Duration,Dominance and Depth in Telephone and Face-to-Face Interviews: A Comparative Exploration

Abstract

Traditionally, methodological textbooks have advised that the telephone mode is not well suited to the task of qualitative interviewing. At the same time, there are well-rehearsed arguments as to why telephone interviews may be a useful option in some circumstances. Despite this debate, there remains very limited systematic empirical exploration of differences in the process and outcomes of qualitative telephone vs. face-to-face interviews. Based on a recent ‘mode comparison’ study that sought to contribute to this gap in methodological knowledge, analysis of the overall duration, dominance and depth of talk between researcher and participant in a small set of telephone and face-to-face interviews revealed the following findings. (i) Despite much variation in individual interview length, telephone interviews were typically, and on average, shorter than those conducted face-to-face. (ii) The shorter duration of telephone interviews was a result of the participant speaking for less time, rather than a proportional reduction in talk from both parties. Additionally, in telephone interviews, participants generally held the floor for shorter stretches at a time. (iii) The researcher did slightly more talking during telephone interviews than in face-to-face interactions. Combined with the reduced amount of participant talk, this meant that the researcher tended to hold the floor for a greater proportion of the time in telephone interviews. (iv) To a moderate degree, the shorter length of telephone interviews could be accounted for by a reduction in coverage of themes. However, the principal explanation appeared to lie in a tendency for telephone interview participants to provide relatively less detail or elaboration. In this article, we consider why these differences may occur, if and how they might matter to the research, and how we might wish to modify interview practices in response.

Keywords

data quality interviews missing data mode comparison telephone interviews

Introduction

This article considers the use of the telephone to conduct interviews in qualitative research. This is an area of discussion that has received relatively little attention in the methodological literature to date. The article emerges from a recent exploratory study that compared a small set of semi-structured qualitative research interviews, some of which were conducted by telephone and some face-to-face. This paper focuses on selected parts of the analysis, including the overall duration of the interviews and the amount and balance of talk in each mode. Analysis of further interactional features, including comprehension, clarification and responsiveness, are presented in a companion paper (Irvine, Drew & Sainsbury, forthcoming).

Background

Traditionally, methodological textbooks have suggested that the telephone mode is not well suited to the task of qualitative interviewing (Gillham, 2005; Rubin & Rubin, 2005; Legard, Keegan, & Ward, 2003). In particular, the lack of face-to-face contact is said to restrict the development of rapport and a ‘natural’ encounter (Gillham, 2005; Shuy, 2003). The absence of visual cues is also considered to affect the depth of meaning that can be conveyed (Fielding & Thomas, 2008; Gillham, 2005). It has been suggested that telephone interviews will be shorter because they are more fatiguing, and it is harder to sustain concentration (Gillham, 2005; Shuy, 2003). Researchers are also advised that telephone interviews will need to be based around more specific questions and more structured interview guides (Berg, 2007; Gillham, 2005; Gray, Williamson, Karp, & Dalphin, 2007; Ruane, 2005). These factors may be disadvantageous when in-depth data is sought. Clearly, the mode will also be impractical where participants do not have access to a telephone or have difficulties hearing (Wenger, 2002; Worth & Tierney, 1993).

At the same time, there are a number of well rehearsed arguments as to why telephone interviews may in some circumstances be a useful option. Most frequent among these is the benefit of resource savings: telephone interviews require less time and money because they do not involve travel (Robson, 2002; Shuy, 2003; Ruane, 2005; Fielding & Thomas, 2008). For similar reasons, telephone interviews allow the inclusion of participants across a wider geographical scale and access to individuals in settings which it may not be feasible, or may be potentially unsafe, for the researcher to enter. There are also ethical arguments as to why telephone interviews might be preferable. These include the greater anonymity and lesser intensity afforded by a telephone encounter, which might be preferable to participants where topics are of a sensitive nature (Chapple, 1999; Kavanaugh & Ayres, 1998; Tausig & Freeman, 1988).¹

Despite recognition of a range of practical and ethical advantages, there remains nonetheless a sense throughout the instructional literature that qualitative interviewing by telephone is something of a methodological compromise, with concerns about the quality of the interaction and of the data that can be generated via this mode. However, in the recent past a number of researchers have published their first-hand reflections on using the telephone to conduct various types of qualitative interview (Burke & Miller, 2001; Carr & Worth, 2001; Chapple, 1999; Dicker & Gilbert, 1998; Garbett & McCormack, 2001; Grant, 2011; Holt, 2010; Miller, 1995; Opdenakker, 2006; Rose, 1998; Stephens, 2007; Sturges and Hanrahan, 2004; Sweet, 2002; Tausig & Freeman, 1988). On the whole, these researchers report that telephone interviews are ‘just as good’ as those conducted face-to-face, achieving successful social interactions and generating useful data. For example, Chapple (1999) reports that telephone interview data were “unexpectedly rich” (p.91), Sweet (2002) concludes that “the quality and quantity of data was not noticeably different between face-to-face and telephone interviews” (p.63), and Stephens (2007) describes his telephone interviews as achieving equally “friendly rapport” as face-to-face interviews and in all instances generating “excellent data” (p.211). Thus, these accounts are to a large extent challenging the received wisdom that telephone interviews are less suitable for qualitative research.

However, in the main, these previous studies appear to have drawn their conclusions through fairly broad and impressionistic comparisons. Moreover, some offer only reflections on telephone interviews per se, rather than an empirical comparison of modes. Few researcher accounts provide detailed descriptions of the analytic approach through which they arrived at their conclusions. Likewise, authors of methods texts invariably provide their cautionary advice on telephone interviews with no reference to empirical evidence. In sum, direct, detailed and empirically-based comparisons of the telephone and face-to-face interview modes seem, to date, to be lacking in the qualitative methodological literature.

