Virtual Fieldwork Using Access Grid

This article discusses the use of Access Grid (AG)—a form of video teleconferencing delivered over computer networks—to perform fieldwork. Interviews and group discussions were conducted with students and criminal court judges at sites remote from the fieldworker. A concept of ‘‘engagement’’ was used to identify distinctive interactional features and provide a first insight into the AG as a fieldwork medium. Technology is increasingly affecting social research methods and a prime current site of innovation is fieldwork and qualitative data analysis. This article discusses the world’s first ‘‘fieldwork’’ using Access Grid (AG) technology. The research explored the use of devices called Access Grid Nodes (AGNs; see www.accessgrid.org) as a medium for interviews and group discussions with participants at sites remote from the fieldworker. Interviews and group discussions were conducted with computer science and social science students at British universities by me and Maria Macintyre. The topic was attitudes toward acting as a criminal trial witness (linked to research reported in Fielding 2006). Subsequently, we conducted group discussions between judges at British and U.S. courts. The criminal justice topic was a vehicle to explore issues relating Access Grid fieldwork to computer-mediated communication (CMC) and video teleconferencing and to mode effects in field methods. The research was, to our knowledge, the first to use AG for virtual fieldwork. Interviews and discussions do not, of course, exhaust the range of field methods. We chose them because they are widely used, but consequently, issues associated with other field methods were not covered (e.g., immersion during participant observation). As a first account of AG-mediated fieldwork, this article, like the research, has focused and limited objectives. It explores situations in which the technology’s affordances might make it attractive relative to physically co-present fieldwork. It relates AG-mediated interaction to that of physically co-present fieldwork, drawing on features of interview/discussion interaction such as rapport, engagement, and paralinguistics, and on themes from the literature on CMC. Among emergent issues were effects on interviewer/respondent power relations, involvement of elite respondents, respondent candor, selfdisclosure, and paralinguistic communication. The article does not claim to give a definitive treatment of AG fieldwork, instead confining itself to a ‘‘proof-of-concept’’ approach and a provisional airing of considerations. It discusses practical, procedural, and methodological concerns in parallel but notes some emergent analytic issues and some perspectives that were helpful. Grid computing is conventionally classified into data grid, computational grid, and AG technologies. Grid computing resources have received major investment from institutional sources in a number of countries, supporting what is called ‘‘e-science,’’ ‘‘e-research,’’ and ‘‘e-Social Science’’ in Europe and ‘‘cyber-research’’ in the United States. Much of the substantial infrastructure now in place is directed to quantitative applications, but this massive capacity for high-performance computing can also support qualitative work (Fielding and Lee 2008), including using AG for field research. AG services are delivered via nodes. AGNs enable images and sound to be exchanged in real time between computers over networks. Multiple cameras, projectors, and microphones at each AGN site relay participants’ images and utterances to other sites. There is no technical limit on how many sites can be linked. Limits on participant numbers at each site depend on room size and node configuration. Visual output is projected onto a wall or screen (see Figure 1). Users can vary image display size. Each image ‘‘tile’’ can be rearranged on the projection wall. Images from a given site will often show the whole room, the current speaker (from different angles), and a computer-generated document like a PowerPoint slide. Like images, AGNs can dynamically display any material that can be shown on a computer screen, such as text or graphical/tabular output. A good installation will ensure all aspects of the room are covered for sound and vision. AGNs provide audiovisual signals without the lag of video teleconferencing, and because participants can be displayed life size or larger, paralinguistic cues are more visible than with video teleconferencing. To enable comparison to conventional interviews/discussions, I draw on a basic conceptualization of participants’ experience of the AG medium using the concept of ‘‘engagement,’’ a set of features gauging whether participants are interacting in a way that is not preoccupied with the medium of their interaction. There is a further point regarding conceptualization. No research technology, from pen and paper through to AG, is a neutral ‘‘carrier’’ to the field. All research technologies reconfigure the field (see, e.g., Hine 2008). There is a rich vein to be mined concerning the point that AG is not a neutral tool but distinctively and substantially mediates between fieldworker and field. That important work is only a modest sub-text in this first foray into AG fieldwork. As a proof-of-concept study, it largely leaves the unpacking of the ‘‘black box’’ to subsequent work, while sketching one possible analytic line via a preliminary conceptualization of what constitutes engagement and its observable signs. The idea is that before exploring the more sophisticated nuances of AG-mediated research, it is sensible to be clear about its fundamentals. Figure 1. University of Melbourne access grid node (AGN). AGNs were originally developed by the Argonne National Laboratory, Chicago, and were initially used for virtual meetings among dispersed participants in international scientific communities, the first being in 1999. Another use is to deliver teaching in educational consortia. By pooling students, specialist material can be delivered more economically than having instructors visit each site. There is a major network in South America, and New Zealand’s universities are all linked via AG. In Britain, specialized physics seminars are run via AG between Birmingham and Wolverhampton universities, and Hull University and University of East Anglia have a joint AG seminar in politics. Dance students at De Montfort