QBRDM #6

Qualitative Research 2015, Vol. 15(6) 722 –737

© The Author(s) 2015 Reprints and permissions:

sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/1468794114567496

qrj.sagepub.com

Q RThe voices heard and the

voices silenced: recruitment processes in qualitative interview studies

Guro Korsnes Kristensen Norwegian University of Science and Technology, Norway

Malin Noem Ravn Norwegian University of Science and Technology, Norway

Abstract This article addresses methodological questions concerning recruitment processes in research using qualitative interviews. The authors suggest that, as an active part of the research process, recruitment influences research results in sometimes unforeseen manners. They argue that recruitment processes should be better attended to – not least in research positioned within the epistemological landscapes of knowledge production and transparent reflexivity. The article draws on six studies in which qualitative interviews of ‘lay people’ were used as the sole or main source of data. Drawing on their own experiences, the authors discuss the ways in which research topics, pre- defined sample, mediators, and the researchers’ positionality and situatedness affect the recruitment of different interviewees, and, hence, also the knowledge researchers are able to produce.

Keywords gender, interview studies, knowledge production, lay people, mediators, recruitment methods, sample

Introduction

Our incentive for writing this article was our contention that, relative to the effect of informant recruitment on knowledge production in qualitative interview studies, theo- retical and methodological considerations and the practical aspects related to recruitment

Corresponding author: Guro Korsnes Kristensen, Norwegian University of Science and Technology, Edvard Bulls veg 1, Trondheim, 7491, Norway. Email: [email protected]

567496QRJ0010.1177/1468794114567496Qualitative ResearchKristensen and Ravn research-article2015

Article

Kristensen and Ravn 723

processes appear to be under-communicated, both in methods textbooks and in research. In this article, we identify and discuss certain aspects of the recruitment process that we think are of crucial importance for knowledge production in research.

In research texts, recruitment methods are often omitted or hidden from methodologi- cal descriptions. This means that the texts convey little information about researchers’ sometimes shifting reasoning in relation to inclusion and exclusion criteria, methods for finding and motivating potential informants, and positive and negative experiences of applying different recruitment strategies.

Similarly, in qualitative methods textbooks, recruitment processes are rarely consid- ered in any detail. For example, in the seminal and much-cited textbook on qualitative methods, An Introduction to Qualitative Research by Uwe Flick, recruitment is only mentioned very briefly (2009: 270). Likewise, David Silverman’s Qualitative Research: Theory, Method and Practice has only one reference to recruitment (2004: 332). Another example is found in Steinar Kvale’s widely used book on interviewing as a research strategy. In this book, Kvale identifies seven stages of qualitative interview investiga- tions: thematizing, designing, interviewing, transcribing, analyzing, verifying and report- ing (Kvale, 1996: 81). However, Kvale’s list does not include recruiting; rather, it jumps directly from ‘planning’ to ‘interviewing’.

In articles that are especially concerned with method, however, recruitment is some- times treated as a specific topic. In most cases, these articles describe research experi- ences in which identifying and motivating informants has proven difficult (McLean and Campbell, 2003; Thomas et al., 2007). For example, researchers working in the fields of migration and integration have provided methodological insights into the challenges that might occur when recruiting informants from ethnic minority groups (Rugkåsa and Canvin, 2011; Wigfall et al., 2013). Others have described challenges related to recruit- ing people who are, or claim to be, very busy (Broyles et al., 2011) or who are involved in stigmatized or illegal activities (Roth, 2012). Still other texts describe research experi- ences in which recruitment has proven difficult because potential informants have mis- trusted the researchers and/or not wanted to take part in the research because of historical abuse (Anthony et al., 2010) or because they considered the research topic too sensitive (Butera, 2006) or private (McCormack et al., 2013; Thomas et al. 2007).

One possible reason for the general lack of focus on recruitment when the identifica- tion and motivation of informants runs smoothly is that recruitment is not an integral part of the established methodological narrative in qualitative interview studies. Another pos- sible reason is that recruitment is considered part of the technical background to a pro- ject, and not of scientific interest in and of itself (McLean and Campbell, 2003: 57). It might also be that most researchers prioritize the publication of research findings over the publication of research methodology (Sanghera and Thapar-Björkert, 2008). We see this as both unfortunate and surprising. Unfortunate, because of the scientific ideal of free and open research that readers can evaluate as valid and reliable; and surprising, when one considers the focus on reflexivity, positionality and situatedness (which Gillian Rose (1997) labeled ‘transparent reflexivity’) that has become increasingly important in qualitative research over the past decades. Also, we claim that this lack of focus on recruitment hinders important methodological discussions that could enhance the strate- gies and lines of action in qualitative research, in general, and qualitative interviewing, in particular.

724 Qualitative Research 15(6)

In the next section, we outline what might be included in the phrase ‘recruitment strategies’. We then present our material – that is, the research projects and processes that we use as examples for our arguments. The subsequent discussion is organized around four aspects that we find particularly important for recruitment and knowledge-produc- tion: research topics, pre-defined samples, mediators and the researchers’ positionality and situatedness. In the last section of the article, we address the paradox that this part of the research process is not more reflected upon in the postmodern epistemological land- scape of knowledge production (versus data collection) and transparent reflexivity.