Qualitative interviews take many forms and are used in many different contexts for a wide variety of purposes. The range of academic disciplines and applied research fields that employ qualitative interviews as a data generation method is vast and varied. As such, there is perhaps no conclusive answer to the question of whether telephone interviews are suitable for some generalized concept of ‘qualitative research.’ Much will depend on a study's overall research aims, the specific questions to be addressed, the nature of participants involved and the analytic approach to be taken, amongst a variety of other things (Holt, 2010; Novick, 2008; Sweet, 2002). Moreover, it is undeniable that, for a variety of pragmatic reasons, telephone interviews are used in contemporary social research. Therefore, rather than debating whether telephone interviews are ‘appropriate’ or ‘as good as’ face-to-face, it seems that the more important concern is to increase our understanding of what difference is made to our research processes and outcomes if we choose, for whatever reasons, to conduct research interviews by telephone and in turn to consider what implications this may have for our research practice. It seems unlikely that telephone interviews will fall out of use; on the contrary, they may be increasingly likely to be considered a viable and practical option as research budgets become tighter.

At present, however, the evidence base on mode effects in qualitative interviews is underdeveloped (Sweet, 2002; Sturges & Hanrahan, 2004; Novick, 2008). As observed by Sweet (2002), “The telephone has found its way into qualitative research processes as a medium for data collection, but its use has not generated the critical discussion that is merited” (p.58). The study upon which this article is based makes a contribution to filling this gap.²

Method

There are several levels at which one might consider the question of ‘difference’ in qualitative research interviews, from the fine-grained interactional detail, to the participant experience, to the substantive content of the data, to the final research findings or outputs. Each of these is worthy of investigation and may be of varying degrees of concern to different researchers at different times. In this mode comparison study, the researchers chose to focus comparative analysis at the level of the spoken interaction, drawing upon the techniques and concepts of conversation analysis.³ There was a deliberate intention to set aside the substantive content of interview discussion and to focus on fundamental interactional features which might appear in interviews on any topic, for example, turn taking and comprehension.

The research questions addressed in the mode comparison study emerged from both existing literature and through working with the data. The specific lines of enquiry were developed, clarified and honed as the study progressed. The key questions focused upon in the final comparative analysis included consideration of differences in:

The duration of interviews

The amount and balance of researcher and participant talk

The ways that the researcher displayed attention and interest

The incidence of misunderstanding or requests for clarification

The patterns of turn-taking and incidence of speaker overlap

In this article, we focus on the first two points in this list, presenting findings on interview duration, speaker dominance and considering the consequent impacts on depth of data. The remaining aspects of the comparison at the more detailed level of spoken interaction (including formulation and completion, clarification and comprehension, acknowledgement tokens and adequacy checks) are reported in a companion article (Irvine et al., forthcoming).

Source data

The interviews used in this analysis were originally conducted for a qualitative study that looked at people's experiences of managing mental health and employment (Irvine, 2008). The study was commissioned by the UK Department for Work and Pensions (DWP) with interviews conducted from January to June 2008. The aim of the study was to explore the experiences of people who considered themselves to have a mental health condition and who had been in continuous paid employment for at least the previous 12 months.

For the original study, the researchers aimed to obtain 30 semi-structured face-to-face interviews. Interview participants volunteered to take part in the study in response to information circulated via workplaces and support organizations. Response rates were somewhat higher than anticipated and in order to include all 38 volunteers, while remaining within planned budgets and timescales, a decision was made to conduct nine interviews by telephone. Decisions about which participants would be asked to take part in telephone interviews were made by the research team, largely on the basis of geographical location and time constraints. Participants were not offered a choice of interview mode but all of those who were asked to take part by telephone agreed to do so. Their willingness to take part in a telephone interview could be taken to indicate a general level of comfort and familiarity with extended telephone communication among this group of individuals.

The 38 interviews in the ‘Managing Mental Health and Employment’ (MMHE) study were carried out by two researchers. One researcher conducted 28 interviews, including all nine of the telephone interviews. This subset of data presented an ideal opportunity to carry out a systematic analysis of potential differences in the interaction that takes place in telephone and face-to-face interviews. The principal features of this data set which made it particularly suitable for comparative analysis of interview mode effects were as follows:

All interviews were carried out by the same researcher, thereby minimizing any differences relating to idiosyncrasies of individual researcher style;

Interviews were part of the same study, using the same interview guide, and were carried out within a relatively short time period;

Interviews were conducted prior to conceiving the comparative project, so any conscious differences in the researcher's interviewing behaviour were minimized; and

Allocation to interview mode was researcher-led, so any conscious preference on the part of the participants was minimized.

Of the 28 interviews conducted by the one researcher, 15 were with participants who also shared a similar employment context, enhancing the potential for systematic comparison. Retrospective consent was therefore sought from these 15 individuals to reanalyze their interview data at a closer level of detail.⁴ Consent was received from eleven of the 15 individuals, five of whom had taken place in face-to-face interviews and six who had been interviewed by telephone. Together, the eleven interviews that were subjected to more detailed analysis (henceforth referred to as the ‘mode effects subsample’) totaled just less than 17 hours of audio data. Table 1 provides a summary of the interview data included in this analysis.

