Which of the following is a characteristics of a typical unstructured interview?

1.2 Levels of Structure

There are two broad types of employment interviews: structured and unstructured. Structured interviews have a fixed format or a fixed set of questions, and the questions are based on a job analysis. Conversely, unstructured interviews do not use a standard procedure for the administration of questions or scoring. In fact, questions may vary from applicant to applicant.

In 1992, Huffcutt reviewed the literature on interview structure and described four progressive levels of structure. Level 1 was the typical unstructured interview characterized by no constraints on the questions and a global evaluation of responses. Level 2 imposed limited constraints by specifying the topics to be covered by the questions and some degree of structure on response evaluation. Level 3 required the prespecification of questions, although applicants were not asked precisely the same questions because different interview forms were used or interviewers were allowed to choose among alternative questions and to probe responses to the specified questions. An example of this kind of interview is the behavioral patterned description interview as described by Janz in 1982. Responses were evaluated using some degree of structured response evaluation. Level 4 involved asking applicants precisely the same questions with no deviation or follow-up probes, and responses were scored according to benchmark answers. In 1980, an example was provided by Latham and colleagues, who discussed situational interviews.

Situational and past behavior are two primary types of structured interviews. Situational questions provide a situation to the interviewee and ask how he or she would likely perform in that situation (e.g., “What would you do if …?”). The situations resemble those in situational judgment tests (SJTs). Past behavior questions provide a broad situation for the interviewee. The interviewee is asked to describe his or her performance in a similar, but more specific, situation (e.g., “Describe a time when you had to give bad news to a client. What was the situation? What did you do? What was the result?”). Both of these kinds of questions derive from critical incidents provided by subject matter experts (SMEs). Critical incidents are structured stories about good and poor job performance. The three-part structure of critical incidents involves having the SMEs respond to three questions: “What was the situation/task?,” “What did you do?,” and “What was the result?”

Interviews can be very appealing to organizations because questions can be created to tap cognitive and noncognitive skills as well as abilities. Interviews can focus on a range of applicant qualities such as communication skills, interpersonal skills, decision making, and substantive knowledge. Examples of situational and past behavior interviews are provided in Tables I and II, respectively.

TABLE I. Situational Interview Example

TABLE II. Past Behavior Interview Example

I.A. Knowledge Acquisition Processes

KA is mentioned as a problem within fields as diverse as machine learning, natural language processing, and expert systems research. However, the most typical KA scenario is where a knowledge engineer engages in an unstructured interview with one or more human experts. After these initial sessions, the engineer may use the results of these sessions to develop tools for structuring more in-depth KA sessions, such as focused lists of questions, simulations, or even expert system prototypes to use as the basis for later sessions. Researchers have used a systems analytic approach to modeling the KA process and have arrived at a variety of models that are generally in agreement. The KA process is typically described as being iterative with loops that depend upon the size of the system to be built, the depth and breadth of the task to be supported, and the quality of the knowledge as it is acquired. Steps in the process are identified as:

Conduct an initial unstructured interview.

Knowledge engineer analyzes results of interview.

Knowledge engineer represents the resulting knowledge.

Experts test, and their comments on the system performance are recorded.

If the knowledge base is complete, then stop.

Acquire the missing knowledge from an expert.

Represent the missing knowledge in the knowledge base.

Return to step 4 until finished.

In reality, KA often involves a number of different KA techniques used in multiple KA sessions or episodes with multiple knowledge sources including texts and various human experts. A classic example of how complicated KA can be in reality is the case of how the ExperTAX system was developed by Coopers & Lybrand. This classic case recorded by Shpilberg et al. describes an expert system that was designed to help guide staff accountants out in the field collecting tax information. This case is somewhat unique even today in that it details the overall activities involved in development and is not limited to illustrating specific KA techniques used or novel features of the system.

