A methodological guide to using and reporting on interviews in conservation science research

Interviews are a widely used methodology in conservation research. They are flexible, allowing in‐depth analysis from a relatively small sample size and place the focus of research on the views of participants. While interviews are a popular method, several critiques have been raised in response to their use, including the lack of transparency in sampling strategy, choice of questions and mode of analysis. In this paper, we analyse the use of interviews in research aimed at making decisions for conservation. Through a structured review of 227 papers, we explore where, why and how interviews were used in the context of conservation decision making The review suggests that interviews are a widely used method for a broad range of purposes. These include gaining ecological and/or socio‐economic information on specific conservation issues, understanding knowledge, values, beliefs or decision‐making processes of stakeholders, and strengthening research design and output. The review, however, identifies a number of concerns. Researchers are not reporting fully on their interview methodology. Specifically, results indicate that researchers are: failing to provide a rationale as to why interviews are the most suitable method, not piloting the interviews (thus questions may be poorly designed), not outlining ethical considerations, not providing clear guides to analysis and not critically reviewing their use of interviews. Based on the results of the review, we provide a detailed checklist aimed at conservation researchers who wish to use interviews in their research (whether experienced in using the methodology or not), and journal editors and reviewers to ensure the robustness of interview methodology use.


| INTRODUCTION
The success or failure of conservation actions often depend on complex decision making made by individuals, from policy makers to resource managers to consumers. Effective management necessitates a firm understanding of how and why decisions are made. Given calls for greater integration of social science methods in conservation research (Bennett et al., 2016), methodologies commonly used in the social sciences such as interviews can be employed to understand the factors which influence decision-maker behaviour. However, lack of robust requirements from interdisciplinary conservation research journals results in poor and inadequate reporting of key points needed to interpret the quality of social research. Maccoby and Maccoby (1954, 449) define an interview as an "interchange in which one person… attempts to elicit information or expressions of opinion or belief from another person or persons." While interviewing has been used for millennia (Babbie, 1992), Charles Booth is recognised as the first scientist to have developed a social survey based on interviewing in 1886 (Converse, 1987). In the early 20th century, interviews were increasingly used in opinion polling, psychological testing, clinical diagnosis and counselling. The interview technique is now used in several fields, through face-to-face verbal exchanges, group exchanges and telephone or Internet surveys (Mason, 2012).
Interviews can range from short exchanges to long detailed exchanges repeated over time. Interviews are now so widespread that researchers have claimed we live in an "interview society" (Silverman, 1993) where "interviewing has become a routine technical practice and a pervasive, taken-for-granted activity in our culture" (Mishler, 1986, p. 23).
Interviewing relies on an interactive method in which mutual learning occurs between those involved in the interview process. In this respect, interviewing is an active research process by which an interview or a "contextually bound and mutually created story" is produced by interviewer and interviewee(s) (Fontana & Frey, 2005, 696). They also allow researchers to focus on the interviewees' perspective of what is important or relevant, thereby potentially highlighting issues that the interviewer might not have considered. Interviews may thereby even help to empower interview subjects themselves, allowing for changes in social policy and improved conditions for interviewees (Fontana & Frey, 2005). Minichiello, Aroni, Timewell, and Alexander (1995) argue that interviews can be preferable to other methods in filling a knowledge gap, particularly if complex behaviours are to be investigated. Previous studies have provided useful advice specifically to conservation scientists about the potential benefits of qualitative social science methodologies (Drury, Homewood, & Randall, 2011;Newing, 2010;St. John, Keane, Jones, & Milner-Guland, 2014).
In this paper, we first review recommendations found in the literature on how to conduct interviews. Then we analyse the use of interviews in the context of decision making in conservation. Through a structured review, we explore where and why interviews were used, before examining how authors reported on their use of the interview methodology in the context of conservation decision making. Based on the results, we provide a detailed checklist aimed at conservation researchers who wish to use interviews in their research (whether experienced in using the methodology or not), and journal editors and reviewers to ensure the robustness of interview methodology use.

