Peter Chen and S.M. Hinton (1999) 'Realtime Interviewing Using the World Wide Web'

Sociological Research Online, vol. 4, no. 3, <http://www.socresonline.org.uk/4/3/chen.html>

Abstract

This paper outlines the adaptation of in-depth interviewing using World Wide Web-based interviewing software between the interviewer and their subject. Through a structured, realtime interviewing process the researcher is able to use the Internet to facilitate communication, recording interviews directly to a file without incurring the costs associated with traditional face to face or telephone interviews. The benefits of this approach are the ability of the researcher to conduct inexpensive interviewing over distances and elimination of transcription costs from the research process, allowing the researcher to undertake a wider range of interviews than may be possible on a limited budget. The interview method has problems associated with the depth of material available from this approach, the loss of paralinguistic cues and the limited size of the available sample, limitations that must be accounted for by any researcher considering using the approach.

Introduction

1.1 One advantage of the convergence of computers and communications is its value to social researchers. With the proliferation of applications associated with the World Wide Web, many students, researchers and academics have begun to embrace communications technology. Electronic mail allows people to communicate quickly and easily regardless of geographic location; academics are increasingly present on the World Wide Web via homepages and through the publication of papers, conference notes and academic data; library catalogues and electronic journals are increasing the availability of research materials available on the Internet^[2]; and an increasing number of studies have directly featured parts of the Internet as a source of data and a medium for research^[3].

1.2 This paper's aim is to introduce a method for realtime interviewing using standard World Wide Web browsing software. This approach was developed in the Department of Political Science at the Australian National University for the study of closed, finite groups of policy makers and political participants (policy "elites" and advocacy coalition members). The reason for developing the method was to meet the requirements for inexpensive interviewing, especially where distance or availability made the cost of undertaking a traditional face to face or telephone interview prohibitive. The method was tested in a small pilot study undertaken at the Australian National University, before subsequent use as a secondary research methodology for ongoing PhD research. This paper has been designed to summarise the method by introducing the interview process and presenting an overview the technical requirements for converting this to an online format. The Hypertext MarkUp Language (HTML) source code of the method, as well as technical explanatory notes are included for other researchers to use and adapt for their own disciplines. As this type of application has had little exposure and critical assessments to date, an overview of similar examples is presented, and additional analysis of the method's advantages and limitations are presented. Additional suggestions are provided for the extension of the method beyond the basic HTML version provided in this paper, such as the use of data management software packages to integrate the research method into other automated data analysis that may be used by the researcher.

Interviewing Using the World Wide Web

2.1 The method presented here is essentially the facilitation of traditional face to face or telephone interviewing using infrastructure provided by the Internet. The essential characteristics of this method are:

• it is conducted in realtime, attempting to capture the spontaneity of traditional interviews;
• it uses existing software in a novel manner, rather than requiring the acquisition or development of new programs;
• it places the technical burden on the researcher, reducing effort required to participate on the interviewee; and,
• the method is essentially an application of an existing technique, rather than a wholly new research approach.

2.2 To examine the online interviewing method we will quickly define traditional interviewing, outline the practical application of the approach and its technical basis, and review examples of computer facilitated research.

Traditional In-Depth Interviewing Defined

3.1 In-depth interviewing is defined by Neuman (1994:246) as "...a social relationship ... a short-term, secondary social interaction between two strangers with the explicit purpose of one person obtaining specific information from the other". Interviewing is essentially a qualitative approach, where the researcher becomes the instrument of the research method. Interviewing, therefore, is a highly personal process where meanings are created through personal interaction. This method is commonly used in political science where a researcher selects a sample of "experts" who can provide information, explanations of observed phenomena, background and contextual material. The scope of these interviews depend on the needs of the research project and the availability of other methods. The value of the interview method is directly related to the type of data that is commonly sought by the social researcher (McCracken, 1985).

3.2 Holstein and Gubrium (1995) argue that interviewing is the key data collection instrument of the social sciences, with ninety per cent of social research utilising the technique in some capacity. Where quantitative research is unhelpful or depth required, the in-depth interview becomes one of a small range of tools for available to the researcher. The method has wide application within the social sciences, and is adapted and modified for use in many studies (from exploratory work to research based on interviews as the primary research tool). Interviewing provides the researcher with flexibility in data gathering, the ability to adjust the research tool to meet their needs and to probe areas be of concern or that may arise during a discussion. Because of the time required to conduct this form of research (establishment costs, travel to and from the interview, equipment, the time required to conduct each interview, transcription of recordings or notes) it is regarded as one of the most expensive forms of survey research available to social scientists, a factor that often limits its application.

The Online Interview

4.1 Online interviewing requires the researcher have access to frame-capable browser software and space on an Internet web server that supports PHP server-side scripting and an interviewee with a similar browser with access to the World Wide Web. Using a webpage as an interviewing "screen" between the interviewer and their subject, the interviewer is able to question the subject in realtime, logging the discussion to a file that serves as a permanent transcript of the interview. Figure 1 illustrates the online interview: two computers linked through the Internet allowing the interviewer and interviewee to communicate through entered text that is sent to a central page (shown in the Message Frame). To communicate one person enters their text and clicks the send button, their message is updated on the webpage displayed in the top of the screen^[4] (and stored in the interview transcript), the other is then free to respond^[5].

Figure 1: The Interviewer and Interviewees' Screens

4.2 A number of screenshots of the software in use are provided:

• Figure 2: The interviewer's screen during the interview
  
• Figure 3: The interviewee's screen during the interview
  
• Figure 4: The interviewee's screen after the conclusion of the interview (inactive)

4.3 The method requires the interview be conducted in rounds (where one participant enters a message and the other responds), and is more ordered than the freeform discussion of a traditional interview. What this establishes is a buffer between interviewer and interviewee, breaking the flow of the conversation into text "chunks" (sentences, paragraphs) and producing a time-lag between either the interviewer or subject being able to see the next statement by the other (as text is transmitted in one block, rather than scrolling onto the screen as the other types). As neither is able to directly see the other person, all non-verbal communication is lost and the method lacks the ability of the researcher to conduct observation based research during the course of the interview^[6].

4.4 A number of criteria should be applied when assessing if this method is useful or appropriate for the research project:

• Are interviewees able to access the technology required?^[7]
• Are interviewees amenable to the use of the technology, as opposed to an alternative method?
• Will the interview examine non-controversial material that does not require a private, or secure form of communication?
• Is the status of the interviewee such that the use of an impersonal interviewing method would not be offensive?
• Does the interviewer have the skills required to use the technology?
• Is observation of the interviewee not important for the analysis of data collected?

4.5 As this approach includes the capacity to automatically transcribe the interview the limitations outlined by Hammer and Wildavsky (1989:70-1) regarding the recording of interviews do not apply (the cost of transcription^[8], background noise, unreliable audio equipment, etc.). Certainly their concerns regarding agreement of the interviewee to recording the interview remain valid, however their proposed method to encourage acceptance of recording (to explain why the record is being made, rather than simply asking if it is acceptable) remains valuable for obtaining consent to transcript the interview^[9]. For interviewees suspicious of surveillance this method is less likely to appeal to them, especially if they lack an understanding of the operations of the Internet and fear that their messages may be recorded covertly (either by the interviewer or by a third party^[10]).