Recruitment strategies in qualitative interview studies

Qualitative methods often rely on interviews with relatively few individuals with special characteristics (Patton, 2002). These characteristics depend on researchers’ pre-defined selection criteria, be they traditional categories such as gender, age, nationality or sexual- ity, or more specific traits and experiences related to people’s personal or professional lives (Morse, 2007). Generally, selection criteria are made according to what researchers believe will best accommodate their study – in the sense that the selected interview sub- jects are assumed to be those who are most suited for shedding light on the research questions. Once selection criteria are defined, researchers must identify relevant research sites and approach potential informants. Whereas some recruitment sites are accessible to all researchers without special permission, others require informal or formal allow- ances. In some circumstances, such an allowance is easily accessible, whereas in other cases it is necessary for researchers to go through quite a complex system of bureaucracy before they can meet probable respondents.

In cases where the sample criteria do not define an already existing group – that is, when the sample consists of individuals who have a characteristic, position or interest in common but who are not part of a specific organization or institution – researchers are more free to approach informants directly, without prior approval. Although this might give researchers more flexibility, it is not necessarily easier for them to recruit inform- ants in this way. Potential informants must still be identified, approached and motivated to participate, often by the researcher him/herself.

There are many ways to contact these ‘publicly accessible’ potential informants. One method is to place written advertisements on the internet or in newspapers, or to print brochures and distribute them in public spaces. Another method is to approach people directly on the street, in cafes, universities and neighborhoods, or on websites and in chat rooms. To be efficient, the direct approach implies that pre-defined recruitment criteria be either identified from physical appearance or strongly related to an activity, appear- ance or site.

Another widespread recruitment strategy involves snowball sampling. As noted by Goodman (2011), the term snowball sampling refers to two very different procedures. In research on accessible populations, snowball sampling is used as a strategy to estimate mutual relations in a given population (Goodman, 2011: 349). In research on less acces- sible populations, however, snowball sampling starts with a set of people (‘seeds’) from the hard-to-reach population who eventually help the researcher contact more respond- ents. When we write about snowball sampling in this text, we refer to the second use of the term – even though the populations in our studies were not always hard to reach.

Kristensen and Ravn 725

Yet another well-known recruitment strategy in qualitative interview studies involves a mediator. A mediator, as we use the term, is a person who uses his/her formal or infor- mal position and relationships to facilitate contact between a researcher and potential informants. In many contexts, the term ‘gatekeeper’ is used to describe this role (Wanat, 2008). As a gatekeeper is seldom considered part of the population under study, but rather someone who grants or denies access to this population, we find the term ‘media- tor’ more illustrative of the ways in which informants are recruited. What gatekeepers and mediators have in common, though, is that they are both very important in a study, in the sense that they can make or break it – or at least make it more or less challenging and successful for the researchers.

Experiences from research projects as ‘material’

In our activities as social anthropologists and gender researchers, we have conducted qualitative interviews in several research projects. We have experienced that this method is highly valuable, in the sense that it provides researchers with in-depth and nuanced understandings of cultural domains. However, we have also found that the entire process – from creating the selection criteria to conducting the actual interviews – is both time- consuming and personally and professionally challenging.

The recruitment phase of a project can be hard to plan in detail, since recruitment is dependent on the responses of others and, as such, is partly unpredictable. Furthermore, the recruitment process is emotional work that should not be underestimated. Researchers must be persistent, must follow-up on calls that are not returned, must send reminders and must repeatedly ‘sell’ their projects to persons they do not know. No matter how professional researchers are, they may suffer personal costs from being repeatedly turned down, and embarrassment and faintheartedness can easily become their daily partners in a slow recruitment process. Besides, the researchers know that, without interviewees, they have no material; the consequences are dire.

The ways in which we as researchers recruit our informants influence who we end up interviewing and, hence, the knowledge we are able to produce. No matter how one looks at it, it is an undisputed fact that the persons whose lives, experiences and mean- ing-making processes researchers are able to study in interview-based projects are those who respond positively to requests for interviews; the rest remain unknown.

By drawing on our experiences of studies in which interviewee recruitment was an important part of the research process, we aim to add to the insights of the recruitment literature and argue that more researchers should include their reflections upon and expe- riences with recruitment in their research texts. Our examples in this article are drawn from six studies in which we used qualitative interviews as the primary method of data production. In these studies, we were responsible – either alone or as part of a research group – for the design and conduct of the interviews.

Research projects

‘One body: Two lives. Pregnancy, the fetus and the pregnant body – an anthropological analysis’ (Ravn), was a longitudinal study of pregnant women’s understandings of

726 Qualitative Research 15(6)

pregnancy, their experiences with their pregnant body and their perceptions of their fetus. Eight women were interviewed – each up to six times – throughout their preg- nancy and the year after their baby was born. The interviewees were recruited through personal networks, mediators and pregnancy websites.