| DESCRIPTION OF KEY STAGES IN AN INTERVIEW PROCESS
Papers by Drury et al. (2011) andSt. John et al. (2014 -see fig. 2 in their paper) outline key steps in qualitative methodologies, including question formulation, ethical review, and techniques to perform them. In a book on the value of social science methodologies in conservation, Newing (2010) similarly outlines key stages for research design, including a chapter devoted to interviews. Here, we briefly outline basic stages for using interviews in conservation research (see Figure 1), mainly aimed at researchers not familiar with the interview methodology, before paying closer attention to how these stages are being reported in scientific publications.
The basic stages in an interview process (see Figure 1) can broadly be defined as the initial project design (Stages 1-3), data gathering (Stages 4-7), and analysis and write-up (Stages 8-9).

| Initial project design (identify research question/s, type of interview and formulate interview questions)
The interview process starts with the identification of research question(s). This is followed by a critical reflection of whether the outlining ethical considerations, not providing clear guides to analysis and not critically reviewing their use of interviews.
4. Based on the results of the review, we provide a detailed checklist aimed at conservation researchers who wish to use interviews in their research (whether experienced in using the methodology or not), and journal editors and reviewers to ensure the robustness of interview methodology use.

K E Y W O R D S
conservation, decision-making, interviews, qualitative methods, semi-structured interviews, social science, structured interviews, unstructured interviews interview is the most suitable methodology to use based on the question and whether the interview should be supplemented with other methods. Key areas to consider at this stage include whether interviews can provide the right kind of data for envisaged outputs or whether other research techniques might be more suitable. In making this decision, researchers could weigh up the advantages and disadvantages of interviews as a methodology in the light of their research question(s), including different styles of interviews (structured, semistructured and unstructured).
Structured interviews are based on a fixed set of pre-determined questions. The same interview script is used in each separate interview, which allows close comparison between different transcripts, but does not allow interviewees to shape the discussion (Punch, 2005).
Conversely, unstructured interviews are not based on an interview script. In this style of interview, the course of the conversation depends on the responses of the interviewee themselves and questions are asked spontaneously based on these answers (Bryman, 2004). While this does allow an in-depth analysis of particular issues that are considered important by an interviewee, and limits pre-conceived researcher bias in shaping the interview, such interviews offer little assurance that all relevant issues will be covered and present a problem for comparative data analysis (Bryman, 2004). Since there are disadvantages of both structured and unstructured interviews, researchers generally prefer to adopt a middle-ground using a semi-structured approach (Dunn, 2000).
These rely on a pre-conceived interview guide, which means that standard questions are asked in each separate interview, allowing comparison and maintaining data quality. Crucially, however, they allow the interviewer to ask additional questions if an interesting or new line of enquiry develops in the interview. This flexibility is important for investigations of complex issues, such as studies of conservation sciencepolicy interfaces, which analyse messy processes that can rarely be foreseen (Rose, Brotherton, Owens, & Pryke, 2016;Young et al., 2014).
Once the type of interview has been selected, the researcher can start to formulate interview questions. Depending on the type of interview selected, formulating questions may result in a structured interview schedule, an interview guide or an aide memoire (Bryman, 2004;Drury et al., 2011). Bias due to poorly constructed questions is a common criticism of qualitative interviews. Therefore, before formulating questions, researchers should seek to build their knowledge of robust question design from methods textbooks and training courses.
In general, it is best to start the interview with relatively easy questions.
Such questions, which may ask the respondent to tell the researcher something about themselves, help the respondent to settle, therefore building a good rapport for the interview. Subsequent questions should be designed in a manner that does not lead or force the respondent to give particular answers desired by the researcher (Bryman, 2004).