Table 1

Overview of interview data included in analysis

	Face-to-face interviews	Telephone interviews
All interviews conducted by the researcher (n = 28)	19 (10 female; 9 male)	9 (6 female; 3 male)
Interviews for which consent was sought to conduct detailed conversation analysis (n = 15)	7 (2 female; 5 male)	8 (5 female; 3 male)
Interviews for which consent was given to conduct detailed conversation analysis: the ‘mode effects subsample’ (n = 11)	5 (1 female; 4 male)	6 (4 female; 2 male)

Participants included males and females, aged between 27 and 63 years. The majority recounted experiences of stress, anxiety and/or depression, whilst a small number had experience of less common mental health conditions. All eleven individuals in the mode effects subsample described themselves as being of white British ethnicity and all study participants were fluent speakers of English. The researcher was female, mixed-race (white British/black African), a native English speaker, and aged 29 during the period when the interviews were conducted.

Findings

In this section, we first present findings on the duration of interviews. We then consider the amount and balance of talk between researcher and participant, which we have conceptualized as speaker dominance. The first part of the analysis (duration) draws upon all 28 interviews conducted by the researcher while the second part (dominance) draws upon the eleven interviews in the mode effects subsample which were examined in closer detail.

Duration

This section presents findings on the duration of interviews conducted in each mode. The durations presented here correspond to what was considered to be the ‘interview proper’ – rather than the total length of the audio recording. The interview proper was defined as starting from the point at which the researcher presented her opening question to the point at which she ‘moved to close’ – typically by way of thanking the participant, stating something to the effect that she had asked all of her questions, and asking the participant if there was anything further he or she wished to add.

Figure 1 shows the duration of each of the 28 interviews (to the nearest minute) in the order in which they were conducted. The pale bars indicate telephone interviews. The durations of the eleven interviews in the mode effects subsample, which are considered in closer detail in later sections of this paper, are boxed for ease of reference. As a general observation, it can be noted that (independent of interview mode) there was substantial variation in the duration of interviews.

To the nearest minute, the average (mean) duration of all 28 interviews was 92 minutes. The average duration across the 19 face-to-face interviews was 96 minutes, while the average duration across the nine telephone interviews was 81 minutes, 15 minutes shorter.

Figure 1

Interview duration to nearest minute

An obvious question, when considering possible explanations for the shorter duration of telephone interviews, is whether the researcher posed fewer questions to the participant. In the eleven interviews considered in conversation analytic detail in this study, this proved not to be the case. In fact, the telephone interviews contained more questions on average (101) than the face-to-face interviews (89) – although it should be noted that there was much variation among individual interviews and some overlap between modes.

The finding that there were in fact more questions asked during shorter interviews seems at first counter-intuitive and suggests that explanations lay in the type or form of questions posed by the researcher. For example, there may have been something about the way in which questions were formulated in telephone interviews which elicited briefer participant responses.⁵ Alternatively, it may have been the participant's reticence that occasioned more intensive questioning, with a greater number of questions required in these telephone interviews in order to elicit the necessary information. Correspondingly, another possibility is that in face-to-face interviews, participants were more ‘spontaneously’ forthcoming, and so spoke at greater length and provided relevant information without the need for explicit prompting. A further finding from this study (see Irvine et al., forthcoming) was that the researcher gave relatively more frequent ‘acknowledgement tokens’ (such as mm hm, yeah, okay) in face-to-face interviews, which – along with any visual feedback such as nods and smiles – may have encouraged participants to continue more extensively with their narratives without the need for explicit prompt questions.

Dominance

This section considers speaker dominance, or the amount and balance of talk between the researcher and participant that made up the overall duration of the interviews. Analysis draws upon the eleven interviews (five face-to-face, six telephone) in the mode effects subsample. These eleven interviews were analyzed to identify the amount of talk (in seconds) produced by each speaker and, from this, the balance of talk between researcher and participant during the interviews. In other words, this part of analysis sought to establish who was ‘holding the conversational floor’ for what amount and proportion of the interview.

The concept of conversational floor holding has been defined and applied in various ways (e.g. Edelsky, 1981). In this study, the concept of floor holding was conceived of as who was ‘in the driving seat’ of the conversation at any given point. A set of conventions was devised to guide, for both the researcher and the participant, what did and did not count as holding the floor.⁶ As a general rule, floor-holding talk by the researcher consisted of speaking turns that in some way steered the conversation (namely questions, prompts and probes), or provided summation, assessment or evaluation of the participant's talk. Very short utterances such as acknowledgement tokens (utterances which invited the participant to simply continue e.g. mm hm, right, okay) were not classed as floor holding by the researcher. As such, in agreement with Edelsky (1981), the present analysis did not necessarily equate any speaking turn with holding the floor.

The measurement of the amount of floor-holding by the researcher was done by reference to the audio recording – calculating the length of the relevant stretch of talk to the nearest second. Floor holding was treated as binary – i.e. researcher or not researcher – and participant floor-holding was not directly measured. However, given that extended stretches of silence or speaker overlap were minimal, the amount of participant floor-holding can be considered as the remainder of the interview's duration, having subtracted the total amount of researcher floor holding.⁷

Figure 2 shows the amount of floor holding (to the nearest minute) by the participant and the researcher in each of the eleven interviews in the mode effects subsample.