4.6 Given the graphical nature of the World Wide Web it is possible for the researcher to tailor the pages used in the interview to meet the needs of their target audience (such as the inclusion of animated graphics, background images, fonts, coloured text, etc.). This may be important when interviewing "non-professional" interviewees (such as youths) or other groups amenable to creatively presented web pages. While tailoring the interview site can allow the researcher to establish better rapport with their target group, more complex the pages (such as those with video or audio clips, interactive maps, Java programming script, or the host of additional available plug-ins^[11]) the more likely the sample size will be reduced because of lack of compatib le software of the interviewee.

4.7 Given the selection of the online interview as a possible method for the research, it is important to consider the debate regarding the use of computer technology in qualitative research and analysis. Kelle (1997) summarises the argument as a conflict between difference and orthodoxy (imposed by computers in research), where the use of computers can be seen to impose a methodological approach on the researcher; defining the research project and the data^[12]. This goes beyond technological distrust to wider concerns about context and the ability to interpret textual data apart from its contextual setting. What can not be ignored, however, is the relationship between research method and context. The availability of technology and scarce resources are a fundamental factor in the consideration of all researchers' methods. Certainly consideration of the technological facilitator for research (be it a telephone, a video camera, or the World Wide Web) needs to be explicitly examined by the researcher prior to the adoption of the method for their study, but this is never placed outside of the context of the researchers felt involvement in the research - their desire to be able to collect desired data ^[13].

A Technical Overview of Online Interviewing

5.1 The online interview described utilises HTML 2.0, PHP/FI (a server-side scripting language) and browser software compatible with HTML 2.0 (this includes versions of Netscape from 2.0 and upwards). HTML 2.0 is widely recognised, as an older format still valid to newer browsing software. The technical requirements that led to the selection of this format are its capacity to produce frames^[14] and the ability of the browser to automatically re-load the message frame at pre-determined intervals. The reasons for selecting this technology (as opposed to other "chat" software^[15]) were the relative ease of use of web browsers, the availability of the software (which is freely available and increasingly commonly provided as standard with computer purchases), and the flexibility provided by HTML and PHP/FI scripting. PHP/FI is a web server based scripting language for data interpretation and manipulation that is increasingly becoming supported by web servers, most notably by the highly popular Apache web server (which is distributed with the capability to interpret PHP scripting). Given these requirements, much of the difficulty of implementing the interview system is placed back on the software and the server, thus reducing the technical requirements of the interviewee's computer.

5.2 One of the major issues involved with creating HTML-based realtime "chat" programs is that the Hypertext Transfer Protocol (HTTP) is a "stateless" protocol. This means that when a user connects to a web page with their browser, the server responds by opening a connection to the requesting computer, transferring the requested information (for example, a web page or graphic image) and then terminating the connection. To draw an analogy with a phone call, this method is like ringing someone, asking them a question and hanging up. The person then rings you back, gives you the answer, and hangs up again. The conversations between caller and callee would then continue in this manner. While this method seems inefficient, it maximises the ability of the server to handle many incoming connections, since it does not need to keep multiple lines open at once, which is a major concern for a web server than may be receiving thousands or millions of requests for information a day. For basic browsing and access of online materials such as pictures and other documents, this method is very effective. However, if one wishes to use HTTP to facilitate communication between two users, the statelessnature of the protocol makes communication difficult^[16].

The HTML Source Code Explained

6.1 The HTML code required for the online interview is available from this page, including instructions about the server requirements, files and changes needed to successfully use the code.

6.2 To describe how the interview code works, it is necessary to understand that data (such as a line of dialogue) can be sent from a HTML web page (which the user sees through their web browser) to a processing program running on a remote server. In the case of the online interviewing system, there are two separate HTML web pages, which transfer data from the user to the processing program - one for the interviewer and one for the interviewee. These can be thought of as "control pages".

6.3 When a user clicks a send button on a control page, all the data in all the fields on the control page is sent to the processing program. The program then makes a decision about how to process the information, based on which button was clicked, the information returned and which user submitted the information (the interviewer or the interviewee).

6.4 By using frames, two web pages can be shown on one screen at once. The conversation log, which appears in the uppermost part of the user's screen shows the most recent exchanges of dialogue, while the control page, displayed in the lower part of the screen simply feeds user input to the processing program. Because of the stateless nature of HTTP the conversation log page must be constantly reloaded to display the most recent discussion. This is accomplished by taking advantage of a feature called "client-pull"^[17], which causes the browser to reload the page after a certain number of seconds have elapsed. This interval can be modified by the interviewer.

6.5 In addition to the dialogue extra information is needed by the processing program. For example, the processing program must know when an interview has started, and when it has finished; it must know the names of the participants; and, know what reload interval to set the interview log page refresh rate to. This information is stored in a configuration file which is only accessed by the processing program. The file contains basic data about the interview, and can be modified by the interviewer through extra form items available on the interviewer's control page. Each time the conversation log page reloads, a script is run which reads the configuration file to check if the values have changed, which allows the reload interval to be changed dynamically, and for the interview to be terminated.

The Pilot Study and Subsequent Use

7.1 During the development of the HTML code, a small-scale study of the method was undertaken with a group of students. These students were interviewed on campus using the method, and a follow-up self-reporting questionnaire distributed. The group also participated in a small focus group, where the use of the computer was discussed. Overall, the group provided useful information about the computer interface and their preferences for and against the method. What the group demonstrated was that, given clear explanation about why the research was being undertaken using the computer and how to use the HTML interface, the interviewees found the method acceptable, if not preferable to other methods. In subsequent use of the method as part of ongoing PhD research, the method has been used for interviewees located in overseas countries.

Assessing the Value on Online Research: Three Examples

8.1 Because of the problem of cost and distance, in-depth interviews have been facilitated by conventional communications technology, such as telephones and audio / video conferencing for some time. The use of the Internet for interviewing in realtime, however, remains very limited to date. Two examples can be identified as practical examples of the method: Marc Smiths' (1998) study of the Virtual Community of the WELL (the "Whole Earth 'Lectronic Link"), and the use of online focus groups by commercial researching companies. Additionally, the use of email surveys and interviews needs to be considered, given their technical relationship to the online method.

8.2 Smith, in examining the social structure of the WELL used commonly accessible "chat rooms" to conduct interviews as a supplement for face to face discussions and his primary research method of selecting material from email postings^[18]. The difference between the proposed method and the study undertaken by Smith was the nature of the respondents (all relatively computer-literate users), and the nature of the interviewing technology (HTML webpage prepared by the researcher, rather than the "standard" chat facilities provided by the WELL^[19]). Focusing on the WELL provided Smith with a discrete, definable sampling universe, and one that was skilled (or at the very least, knowledgable of) realtime electronic communication. Smith's method^[20] was suited to areas in which respondents are by nature technically literate, but limits the approach to the study of virtual communities or computer users in general.

8.3 The focus group is a method that remains extremely popular in commercial research and is commonly used in product testing, ideas generation or where broad impressionistic responses are desired from a small group of subjects^[21]. A focus group consists of a small number of subjects (four to twelve) and a moderator who introduces topics of discussion, monitors and feeds back responses, and guides the group through the issues under consideration. Focus groups are more expensive than simple face to face interviewing, with the costs of room hire, transportation, food and other incentives adding to the normal expenses of the interview method^[22].