As is usually the case, the development of ExperTAX began with an unspecified number of unstructured exchanges with multiple experts (20 experts participated overall). After this, the knowledge engineering team decided to set up a simulated tax information gathering case with three different accountants playing roles. The team videotaped this session and several subsequent variations on it. Other interactive KA sessions were held using a computer to help in the design of the user interface. After these different sessions, a prototype of Exper TAX was developed. This then served as the basis of all the KA sessions that followed with individual experts, leading to significant additions. The whole development process spanned about 2 years and almost 6 person-year's labor. This particular case describes in unusual detail how varied KA can be in reality with respect to the combinations of KA techniques and participants at each stage of the project. Because KA is such a mix of activities, in actual practice it has been difficult for researchers to evaluate how these activities may be best implemented.

Triangulation

There are two major types of triangulation: methods and sources. Methods triangulation is the checking of the consistency of findings by using different data-collection methods. The methods may be all qualitative, for example unstructured interviews and observation, or all quantitative, for example questionnaires and structured interviews, or a mixture of the two. Sources triangulation, on the other hand is the cross-checking for consistency of the information derived at different times and from different people, for example managers and para-professionals in a library situation.

The advantage of triangulation is that conclusions are likely to be more reliable if data are collected by more than one method and from the perspective of more than one source. If different methods are used, the researcher can take advantage of the strengths and offset the weaknesses of each. Both the broad and narrowly focused (or macro- and micro-level perspectives) can be combined in a single project (Gorman and Clayton 1997, p. 32).

Sociological work on identity

Early social psychological work aimed at tapping ‘social identification’ used qualitative methods to conceptualise connections between the topic guides administered to homeless participants and the structures used to guide the analyses. For example, Farrington and Robinson (1999) conducted unstructured interviews with 21 homeless people from a single sheltered accommodation service near Bristol, United Kingdom. On the basis of their qualitative analysis, they concluded that the role of social identification, both as a part of a highly stigmatised social group of ‘homeless people’, and with particular groups, diminished the longer a person had been homeless. Likewise, Osborne (2002) explored more direct measures of social identity amongst 97 homeless people in Austin, Texas (United States). Using self-report data and no objective measure of behaviour (i.e., no method to verify attendance), Osborne found that homeless people who identified highly with a ‘homelessness identity’ were less likely to use services but reported higher self-esteem scores. He suggested that this, as well as his prior research, supports the notion that homeless people actively seek to protect their personal identities – that is, identity as an individual person – by not using services or associating with other homeless people. Although on the surface this might explain the absence of service-seeking behaviour, there may also be a more complete explanation to the way in which homeless people view other homeless people and the stability of their friendships, particularly as a motivating factor affecting service use.

III.D.2.b. Interviews

Interviews are, by far, the most commonly used approach to knowledge acquisition. In an examination of U.K.-based expert systems, Doukidis and Paul found that of 166 expert systems, 100 relied on interviews as the main source of expertise. Unstructured interviews are usually more exhaustive than structured interviews. However, such interviews may deviate the KE and the expert from the main task, thereby making the interpretation of transcripts challenging. Structured interviews, on the other hand, can reduce the time spent relative to unstructured interviews. Furthermore, if the KE is well trained, structured interviews can be more effective in eliciting expert knowledge. The most common risk of such an interviewing technique is that it is not possible to probe an expert's reasoning process in depth. A combination of structured and unstructured interviews could be more productive since the KE can begin the interviews with a set of predefined questions but may build sufficient flexibility into the interviews to follow the expert's reasoning process to closure.

Interviews

Interviews are one method of assessing social validity that can provide a large amount of information in a short period of time. While information from interviews may be plentiful, it may also not be highly relevant toward determining the social validity of a treatment. An unstructured interview may lead to discussions that are not directly related to the assessment of social validity unless the interviewer is highly skilled at redirecting the person being interviewed. Another interview option may be a structured or semistructured interview which delineates questions relevant to specific areas of social validity. Gresham and Lopez (1996) provided an example of a semistructured social validity interview which allocated questions to three areas of social validity. See Table 3.2 for the semistructured interview created by Gresham and Lopez. The interview consists of 21 base questions with additional follow-up questions. The first nine questions focus the significance of treatment goals, the next six questions relate to acceptance of treatment, and the last six questions are concerned with the importance of treatment effects. Their interview may be a useful approach to gathering information on social validity, but extended applications of the usefulness of the interview have not been disseminated.