Figure 2

Amount of floor holding by participant and researcher

In the previous section, it was noted that (on average) face-to-face conversations across the full set of 28 interviews were some 15 minutes longer. Taking just the eleven interviews in the mode effects subsample, this difference was somewhat larger, at 21 minutes. Figure 2 allows us to answer a supplementary question to this finding: who is contributing this substantial amount of ‘extra’ talk? In this data subset, it emerged that the additional length of the face-to-face interviews came from the participant saying more. On average, there was 25 minutes more participant floor holding in face-to-face interviews compared to telephone interviews. The amount of researcher floor holding across modes varied relatively little in comparison. However, in telephone interviews, there was an average of four minutes more researcher floor holding compared to face-to-face interviews. The shorter duration of telephone interviews was therefore a result of the participant saying less, rather than a proportional reduction in talk from both parties.

Using the above figures on the absolute amount of talk from each party (Figure 2), it is also possible to calculate the proportion of time for which each held the conversational floor. Figure 3 shows the proportion (percentage) of time that the researcher held the floor in each interview.

Figure 3

Proportion of researcher floor holding (%)

Although there are clear exceptions in F4, there is an apparent tendency for a smaller proportion of researcher floor holding in face-to-face interviews and a larger proportion in telephone interviews. In face-to-face interviews, the researcher tended to hold the floor for around 10–13 per cent of the conversation while in telephone interviews the researcher typically held the floor for somewhere between 16–24 per cent of the time. In sum, this greater occupation of ‘airtime’ by the researcher during telephone interviews could be conceived of as greater researcher dominance in the telephone mode.

A further way in which to consider dominance in the conversation is the duration of each ‘stretch’ of floor holding, in other words, how long the participant continued to speak uninterrupted or without actively ceding the floor. As a subgroup, the average (mean) duration for which telephone participants held the floor ‘at a stretch’ was substantially shorter than in face-to-face interviews, the figures being 31 seconds and 47 seconds respectively. The combined outcome of telephone interviews being (typically) shorter and of participants holding the floor for shorter stretches at a time is that the ‘to and fro’ of conversation during telephone interviews was generally more rapid.

A possible explanation for the shorter contributions of participants in the telephone interviews is that the researcher somehow ‘cut off’ responses. However, analysis of turn taking and transitions between speakers in this data set suggests that this was not the case. There were very few instances where the researcher might be considered to have ‘interrupted’ a participant in mid-flow. As such, it seems that the tendency was for participants to offer less over the telephone.

Discussion

A defining characteristic of qualitative interview research is the quest for data that is rich, detailed or otherwise ‘in-depth.’ The above findings, that telephone interviews tend to be shorter and generate substantially less participant talk, might therefore be treated as a matter of concern. What is contained within those additional minutes in face-to-face interviews that is potentially ‘missing’ from the telephone interviews?

Loss of data could occur in two dimensions: breadth of coverage or depth of detail. As explained in the introduction to this paper, a detailed comparison of the substantive content of the data was not within the aims or scope of the present study. However, drawing upon the thematic analysis conducted for the original MMHE study (Irvine, 2008), it is possible to assess the extent to which there was comparable breadth of data and to make some tentative suggestions about differences in depth.

For the original study, the two researchers who conducted the thematic analysis moved quickly from the full transcripts to working with syntheses of the interviews, which were produced using a combination of ‘notes and quotes’. As such, there was already some loss of detail at an early stage in the analytic process.⁸ The original thematic analysis shows that, for a majority of the main thematic categories, there was coverage of each topic in each of the six telephone interviews included in the mode effects subsample. Among the mode effects subsample, all but one of the telephone interviews were conducted after the face-to-face interviews (see Figure 1). Where the researchers' synthesis suggested that a topic had not been covered during a telephone interview, this could usually be attributed to a narrowing of focus in later interviews, as certain themes emerged as more central and others less pertinent. A progressive honing and re-focusing of lines of inquiry is characteristic of qualitative research and so, to some degree, we might expect to find such alterations in scope and focus when comparing any set of interviews, regardless of mode.

However, this refining of research themes cannot fully explain the shorter duration of the telephone interviews, particularly as subsequent face-to-face interviews conducted after the bulk of the telephone interviews were again of a longer average duration (see Figure 1). Thus, the explanation appears in part to lie in a lesser degree of detail or elaboration offered by telephone interview participants.

In this study we were not able to analyze in a systematic way the social interactions that occurred off tape prior to the interview. However, there were clearly differences in the amount and nature of small talk that occurred when being greeted and invited into a private home or workplace to conduct a face-to-face interview as compared to when opening an interview by telephone. The common practices of establishing rapport such as offering/accepting a drink, making informal conversation about the researcher's journey, and admiring family photos or pets did not occur in the lead-in to telephone interviews. Although there were polite greetings and checks that the timing of calls was convenient, the telephone interviews tended to ‘get down to business’ much more quickly. It is possible that these differences in the initial moments of the research encounter set a different pace and atmosphere for the interview which followed. Alongside the finding (noted earlier) that the researcher gave fewer acknowledgement tokens during telephone interviews, these aspects of the interaction are potential explanations for a more succinct and somewhat more brisk data generation process during telephone interviews.

However, it is important to bear in mind that the 28 interviews conducted by this researcher for the MMHE study showed a great deal of variation in length regardless of interview mode, and that despite an overall tendency for telephone interviews to be shorter than face-to-face, there were exceptions among both modes (in particular, interviews identified as F4 and T1: see Figures 1, 2 and 3). There are numerous factors other than mode that could contribute to interview length, which the present analysis has not addressed. These may include influences such as time of day, energy/mood of the researcher or participant, time available (due to other commitments) and indeed the basic matter of how much the participant has to say on the subject or the complexity of the experiences they choose to share. As suggested by a commentator during the dissemination of this study (personal communication, October 2009), it may be that to some extent, the amount that a participant says during a research interview is ‘just down to personality.’ Nevertheless, the present data suggests that mode did have at least a partial influence on the duration and depth of participant talk, with a resultant effect on speaker dominance.

