|Also available as an Acrobat File|
Review of graphical environments on the WWW as a means of widening public participation in social science research
3. Literature review
Existing published literature on the use of computer graphics for improving public participation in social science research is rather limited, but even more limited when specific to the potential role of the WWW. The literature that is available in this field falls into one of two emerging categories. The first, which appears to dominate the current focus of research, relates to the debate surrounding the benefits, pit-falls and practicalities of using the WWW as a tool for carrying out social surveys. The second relates to research which has either taken place or is underway and illustrates actual cases where the WWW has been used for survey research.
Exhaustive use of various search engines on key word searches produces very little evidence of existing web sites. However, it is noted that some surveys have limited shelf-lives and unless you are actually using the WWW at the time a survey is taking place there is the possibility of missing its existence. The majority of sites found focus on surveys of web users with details of who uses the web, the type of connection and software used, and reports of its ever increasing size as opposed to undertaking a survey on other specific subjects. Web-based social survey work appears to be most active in psychology where a number of on-line questionnaire surveys have been found. One such example is given in the case studies.
The categorisation of the literature is demonstrated in the research undertaken by a number of authors. Schmidt (1997), a psychologist, discusses the benefits, potential problems and solutions to using the WWW for survey research, while Coomber (1997) reports on his work of conducting a survey on the Internet. While it is apparent that there was no use of graphics in the survey, the research at least demonstrates the usefulness of such an approach. Mumford (1996) goes on to discuss the technical issues of using graphics on the WWW although not necessarily as part of on-line surveys. The work undertaken by Wherrett (1996, 1997) in landscape preference and Carver et al. (1997) on the siting of Britain's nuclear waste offer insights into the use of graphics on the web for social science based research. Finally, the potential of 3D virtual environments and virtual reality (VR) for social science research are reviewed, highlighting the benefits that this new technology presents.
It is recognised that the WWW provides immense potential for increasing public participation is social science research. Schmidt (1997) notes:
"The World-Wide Web presents survey researchers with an unprecedented tool for the collection of data. The costs in terms of both time and money for publishing a survey on the web are low compared with costs associated with conventional surveying methods. The data entry stage is eliminated for the survey administrator, and software can ensure that the data acquired from participants is free from common entry errors. Importantly, web surveys can interactively provide participants with customized feedback. These features come at a price of ensuring that appropriately written software manages the data collection process. Although the potential for missing data, unacceptable responses, duplicate submissions, and web abuse exist, measures can be taken when creating the survey software to minimize the frequency and negative consequences of such incidents." (p.274)
While Schmidt provides no current example of on-line questionnaires, he has developed software which allows prospective researchers to up-load their questionnaire to a web site where it can administered. The potential researcher can choose from various questionnaire packages allowing the automatic creation of HTML and CGI (Common Gateway Interface) files along with the ability to display the survey results on-line as they appear (see http://or.psychology.dal.ca/~wcs/).
The use of the WWW for survey research presents the researcher with an opportunity to reach much wider populations than the previous scope of traditional surveys. As the available literature in this field illustrates, there are several noticeable benefits, coupled with issues to bear in mind. The benefits of publishing a survey on the web range from increased population access to saving time and money and providing dynamic and interactive surveying. As the web community continues to increase, sample populations will become larger and therefore biases will, in time, decrease. Surveys can also be focused at particular groups who are interested in very specific issues. The administration of an Internet based survey eliminates many obvious processes which a paper based survey entails. These include the elimination of traditional data input, distribution, collection and checking processes. Summary statistics of the respondents input can be provided instantaneously and provide immediate response for their time and effort in completing the survey.
The most obvious requirement of using web-based surveys is access to a computer with an Internet link so that participation or the execution of a web-based survey can take place. There is also a minimum level of software required to be able to view and run on-line questionnaires and this is discussed in great detail by Schmidt. As with any new technology there is bound to be the potential for problems and, as with any good idea, the suggestion of solutions. The potential problems which Schmidt identifies range from incomplete response forms where questions may be left blank, unacceptable responses due to mis-understanding and problems associated with multiple submissions. This creates the problem of having biases in the data which are being collected through deliberate intent to hamper the survey by multiple submissions. Further issues concerning security and data integrity are discussed, but, while this may appear to be a problem to the novice researcher, it appears to be easily overcome through the use of secure CGI programming. It would appear reasonable to assume that a researcher considering using the web for survey work would be well advised to consult with someone who has knowledge of this computer language to safe guard against some of the potential problems that Schmidt identifies.