‘Family planning – behind the numbers. Narratives on family planning among immi- grants in contemporary Norway’ (Kristensen) was a qualitative research project in which a total of 21 women and men from Iran and Iraq who were living in Norway were each interviewed once or twice. The interviews centered on their understandings of the ideal family and on fertility decisions, including perceptions of and experiences with contra- ceptives and induced abortion. The pre-defined recruitment criteria in this project were nationality (Iran- or Iraq-born) and age (18–45). The interviewees were recruited through different organizations working with immigrants and through contacts from a former research project, then later through snowball sampling.

‘The social meaning of children’ (Ravn), was a research project aimed at understand- ing some of the mechanisms underlying Norwegian fertility patterns through a multi- level analysis of individual reproductive ‘choice’.1 Five researchers worked on the project, and several more were involved in recruitment, interviewing and transcription. The project group built up a common base of qualitative data, consisting of 90 interviews with young men and women, with and without children. The interviewees were recruited through mediators, personal networks and workplaces.

‘Reproductive relations: Gendered meanings in the field of reproduction’ (Ravn) was a study of cultural conceptions of motherhood and fatherhood within the context of reproduction and gendered relations. Ravn’s sub-project focused on heterosexual fathers- to-be. Ten people were interviewed, and these participants were recruited through web- sites, pamphlets distributed in healthcare centers, personal and professional networks and snowball sampling.

‘Someone to help out’ (Kristensen) is an ongoing qualitative research project in which a total of 35 persons have been interviewed about unpaid and paid domestic work and family–work balance. The interviewees are all parents of relatively young children, and all but five single mothers are married/co-habitants and in dual-earner relationships. The interviewees were recruited through the researcher’s personal network, then conse- quently through snowball sampling.

‘Same situation – different solution’ (Ravn and Kristensen) is the sixth study we draw on here. In this study, the authors worked together to explore how middle-aged couples who employed a domestic cleaner perceived this arrangement and what meanings they attached to it. The data material included 10 interviews. The interviewees were recruited through acquaintances and snowball sampling.

What all of these studies have in common is that the interviewees were ‘lay people’ – that is, people who were chosen because of their association with a specific categori- zation and/or life situation. Besides, all projects were conducted in Norway, a context with which the researchers, themselves, were familiar. Further, all of the projects were organized as semi-structured and unstructured interviews (Kvale, 1996), performed as active interviews (Holstein and Gubrium, 1995) and analyzed with an explicit focus on social interaction, co-construction and narratives (Järvinen, 2005; Staunæs and Søndergaard, 2005).

Kristensen and Ravn 727

The present text was generated from discussions between the authors over several years. Through informal debriefing activities to discuss our research and research pro- cesses, we realized that our discussions repeatedly returned to the challenges of recruit- ment. We decided to explore this aspect of our research processes together, and more systematically. Our analytical procedure for the present text was to review field notes and method logs and to retrospectively systematize our experiences, focusing on com- monalities and differences.

Through this joint work, we have identified four aspects of recruitment that we find particularly influential with respect to knowledge production: research topics, pre- defined samples, the use of mediators, and the researchers’ positionality and situated- ness. We will present and discuss each of these aspects, in turn.

The meanings and effects of research topics

Our first contention is that the research topic of an interview study has a significant impact on recruitment and, hence, the lives, experiences and meaning-making processes that are most likely to be included in the data material – and those that are not. A telling example of how a research topic can affect the recruitment of informants is found in the study of expectant mothers’ experiences with pregnancy. Although this study employed the highest number of interviews with each informant and involved both personal and intimate questions, it was by far the easiest project to recruit informants for. With very few exceptions, the approached potential informants not only responded positively to the request for an interview, but also expressed gratitude and a strong interest in taking part in the study. The participants’ positivity and willingness was maintained throughout the year and a half of fieldwork.

Our interpretation of this successful recruitment experience is that the interviews filled a purpose not only for the interviewer – who wanted to know more about women’s perceptions of and experiences with pregnancy – but also for the interviewees. By this we mean that the interviews functioned as a way for the women to process their preg- nancy and reflect upon different aspects of it. Some of the interviewees explicitly said that the research project and their participation in it functioned as a way of letting them ‘be’ pregnant in a proper way and that they looked forward to the interviews. In other words, pregnancy was a research topic that the pre-defined sample found interesting, relevant and meaningful to talk about and to share their reflections of. Thus, the recruit- ment process and the interviewees’ eagerness to participate in the project both supported and helped shape the thesis’s argument that pregnancy is perceived as an important expe- rience in a woman’s life and that it has become ‘something in itself’, which must be performed in certain ways (Ravn, 2004).

The research topic can also help explain why immigrants, who are often reported to be difficult for researchers to recruit (Rugkåsa and Canvin, 2011; Wigfall et al., 2013), were rather easy to motivate into participation in the study on family planning. Although the researcher had only a few personal contacts within so-called migrant communities and no research sites for approaching potential informants when the project started, all interviewees were recruited in just a few months’ time. Furthermore, very few of those who were asked for interviews turned the request down. As we will return to later in this

728 Qualitative Research 15(6)

article, this rather unproblematic recruitment process could have several explanations. However, we suggest that one plausible explanation is that, unlike many other topics, family planning (which, in this project, was defined as ideas and practices related to hav- ing and not having children) is something that immigrants are just as entitled and com- petent to talk about as the general population; furthermore, the topic is not necessarily related to migration and being a migrant. Thus, participating in an interview study on family planning might have allowed the informants to feel ‘normal’. This interpretation is supported by the easiness, and sometimes even pleasure, that was put forward in sev- eral of the interviews. A related interpretation is that some of the informants might have considered the interviews an opportunity to tell their own story and, in so doing, to chal- lenge a well-established myth that immigrants are ‘breeding like rabbits’ (Luibhéid, 2004). In support of this explanation, quite a few of the informants in the interview made a point of expressing their desire to not have many children (Kristensen, 2011).