8.4 Over the last few years a number of companies have begun to offer commercial information (especially marketing intelligence) based on the concept of an "online focus group" (Strategic Focus Inc., Insights Online, CLT Research Associates, Common Knowledge, Cyberdialoge, Greenfield Online, and the Xerox Business Research Group). The online focus group is facilitated using the Internet (commonly via a HTML interface) to gather a group of subjects "together" for discussions and moderated conversation. With a number of minor exceptions, the approach undertaken by these organisations is essentially similar: through the use of Internet or Bulletin Board System (BBS) technology, the companies run focus groups using the computer technology to facilitate the conversation. Summarising the available literature provided by these organisations^[23], online focus groups have a number of advantages and disadvantages:

8.5 These approaches, regardless of small variations, are predominantly based on original sampling via the Internet - either through random sampling or the establishment of a database of users that can participate depending on their personal characteristics (normally on a pay per participation basis). This research relies on participants that have: ongoing Internet access^[25], computer literacy, and interest in the process^[26]. Thus, they are similar characteristics to the work of Smith^[27]: research consciously based on computer network users.

8.6 What needs to be recognised, however, is these assessments of online focus groups are limited in their analysis of the relative pros and cons of the approach, considerations that will need to be developed for the method to gain favour beyond use in market research firms.

8.7 A different application of computer-mediated communication that has attracted attention is the use of standard electronic mail for social research. This method has been gaining popularity and is an increasingly common journalistic practice by traditional media and online magazines (ezines). Of all the applications of computer-mediated communication in social research, this is the oldest. Examples of the approach exist from pre-Internet research (such as Myers use of electronic mail on bulletin board systems in 1987), and Smith's study of the WELL included email interviews as an adjunct to realtime discussions.

8.8 Email Interviewing is an electronic version of mail interviewing, using written questions and answers exchanged between the interviewer and subject, with follow up questioning undertaken through ongoing correspondence. Like face to face interviewing the approach can form an ongoing social relationship, however mediated through the process of formal letter writing. The method allows for very inexpensive interviewing to be undertaken, and, because of the speed to which electronic mail can be delivered, it can be relatively fast and convenient (especially for the researcher, who can store and collate the messages in electronic form).

8.9 This concept is discussed in some detail by Selwyn and Robson (1998), examining some of the issues associated with computer-mediated communication, such as the loss of paralinguistic cues. These issues are, by nature, very similar to those experienced by the realtime online interviewer. While the methods are similar, the online interview and the email interview differ in the realtime aspect of the interview, which can limit the immediacy of email communication. In a realtime interview immediacy provides two key advantages over interviews by correspondence:

• It aids in the relationship-development process between interviewer and their subject (Weiss, 1994:65-6) by demonstrating greater attention and commitment on the part of the interviewer to their interlocutor, and
• It produces greater spontaneity which is one element in a quality interview (Kvale, 1996:145), producing top-of-mind responses that are not subject to detailed, considered revision.

8.10 While email interviews can be undertaken faster than mail interviews, the structure of the approach (providing the interviewee with a opening set of questions, rather than a discussion in "rounds") and the nature of the communications channel can stretch the discussion out over time. This has been discussed by Christensen (1999) in his use of email interviews to study the Inuit people, stating:

The possibility of quickly reformulating ones question does not always exist in the use of email, and so it can become a tedious business for both researcher and researched to conduct this type of interview. In many occurrences of my study people would read their mail three or four times a week and thus it could easily take a week for a reply to arrive. Add that to the repetition of clarifying questions, and one was looking at 3 to 4 weeks in extreme cases to get a question and its surrounding areas answered and clarified.

8.11 By nature the realtime aspect of the online interview precludes this delay, with both the interviewer and interviewee conversing simultaneously. While the use of delayed responses allow the interviewee to consult records or other people in formulating their reply (Suduau and Brown, 1984:40-1), this element does shift the interview by correspondence further away from the immediacy of traditional interviewing that may be a section criteria for researchers using the interview method.

8.12 What the work of writers like Selwyn and Robson and the online market research companies show is the range of research techniques that can be adapted for use with computer communications, ranging from simple email surveys and interviews, to realtime interviewing and round table discussions held between moderated groups. What these approaches have lacked to date is detailed discussions of their method and critical assessments of their limitations and advantages to justify their use by the researcher, but also inform technical development of the various methods.

Online Interviewing: A Synthesis of Advantages and Limitations

9.1 The online interviewing method outlined can be an extremely cost-effective way of undertaking and recording interviews. The approach is limited by sampling problems, where non-specialised interviewing is required (such as the general evaluation of a group which is not defined in terms of their access to computer network technology). This limitation may decline in first and second world nations, however, with the proliferation of computer network technology. While the use of computers to facilitate the interview has some advantages for researchers unfamiliar with ways to interpret non-verbal communication or where the interviewer effect must to be carefully avoided, the data presented by the method must be considered as distinctly different from that of a transcribed conversation. Overall, the online interview tends to be shorter and may need to be restricted to the gathering of low-cost supporting evidence. The incorporation of additional technologies and software applications into the method may enhance the work of their researcher, however, through automation and computer-assisted data management.

9.2 This section outlines the advantages and limitations of online interviewing (summarised below in table 2) and is drawn from a wide range of secondary sources, the pilot study and practical application of the method. Many of the issues are, by nature, not constrained to computer-facilitated interviewing (the loss of paralinguistic cues and editorial revisions are shared with email interviews and surveys, and mail interviews). What needs to be considered is that the aim of the method is to provide a means of engaging in research where it would not otherwise be possible. To date the method has only been used as an adjunct to existing, traditional interview methods because of these facts.

Resource Implications

10.1As commercial rates for Internet access are low or where the service provided the researcher by their host institution, the cost of carrying out an online interview can be very low compared with face to face or telephone interviewing, especially where the interview is conducted at great distance from the researcher.

10.2 The method eliminates costs associated with post-interview record keeping. As the complete text of the interview can be logged to a file accessible by the interviewer, the method removes the necessity for the interviewer to transcribe (or pay for the transcription of) hours of audio or video tape. Additionally, the storage of interview material electronically allows quotes to be "dragged and dropped" directly into written work.

10.3 These advantages make the online interview a low cost method of gathering primary research material that could allow for the conduct of interviews where funding or time constraints would not allow, or as a supplement to other methods to increase the ability of the researcher to cross-verify the value of information gathered from other sources.

Sampling Concerns

11.1Online interviewing may include a range of minor and major sampling biases that should be considered by any researcher considering the method. Research methods, as constructs, are limited in their ability to survey members of the researcher's target population. Because:

• the willingness of potential interviewees to agree to being studied using the proposed method; and
• the ability of the method to include selected interviewees ^[28].

11.2 The identification of these sample biases^[29] is a integral part of the development of any research methodology, allowing for recognition of the limitations of analysis based on the method or the inclusion of additional methods to compensate for a distorted sample. We can identify two factors that may limit the sample of the online interview: the lack of physical presence of the interviewer, and the use of computer technology to facilitate communication.