Table 3.2. Semistructured interview for social validation.

Source: Gresham, F. M., & Lopez, M. F. (1996). Social validation: A unifying construct for school-based consultation research and practice. School Psychology Quarterly, 11, 204–227. Reprinted with permission.

Lane (1997) offered another example of a semistructured interview to assess social validity that focused on perspectives of students’ targeted for treatment of academic difficulties titled the Children’s Social Validity Interview (CSVI). The CSVI consists of Yes/No questions, Likert-type rating scale responses, and open-ended items. The CSVI is interesting in that it asks the child to evaluate how often they use the intervention in various settings. This touches on another method for assessing social validity which focuses on how a treatment is used by consumers. Recurring use of a treatment is an indicator of social validity (Gresham & Lopez, 1996; Kennedy, 2002a). Determining the recurrent use of treatment may be assessed by asking consumers or via direct observation of the use of a treatment. Treatments that are used more often may be considered to have higher social validity than treatments that are available but used less frequently. This aspect of measuring social validity may be a useful indicator but should be evaluated within the larger scheme of social validity. Some treatments that are frequently used may also have limited acceptability such as punishment procedures. Punishment procedures may be frequently used but may be considered highly unacceptable by others.

Some information collected within functional behavior assessment questionnaires and interviews can provide evidence of social validity, and it might be beneficial for future research to consider correlating these types of assessments with measures of social validity. An example of a functional assessment interview that provides a significant amount of information that could be indicative of social validity is the Student-Assisted Functional Assessment Interview (Kern, Dunlap, Clarke, & Childs, 1994). This interview consists of four sections of items that deal with identifying and evaluating aspects of a classroom that a student likes or dislikes (see Table 3.3). The first section includes a three-point Likert-type scale that concerns some broad features of a classroom environment as well as some more specific evaluations of such things as availability of rewards. The second section consists of some fill in the blank questions and some open-ended items that focus on classroom problems and personal interests of the student. The third section consists of a five-point Likert-type scale that asks the student to rate how well they like a number of academic subjects. The fourth section directly relates to section three by asking follow-up questions concerning specifically what the student likes or dislikes about a certain academic subject. This functional assessment interview seeks to obtain a great deal of information from a student about their personal preferences in general and as they relate to specific academic subjects. The information that is obtained could be useful toward identifying certain aspects of treatment that may prove effective, but possibly even more importantly the interview may allow for some of the likes and dislikes of the student to be incorporated into a treatment. By including definite preferences of the student into the treatment, it may be possible to increase the social validity of the treatment components as evaluated by the student. Obtaining similar information from treatment mediators may also prove beneficial toward developing treatment procedures with high acceptability.

Table 3.3. Student-assisted functional assessment interview.

Source: From Kern, L., Dunlap, G., Clarke, S., & Childs, K. E. (1995). Student-assisted functional assessment interview. Diagnostique, 19, 29–39. Reprinted with permission.

Types of interviews

Structured (standardised or scheduled) – with these each respondent is asked exactly the same questions. The sequence of the questions is also fixed. Unless there is some freedom for respondents to express their own views, unconstrained by the researcher’s agenda, this is simply a survey questionnaire administered by interview.

Structured interviews are used where it is important to be able to compare results across respondents: for a survey, for a number of respondents within a case study, or for multi-site case studies. There can be the flexibility for individual expression within a more formal structure.

It is important to note that if a structured interview schedule is used, it must be tested in the same way as a self-administered questionnaire.

Unstructured (non-standardised, non-scheduled, or in-depth) – with this approach each answer basically generates the next question. This type of interview is useful for exploring a subject or for gaining insights into people (the kind of exploratory interview, mentioned above). Unstructured interviews are sometimes used in case studies to collect extensive data from key people. These are the interviews which are seen as wholly acceptable in inteipretivist research. They are also often used before compiling a structured interview schedule or a self-administered questionnaire.