Interestingly, this study's findings on duration are in contrast to other empirical accounts, which report that the length of telephone and face-to-face interviews did not differ notably (Sweet, 2002; Sturges & Hanrahan, 2004). Previous empirical mode comparisons have tended not to provide specific details of interview length or how these were calculated (Sturges & Hanrahan being a partial exception). As such, it is difficult to comment in any detail on these apparent differences between studies. However, in itself, these contrasting findings signal the value of further mode comparison studies to increase the evidence base.

To summarize the findings of the present study, an analysis of the overall duration, dominance and depth of talk between researcher and participant in this set of interviews revealed the following key findings:

Despite much variation in individual interview length, telephone interviews were typically, and on average, shorter than those conducted face-to-face.

The shorter duration of telephone interviews was a result of the participant speaking for less time, rather than a proportional reduction in talk from both parties. Additionally, in telephone interviews, participants generally ‘held the floor’ for shorter stretches at a time.

The researcher did slightly more talking during telephone interviews than in face-to-face. Combined with the reduced amount of participant talk, this meant that the researcher tended to hold the floor for a greater proportion of the time in telephone interviews, or was more ‘dominant.’

To a moderate degree, the shorter length of telephone interviews could be accounted for by a reduction in coverage of themes. However, the principal explanation appeared to lie in a tendency for telephone interview participants to provide relatively less detail or elaboration.

Two key questions arise from these findings: (i) What are the implications for data quality? And, (ii) how might we wish to modify interviewing approaches in response? While this small-scale exploratory study cannot provide firm conclusions, the remainder of this section offers some tentative suggestions.

What are the implications for data quality?

The concept of ‘missing data’ in qualitative research has been discussed by Hinds, Vogel, and Clarke-Steffen (1997) and Crow and Powell (2010). Hinds et al. (1997) focus particularly on the matter of topic coverage:

Missing data in a qualitative study results when an issue is explored by a researcher in one interview (or other data source) but is not addressed in all other interviews. This could happen because the topic emerged spontaneously and unexpectedly and thus had not been addressed in previous interviews or other sources, or it could result from a narrowing of the study's scope and a refining of the interview questions such that the topic is not subsequently addressed. (p.411–412)

Hinds et al. (1997) focus on data loss at the point of conducting a research interview. However, throughout the qualitative research process, data may be reduced in depth and detail at several stages, as shown in Figure 4.

Figure 4

Data reduction in thematic qualitative analysis

At the top level, failure to generate data on a given theme during an interview results in data lost to the study altogether (assuming there is only one interview opportunity). At the level of transcription, the choice of conventions used to convert an audio recording into a textual representation may result in the discarding of details such as pauses, speaker overlap and pitch. In the interests of brevity, utterances such as ‘erm’ or ‘you know’ may also be omitted in the transcription process. For conversation analysts, this would be perceived as a loss of essential information. However, for a more typical thematic analysis, such omissions may not be critical (see Hammersley, 2010, for further discussion). At the remaining levels, a distinction can be drawn between whether the reduction of data at any given point in the process is through a condensing or a discarding of information. For example, at the level of synthesis, researchers will make decisions both as to how to condense lengthy participant accounts into more succinct paragraphs and also perhaps decisions to discard altogether certain themes or comments that are not deemed central to the research questions. At the level of analysis, methodological and pragmatic drivers (for example, time constraints or the interests of research funders) may lead researchers to focus on only a selection of the themes covered in interviews. Thus, whole tranches of the data may be set aside. Finally, at the level of reporting, constraints on word length or considerations of audience may lead to a narrower still selection of the data being presented or the use of very pithy ‘key points.’

The process depicted in Figure 4 is typical of a semi-structured interview study following a thematic approach to analysis, as was conducted for the MMHE project upon which the present study draws. However, other analytic approaches (not least conversation analysis) will not involve this process of synthesis and will work throughout the analysis with a much more detailed and comprehensive version of the original interview. This brings us to perhaps the most pertinent question when considering the implications of mode-related difference: To what extent do differences in length, breadth or depth matter within a given research project?

Different research studies will have different aims and objectives, resulting in different perspectives on the implications of the above findings and the consequent appropriateness of each interview mode. In the MMHE study, the primary audience was the funding government agency. Their requirement was for a relatively short report, which presented thematic findings in response to a series of pre-determined and fairly focused questions that were of interest to the policymakers. There was little scope within that study to explore in detail aspects such as participants' ‘illness narratives’ (Good, 1994) or socio-cultural constructions of mental health. For the purposes of the research project as commissioned, and taking into account the process of data reduction depicted in Figure 4, the consequences of having generated 15–20 fewer minutes of participant talk in telephone interviews might be considered to be minimal. In all interviews, telephone and face-to-face, key themes were covered to an extent that met the project's requirements. Bearing in mind the possibility that additional duration of face-to-face research encounters may be composed of ‘superfluous’ discussion (Nicholas et al., 2010), greater quantity of data may not necessarily imply greater quality of data for a given analytic purpose. On the other hand, an approach that seeks to analyze participants' narratives or to conduct a much more in-depth phenomenological exploration of an individual's experience may be hampered by more succinct accounts.