11.3 The lack of interviewer presence may preclude interviewees who are affronted by the inability of the interviewer to attend them in person. This may be due to:

• Social factors - the perceived or actual status of the interviewee;
• Contextual factors - perceived or actual status of the issue; or,
• Methodological bias on the part of the interviewee - where traditional research methods are preferred.

11.4 Conversely, the lack of the physicality may be useful in reaching people who wish to retain some degree of anonymity, are adverse to physical meetings, or who have not the time to be physically accessible to the researcher. Additionally, the approach may serve to overcome some of the problems associated with cross-cultural interviewing (especially where interaction between people of different genders, racial, or religious groups are culturally inappropriate). Some interviewees may be intrigued or interested in the novelty of the approach (the innovation function), limiting sample bias to some degree.

11.5 When undertaking general interviewing (rather than the specialised interviews common to political science research) the use of computer facilitated communication will affect the sample in a number of ways:

• Primarily, it limits the sample to those who own or can access a computer (this will be especially relevant to studies in societies or groups who are especially adverse or deprived of computer access);
• Secondly, while effort has been made to keep the method as simple as possible, it is likely that some interviewees who could access computers will be hesitant or unable to do so, because of either technological aversion, lack of computer literacy, reading / typing problems, literacy or some other factor;
• Alternatively, researchers are able with this method, to reach any person able to access the World Wide Web and thus could be used where the scope of the research requires contacting individuals with a wide geographic scope, or who reside in areas remote to the researcher.

11.6 These factors are illustrated by Coomber (1997)^[30] who discusses the use of Internet services for the collection of survey data and the problems of sample bias. One of the reasons for using computer networks to collect primary data has, traditionally, been because of the representativeness of network users for the study in question (where studies have deliberately attempted to sample computer network users^[31]). Where network users are not explicitly the population the, still limited, uptake of computer network technology and demographic spread (the preponderance for white, male, first-world, educated users^[32]) are limiting factors for any sample. Thus, while it is possible to use the Internet to find self-defining groups (for example the Newsgroup Fan.Jackie-Chan is comprised of people talking about the Hong Kong action star), it is impossible to preclude the fact that these people will still be Internet users and fit within some variation of the demographic pattern of the overall Usenet community. In this way the use of online services as a tool of research can be compared with that of the telephone following its introduction: a method that had a reducing level of bias^[33] as technological uptake increased (Bas�nez, 1997:55-7).

11.7 Factors affecting sample bias of online interviewing are also influenced by the commitment of interviewees to the research project and method. Commitment is especially relevant in this case because of the additional requirements placed upon the interviewee undertaking the online interview. Rather than being faced by a researcher at a pre-arranged time for an intensive discussion, an online interviewee must accept to use their computer for the research (or seek out an acceptable machine) and engage in a lengthy process of typing and reading^[34]. While McCracken (1988:25) argues that interviewees are surprisingly willing to engage in long interviews with "energy and involvement", the online interview is not just potentially long, but also quite physically demanding (typing, looking at a monitor, reading, etc.).

11.8 The observation of McCracken is somewhat supported by the results of the pilot study, however, where the participants accepted the method. To quote one participant:

"... we only had half an hour so there's of course an adjustment, but by the end of the half hour I was getting into the swing of it.".

11.9 This was a common response, and certainly the novelty of the method seemed to engage the interest of the interviewees. What the study group expressed was a need to know why the researcher utilised the computer^[35] and a clear, straight-forward explanation of the use of the interface (in one case during the study a major difficulty was encountered with the method, however some of the interviewees reported minor, self-correcting difficulties). In the pilot study interviewees have felt a commitment to complete the process, however this was influenced by the presence of the interviewer, which would not be a factor in its real-world application, sans interviewer.

11.10 To further explore the concept of commitment, a comparison can be made with the tendency for respondents to engage in Computer-Assisted Survey Interviews (CASI^[36]). CASI allow respondents to self-report or be assisted in completing a survey using computers. In a study by Couper and Rowe (1996) the method and the response rate to the technique is detailed. In their assessment a number of factors were most pertinent in increasing the response rate to their self-administered surveys: age, familiarity with computers, vision, education, and literacy. The survey identified white, educated, younger people with past familiarity with computers who had good eyesight as most likely to respond and complete the survey using the computer. Overall, the researchers received a response rate of seventy per cent from a broad range of interviewees (age, gender, eduction level, ethnicity, employment status, computer experience). However, a proportion of these responses (some twenty per cent) required the assistance of a present interviewer to read questions, key responses, or both. What is apparent, however, is that the presence of the interviewer (as opposed to a purely automated system) encourages increased response, however the requirement for this presence to be physical (as opposed to virtual) is unclear^[37].

Paralinguistic Cues and Language

12.1Hamman (1996) identified virtual environments that depend on the use of textual messages as "Narrow-Bandwidth Space", eliminating paralinguistic cues and communication. The face to face interview can be an intensely personal activity, with the interviewer engaging a range of senses as their data collection tools, the lack of physical presence in the online interview will have two key effects:

• it removes the ability of the interviewer to use and interpret paralinguistic communication; and,
• it is dependant on written rather than spoken language.

12.2 The loss of paralinguistic cues is discussed by Mizrach (1998)^[38]. These cues, which Oldfield (1951:99-101) has categorised into five groups: morphological, dynamic and static aspects of posture or movement, facial expression, vocal qualities and miscellaneous (habits of dress, general tidiness, etc.), are seen to be important elements of interviews, both as data for, and as a tool of, the interviewer^[39].

12.3 In considering the importance of paralinguistic cues what is important to question is the facility of interviewers to interpret, or, having done so, to utilise the information afforded from this non-verbal communication. As the interviewer is the instrument of data collection in the interview process, a keen assessment of their personal strengths and weaknesses is required before undertaking a study. Douglas (1985:39-54) outlines the need for researchers to understand their own strengths and weaknesses before undertaking research and quotes from classic examples of error based on the researcher failing to examine their own limitations (such as Mead's Samoan study). While many social research texts stress the importance of observation and non-verbal cues, few provide the information required to fully interpret them^[40]. Additionally, the production of work based on interviews often demands that the text of these discussions be translated onto paper in the form of a transcript, which, while often including comments or annotations outlining some aspects of non-verbal communication (such as shrugs, nods, etc.) seldom includes extensive records or interpretation of paralinguistic cues.

12.4 The lack of physicality prevents the interviewer from being able to "control" the place in which the interview takes place^[41]. This limits Hammer and Wildavsky's emphasis on a number of important considerations that the interviewer should consider when negotiating the location of the interview (such as the ability to see and assess the work environment of the individual or the need for privacy for the interview). Additionally, in losing the physical presence of a traditional interview the interviewer has less control over the interviewee deciding to terminate the interview (which can be done at the press of a button). Whereas, an interviewee who may wish to terminate a traditional interview has to physically leave or request the interviewer to leave, the social nature of this process is less immediate and does allow the interviewer the possibility of recovering the situation (such as apologising for an inappropriate question, requesting the interviewee remain, retracting a line of questioning or simply through the process of showing attentiveness through non-verbal communication^[42]). The study group discussed this question in some detail and responses were varied. Certainly none felt the need in the pilot study interviews to terminate, however the topic under discussion was not particularly sensitive. The group resolved that they felt they had some control of the process, some commenting that they felt "less nervous" talking to the computer.