Semi-structured – these interviews have a standard list of questions, but allow the interviewer to follow up on leads provided by participants for each of the questions involved. The semi-structured interview is closer to the unstructured, in-depth interview, than to the structured, standardised form.

The purpose of in-depth interviewing is basically ‘… to capture the respondent’s perspective on a situation or event under study’ (Mellon 1990, p. 55). This is in keeping with the central precepts of interpretivism. As Slater (1990, p. 114) says, ‘… the respondent is allowed, in fact, encouraged, to talk expansively on the main subject, raising topics within it in any order s/he wishes’.

Usually in-depth interviews are tape recorded (with the permission of the interviewee). Slater (1990, p. 114) says that ‘the more fully and accurately any interview can be recorded in the respondent’s own words, the better analysis and conclusions will be’. Tape recording of interviews also avoids the disruption to interviews which results from the need to take notes.

Both structured and unstructured interviews have been used extensively in both information management and information systems. Issues concerned with the introduction of new technology into libraries and with information seeking behaviour are only two of the many research areas in which interviews have been used in information management. Both unstructured and structured interviews are often used during information systems development to gather information about potential users’ requirements for a proposed system. Unstructured interviews allow users to articulate their specific concerns and needs while structured interviews are useful for detailed or specific information about current operations and future requirements. Unstructured interviews were used extensively for requirements gathering in a project to develop a module of an EIS (Executive Information System) to monitor and report on the financial and physical performance of the projects undertaken by a government authority. Senior and middle-level executives were asked about their program monitoring and reporting needs in a relatively informal interview process. It was felt that executives would lose interest in the project if obliged to participate in a formal process, that they would not have time to attend formal meetings, and that the overheads involved in organising committees and formal meetings would slow down the process and lead to a loss of interest in the project.

Semi-structured and unstructured interviews

Although a scheduled list of questions, and often prompts as well, are used in semi-structured interviews, they are usually flexibly administered in order to capture the perspectives of participants as far as possible while ensuring that interviewees focus on issues relevant to the study. A key characteristic of this kind of interview is the opportunity offered to follow up on leads offered by participants. Unstructured interviews provide the same opportunity but without a scheduled list of questions. Each participant answer basically generates the next question.

An unstructured interview would be appropriate in a phenomenological study of one person’s experience of phenomena (see Chapter 1: Research concepts). It might also be used in a case study to collect extensive data from key informants. Patton (2015) saw unstructured interviews as occurring naturally as part of participant observation. An unstructured interview could also be useful when a systems developer cannot predict user reactions to a prototype of a new design feature, and has no preconceptions about its effect. The developer may want to elicit as many free-flowing thoughts as possible about a first-time contact with the feature, in order to incorporate fresh ideas into the system. Another purpose could be to gain an understanding of a topic, in-depth, before compiling a structured or semi-structured interview schedule or self-administered questionnaire.

The semi-structured interview has more in common with the unstructured, in-depth interview, than with the standardised, structured interview. Indeed, semi-structured interviews can also be regarded as being ‘in-depth’. Some of the literature does not make a difference between the two, referring simply to ‘qualitative interviewing’ (e.g., Baker, 1999; Brinkmann and Kvale 2015). Baker (1999, p. 247; citing Schwandt, 1997) postulated that “in a qualitative interview, the interview is constructed more as a discourse between two or more people which is not so fully controlled by the interviewer’s questions but is in fact constructed by both the interviewer and the respondent over the course of the discussion”. This resonates with Charmaz’s (2003) opinion that “the viewer creates the data and ensuing analysis through interaction with the viewed” (p. 271). It means that interviewers will almost certainly receive multiple and often contradictory views from interviewees, as is consistent with key precepts of the interpretivist approach. Williamson and Kingsford Smith (2010) spoke of looking for ‘consensus and dissonance’ in their interview data.