| Data gathering (sampling, ethical review, piloting/refinement of interview and undertaking interview)
Data gathering begins with the identification of interviewees. A robust sampling strategy should be developed to ensure informed coverage of the population of interest. St. John et al. (2014) suggest first considering the proportion of the interest population that can realistically be sampled, before considering whether there are any important subgroups that could be under-represented. Once the practicalities of sample size are known, a strategy can be devised to ensure that meaningful and robust data are collected to answer the research question/s. Several sampling techniques exist (Newing, 2010), including, (1) snowball sampling-where initial informants are identified and the subsequent sample is built by asking for key recommendations from these informants, (2) theoretical sampling-where you interview a few informants, transcribe, analyse and look for key patterns, and then identify further participants based on emergent themes, (3) key informant sampling-where you target key people that are knowledgeable about the issue, (4) representative sampling-where a sample is chosen to be representative of the total population (involves stratification), (5) F I G U R E 1 Basic stages in an interview process: initial project design (stages 1-3), data gathering (stages 4-7), analysis and write-up (Stages 8-9) random sampling-where people are spoken to at random. Repeats may need to be undertaken depending on the initial response rate.
The next stage is to apply for ethical clearance (Silverman, 2005). St. John et al. (2014) argue that many conservation researchers are undertaking research on people without being properly informed on issues such as informed consent (checking that participants understand the aims of the project and how their data will be used), anonymity (protecting the identity of informants) and compensation (providing some form of compensation for time disruption, e.g. financial, otherwise the process could be coercive). Other ethical considerations include level of personal intrusion, including the sensitivity of questions, vulnerability of participants groups and the storage of confidential data. Although most organisations, particularly universities, as well as non-governmental organisations (NGOs) and government departments, have ethical review committees, St. John et al. (2014) argue that members of biological departments are inadequately trained to scrutinise human research. Some journals also require that authors report that ethical clearance for the study has been gained during the submission process. Ethical guidelines should be followed voluntarily in cases where official ethical standards are non-existent. In this review, we made the explicit assumption that ethical clearance had been approved, and focussed on identifying whether any ethical concerns had been raised by authors.
When an initial list of questions has been designed, it is useful to pilot or test the interview on colleagues or a subset of the target population (after ethical clearance) in order to check for length, language suitability and potential sources of bias (e.g. leading questions).
The pilot interview can be transcribed to check that it has produced enough relevant data to answer the research question; if not, changes are needed. Interviews may be carried out face-to-face, in person, over the telephone or increasingly using video technology.

| Analysis and write-up (including dissemination and result feedback to participants)
For semi-structured or unstructured interviews in particular, the analysis of interviews often involves a process known as "coding" (structured interviews may follow a pre-selected coding pattern). Codes may be described as "tags or labels for assigning units of meaning to the descriptive or inferential information compiled during a study" (Miles & Huberman, 1994, 56), and they are usually attached to pieces of data of varying size. For this reason, codes can be assigned to individual words, phrases, sentences or whole paragraphs in each transcript, in order to aid the interpretation of meaning. Central to the use of coding is the notion that words themselves do not matter. Instead, their meaning is more important, and this allows the clustering of key issues in the data. Researchers might find the use of computer software, such as NViVo or AtlAs.ti, useful in producing "code maps" and organising the codes in a logical way. The skills needed to operate these software packages are relatively easy to learn, with training courses in most universities and good online guides available.
After analysis, findings are written up, including a critical evaluation of the advantages and disadvantages of using interviews and how their use could have been improved, and efforts made to provide feedback to participants.

| Key data to provide when reporting on interviews
In order to allow reviewers and readers to make an informed judgement about the quality of data collection and suitability of conclusions, key data on the application of interviews should be provided in scientific publications. Figure 2 provides a checklist of key data which need to be included in publications, and the subsequent review described in this paper assess whether these data are currently being reported adequately.

| OUTLINE OF REVIEW METHODOLOGY
The review had two purposes, first to review where interviews have been used in conservation decision-making research, and second, to assess whether key data on the application of interviews are being reported. We conducted a structured literature review in Scopus for the years 1996 till 08-09-2016. The exact search terms were: Unlike the other papers in this special issue, the search term "decision" was specifically added to the search string to narrow the focus to decision making. This resulted in 676 documents, 40 of which were reviews (none of which were on the interview method itself). We excluded the reviews as these were not primarily focussed on the use of interviews as a method, and shortlisted 636 documents (articles only), which are provided in the supplementary material.
The 636 documents were manually screened based on titles and abstracts, resulting in 382 relevant articles (i.e. they were relevant to conservation and used interviews-see Appendix S2, Supporting Information). Over half of these articles had been published in the last 5 years alone. Four papers were inaccessible or not accessible in English. We therefore focussed on 227 articles that had been published since 2011 (Appendix S3). Articles were screened against a set of pre-determined criteria (see protocol in Appendix S1). This firstround screening produced an initial set of results, which were then cross-checked for consistency by a co-author (HSM). The checking process consisted of selecting 20 random articles, then testing them against the same criteria to see whether HSM's interpretation resulted in the same input as the other co-authors. Based on slight differences in specific categories, namely sampling technique and coding methods, the protocol was clarified to limit any vagueness of interpretation. The articles were then re-screened to ensure consistency (Appendix S3).