12.5 Another key difference between traditional face to face interviews and the method outlined is the reliance upon written, rather than spoken, language. Slaughter (1985) discusses the differences between the written and the spoken, stating (122-3):

... written language isolates the individual from the group and is the communication medium of isolated individuals. ... It allows communication over great differences and long stretches of time with people who are unknown and may never be seen. Because of this the writer cannot assume that his readers have the same background as he does - the same values, shared experience, language, assumptions, definitions. As a result his discourse must be explicit, autonomous, and spelled out. (sic)

12.6 Slaughter argues that a number of important differences distinguish the two forms of communication: First, there are no words in the written language that indicate specific nuances of context; Second, because writing is an activity that has to be formally learned, written text contains more structure and explicitness than spoken. As the writer needs to add context and meaning that are lost, there is the tendency to ensure meaning through an explicitness not found in spoken language; Third, in writing, the individual becomes separate to the text, standing apart from it and objectifying their experiences. In this way Zito (1984:55-6) identifies the immediacy of spoken communication, stating "... we most often cannot 'think before we speak,'..." - spoken communication has not be systematically prepared, nor had the application of rules and formality that come with the learned process of writing.

12.7 From the pilot study, however, it is clear that the interviewees felt rushed to respond. As one interviewee stated "[I tended to] speak more freely [because of the lack of physical interviewer], but the problem is you tended to speak a little earlier.". While the group could not come to any agreement on the issue of "speaking freely"^[43], there was general agreement that the absence of the interviewer prompted them to respond more quickly than they would in a face to face interview. This appears partly caused by the plain interface and the absence of a dynamic processes of interview (the interview, broken into "rounds" leaves some seconds or minutes waiting for a response), but also because of the lack of reassurance that the interviewer was being attentive. One interviewee described the process as "exam conditions", while another stated that they "had no-one to please". Certainly the group would have liked a clear indication of how much was required, but more importantly, and contrary to the statement by Zito, the group felt compelled to type before they thought^[44]. In doing so, however, the study group produced, in general, transcripts that were more ordered and structured than would typically appear on the transcript of a taped interview.

12.8 The researcher, therefore, must consciously recognise the differences between the transcribed text of spoken interviews and the written communication of the online method. There is doubt about the spontaneous nature of responses^[45]; where an interviewer would, in a traditional interview, see an interviewee revise their statement (to clarify or to conceal their initial response), the online interviewer is unable to know how much editing has been undertaken in any text "chunk" received. Therefore, while the automatic transcript may be seen as entirely reflective of the content of the interview, rather than a simple (and, by nature, modified) version of the original spoken dialogue that has been decontextualised in the process of transcription, it may not be entirely reflective of the thoughts of the interviewee.

12.9 Related to this is the process of questioning and responding in rounds, rather than the freeform nature of common conversation. The online interviewer is less able to "cut in" on their interviewee. This can be a distinct advantage for those interviewers who find themselves constantly talking over their interviewee, but limits the interviewers ability to control the process. Similarly, the structure of the online interview does tend the make the interview more "Rat-A-Tat-Tat", as Dexter (1970:56) would say. The interview is less like a conversation and more like a series of questions "fired off" by the interviewer. This problem can be limited by the use of more informal questioning or through rapport establishment at the beginning of the interview process^[46], but needs to be consciously controlled by the interviewer. In the case of the online interview, the concept that "you are your words" must be remembered. The interviewer must consider the manner in which, devoid of personalising features and non-verbal characteristics, their messages will be interpreted by the interviewee.

General Considerations

13.1To conclude our assessment of the strengths and weaknesses of the online method, two final considerations need to be made: the advantage of "distance" between interviewer and their subject and the use of additional software applications as adjuncts to the interview process.

13.2 While the online interview is a process that removes the immediate presence of the interviewer from their subject, this can assist the researcher through the manufacture of distance between interviewer and interviewee (McCracken, 1988:22-3). The notion of distance is important when assessing the manner in which the culture we reside places assumptions over every day events and phenomena. Interviewers have been keen to identify these issues and attempt to maintain a neutral standing in regards to the issues and individuals they are studying as to be better able to identify unspoken social norms and not allow cultural paradigms from influencing the answers of interviewees (Kahn and Cannell^[47], 1957:208).

13.3 While this neutrality must be developed in the way questions are asked and conversation directed, it can often be also associated with the interviewer as instrument^[48]. The way an interviewee perceives the interviewer is acknowledged as an important influence on the interview process and the outcome of the interview itself. Many of the basic ways that one individual would assess the other is by looking at them and assessing them on that basis: the style of their hair, clothing, gender, attractiveness, age, and a host of other factors will subtly influence the interviewees approach to the interviewer^[49]. Therefore, the online interview does have the potential to be the "great equaliser", reducing the number of factors the interviewer need control to the way they use the written language and the pertinence of the questions they ask^[50]. Seidman argues that (1991:74), while friendly relationships can be useful in the interview process, it is wise to err on the side of formality, and the utilitarian nature of the online interview can assist the interviewer in this task.

13.4 While this does appear to be played out in the results of the pilot study, the question of verification of the interviewer's identity was raised by the study group. The group, while accepting of the study (they were fairly sure that the study was being undertaken "legitimately"^[51]), were less convinced that they would be convinced of the identity of a researcher they had not met (ie one who wanted to survey them remotely). In this case letters of introduction (with official letterheads) and the identification of an "official" Uniform Resource Locator (URL) (such as the http://www.anu.edu.au/ that designates the Australian National University) were not seen by the group as guarantees of legitimacy. The group expressed a strong need to feel that the research was legitimate as part of the justification of the method. Therefore, a researcher using the method may need to explain the financial or logistic reasons behind the use of the study as a means of stating that the study, while being undertaken using the World Wide Web, was a legitimate part of an academic study.

13.5 Finally, because of the inclusion of the computer as a key element in the interview process this method does allow the integration of additional software packages into the interview and post interview process. A number of examples can be suggested (for an more detailed overview of these examples click here):

• Question management software: at the most basic level a semi-structured interview could be managed by having a series of prepared questions that are copied and pasted into the web browser from a simple text editor (such as Word or Simpletext/Notepad) from a database of options. More sophisticated systems, such as CATI could be adapted for this approach, especially where a large number of very similar interviews are required.
• Post-interview data analysis: information gained from the interview can be incorporated into data management packages by directly setting the recording parameters of the transcript log to include the necessary heading indicators. One excellent example of this would be the NUD*IST software package which is very familiar to social researchers.
• Translation: while automatic translation remains primitive, there are already online translation services and desktop software packages that could be incorporated into the interview process, expanding the sample size for researchers who lack the capacity to hire translators.
• Privacy: developments in personal encryption can be incorporated into the web interviewing process, using "public key encryption". This software could assist in increasing interviewee confidence in the privacy of the interview process.

13.6

Generally, these packages can be run in parallel with the web-based interview, or through the modification of the source code through the inclusion of, for example, CGI scripts. As increasing numbers of software packages are developed, and the trend to web-integration of software expands, the range of available options will increase.

Conclusion

14.1The method outlined in this paper has the possibility of becoming a useful low-cost adjunct to existing interview and survey methods in the social sciences. Overall the approach is an adaptation of a well accepted research strategy to technical innovation, very similar to the adoption of the telephone as a research tool where distance and cost are limitations placed on the researcher.