One of the advantages of semi-structured and unstructured interviews is the opportunity they provide to support research findings with direct quotations from participants. It is therefore important to record both semi-structured and unstructured interviews so that researchers can support their findings by correctly reporting what participants have said. This means that the “… audience is at least partially able to project themselves into the point of view of the people depicted” (Patton, 2015, p. 33, citing Lofland, 1971, p. 4). However, recording must be done only with the permission of the interviewee. Before the interview begins, informed consent needs to be obtained from the interviewee and permission to record the interviewee can be obtained as part of this. (See Chapter 20: Ethical research practices, for further discussion of informed consent.)

For the kinds of research I have undertaken in the field of LIS, I have found that the semi-structured interview has worked very well. I have often started with unstructured interviews with key informants, as a way of understanding the topic, particularly if my partners and I have insufficient knowledge of the topic involved. For example, before we undertook a study with the Australian Plants Society Victoria, about data and information management and sharing in environmental voluntary groups (Kennan, Williamson & Johanson, 2012), we conducted two very long interviews with key members of the organisation to gain understanding of the types of questions and prompts we should use with other participants.

Chapter 2: The fundamentals of research planning discusses an ARC Discovery project (Williamson & Kingsford Smith, 2010) in which Kingsford Smith and I investigated, inter alia, the role of information in online investment. Box 16.1 returns to that project and provides a list of questions and prompts that Kingsford Smith and I used to explore one of the research questions outlined in Chapter 2.

Box 16.1

Example of the use of questions and prompts in a semi-structured interview: Information seeking by online investors

As indicated in Chapter 2, the online investment project used mixed methods. The initial online survey was followed by the qualitative (interpretive) component, which provided the theoretical drive (Morse 2003). The theoretical drive is determined by the component (usually positivist/quantitative or interpretivist/qualitative) which will play the more important part in answering the research questions. Because the interviewees were purposively selected from the almost 200 survey respondents who offered to be interviewed (from a sample of 520), we did not need to collect demographic data; it had already been provided in the questionnaire responses. Had we not had this survey data, we would have used a demographic, self-administered questionnaire at the beginning of the interview. Unless your interviewees are very old, very young, or disabled, a good way to collect demographic data about interviewees is by self-administered questionnaire at the beginning of an interview or focus group. This approach saves time and affords privacy to interviewees.

While there is not the same need for consistency in questioning for in-depth interviewing as there is for structured interviewing, interviewers for all types of interviews should be “flexible, objective, empathetic, persuasive, a good listener” (Fontana and Frey, 1998, p. 55). It is mandatory for interviewers to seek the views of participants, rather than to express their own. High levels of concentration are required so that the most appropriate follow-up questions can be asked and so that all required areas are covered in the level of depth necessary to answer the research questions. Analysis of structured interviews will mostly be similar to that of self-administered questionnaires, discussed in Chapter 18. Analysis of semi-structured and unstructured interviews is discussed in Chapter 19: Qualitative data analysis.

This article focuses on the usefulness and reliability of the case study approach in both research and teaching in human geography. Examples of some of the ways that local case studies can be used to help inform larger questions, issues, and theories are provided, along with a discussion of selected mixed methods used by case study practitioners in the field. Triangulation of multiple methods such as structured and unstructured interviews, participant observation, field observation, textual analysis, survey questionnaires, and spatial analysis are discussed and encouraged within the context of a few selected perspectives used by researchers.

Particular emphasis is placed on the effectiveness of the case study approach in the teaching of human geography. The case of the recently approved US advanced placement human geography course is discussed to provide an example of a classroom context rich in opportunities for using the case study approach to create a more student-centered, issues-based, and collaborative classroom. It is argued here that the use of case studies as discussion points for comparison and critique make especially effective center points for conversion of traditional classrooms into sites of interactive learning and teaching.

This article concludes with a brief discussion of some of the limitations of using the case study approach in research and teaching human geography with a few suggested solutions for helping mitigate these challenges.

Methods of Collecting/Generating Qualitative Data