Alongside the above considerations, it is also important to remember that our research questions or the requirements of a research funder are not our only concerns as qualitative researchers. The differences in duration and dominance that have been revealed through an analysis of the interview recordings, placed in the wider context of the overall interview encounters, suggest that the experience of participating in the research will have been quite different for those people who were visited at home or in their workplace compared to those who took part by telephone.⁹ While in this case the telephone mode provided satisfactory data for the research team's analytic purposes, in considering the participant experience, the present findings suggest that there may be aspects of the interaction that researchers might pay particular attention to when conducting interviews by telephone. These are considered in the final part of this discussion.

How might we wish to modify interview practice?

As has been noted, depending on the aims and scope of a given research project, the added value of an additional (as it may be) 15–20 minutes of talk from the participant will vary. For some more tightly focused qualitative research purposes, a more succinct account might be entirely adequate. However, in many cases researchers taking a qualitative approach will be concerned to create an atmosphere where participants do feel able to be expansive and detailed in their responses. As such, this final section suggests some possible ways to encourage this in telephone interviews, where (based on the present findings) there may be a tendency towards greater participant reticence or parsimony.

Burnard (1994) suggests that, when conducting telephone interviews:

…it is also useful to make some more general enquiries about the interviewee to set him or her at their ease. Many people take a little time to ‘warm up’ on the phone and such general conversation is equivalent to the initial questions that are asked at the beginning of a face-to-face interview to make the interviewee feel more comfortable. (p.70)

As noted earlier, there was far less of this ‘warm up’ type discussion during the telephone interviews in this study and, in light of the above findings, the advice of Burnard (1994) seems worth reinforcing. Researchers seeking elaborate or in-depth information via the telephone interview mode might want to consider ways of establishing a more relaxed conversational style prior to asking specific interview questions. A separate preliminary and more informal phone call to make personal introductions before the ‘research encounter proper’, may be helpful in this respect. Establishing some common ground with a participant early in an interview may also facilitate rapport and the opening up of conversation.¹⁰

In light of the findings on turn length, there may also be a need to ‘prime’ telephone interview participants in a more explicit way for the type of in-depth responses that are desired in qualitative interview genres (see also Holt, 2010). Perhaps due to associations with telephone survey research, the present data showed signs of more succinct responses from telephone participants. As such, it may be beneficial at the opening of telephone interviews to explain or reiterate that detailed, reflective or otherwise expansive responses will be welcomed. Researchers may also wish to be more proactive and explicit in (vocally) encouraging elaboration if telephone participants tend towards the concise.

Conclusion

Decisions on whether or not to use telephone interviews in qualitative research will be influenced by many factors, including physical capacities, safety considerations, the importance of the physical/visual context, available budgets, and time constraints. For a range of practical and ethical reasons, there are occasions when the telephone may be a preferable or entirely justified methodological choice for a given project or specific interview. As such, it is important to enhance the evidence base on what difference is made to the interview interaction and the data that results when, for whatever reason, we conduct interviews by telephone rather than face-to-face. On the basis of such evidence, researchers will be in a better position to assess the implications for any given project and, if necessary, modify their approach to interviewing in ways that address any perceived risks to the quality or usefulness of the data.

This paper has presented findings of a systematic analysis of interview duration and speaker dominance and has offered some more tentative findings regarding the depth of data generated. A companion paper based on the same study presents findings on more fine-grained interactional details including comprehension, clarification and acknowledgements (Irvine et al., forthcoming). As has been emphasized, the present study took a specifically interactional perspective on the data, rather than analyzing the substantive content of interviews. However, the matter of difference in substantive content has been raised by all audiences to which this research has been presented. We therefore recognize that this must be a feature of any future investigation into mode effects in qualitative interviews. Nevertheless, there may be methodological challenges inherent in establishing a meaningful way to assess qualitative interview data for equivalence in breadth and depth. We cannot know precisely what information was held back or otherwise left unelicited in a shorter interview, or specifically which parts of a longer interview constitute the ‘extra’ data that emerged due to interview mode. Moreover, while there are sound methodological reasons for attempting to cover a similar range of key themes with each participant, seeking to demonstrate consistent breadth and depth across an interview study as a mark of ‘quality’ may be fundamentally at odds with qualitative approaches that privilege flexibility and responsiveness over uniformity.

Despite these challenges, the research findings presented in this article make an initial contribution to generating a more robust evidence base on what differences we may find in semi-structured telephone and face-to-face interviews. While the study was based on only a small number of interviews conducted by one researcher, its systematic approach to analysis offers a model which others can replicate with further data sets, thus expanding the knowledge base.

We deliberately decline to offer any concluding comment about whether, based on the findings above, telephone interviews can be deemed ‘suitable’ for qualitative research. This is a question that will need to be addressed afresh for any given study, but (it is hoped) with decisions based upon an increasingly strong body of empirical evidence about the variations in the process and outcomes of interviews conducted via different modes.

Footnotes

1.

Further personal experiences relayed to the author during dissemination of this study included a researcher who had used telephone interviews in order to continue with a phase of interview research while breastfeeding, and a researcher who had a physical disability which would have made travelling to meet with participants particularly challenging (see also Galvin, 2005).

2.

It is relevant to mention briefly the use of online and email interviewing, an area within which a mode comparison debate is also ongoing (e.g. Ison, 2009; McCoyd & Schwaber Kerson, 2006; Meho, 2006; Seymour, 2001). Interestingly, the more recent expansion in online research methods seems already to have captured the critical interest of qualitative researchers to a much greater extent than telephone interviewing has ever done. For example, Fontana and Frey (2005) devote several paragraphs to electronic interviewing while Flick (2009) provides a full chapter on qualitative online research. Novick (2008) also makes this observation rather eloquently: “It seems that telephone interviews neither have the endorsement enjoyed by face-to-face interviews, which are seen as the gold standard, nor the aura of excitement generated by Internet interviews” (p.397). As the use of video-phone technology (such as Skype) increases, a further set of methodological questions presents itself in the way that such interactions compare to the existing suite of possibilities (Sedgwick & Spiers, 2009).