14.2 The method has four key advantages: the ability to conduct realtime interviews where distance and cost prohibit the use of face to face or telephone interviewing; the ability to log transcripts directly to file avoiding transcription error and the cost associated with the process; distancing of interviewees from the researcher; and, the use of additional software packages to assist in the process of data collection and post-interview data analysis. As technology develops, the use of additional techniques and software packages will definitely expand the range of options available to the researcher. A number of limitations of this method need to be considered in any research design including the online interview: the loss of paralinguistic cues and observation, questions of data reduction and sample bias limitations.

14.3 Overall, the approach can be seen as a specialised tool useful in some areas of social research, specifically where closed, finite groups that have computer access and literacy are the subject for study, or where subjects can not be accommodated with traditional research methods. For researchers with limited budgets or time to travel, the method is recommended as a means of conducting interviews that would not be included in the research project without the use computer-facilitated communications technology.

Notes

⁰¹ With thanks to those who have provided comments and suggestions: Maria Maley Department of Political Science, ANU), Marian Simms (Department of Political Science, ANU), Sam Frangiamore (Insights Online), Sally Hooper (Xerox Business Research Group), and Cameron Robinson (Psychosoftware). Additionally, thanks to those students to undertook the pilot study and mini-group interview and Pam Kennahan of the Council of American Survey Research Organisations for her kind assistance.

⁰² Estimates place the number of pages of information available on the World Wide Web at 320 million (Reuters, 1998).

⁰³ These studies, however, (for example Smith's sociological study, Voices from the WELL, or Rimms Marketing Pornography on the Information Superhighway) have mainly been concerned with the World Wide Web or other online service as the focus of the study, rather than simply a tool for the study of an unrelated issue or subject.

⁰⁴ Figallo (1995:54) argues that continuously scrolling documents in conference discussion are more coherent than separately displayed messages (such as email or Usenet postings) and draw users more deeply into the discussion.

⁰⁵ It is important to distance the technique outlined above and Computer-Assisted Personal Interviewing (CAPI). CAPI is a interactive interviewing method that replaces the human interviewer and substitutes a predeveloped system of tree-structured coding to allow for the substitution / insertion of questions based on a number of predefined responses (Saris, 1991).

⁰⁶ Such as observing the interviewee in their home / work environment, observing the nature of the office (for example, noticing piles of uncompleted work, etc.).

⁰⁷ Or: Aare a significant proportion of the interviewees able to access the technology required?

⁰⁸ Hammer and Wildavsky estimate nine hours of transcription are required for each hour of tape.

⁰⁹ It should be noted that this consent my be worked into the interview transcript itself, through some form of "gateway" page that includes an online informed consent document suitable for electronic "signing". If this is used, however, then one should examine Bradburn and Sudman's (1980) chapter regarding the sampling effects of informed consent. To summarise: written consent forms can lower the response rate (by some eight per cent overall), and those who reject signing such forms provide lower quality responses than those who do.

¹⁰ The legality of employers reading the electronic communications of the employees, for example, remains uncertain. Some cases exist where employers have take action over their employees use of computer networks (example, in 1997 a Victorian university staff member was dismissed for misuse of computer facilities [Healy, 1997:33]). Additionally, an interview conducted in a relatively public place (such as an open-plan office or Internet cafe) may allow others to see the text as it appears on the computer screen. Where some interviewees are concerned about eavesdropping may request the interview not occur in such a place, the online interview is limited by the access of the participants to networked computers.

¹¹ A "plug-in" is an additional piece of software that adds capabilities to the original browsing software. These programs extend the usefulness of the browsing software, but are not universally used (many of the latest developments are beyond the capacity of older hardware or software).

¹² Is should be noted that this debate concerns the use of qualitative data analysis software specifically, however these concerns have relevance to the data collection tool as well. HTML does allow greater flexibility in the design and functions of the webpage interviewing screen (one of the reasons HTML is preferred to other software packages), however the fundamental nature of the communication: using computers, and the parameters set by HTML and the browsing software do limit the researchers ability to modify the method ad infinitum.

¹³ Thus while Platt (1996) sees funding issues as not affecting the underlying approach of research conducted (qualitative versus quantitative), it does facilitate research and the amount of data collected. Research is not exempt from the context of equipment availability or funding considerations, a context that may need to be explicitly recognised by researchers when justifying their methodological choices.

¹⁴ Split screens containing two different pages of information.

¹⁵ The most obvious example of a rival approach would be the use of Internet Relay Chat (IRC). IRC is a realtime conversational medium that utilises the Internet network to allow for group and "coupled" conversation through a very similar medium as outlined in the section above. Compared to uptake of the World Wide Web, however, IRC remains a subset of the larger Internet community. Accessing IRC can be a technical task requiring some skill and additional software that must be used by its participants. While IRC would probably be suited to "random" interviews (especially where the research wishes to take advantage of the self-selected nature of participants in different IRC "channels" [topic areas]), it is likely to also include the sample bias of a medium requiring more computer literacy than the average Internet user. However, as Rose (1995) points out, most users will have some facility with the medium (having been able to establish contact with IRC hosts) and will be more familiar with the concept of online communication, using non-paralinguistic cue substitutes (such as the ubiquitous ":)" smiling face symbol, and other facial expression substitutes.

¹⁶ "Server push" is one solution devised to circumvent problems caused by the stateless nature of HTTP. Server push keeps the connection between the web server and the browser open indefinitely, allowing data to be "pushed" across the Internet to the user. This method was not used, since it requires a browser at the other end which knows how to process the incoming stream of data.

¹⁷ Unlike server-push, client-pull is simply a directive written into the HTML source which tells the browser to load an arbitrary web page after a certain number of seconds. This "client-push" has been implemented by browsers for some time, and will work with even early versions of Netscape.

¹⁸ Smith does highlight the advantages of this face to face discussions (conducted at a social function for WELL users), where he gathered impressions of the "...social status of WELL members that could not be easily derived from contact via the WELL itself.".

¹⁹ While Smith is not specific about the nature of this email interview forum (and certainly some of the interviews were undertaken as correspondence, rather than as a realtime interview), it is certain that Smith accessed chatting facilities commonly used by WELL users. These users would have had a degree of expertise in using these facilities that is not required in the method outlined in this paper.

²⁰ Which is not well articulated in Smith's piece and lacks a keen assessment of the advantages and limitations of the approach used. Similar discussions on the topic of computer moderated communications and the research implications for social researchers (such as Thomas, 1995) also fall into this category.

²¹ As a tool of the marketing profession, the focus group is often used with a individuals who are deemed as either users (the market) or potential users (target market) of a product or service. The group interview allows the researcher to introduce issues and questions for discussion, with the aim of developing a view of consensus opinion, or where discordant views exist (for the purposes of "niche" marketing or product modification).

²² This is due to the incentives requires to get participants to come to a central location (normally a conference facility or similarly appointed room), unlike in the interview environment where the interviewer often travels to the interviewee. Participants are usually provided with gifts or remuneration for their attendance, which is not common in semi-structured, in-depth interviewing. However, as groups are "interviewed" in the focus group method, the cost per interviewee (individual) may be lower. However, it is likely that cost comparisons are not really very meaningful.