| Where are interviews being conducted?
Based on the review results, use of interviews was reported most often in the US, Brazil, Australia, and western and northern Europe (see Figure 3). Interviews were reported to be used to a lesser extent in certain African countries, India, China, South America, and eastern and southern Europe. No papers in the final selection of 227 were found of interviews being used specifically in the Middle East, Russia and northern Africa.

| Why are interviews being used?
Interviews were used for a variety of different purposes. These can be broadly categorised under three main headings: 1. Gaining ecological and/or socio-economic information on specific conservation issues (58.6%).
There was overlap between some of the categories. For example, understanding knowledge of stakeholders in some cases overlapped with gaining information on a specific issue. However, there is a clear distinction between the first two headings-the focus of the first is extracting specific conservation-related information from stakeholders, rather than understanding the knowledge held by those stakeholders.

| Gaining ecological and/or socio-economic information on specific conservation issues
Based on the review, interviews were primarily used to gain specific ecological and/or socio-economic information on conservation issues.
Issues under this category included interviews aiming to gather information on specific species or habitats of conservation interest (23%), governance (20%), hunting/farming and other extractive uses of biodiversity (7%), and conservation conflicts (4%).  (167). In addition to ecological data, this category also F I G U R E 2 Key data to provide when reporting on interviews includes socio-economic data, for example, understanding household livelihoods of coffee growing cooperative members (290).
Another use of interviews was to gain information on conservation governance aspects (20%). This category included the use of interviews to understand perceptions of conservation governance (e.g. from the perspective of individuals whose livelihoods depend on a protected species, 107), to understand challenges hampering the effective implementation or enforcement of conservation governance (96,149) and to evaluate the governance outcomes of conservation actions (7).
Interviews were also used to better understand governance systems to improve conservation outcomes (e.g. 258) and to better understand the (potential) role of actors within governance structures (e.g. 173, 303).
Finally, interviews were carried out to understand better the impact of or the relationship between human activities and conservation, such as hunting (10, 64) and other extractive resource use (e.g. fisheries, use of specific tree or plant species, 184, forest clearing 269) as well as the conflicts between conservation and other human activities (e.g. 204, 216). This category also included using interviews to understand practices that could impact on species or habitats of conservation interest and vice versa (e.g. 55, 153, 245).

| Understanding knowledge, values, beliefs or decision-making of stakeholders
Three different categories could be identified under this heading.
Over a quarter (26%) of interview papers focussed on understanding the knowledge, viewpoints, values beliefs or decision making of specific groups of stakeholders. While all these papers interviewed stakeholders with an ecological role, or an ecological issue of conservation concern, the focus was not specifically on the conserva-

F I G U R E 3 Global heat map for interview use in conservation decision-making research
A third category explored the differences and similarities in knowledge, views, values and decision making across different types of stakeholder groups (10%). These papers included using interviews to investigate how different stakeholders perceive conservation challenges, e.g. problems related to marine finfish aquaculture (120) and eutrophication (268). Interviews were also used to understand how different groups understand and/or value certain ecosystem services (e.g. 61, 26) or areas providing a range of services (e.g. marine areas, tourism in specific areas, etc.; see 255, 46, 345). Other authors used interviews to explore how and why different groups of stakeholders make decisions relating to conservation issues. Examples here included interviews eliciting stakeholders' mental models in a water management system (75) and exploring the role of knowledge exchange of visual products in terms of influencing decisions (132, 321).

