners are selected by a small committee of ten members picked from Swedish academia; most, but not all, are academic economists. Membership in the committee is both an acknowledgement of excellence and a uniquely powerful position when it comes to influencing the future of the economics discipline internationally. Therefore, I have also included informants that are or have been members of the Nobel committee. Finally, I included vii) representatives from both ends of the academic career trajectory, which means that informants were selected not only from among well-established professors, but the also doctoral students who will form the next generation of the profession. Some informants also represented career steps in between these two extremes, i.e. lecturer or similar.
Table 1. Selection criteria for informants
REPRESENTATIVES OF THE MAINSTREAM REPRESENTATIVES OF HETERODOXY
i) All from top-5 university departments
Selection based on expressed critique or heterodox self-identification, or identification by others as to some extent heterodox.
Selection not limited to top departments.
At least one informant matches each of the following criteria:
ii) Top-ranked researcher iii) Textbook author
iv) Doctoral programme director v) Public intellectual
vi) Current or former Nobel committee member vii) Different career stages (from doctoral students to
A second group of informants comprised those who could, in any sense of the term, be thought of as “heterodox”. Almost all informants in this group held an economics doctorate; two exceptions were included because of their role in heterodox circles and close relation to the discipline. However, only a few in this group belonged to any of the top five university departments. Some held, or were on their way towards, positions at smaller universities or even in other disciplines.
These informants were selected based on my previous knowledge of them and their work, often explicitly presented as critical of mainstream economics in some
sense. In a few cases, informants whom I thought of as borderline cases or perhaps
“mainstream heterodox” were also selected. Some were added as a result of snowballing in the earlier interviews. Of all informants, about two thirds, or 14 persons, were selected in the first group, and a third, or six informants were selected in the second, heterodox group.
Informants on the wish list were contacted with an invitation to participate in the interview study, with brief information about the study and confidentiality. The positive response rate to my invitation was high, with only five out of 25 (20 per cent) of requests turned down. Of the twenty informants in my sample, seven were doctoral students or similar, nine were full professors (including senior and emeritus professors), and four were researchers at intermediate positions as lecturer or similar.
Apart from the twenty researchers in economics, I also did a few brief background interviews with economics students active in two student organisations that had in some way promoted pluralism. However, these interviews have only been used for background understanding of these movements, and are not used in the main analysis. The semi-structured interviews took place between May 2015 and February 2016, in the majority of cases at the respondent’s home department, in a few cases at a café or similar, the interviewee’s home, and in one case over a Skype videolink. The twenty researcher interviews lasted between 45 minutes and two hours, with most ninety minutes or longer.
The interviews were recorded for later transcription and analysis, with a total of more than thirty hours of recorded material.
The format of the interviews can be seen as a hybrid between life-world interviews, where the purpose is to get a grasp of the informant’s point of view, and expert interviews, where the interviewees report facts that the interviewer inquires about. I used a flexible interview guide with a range of themes and questions to potentially use, depending on the turns of the interview conversation.
The interview guide functioned as a resource to be used flexibly, following up as topics evolved, breaking as necessary the order of questions in the interview guide.
The guide was continually updated with refined questions as I learned during the process which questions were productive and which were not. Before each interview I also did some background research and read or skimmed some of each author’s work or other relevant sources as preparation. The interview guide was also adjusted to include questions specific to the particular interviewee, sometimes relating to their position, career stage, or specific issues in their work.