²³ Which are surprisingly homogenous in their assessments of the strengths and weaknesses of the approach.

²⁴ Of the material reviewed, only two companies listed limitations (Insights Online and the Xerox Business Research Group).

²⁵ Often selection is via a direct relation to computer usage (such as the access of certain websites, use of particular facilities, etc.). Additionally, initial contact to request participation is often through email (Frangiamore, 1998:1).

²⁶ Either through interest in the novelty of the process (Xerox Business Research Group, 1998) or through incentive payment (Strategic Focus Online, 1998).

²⁷ It would be misleading to draw to close a comparison between online focus groups and online interviewing, however, because of the differences of the two methods. As Krueger (1988:19) points out, interviewing is an active process (where the interviewer seeks information), while a focus group is more passive (the facilitator attempts to encourage and guide the discussion of the group, rather than be a key part of the discussion themself). In this way the focus group has been compared to a compromise between the semi-structured, in-depth interview and participant observation, placing a distinctly different emphasis on the role of the researcher in the process.

²⁸ These two factors are importantly different: the first relates to individuals acceptance of the method (understanding its workings and agreeing that they wish to be part of a study based on the method); the second relates to the physical ability of a researcher using a particular tool to reach interviewees (an example being the telephone, which is an unsuitable research tool where interviewees have no access to it).

²⁹ King et. al (1994) list sample bias as one of the key questions of social research with the consideration of factors leading to bias being one of the earliest considerations in the process of constructing social inquiry.

³⁰ Coomer's work centres around the use of the Internet for the collection of written surveys, rather than interviewing, however his use of methods (writing to Listservers and Newsgroups and the establishment of an online, web based surveying tool) have relevance for this method, especially if the interviewer were attempting "vox pop" interviews, where the initial point of contact with interviewees were through mass emailings to wide samples.

³¹ A "classic" example being Donath's (1996) study of identity in virtual communities where the sampling universe was Usenet Users.

³² This trend, however, is slowly changing due to the increased usage of the Internet, it's popularism and decreasing costs associated with the uptake of the technology required to use its services.

³³ Frey (1983), in recognising the declining level of sample bias in telephone research (in the United States), does identify problems associated with locating certain types of respondent via telephone. The telephone, therefore remains limited, as does mail survey, face to face (geographic sampling has always remained difficult to logistically manage and is filled with compromises based on geographic access and cost issues).

³⁴ This is also relevant in most survey methods where the respondent is required to invest in the process (for example: the investment of the time required to undertake the interview, effort in completing a survey form, or the money to return post a mail survey).

³⁵ The interviewees spoke of being "surprised" at the use of the computer.

³⁶ The term "interview" in regards to CASI can be misleading, CASI is a survey approach using predeveloped questions and answers (designed for larger studies for ease of tabulation and analysis). Therefore the CASI approach is distinctly dissimilar to a semi-structured in-depth interview in the number of possible responses that can be gathered via the method.

³⁷ This issue is also relevant to the pilot study.

³⁸ In this regard Mizrach is writing speculatively, rather than from experience. He outlines a number of possible methods: IRC, video conferencing, and standard email interviewing.

³⁹ As Dexter (1970:63-4) outlines, the interviewer can also use significant paralinguistic cues (a raised eyebrow, a significant pause) to prompt the interviewee for more information.

⁴⁰ Possibly because of the difficulty of the task. The researcher is often faced with the question "What does that gesture mean?", while the gesture is important, simply observing these cues is not interpretation and interpretation based on casual inference can lead the researcher beyond the scope of their competence.

⁴¹ To the extent that an interviewer attempting to arrange a face to face interview, is able to suggest locations and veto suggestions of the potential interviewee.

⁴² Such as leaning forward, appearing interested, and showing distress at an interviewee wishing to terminate the interview.

⁴³ Some considered that they would talk about the same issues in face to face while others felt that they would be able to speak more freely.

⁴⁴ To the extent that one interviewee was so concerned at their text being too off the cuff they edited their response before sending.

⁴⁵ Note here that this is neither identified as a valuable or problematic issue. This depended on the nature of the interview and the issue under consideration. Generally the loss of preconscious immediacy may be seen as problematic to the extent that the interviewee is more measured in their response to questions, however this may increase the clarity of the written transcript in better conveying the true meaning of the interviewees intended replies. The study group did recognise the "formality" of the online interview - a factor that may have tempered their responses. This could conform to Labov's view of "careful" and "casual" speech, where the interviewee recognises the formality and seriousness of the interview, and changes modes of conversation for the different occasion (as examined in Briggs, 1986:18).

⁴⁶ This issue was highlighted by the study group, who would have preferred a period of rapport building (which was called a "practice period" by the group). This practice period would have allowed the interviewees to gain confidence with the method, as well as trust in the interviewer as an attentive person, rather than a "ghost in the machine".

⁴⁷ Kahn and Cannell call this the non-directive technique, where the interviewer must balance their desire to elicit additional responses or more detailed information, without introducing unwanted or unplanned influence upon their responses.

⁴⁸ For an interesting and relevant refutation of the overemphasis on bias reduction within the positivist school of research, see Briggs (1986:21-3) discussion of bias theory and individual true value. In this piece Briggs argues bias can be eliminated, and b) that through this process distortions in research responses can be eliminated and absolute truth gained. While this may be relatively straightforward, it is an unspoken feature of research that the concept of an "individual true value" is achievable and that research can be validated through its attempts to achieve this ideal type.

⁴⁹ Seidman (1991) lists a number of influencing factors: Race, Gender, Class, hierarchy and status, and Age as influences in developing an "equitable interviewing relationship".

⁵⁰ The interviewee will make assessments of the interviewer through their contact in the pre-interview process (such as on the telephone) and by determining (or possibly misinterpreting) their gender, ethnicity, and/or nationality based on their name.

⁵¹ The main concern regarding the legitimacy of researchers was based on their preponderance to encounter commercial researchers who were seen as unsavoury, time consuming and intrusive.

⁵² Not that this is impossible in the traditional face to face interview, but more easily achieved using on online method.

⁵³ Additionally, while the issues outlined by Donath may not be relevant to a semi-structured, in-depth interview because of the lack of anonymity between interviewer and interviewee, this may reduce the need to consider the media impact on responses, such as those discussed by MacElroy in The Effect of E-Personality on Research Results where communications with anonymous interviewees, or interviewees who is only known in the virtual environment (such as may be encountered in vox pop interviewing), is seen as encouraging more forceful or aggressive behaviour, stronger opinions, intense candour, cynicism or mechantilism.

References

BASANEZ, M, 1997, "Is the Web the Telephone of the Future", Globalisation and the Worldwide Web of Survey Research, CASRO 1997 Annual Journal, The Council of American Survey Research Organisations, New York.

BRADBURN, NM and S SUDMAN, 1980, Improving Interview Method and Questionnaire Design: Response Effects to Threatening Questions in Survey Research, Jossey-Bass, San Francisco.

BRIGGS, CL, 1986, Learning How to Ask: A sociolinguistic appraisal of the role of the interview in social science research, Studies in the Social and Cultural Foundations of Language No. 1, Cambridge University Press, London.