| Strengthening research design and output
A small proportion of papers used interviews to strengthen research design, either using interviews to identify stakeholders and/or design research (4%) or using interviews to validate existing ecological or social data (3%). Examples under the first category included using interviews to identify key stakeholders in preparation for a spatially explicit scenario development process to explore policy implementation options (34), or interviews to inform the design of a quantitative survey to explore the values, beliefs and attitudes of farmers (266). In some cases, interviews were used at all stages of the research: exploratory expert interviews contributing to the specification of the research focus and to the operationalisation of the research design, problem-centred interviews followed by expert interviews with project managers (161). As mentioned earlier, interviews were often used in conjunction with other methods.
However, very few papers (3%) mentioned explicitly the use of interviews to validate existing data, for example, models (349), roleplaying games (136), or to explain quantitative results of previous surveys (160; 317). Figure 2)

| Initial project design
The rationale for using interviews was specifically mentioned in only 27% of papers reviewed. These papers mentioned the usefulness of interviews to gather sensitive (10) or hard to obtain information, and understanding processes such as information-processing and decision making (61). Authors also highlighted the use of interviews to obtain information from a wide range of stakeholders (92), or to build on, validate or complement results gained from other approaches (e.g. questionnaires, focus groups-135, 322). The flexibility afforded by interviews was also mentioned as a rationale for their use.
In 70% of papers reviewed, interviews were not the sole method used. In cases where interviews were used in conjunction with other methods, these included a range of social science methods including Interviews were also used in conjunction with more standard ecological methods including recall or repeat surveys, conceptual modelling and process-based computer modelling, satellite image analysis, inventories, probability and uncertainty assessments, field-based ecological studies and overarching methods such as literature reviews, analysis of secondary data and documents, online discussions.
Of the papers examined, only 11% indicated the use of pilot interviews to refine the interview guide prior to carrying out interviews.
While 4% stated explicitly they had not used pilot interviews, 86% either did not use pilots or did not say whether pilots had been used.

| Data gathering
Of the papers reviewed, 17% did not specify their chosen sampling strategy. Of those that did, the most popular sampling strategy was through key informants (46%), followed by snowball sampling (23%), representative approach (13%) and theoretical sampling ( For the review, we assumed that ethical approval had been con- Of the papers reviewed, the majority (>60%) were carried out face-to-face, with a much lesser percentage being carried out over the phone (<10%) and over the Internet (<5%). A third of papers did not mention how interviews were carried out.
Interview length was reported in 90% of papers reviewed.
Interviews ranged from 3 min to 5 hr in duration, with a mean of 64 min per interview.

| Analysis and write-up
A total of 25% of papers stated they transcribed interviews verbatim, while 66% of papers did not mention their method of transcription.
In terms of coding, whilst it was not always clear from the papers, it appeared that 39% of papers did not specify how coding was carried out, 19% used pre-selected codes, 39% coded using grounded theory (the themes emerged out of the raw data itself) and 3% did not code. Interview or topic guides were provided in only 18% of the papers reviewed.
Only 14% of the papers reviewed highlighted a critical evaluation of advantages of using interviews. Of these, authors mentioned the benefits in terms of providing high quality data on complex problems and issues. These include processes such as decision making, preferences and perceptions. Other advantages included validating or explaining existing (mainly qualitative) data and improving the design of research processes. In papers in which advantages were outlined, authors referred to the practical nature of interviews, being a method that was flexible (283), less time consuming (for researcher and researched) than participatory methods (318), relatively inexpensive and rapid in comparison to other methods (e.g. methods for detecting rare species, 19), and an effective and accurate way to obtain detailed information (265). Finally, some authors mentioned that interviews had allowed for relationship and trust building.
Only 14% of papers reviewed mentioned the disadvantages of using interviews. The main disadvantage, highlighted in 50% of the papers in which disadvantages were mentioned, was bias in terms of sampling (e.g. 340, 341), and interviewer and interviewee bias (e.g. 125, 265). In 30% of the papers that mentioned disadvantages, authors described the inability of interviews to produce the data required. Some authors (13%) highlighted that interviews had not allowed for generalisations, either statistical (197) A limited number of authors (14%) recommended follow-up of interviews. Suggestions included the addition of a protocol to assess respondents' learning as a result of the interviews. Another suggestion was the use of the "member checking" technique that involves the lead researcher meeting with a number of interviewees multiple times following the primary interviews to share interview transcripts and clarify uncertainties in recording and transcription. The authors argue that these interactions serve to both increase the researcher's understanding of participant's perspectives and to increase rapport between researchers and study communities (87).