In contrast to the projects on pregnancy and family planning, the study on paid domes- tic labor proved to be more challenging to recruit for than had been expected – despite the fact that the selection criteria were so wide that there should have been many poten- tial informants from which to choose and the fact that the researchers had personal con- tacts who had volunteered as mediators. Generally, it was difficult for us to identify potential informants, as many of our open requests were met with silence. Further, when we did identify and approach persons who met the pre-defined selection criteria, the decline rate was surprisingly high. There are several ways to understand this; however, as we see it, the research topic might be the most important of these reasons. This claim is mainly based on two arguments. First, quite a few of those who were asked to partici- pate in the study gave the impression that they did not have strong feelings about the topic, and hence they did not want to spend precious time discussing it. Second, the practice of paying a person to perform domestic services in the private home seemed to be surrounded by social shame. This social shame had at least two consequences: first, it made people unaware of those in their social milieu who had hired a domestic cleaner; second, those who had hired domestic cleaners were uncomfortable with the idea of being interviewed about it. For example, some of the text messages and emails sent to people who we had been told were employing a home cleaner were answered in surpris- ingly resistant tones, which informed us that the senders’ participation in the study was out of the question. Others replied in a more accommodating manner, though still giving the clear impression that they were not very eager to participate without any further explanation. One person was even affronted by being asked, saying ‘of course I do not have a home cleaner!’. As we see it, these mechanisms and experiences contributed to the study by emphasizing that, although the use of domestic cleaners in Norway had recently increased, the cultural evaluation of outsourced domestic work had changed more slowly, and there was still a cultural-specific taboo related to the idea of having servants, not cleaning up one’s own dirt and not fending for one’s self (Kristensen, 2015).

Another experience from the studies of paid domestic labor was that women were easier to recruit than were men. Many of the men we asked did not know of anyone who was employing a home cleaner, and we were often told that ‘this is not something we talk about with our friends and acquaintances’. As some of the women interviewed claimed to talk quite a lot about having and not having a home cleaner, we suggest that our difficulty

Kristensen and Ravn 729

recruiting men could have also been related to the research topic, as women (at least tra- ditionally) are more engaged in domestic work than are men and, as such, are more likely to be able to, interested in and willing to talk about domestic work in an interview. In spite of the trend towards greater gender equality in Norway, the recruitment experience in this project clearly illustrates that housework is still thought of as mainly women’s work. This gender-related argument brings us to the last example we use in this part: the study of men’s experiences as expectant fathers.

In contrast to the study of women’s experiences with pregnancy, the somewhat paral- lel study conducted some years later was much more difficult to recruit for. Despite the active use of different recruitment strategies directed strategically towards men who were expecting a baby, the first 20 requests were turned down, either silently (with no response at all) or with more or less convincing excuses. Fortunately, some of the men who were approached in the second round responded positively, but, all in all, the recruit- ment process in this study was both challenging and time-consuming. One way to under- stand this – which aligns with the argument posed above in relation to paid domestic work – is that expecting men, in contrast to expecting women, do not find the research topic interesting enough to be worthy of an interview.

Furthermore, interview requests were sometimes met with surprise, or even disbelief, as depicted in responses such as: ‘Are you sure that it is me you want to talk with and not my wife?’. This indicates that some men do not consider their perspectives and experi- ences interesting enough to be worthy of an interview. It is also possible that men, being less used to talking about matters of fertility to persons other than their partner (Jensen, 2013), feel more uncomfortable speaking about such topics with a stranger. In other words, it could be that quite a few of the men we approached felt that they lacked com- prehension of narratives that would make sense in an interview study of pregnancy; this underscores the traditional cultural contention that pregnancy is a topic on which women rightfully play the ‘main lead’ and men feel ‘one step removed’ (Draper, 2000). In terms of knowledge production, the consequence of the described difficulties was that the men who were finally included in the study either showed above-average interest in gender equality or were explicitly determined to show the ‘particular male’ side of the matter. In other words, as so many expecting men turned down the request for an interview, it is possible that perceptions and experiences with pregnancy that were relevant to many Norwegian men were not included in the study.

The effects of the described experiences with different research topics are that some topics are easier to explore than are others, and the easiest topics to explore are those that people see as both possible to talk about and important, to their selves and to society. In contrast, the most challenging topics are those that people cannot see the importance of or do not feel comfortable talking about. This means that we have more knowledge of what informants consider interesting and important and less knowledge of what potential informants consider ‘boring and difficult questions’, even though these questions may still have social and scientific interest. Furthermore, the above description underscores that people who – by convention or cultural praxis – perceive themselves as relevant contributors to illuminating certain topics are easier to recruit than are those who per- ceive themselves as irrelevant. This leads, again, to a strong circular motion, wherein research in a particular field easily includes the most known voices but is less likely to include less known voices that could expand our scientific understanding of the field.