BUSINESS RESEARCH CENTRE, 1998, Business Research Centre, http://www.xbrg.com/centerbr.html, Xerox Corporation (The Xerox Business Research Group), 29 May [linkdate].

BUSTON, K, 1997, "NUD*IST in Action: Its Use and its Usefulness in a Study of Chronic Illness in Young People", Sociological Research Online, Vol. 2 No. 3, http://www.socresonline.org.uk/2/3/6.html.

CHRISTENSEN, NB, 1999, Inuit in Cyberspace: Practising and Constructing Computer-mediated Space, Paper Prepared for the 3rd Annual ARCUS Award for Arctic Research Excellence, http://home.worldonline.dk/~nbc/arcus.html.

COUPER, MP and B ROWE, 1996, "Evaluation of a Computer-Assisted Self-Interview Component -Assisted Personal Interview Survey", Public Opinion Quarterly, Vol. 60, pp. 89-105.

COMMON KNOWLEDGE, 1997, Virtual Focus Groups, http://www.commonknowledge.com/html/virtual_focus_groups.html.

COOMBER, R, 1997, "Using the Internet for Survey Research", Sociological Research Online, vol. 2 no. 2, http://www.socresonline.org.uk/.

DEXTER, LA, 1970, Elite and Specialised Interviewing, Northwestern University Press, Evanston.

DONATH, J, 1996, Identity and Deception in the Virtual Community, http://smg.media.mit.edu/people/judith/Identity/IdentityDeception.html,

DOUGLAS, JD, 1985, Creative Interviewing, Sage, Beverley Hills.

FIGALLO, C, 1995, "The WELL: A Regionally Based online Community on the Internet", in Public Access to the Internet, B. Kahin and J Keller (editors), A Publication of the Harvard Information Infrastructure Project, MIT Press, Massachusetts.

FRANGIAMORE, S, 1998 [unpublished], "Qualitative Analysis", Extracts of Material Included in Research Proposals for Insights Online.

Frey, JH, 1983, Survey Research by Telephone, Volume 150 Sage Library of Social Research, Sage, Beverley Hills.

HAMMAN, RB, 1996, Cyborgasms: Cybersex Amongst Multiple-Selves and Cyborgs in the Narrow-Bandwidth Space of America Online Chat Rooms, MA Dissertation, Department of Sociology, University of Essex, http://www.socio.demon.co.uk/Cyborgasms.html, 30 September.

HAMMER, D and A Wildavsky, 1989, "The Open-Ended, Semistructured Interview: An (Almost) Operational Guide", Craftways: On the Organisation of Scholarly Work, Transaction Publishers, New Brunswick.

HEALY, G, 1997, "Dismissal Sparks Net Warning", The Australian, June 11.

HOLSTEIN, JA and JF GUBRIUM, The Active Interview, Qualitative Research Methods Series, Sage, Newbry Park.

INSIGHTS ONLINE, 1998, Insights Online - online focus groups / moderated chat sessions, http://www.insightsonline.com/olg.htm, 29 May [linkdate].

KAHN, RL and CF Cannell, 1957, The Dynamics of Interviewing: Theory, Technique and Cases, John Wiley & Sons, New York.

KELLE, U, 1997, "Theory Building in Qualitative Research and Computer Programs for the Management of Textual Data", Sociological Research Online, Vol. 2 No. 2, http://www.socresonline.org.uk/2/2/1.html.

KING, G, RO KEOHANE, and S VERBA, 1995, Designing Social Inquiry: Scientific Inference in Qualitative Research, Princeton University Press, Princeton.

KRUEGER, RA, 1988, Focus Groups: A Practical Guide for Applied Research, Sage, Newbury Park.

KVALE ,S, 1996, InterViews: An Introduction to Qualitative Research Interviewing, Sage, Thousand Oaks.

MARTIN, J and T MANNERS, 1995, "Computer Assisted Personal Interviewing in Survey Research", in Information Technology for the Social Scientist, Lee, RM (ed.), UCL Press, London.

McCRACKEN, G, 1988, The Long Interview, Qualitative Research Methods, Volume 13, Sage Publications, Newbury Park.

MIMICHIELLO, V, et al, 1995, Indepth Interviewing, 2nd Edition, Longman, Melbourne.

MIZRACH, S, Using Computers in Qualitative Research, http://www.clas.ufl.edu/users/seeker1/scholarly/comp-in-qual-research.html , 31 March 1998 [linkdate].

MYERS, D, 1987, "'Anonymity is Part of the Magic': Individual Manipulation of Computer Mediated Communication Contexts", Qualitative Sociology, Vol. 10, No. 3, Fall, pp. 2521-66.

NEUMAN, WL, 1994, Social Research Methods: Qualitative and Quantitative Approaches, Allyn and Bacon, Needham Hights.

OLDFIELD, RC, 1951, The Psychology of the Interview, Methuen, London.

PFAFFENBERGER, B, 1988, Microcomputer Applications in Qualitative Research, Qualitative Research Methods Series, Sage, Newbry Park.

PLATT, J, 1996, "Has Funding Made A Difference To Research Methods?", Sociological Research Online, Vol. 1, No. 1, http://www.socresonline.org.uk/1/1/5.html.

PRETTY GOOD PRIVACY INC., 1997, PGP for Personal Privacy, Version 5.5, Macintosh Users Guide, http://www.pgp.com/, San Matco.

REUTERS, 1998, "Study: Search Engines Fall Short", CNET News.Com, http://www.news.com/News/Item/0,4,20728,00.html, April 2.

ROSE, B, 1995, Internet Chat Quick Tour: Real-Time Conversations & Communications Online, Ventana, Chapel Hill.

SARIS, WE, 1991, Computer-Assisted Interviewing, Quantitative Applications in the Social Sciences Series, Sage, Newbry Park.

SEIDMAN, IE, 1991, Interviewing as Qualitative Research: A Guide for Researchers in Education and the Social Sciences, Teachers College Press, New York.

SELWYN, N and K ROBSON, 1998, "Using E-mail as a Research Tool", Social Research Update, Issue 21, Summer.

SLAUGHTER, MM, 1985, "Literacy and Society", International Journal of Language, 56, pp. 113-39.

SMITH, MA, 1998, Voices from the WELL: The Logic of the Virtual Commons, http://www.sscnet.ucla.edu/soc/csoc/papers/voices/Voices.htm, 2 March [linkdate].

STRATEGIC FOCUS ONLINE, 1998, Strategic Focus Online - Information About Online Focus Groups, http://www.sfionline.com/, 29 May [linkdate].

SUDUAU,S and N BROWN, 1984, "Improving Mailed Questionnaire Design", Making Effective Use of Mailed Questionaries, Lockhart (ed.), Jossey-Bass, San Francisco.

SUMMITT, PM and MJ SUMMITT, 1996, Creating Cool Interactive Web Sites, IDG Books, Foster City.

THOMAS, R, 1995, "Online Services and CD-ROM", in Information Technology for the Social Scientist, Lee, RM (ed.), UCL Press, London.

WEISS, RS, 1994, Learning from Strangers: The Art and Method of Qualitative interview Studies, The Free Press, New York.

ZITO, GV, 1984, Systems of Discourse: Structure and Semiotics in the Social Sciences, Contributions to Sociology, number 51, Greenwood, Westport.