3.

The methodological approach is described in more detail in a companion paper that presents further aspects of the study's findings (Irvine et al., forthcoming). For the purposes of the present paper, which reports on overall duration, speaker dominance and depth, specific details of the conversation analytic approach are not essential and are omitted in the interests of brevity (for an introduction to conversation analysis see Drew, 2005; Sidnell, 2010)

4.

For the mode comparison study, consent was sought via a personal letter to each participant, explaining the aims and purpose of the new study. A printed consent form (offering the option to either provide or refuse consent) and a prepaid return envelope were included with the letter. For the original MMHE study, all participants received the same written and verbal explanations of the project's aims, methods, data handling and reporting processes. Participants who were interviewed face-to-face then completed and signed a consent form. For the telephone interviewees, consent was recorded orally (on a digital recorder). The researcher read out the statements from an identical consent form as used face-to-face and the participant indicated their agreement verbally on the audio recording.

5.

Detailed investigation of questioning style in research interviews lies beyond the scope of the present study, but is a key area for further empirical research. The author thanks colleagues who took part in a dissemination workshop as part of the present study, in particular Professor Jonathan Potter and Dr Tim Rapley, for highlighting the value of further research into question construction and the ‘practical enactment’ of a topic guide.

6.

Due to limitations of space, these conventions are not detailed in full here, but are available from the author on request.

7.

This approach to measuring the balance of talk during interviews is in contrast to the approaches taken by Ezzy (2010) and Roulston, deMarrais, and Lewis (2003). Ezzy (2010) counted the number of words attributed to researcher and participant in interview transcripts and counted the number of lines of the transcript attributed to researcher and respondent. While these approaches are two of the more systematic to be presented in the literature, they nevertheless convey limited information about the actual duration of turns at talk. Moreover, the results of a word or line count may vary substantially depending on how detailed an approach is taken to transcription.

8.

It is also worth noting that there were differences in the two researchers' approaches to synthesizing the interview transcripts. One researcher included considerably less detail in her syntheses, being more succinct and including fewer verbatim extracts, while the other was more expansive in her representation of the interview content and included several long verbatim extracts. This observation may in itself be worthy of further reflection in the context of team research in qualitative inquiry.

9.

While drawing this inference, it must be acknowledged that the present study did not consult participants on their mode preferences or experiences of research participation. This would be an important facet of any future study of mode effects in qualitative interviews. Holt (2010) and provide brief commentary in this regard.

10.

This suggestion is made based on the author's experience of the longest telephone interview in this data set (T1), which in duration and balance resembled the face-to-face subgroup more closely than the other telephone interviews. This encounter seemed to get off on a particularly good conversational footing due to a number of coincidental shared experiences between researcher and participant that emerged in the opening stages of the interview.

References

Berg

B.L.

(2007). Qualitative research methods for the social sciences. Boston, MA: Pearson Education Inc.

Burke

L.A.

, & Miller

M.K.

(2001). Phone interviewing as a means of data collection: Lessons learned and practical recommendations. Forum: Qualitative Social Research 2. Retrieved from http://www.qualitative-research.net/index.php/fqs/article/view/959

Burnard

(1994). The telephone interview as a data collection method. Nurse Education Today, 14, 67–72.

Carr

E.C.J.

, & Worth

(2001). The use of the telephone interview for research. Nursing Times Research, 6, 511–524.

Chapple

(1999). The use of telephone interviewing for qualitative research. Nurse Researcher, 6(3), 85–93.

Crow

, & Powell

(2010). What is missing data in qualitative research? In NCRM Research Methods Festival 2010, 5–8 July 2010, St. Catherine's College, Oxford. Recording available http://eprints.ncrm.ac.uk/1494/

Dicker

, & Gilbert

1988. The role of the telephone in educational research. British Educational Research Journal, 14, 65–72.

Drew

(2005). Conversation analysis. In Fitch

K.L.

& Sanders

R.E.

(Eds.), Handbook of language and social interaction (pp. 71–102). New Jersey: Lawrence Erlbaum.

Edelsky

(1981). Who's got the floor? Language in Society, 10, 383–421.

10.

Ezzy

(2010). Qualitative interviewing as an embodied emotional performance. Qualitative Inquiry, 16, 163–170.

11.

Fielding

, & Thomas

(2008). Qualitative interviewing. In Gilbert

(Ed.), Researching social life (3rd ed.) (pp. 245–265). London: Sage Publications.

12.

Flick

(2009). An introduction to qualitative research. London: Sage Publications.

13.

Fontana

, & Frey

J.H.

(2005). The interview: From neutral stance to political involvement. In Denzin

N.K.

& Lincoln

Y.S.

(Eds.), The sage handbook of qualitative research (pp. 695–727). Thousand Oaks: Sage Publications.

14.

Galvin

R.D.

(2005). Researching the disabled identity: Contextualising the identity transformations which accompany the onset of impairment. Sociology of Health & Illness, 27, 393–413.

15.

Garbett

, & McCormack

(2001). The experience of practice development: An exploratory telephone interview study. Journal of Clinical Nursing, 10, 94–102.

16.