University conduct rehearsals via AG. An 18-university consortium led by Sheffield University delivers advanced mathematics teaching via AG to schools as well as universities. Full AGN installations grew from about 220 in 2004 to over 550 in 2005. Inexpensive AG software can also be installed on a standard personal computer equipped with a webcam, providing personal access to the grid (PAG). Positioning the Research Our research explored using AGNs to conduct semistandardized interviews and moderator-led group discussions using conventional offline comparator methods, asynchronous online interviews, and conventional video teleconferencing. I selected these fieldwork modes because they are widely used. The substantive topic was deliberately drawn from my previous research to enhance authenticity and comparative analysis, add value to earlier data, and, regarding the ‘‘elite respondent’’ element, facilitate access via existing contacts. Methodological research indicates difficulties in obtaining elite respondent participation in group discussions (Zuckerman 1972; Ostrander 1993). Factors include busy schedules, legal/professional constraints on participation, and reluctance to travel to participate alongside less-elevated respondents. These constraints are not all addressed using AG, but the logistical problems may be. We therefore tested whether AG-mediated group discussions could be convened between U.S. and British judges, as it would be unlikely they could be assembled at a common location. All methods and modes of administering methods have advantages and disadvantages. This article notes both promising and problematic aspects of AG-mediated fieldwork. The primary research question was whether it was feasible to conduct interviews and group discussions via AG. Tied to this were issues like ‘‘Would informed consent be forthcoming?’’ ‘‘Would we achieve a sample?’’ ‘‘Of students?’’ ‘‘Of judges?’’ and ‘‘Would the system deliver at a technically adequate level?’’ These issues raised corollary concerns relevant to preliminary evaluation of an AG-mediated fieldwork environment and process, including facilitators and barriers to communication, rapport, the effect of technical glitches, and the role of paralinguistics. Rather than being exhaustive, this work is provisional and tentative. I refer to AG’s ‘‘potential utility,’’ not to its utility as accomplished fact. As the principal comparator, we took the physically co-present (conventional) face-to-face interview or group discussion, frequently applying this to particular interactional features of AG-mediated communication. Moreover, in the participant debriefings that form an important part of our data, we explicitly asked participants to compare AG-mediated interviews/discussions to face-to-face ones (after checking they had indeed experienced face-to-face interviews/discussions). It should be noted that among the constraints applicable to AG-mediated fieldwork there are topical constraints. This would probably not be an appropriate medium for research on highly sensitive topics. As discussed later, self-disclosure does not work as one might predict in AG fieldwork (briefly, for some respondents, the AG had a disinhibiting effect that was the opposite of what we expected). Standard ethical procedures, like having a counselor on call if participants become emotionally upset, are impractical because one is interacting with participants at remote sites. AG’s potential fieldwork utility must be weighed against investment in facilities (2008 full installation was about $38,000), technicians, maintenance, and participant expenses for attending at an AG site (one cannot expect participants to commit to more than local travel even if compensated). However, installation and participant travel costs will decline as the technology spreads (and with availability of laptop/desktop AG software), and conventional focus groups also rely on participants’ willingness to travel. Procedures and Ethics Participants received an information pack about AGNs, our project, and their session. Interviews took a semistandardized format (freeform probes were allowed after standard main questions). The initial interview explanation prompted completion of consent forms if not already done at the information pack stage and noted that there were questions on a criminal justice topic and about being interviewed via AG. To test the value of AG’s ability to display contextual material, we compared response in interviews with and without display of the interview schedule. The display screen was positioned next to the screen showing the interviewer and sized slightly larger than the interviewer screen. Respondents indicated modest benefits, thus meriting further investigation with more complex response stimuli. For instance, background statistics or vignettes (textual or video) may be more valuable than displaying schedules. Following the criminal justice section, we invited comments on the AG as a research medium. We asked for impressions of audiovisual and technical quality and whether participants found the technology distracting. We asked them to compare the AG interview with a physically co-present interview, and which, given the choice, they would prefer, assuming similar travel distance. Participants then completed online questionnaires with items on gender, age, education, and experience of information technologies. The latter asked about use of computers and virtual communication (online message boards, video conferencing, and video phones) and experience of research participation. In AG fieldwork, one must allow for the possible presence of third parties—technicians—and agreements are needed with bodies like the AGSC (AG Support Centre) when recordings reside elsewhere than one’s own institution. Regarding technicians, basic operation can be learned in a few minutes and, unless there are network problems, researchers can operate the AGN solo during fieldwork; but one should assume that technicians may be present at remote sites. Consent forms should accommodate these points. An alternative is encryption, which also prevents access by AG users at sites not involved in the research. We assured participants that access to AG footage would be confined to legitimate researchers and that they could optionally restrict it to the immediate research team. Anonymity and confidentiality rights were otherwise standard.