730 Qualitative Research 15(6)

The meanings and effects of the pre-defined sample

A qualitative sample is strategic and often cannot be selected in advance of fieldwork (Grimen and Ingstad, 2007: 285). This means that the recruitment process can be described as goal-oriented, but simultaneously explorative, work. The process of select- ing whom to approach is not a scientifically neutral process, but, rather, one that is imbued with subjective decisions. The sample must make sense, but the best material does not necessarily come from the most stringent or clearly defined sample. Similar to research topics, the selection criteria behind a pre-defined sample seem to greatly impact the recruitment process. We have experienced that some samples are more challenging to recruit for than are others. As the pre-defined selection criteria in our projects implied that potential informants would be possible to reach without formal allowances, our recruitment challenges mostly related to identifying and motivating potential informants to participate in different kinds of interviews.

This can be illustrated by the study ‘The social meaning of children’, for which we identified eight categories of informants according to gender, family status and employ- ment status, and aimed at achieving approximately equal distribution of 90 interviewees in the eight categories (11–12 in each group). Furthermore, age-based criteria were added – the interviewees would ideally be between 25 and 35 years old and, if they had chil- dren, the children would ideally be no older than 3 years.

In this project, recruitment proved to be more challenging than we had anticipated. The challenges were gender- and class-related, and the groups we struggled to recruit for were working class men, in particular, as well as working class women. The ‘easiest’ category to recruit for was upper middle class women, both with and without children. When conducting the interviews, we also found that marital status, which was not one of the pre-defined criteria, also influenced the material. The final material showed that we had little success recruiting men in relationships; that is, most of the working class men we interviewed were single. This again meant that we had little access to some of the men’s thoughts on when they should have children, as it seemed unachievable for them to have a child without first having a partner (Ravn and Lie, 2013).

With so many pre-defined criteria to fulfill, it was difficult for us to achieve a ‘pure’ sample. For example, although we wanted informants between the ages of 25 and 35, we found that participants who were younger than 25 with children or older than 35 without children were also relevant. Likewise, class was a slippery concept. We had determined class from occupation, with certain occupations identified as working class and other occupations identified as upper middle class. However, because of the high level of edu- cation in Norway, and as university education is free, we encountered situations in which those we had recruited on the basis that they were working class turned out to have had a year or two of university study.

A potential effect of selection criteria in the recruitment phase of interview studies is more research on some groups than on others, and more research on women and middle class people/women and less on working class men – at least in relation to the kind of research questions we used. Other areas of research might have other groups that are over- and under-represented. Another potential effect is that the samples – often for pragmatic reasons – are stretched to fit those who are willing to participate. A consequence is that not all questions are explored and not all relevant groups are given voice or representation.

Kristensen and Ravn 731

This leads to research that reproduces cultural ideas about which persons are relevant for certain topics, and hence reifies, rather than explores, these cultural fields.

The meanings and effects of mediators

In our projects, we used a wide range of strategies to approach potential informants, including indirect approaches (such as leaflets, Facebook and open requests on email lists and websites) and more directly targeted approaches using SMS, phone calls and face-to-face communication. Our experience has been that indirect approaches can func- tion as sources of background information for people who are contacted through other means, but rarely work by themselves. That is, in our projects, only one potential inform- ant got in touch with us as a result of having found a leaflet. Likewise, only four potential informants responded to open requests on email lists and websites. Those who did con- tact us seemed to have either a particular interest in the research topic or much spare time on their hands.

Generally, we have experienced that more potential interviewees respond positively to requests when contacted through well-functioning mediators. In the projects on paid domestic labor and expectant fathers, all informants were recruited through personal contacts, often with the assistance of mediators. But what makes some mediators ‘well- functioning’ and efficient for use on a particular project? Further, what may go – if not wrong – then at least unexpectedly?

In fertility-related research we have found that gender is an influential factor. In some aspects, women are more likely to work well as mediators, both because they are cultur- ally more closely associated with fertility than are men, but also because of the social pattern that women talk more among themselves about fertility-related questions (Jensen, 2013) and hence know more about each other in this respect. Knowledge about others is a crucial factor in the successful work of a mediator. Then again, for recruiting men to fertility research, we have found that use of a female mediator is not always the most effective way to reach the most interesting interviewees. In the expectant father project, we started by using women as mediators, but this did not work out very well as the decline rate was very high. The mediators’ networks were primarily linked by females; that is, participation requests were often placed by pregnant women to their partners. We suspect that part of the reason for the high decline rate was that the female mediators (and female researchers) reinforced the understanding of pregnancy as a female issue.

Use of male mediators for recruiting expectant fathers functioned much better, and all of the men who were approached by male mediators accepted the request to join the project. However, the problem we found was that quite a few men were reluctant to take on the responsibility of mediating contact. Most often, they explained this by claiming that they did not know any men who were expecting children at the time; we suspect this is related to the abovementioned social trait that men talk less among themselves about fertility, and hence know less about each other.