| The use and reporting of interviews in conservation
This review confirms that interviews are widely used in conservation research, although it suggests a disparity in where interviews are used. Based on the review, we found no evidence of interview use in many parts of the world. We further note that providing data at a country level may mask regional variation within countries. The geographical bias found in our review may be influenced by our search being carried out in English (see Amano, Gonzalez-Varo, & Sutherland, 2016), and focussing on academic rather than grey literature. It may (partially) also reflect the geographical distribution of conservation research more generally.
Interviews were characterised as a flexible method, useful in generating high quality data on complex problems and issues, including processes of information-processing and decision-making. A significant proportion (c. 70%) of the papers used interviews alongside other methodologies, which is useful since a multimethod approach enables data triangulation and can limit the bias associated with any one single method. Overall, however, there were relatively few cases where the main rationale for using interviews was stated (27%). It was not clear why the researchers had chosen to use interviews, which makes it difficult for the reader to judge whether or not it was the most suitable method.
Furthermore, if the main rationale for using interviews is not widely outlined, then it makes it more difficult for other researchers to ascertain whether it is a suitable method to answer their own research question(s).
Perhaps the most important finding of this review is the limited information provided in papers on how interviews were used and analysed. Good examples providing clear methodologies were the exception (e.g. 204-see Zappes et al., 2014), and it was often impossible to ascertain exactly how the researcher/s had applied interviews in their study. This is an important issue in the future application of this methodology in conservation decision making if the robustness and credibility of process and outputs of interviews cannot be ascertained.
In part, the lack of care taken to outline methodologies in full when using interviews may be the result of the review process. Based on the lack of reporting of interview data in the review papers, reviewers would rarely seem to insist on a robust adherence to a checklist of how interviews were carried out and analysed (see Figure 2). Rather, they may actually suggest that detailed information in a methods section is superfluous. Thus, reviewers may allow a methods section to pass without, analysis further suggests that reporting on interviewing is not comprehensive. The latter factor is of particular concern since the conclusions generated from each study are being based on interview data; if the researcher is not clear about the analysis process used, then it is difficult to ascertain whether the conclusions are evidence-based.

| Suggestions to strengthen the interview methodology for future applications in conservation
Researchers should consider whether they have the skills and resources to design, conduct and analyse interviews carefully, so that key stages are carefully considered in the interview process. We suggest the checklist illustrated in Figure 2 could be a useful guide for researchers, reviewers and journal editors, about what data to include in scientific publications. Researchers could use this checklist to guide decision making and subsequent reporting for each key stage.
For data gathering, to ensure robustness and credibility, the sampling of interviewees must be justified. When selecting a sample, therefore, researchers must be able to show that they have consulted the full range of views needed to answer the questions adequately.
Greater transparency is needed with the inclusion of information such as the reason for inclusion and response rate. It is insufficient, for example, to describe a snowballing method without including the reason why each additional participant was recommended and targeted.
Researchers also need to discuss ethical considerations, the method of carrying out interviews, and the process of analysis and write-up, including providing feedback to participants. Supplementary material may be used to keep within word count restrictions.
A simple way of encouraging researchers to provide a clear stepby-step guide in the Methods section, or supplementary material, is for editors and peer reviewers to ensure that providing the data listed in the checklist ( Figure 2) is a prerequisite for publication. This is information that researchers are expected to collect during interviews. Such information would allow peers and decision-makers to ascertain the credibility and robustness of the interview methodology and its conclusions.
Journals will need to provide clear guidance to submitting authors about these requirements, including best practice guides on specific considerations. One such area in need of clear guidance from journals is on ethical matters, which may not be familiar to conservation researchers. Here, we suggest that journals follow the recommendations provided by St. John et al. (2016). They argue that all conservation and ecology journals should provide clear ethical guidelines, require an ethics statement and ensure submitted papers on human research are scrutinised with the same rigour as ecological science.

| CONCLUSIONS
This review provides an overview of some of the places and ways in which interviews have been used in research on conservation decision making. While the review highlights a number of concerns, it also provides a basis for recommendations to strengthen the interview methodology for future applications in conservation. These recommendations are not only aimed at researchers using the interview methodology, but also at researchers working in conservation research, journal editors, reviewers and decision-makers using information from peer-reviewed papers. Interviews can be a very useful method in conservation research, allowing for in-depth understanding of processes and issues, often based on a small sample size. The usefulness and credibility of this methodology, however, would benefit from a more strategic approach, as described in this paper, including better justification for its use over and/or alongside other methodologies, and more detail in terms of how interviews are undertaken and interpreted.