Gillham

(2005). Research interviewing: The range of techniques. Berkshire: Open University Press.

17.

Good

B.J.

(1994). Medicine, rationality and experience: An anthropological perspective. Cambridge: Cambridge University Press.

18.

Grant

(2011). Fear, confusion and participation: Incapacity benefit claimants and (compulsory) work focused interviews. Research, Policy and Planning, 28, 161–172.

19.

Gray

P.S.

Williamson

J.B.

Karp

D.A.

, & Dalphin

J.R.

(2007). The research imagination. An introduction to qualitative and quantitative methods. Cambridge: Cambridge University Press.

20.

Hammersley

(2010). Reproducing or constructing? Some questions about transcription in social research. Qualitative Research, 10, 553–569.

21.

Hinds

P.S.

Vogel

R.J.

, & Clarke-Steffen

(1997). The possibilities and pitfalls of doing a secondary analysis of a qualitative data set. Qualitative Health Research, 7, 408–24.

22.

Holt

(2010). Using the telephone for narrative interviewing: A research note. Qualitative Research, 10, 113–121.

23.

Irvine

(2008). Managing mental health and employment. London: Department for Work and Pensions.

24.

Irvine

Drew

, & Sainsbury

(forthcoming). “Am I not answering your questions properly?” Clarification, adequacy and responsiveness in semi-structured telephone and face-to-face interviews. Qualitative Research.

25.

Ison

N.L.

(2009). Having their say: Email interviews for research data collection with people who have verbal communication impairment. International Journal of Social Research Methodology, 12, 161–172.

26.

Kavanaugh

, & Ayres

(1998). “Not as bad as it could have been”: Assessing and mitigating harm during research interviews on sensitive topics. Research in Nursing and Health, 21, 91–97.

27.

Legard

Keegan

, & Ward

(2003). In-depth interviews. In Ritchie

& Lewis

(Eds.), Qualitative research practice: A guide for Social Science students and researchers (pp. 138–169). London: Sage Publications.

28.

McCoyd

J.L.M.

, & Schwaber Kerson

(2006). Conducting intensive interviews using email: A serendipitous comparative opportunity. Qualitative Social Work, 5, 389–405.

29.

Meho

L.I.

(2006). E-mail interviewing in qualitative research: A methodological discussion. Journal of the American Society for Information Science and Technology, 57, 1284–1295.

30.

Miller

(1995). In-depth interviewing by telephone: Some practical considerations. Evaluation and Research in Education, 9, 29–38.

31.

Nicholas

Lach

King

Scott

Boydell

Sawatzky

& Young

(2010). Contrasting internet and face-to-face focus groups for children with chronic health conditions: Outcomes and participant experiences. International Journal of Qualitative Methods, 9, 105–121.

32.

Novick

(2008). Is there a bias against telephone interviews in qualitative research? Research in Nursing and Health, 31, 391–398.

33.

Opdenakker

(2006). Advantages and disadvantages of four interview techniques in qualitative research. Forum: Qualitative Social Research 7(4). Retrieved from http://www.qualitative-research.net/index.php/fqs/article/view/175

34.

Robson

(2002). Real world research. Oxford: Blackwell.

35.

Rose

K.E.

(1998). The telephone as a data collection instrument in a qualitative study of informal carers of terminally ill cancer patients. European Journal of Oncology Nursing, 2, 59–61.

36.

Roulston

deMarrais

, & Lewis

J. B.

(2003). Learning to interview in the social sciences. Qualitative Inquiry, 9, 643–668.

37.

Ruane

J.M.

(2005). Essentials of research methods. A guide to Social Science research. Malden, MA: Blackwell.

38.

Rubin

H.J.

, & Rubin

I.S.

(2005). Qualitative interviewing: The art of hearing data (2nd ed.). Thousand Oaks: Sage Publications.

39.

Sedgwick

, & Spiers

(2009). The use of videoconferencing as a medium for the qualitative interview. International Journal of Qualitative Methods, 8, 1–11.

40.

Seymour

W.S.

(2001). In the flesh or online? Exploring qualitative research methodologies. Qualitative Research, 1, 147–168.

41.

Shuy

R.W.

(2003). In-person versus telephone interviewing. In Holstein

J.A.

& Gubrium

J.F.

(Eds.), Inside interviewing: New lenses, new concerns (pp. 175–193). Thousand Oaks: Sage Publications.

42.

Sidnell

(2010). Conversation analysis: An introduction. Oxford: Wiley-Blackwell.

43.

Stephens

(2007). Collecting data from elites and ultra elites: Telephone and face-to-face interviews with macroeconomists. Qualitative Research, 7, 203–216.

44.

Sturges

J.E.

and Hanrahan

K.J.

(2004). Comparing telephone and face-to-face qualitative interviewing: A research note. Qualitative Research, 4, 107–118.

45.

Sweet

(2002). Telephone interviewing: Is it compatible with interpretive phenomenological research? Contemporary Nurse, 12, 58–63.

46.

Tausig

J.E.

, & Freeman

E.W.

(1988). The next best thing to being there: Conducting the clinical research interview by telephone. American Journal of Orthopsychiatry, 58, 418–427.

47.

Wenger

G.C.

(2002). Interviewing older people. In Gubrium

J.F.

& Holstein

J.A.

(Eds.), Handbook of interview research: Context and method. Thousand Oaks: Sage Publications.

48.

Worth

, & Tierney

A.J.

(1993). Conducting research interviews with elderly people by telephone. Journal of Advanced Nursing, 18, 1077–1084.