In this study, then, interviewees were secured with a few good and willing male medi- ators. However, the male mediators also influenced the material in several ways. First, all of the men who agreed to be mediators were middle class, and they recruited only middle class men. Second, the male mediators clearly added some selection criteria of their own;

732 Qualitative Research 15(6)

they searched their social circles for expecting men whom they thought would be verbal and comfortable discussing matters of fertility, while leaving out those they thought of as too shy. A typical explanation for few positive responses would be ‘most of the expecting fathers I know are not the kind of people you are looking for, as they are too timid, but I know one man who is rather talkative’.

In our projects on paid domestic labor, women functioned better as mediators than did men. As we wanted to interview both women and men, preferably together as a couple, we tried to activate both female and male mediators. However, whereas some of the women gave us lists of names of willing informants, very few men were able to come up with even a single name. As the names given by the female mediators were exclusively female (and these females eventually had to motivate their husbands/partners to take part in an interview), it might be that the couples we eventually did interview had some char- acteristics in common. For example, it might be that these couples wanted to take part in the study because they were satisfied and proud of the way they were organizing their everyday life. It might also be that the relations in these couples were of such a kind that the women could actually persuade their husbands to participate – even though the hus- bands may not have felt very eager to do so.

In ‘The social meaning of children’ project, use of mediators was necessary, but dif- ficult. The project was run by several researchers and even more mediators, and it was a challenge to preserve the limits of the chosen sample. The general problem pertained partly to the difficult task of specificity in the recruitment phase: we asked for interview- ees of a certain sex and within a certain age range, we specified whether they should have children or not and what age the children should be, and we asked for people in specific occupational positions. We found it difficult to be even more specific in our requirements without making the task too complicated for those who were helping us recruit and who were mediating our requests. The mediators, who had no part in planning the study, sometimes found it difficult to remember all of the specifics and, sometimes, the people who were recruited were older/younger, had older children or had occupations that dif- fered from our ideals. Some potential mediators were vaguely worried that the people they knew did not fit all of the criteria they were given.

Mediators directly influence the material and, hence, the knowledge produced through a project. Sometimes a researcher ends up interviewing people who are dissimilar to what was originally conceived of as the target group. This stretches the material, which, in some cases, dilutes the research; in other cases, it makes the research more viable and in touch with reality, due to the feasible sample.

The meanings and effects of the researchers’ positionality and situatedness

Working in the fields of anthropology and gender research, we are well-accustomed to the idea that the researcher’s positionality and situatedness inform the research findings (Clifford and Marcus, 1986; Harding, 1987; Haraway, 1988). However, when collecting and systematizing our recruitment experiences, we came to realize that these reflections are not only applicable to the knowledge-production that occurs in interviews, but also for the preceding recruitment of relevant and motivated interviewees. Thus, we will

Kristensen and Ravn 733

present some examples of how factors such as the researcher’s gender, social class, nationality, age and way of organizing household chores seem to impact the recruitment process and the resulting group of interviewees.

The first example is found in the project ‘The social meaning of children’. As already described, there were challenges connected to the researchers’ goal of covering a broad spectrum of the reproducing part of the population. Whereas the quota of women from the middle class was filled rather quickly, working class men turned out to be more difficult to recruit. As the research team consisted entirely of women from the upper middle class, we think it is highly relevant to reflect upon the possibility that the intersection of gender and class in this context had an impact on the (un)successful recruitment of different groups of informants. Another example is to be found in the study on expecting fathers, which has already been described as particularly challenging in relation to the recruitment of relevant and motivated interviewees. In this study, the researcher was female, and, moreover, a gender researcher and a mother, herself. Although none of the potential inter- viewees explicitly said that this had influenced their decision to participate in the research, it is likely that it had an effect. Indications of this can be extracted from some of the inter- viewees’ utterances, such as ‘You probably don’t agree with me, but I think gender equal- ity has gone too far’. This quote shows, among other things, a pre-conceived idea of the researcher’s own positioning within the field. Further, it indicates that this particular inter- viewee had joined the project precisely to argue against the pre-conceived positioning of the researcher. From comments such as these, and from the sample as a whole, we are under the impression that this particular project attracted interviewees who were either in agreement with pre-conceived ideas about the researcher (i.e. that she was pro-gender equality) or who explicitly challenged gender equality ideals.

Lastly, we will look at some experiences from the research projects on paid domestic labor. As already described, recruiting people who were engaged in this activity as employ- ers turned out to be both time-consuming and emotionally hard work for the researchers. As we see it, one plausible explanation is the already mentioned fact that, despite being a growing phenomenon, the practice is somewhat controversial in contemporary Norway. In addition, we suggest that the researchers’ lack of experience with employing home cleaners might have made recruitment more challenging than it would have been had they had such experience: first, because they would have had a better overview of potential informants and qualified mediators; second, and maybe more importantly, potential informants might have been more likely to accept the request for an interview, as they would have felt less uncomfortable talking about the culturally and socially questionable decision to pay some- one to perform housework in the private home. If this is the case (and we believe it to be), it is possible that the researchers ended up interviewing those who, for some reason, were either unaware of the cultural controversies or did not care about them, whereas those who were more ambivalent and ashamed were not included. Alternatively, and even more likely (judging from the narratives identified in the interviews), those accepting the request for an interview may have been absolutely aware of the controversies and eager to explain and justify their decision, and maybe also to counter the criticism and argue for a different understanding. Following from this, we could say that the difficulty of recruiting inform- ants for these projects can be explained both by the controversies and, at the same time,

734 Qualitative Research 15(6)

assisted by the controversies, as positive and negative replies to interview requests helped to reproduce these controversies and ambivalences in the research findings.

Recruitment in times of reflexivity, positionality and situated knowledges

An important and rather predominant trend in qualitative research over the past decades has been the departure from a more positivistic view, in which researchers ‘collect’, ‘dis- cover’ or ‘unearth’ data that exists independently of researchers’ presence and activities, to a perception of data as something that is produced through the dynamic and active interaction between researchers and research ‘objects’ (Gorden, 1987; Holstein and Gubrium, 1995; Järvinen, 2005; Staunæs and Søndergaard, 2005). This implies that meth- odological considerations have changed from fears of biases and source errors in qualita- tive data to an increasing consciousness of and reflexivity about aspects of the research process, including the position of the researcher and his/her situated knowledges.

As we see it, these postmodern perspectives on interviewing and interview data have important implications on the way in which we think about recruitment and the effects that different aspects of recruitment have on knowledge production. These implications should be included in descriptions of methods and in methodological reflections, as well as in analyses. For example, it is even more important for researchers to outline the rea- soning behind their definitions of selection criteria and their choices of time, place and strategy for recruitment. Likewise, it is critical for researchers to reflect openly on their own positioning in the cultural landscape and their relationship with potential interview- ees, and on the role of the mediator and the relationship between the mediator, the researcher and the informants. Other relevant considerations to reflect upon are: What kinds of information were the mediators given and what choices might they have made? What were their motivations for recruitment and how might these have affected the recruitment process and, hence, the data material?

In this article, we have demonstrated some of the potential complexities involved in recruitment processes in qualitative interview studies. By providing information from six different research projects, we have drawn attention to some of the complexities that arise in the recruitment process yet often remain hidden from the final presentation of research. Further, the empirical examples have demonstrated a need for a more nuanced understanding of recruitment and a more critical and reflexive perspective towards this part of the research process.

In our experience, recruitment often proves to be more challenging than anticipated and more consequential than generally acknowledged. Important insights can be drawn from the recruitment process itself, in terms of which groups of people feel that disclos- ing their experiences of a certain research topic is relevant. Similarly, resistance to par- ticipation can inform researchers about taboos or topics that are evaluated as irrelevant. This adds to the knowledge production and underscores or challenges other findings in the project. Recruitment to some research projects shows clear gender and class distinc- tions; in our particular cases, we had easy access to middle class women, while both men and those in the working class – particularly working class men – were left (or left them- selves) in silence. The result was a reinforcement of the middle class as the main voice

Kristensen and Ravn 735

in society and women as the main voices on fertility and domestic affairs. We believe that other research topics have different, but similarly structured, ‘main voices’.

Further, it is important for researchers to outline the reasoning behind the selection criteria for a sample and to make any stretches of the sample in different directions both visible and explicit. The relationship between the researcher and the mediator – and par- ticularly the instructions given to the mediator – must also be described. We suggest that researchers ask mediators to describe their subjective evaluation of the sample criteria and their reasoning for choosing potential interviewees. This could offer valuable insight into both the actual sample at hand and the mechanisms of recruitment.

The recruitment process, in itself, contributes to the study, in that the researcher grasps the opportunities that offer themselves and forms the final material in an interaction between the pre-defined criteria, the mediators, the development of the project and the approached interviewees who actually agree to be interviewed. Recruitment is an impor- tant part of the research process in qualitative interview studies, and it should be more explicitly included in the end products of research and more thoroughly discussed in the research community.

Acknowledgements

Siri Øyslebø Sørensen and two anonymous reviewers read the article in draft and provided invalu- able comments. They are excused for subsequent failings.

Funding

This research was funded by the Research Council of Norway, the Norwegian University of Science and Technology and the Nord-Trøndelag University College.

Note

1. The project was headed by Professors Anne-Lise Ellingsæter, An-Magritt Jensen and Merete Lie.

References

Anthony JS, Lee RC, Barry DG et al. (2010) Recruiting and keeping African American women in an ethnographic study of pregnancy: the community-based partnership model. Field Methods 22(2): 125–132.

Broyles LM, Rodrigue KL, Price PA et al. (2011) Overcoming barriers to the recruitment of nurses as participants in health care research. Qualitative Health Research 21(12): 1705–1718.

Butera KJ (2006) Manhunt: the challenge of enticing men to participate in a study on friendship. Qualitative Inquiry 12(6): 1262–1282.

Clifford J and Marcus GE (1986) Writing Culture: The Poetics and Politics of Ethnography. Oakland: University of California Press.

Draper J (2000) Fathers in the Making: Men, Bodies and Babies. PhD Thesis, Hull University, UK.

Flick U (2009) An Introduction to Qualitative Research. Los Angeles, CA: Sage. Goodman LA (2011) Comment: on respondent-driven sampling and snowball sampling in hard-

to-reach populations and snowball sampling not in hard-to-reach populations. Sociological Methodology 41(1): 347–353.

736 Qualitative Research 15(6)

Gorden RL (1987) Interviewing: Strategy, Techniques, and Tactics. Homewood, IL: Dorsey. Grimen H and Ingstad B (2007) Qualitative research. In: Laake P and Benestad HB (eds) Research

Methodology in the Medical and Biological Sciences. Amsterdam: Academic Press. Harding S (1987) The method question. Hypatia 2(3): 19–35. Haraway D (1988) Situated knowledges: the science question in feminism and the privilege of

partial perspective. Feminist Studies 3(14): 575–599. Holstein JA and Gubrium JF (1995) The Active Interview. Thousand Oaks, CA: Sage. Järvinen M (2005) Interview i en interaktionistisk begrepsramme. In: Järvinen M and Mik-Meyer

N (eds) Kvalitative Metoder i et interaktionistisk perspektiv. København: Hans Reitzels Forlag.

Jensen A-M (2013) Rising fertility, fewer fathers: crossroads of networks, gender and class. In: Ellingsæter A-L, Jensen A-M and Lie M (eds) The Social Meaning of Children and Fertility Change in Europe. London: Routledge.

Kristensen GK (2011) Familieplanlegging - bak tallene. Fortellinger om reproduksjon i det flerkul- turelle Norge. PhD Thesis, Norwegian University of Science and Technology, Norway.

Kristensen GK (2015) A fair deal? Paid domestic labour in social democratic Norway. In: Triandafyllidou A and Marchetti S (eds) Employers, Agencies and Immigration: Paying for Care. Farnham: Ashgate.

Kvale S (1996) Interviews: An Introduction to Qualitative Research Interviewing. Thousand Oaks, CA: Sage.

Luibhéid E (2004) Childbearing against the state? Asylum seeker women in the Irish republic. Women’s Studies International Forum 27: 335–349.

McCormack M, Adams A and Anderson E (2013) Taking to the streets: the benefits of spontaneous methodological innovation in participant recruitment. Qualitative Research 13(2): 228–241.

McLean ACA and Campbell CM (2003) Locating research informants in a multi-ethnic commu- nity: ethnic identities, social networks and recruitment methods. Ethnicity and Health 8(1): 41–61.

Morse JM (2007) Sampling in grounded theory. In: Nryant A and Charmaz K (eds) The SAGE Handbook of Grounded Theory. London: Sage.

Patton MQ (2002) Qualitative Research & Evaluation Methods (3rd ed.). Thousand Oaks, CA: Sage.

Ravn MN (2004). En kropp: To liv. Svangerskapet, fosteret og den gravide kroppen - en antro- pologisk analyse. PhD Thesis, Norwegian University of Science and Technology, Norway.

Ravn MN and Lie M (2013) The cultural ideal of the joint decision: illuminating values of indi- viduality and relationality of the child choice. In: Ellingsæter A-L, Jensen A-M and Lie M (eds) The Social Meaning of Children and Fertility Change in Europe. London: Routledge.

Rose G (1997) Situating knowledges: positionality, reflexivities and other tactics. Progress in Human Geography 21(3): 305–320.

Roth AM (2012) A methodological approach to improve the sexual health of vulnerable female populations: incentivized peer-recruitment and field-based STD testing. Journal of Health Care for the Poor and Underserved 23(1): 367–375.

Rugkåsa J and Canvin K (2011) Researching mental health in minority ethnic communities: reflec- tions on recruitment. Qualitative Health Research 21(1): 132–143.

Sanghera GS and Thapar-Björkert S (2008) Methodological dilemmas: gatekeepers and position- ality in Bradford. Ethnic and Racial Studies 31(3): 543–562.

Silverman D (2004) Qualitative Research: Theory, Method and Practice. London: Sage. Staunæs D and Søndergaard DM (2005) Interview i en tangotid. In: Järvinen M and Mik-Meyer N

(eds) Kvalitative Metoder i et interaktionistisk perspektiv. København: Hans Reitzels Forlag. Thomas M, Bloor M and Frankland J (2007) The process of sample recruitment: an ethnostatistical

perspective. Qualitative Research 7(4): 429–446.

Kristensen and Ravn 737

Wanat CL (2008) Getting past the gatekeepers: differences between access and cooperation in public school research. Field Methods 20(2): 191–208.

Wigfall VG, Brannen J, Mooney A et al. (2013) Finding the right man: recruiting fathers in inter- generational families across ethnic groups. Qualitative Research 13(5): 591–607. DOI: 10.1177/1468794112446109.

Author biographies

Guro Korsnes Kristensen is a Post-doctoral Fellow at the Norwegian University of Science and Technology’s Department of Interdisciplinary Studies of Culture. She holds a MA in social anthro- pology and a PhD in Interdisciplinary Studies of Culture. Kristensen specializes in gender research, with a particular focus on the intersections of gender, ethnicity/nationality, religion and social class. Her research is broadly focused on migration and family life.

Malin Noem Ravn is a social anthropologist and gender researcher at the Norwegian University of Science and Technology. Her main fields of research are reproductive issues, biotechnology, and the body.