After taking the time and effort to compose an on-line questionnaire several issues need to be addressed so that the survey is completed and operates successfully. Matters relating to ethics, the validity of the results (especially considering the current web user population) and the fact that not all browsers operate equally need to be accounted for. Finally it is also necessary to publicise the survey, either focusing it to particular Internet groups or to a more general audience.
Large market research organisations such as NOP are also starting to carry out limited web-based research, usually amongst specialist target audiences. They appear to have no plans in the near future to extend this to `traditional' social/political research for the time being as they see "the problems of representativeness and self selection still remain" (personal communication, NOP, 6th February 1997).
Of the web-based surveys that they have conducted, most have used graphics, either to demonstrate concepts, or to improve attractiveness to respondents. In the future, it would appear that sound and video will play some part in on-line surveys. Survey authors also realise the convenience and accessibility of Internet based research being of major help with co-operation rates, although in the short term they believe that there is unlikely to be any significant impact.
A recent web-based survey discussed by Coomber (1997) takes the concepts and ideas discussed previously a step further and reports on research where the Internet was used to survey drug dealers. Recognising that the `target' population would be difficult to identify through traditional survey methods, the work by Coomber highlights how surveys using the WWW can benefit from a respondent remaining anonymous. The author highlights how evidence is emerging which will overcome the problems associated with sample biases.
"While Internet users still tend to be upscale, their overall characteristics are coming more in line with general population averages", and, "Internet access and use are becoming increasingly mainstream" (CommerceNet and Nielson Media Research, 1996).
Once again, the actual questionnaire for this specific piece of research is no longer on-line to view, something which appears to be the norm whereby once a questionnaire has been completed its virtual presence seems to disappear from the web. The work by Coomber concludes that using the web for this type of research offers "exciting new possibilities to the research" (Coomber, 1997) but the limitations of such work, similar to those identified earlier, must be understood.
The literature identified until now in this study has lacked any evidence of the use of graphics within the limited number of web-based surveys. Literature which discusses the use of graphics on the WWW, but not necessarily as part of on-line surveys, focuses attention on the need to widen the data formats which current and future web browsers are capable of handling (Mumford, 1996). Graphics need to be more than just add-ons to web pages and serve more than just an aesthetic purpose. They need to be functional, but at the same time not too large in file size so as to cause long download times which can be off-putting and cause participants, particularly those using a modem, to quit a particular web site. Mumford believes that there needs to be a wider vision with regard to graphics on the web for two reasons:
"graphics can indeed be used to illustrate a point and add interest. We can however move beyond this and have graphical interfaces to information. In the same way as we can display text in a range of styles, we have the technology to allow graphical interfacing to, for example numeric, information to allow it to be displayed in different ways and to be manipulated by the user.""it is also necessary to think of the information being transported across the network and to consider the relationship between the size of that transferred information and the processing carried out locally. We currently have a situation where a great deal of network traffic is being taken up with simple pictures transferred using a format which minimises local processing needs. Would we be better reducing file size and increasing the processing required of the server? The demands on the network do require some serious thinking about a range of other possibilities." Source: http://www.agocg.ac.uk:8080/TechReports/Thirty/Intro.html
These ideas have great potential for on-line surveys. Graphical images can be used within a survey, particularly qualitative research, rather than simple multiple choice or preference choices. Problems generated by graphical file size can also be overcome through local rather than remote processing of graphics and image files. Survey based web pages which are currently making extensive use of graphics, while limited, appear not to be facing major problems with download times. Work being undertaken at the Macaulay Land Use Research Institute on landscape scenic preference makes extensive use of graphics as an integral part of the survey. Participants in the survey are presented with a series of photographs of rural landscapes and asked to give them a preference ranging from a low scenic preference of 1 to a high scenic preference of 7. Further details of this work are discussed in the third case study of this report (Section 4.3). Carver et al. (1997) discuss the need to develop "easy-to-use graphic user interfaces" if participation in social science research over the Internet is to gain any momentum. Sophisticated Internet based systems will only be popular if they provide the participants with the types of interfaces which they can understand. Carver et al. go on to explain the that "different front-ends... ...can be engineered depending on the background of the individual (e.g. layperson, professional, academic, etc.)." By tailoring the front end to particular audiences, research can be directed and targeted to specific sectors. The forth case study (Section 3.4) discusses of this work in greater detail. An area within the Internet community which is receiving increasing amounts of attention and research is that of 3D virtual worlds. The abundance of web sites and literature available in this area highlights the vast array of activity currently taking place in this field to an extent whereby a review of this work justifies a separate report. A virtual world can be describes as "an interactive computer simulation which lets its participants see, hear, use and even modify the simulated objects in a computer-created world." (Hughes and Moshell, 1995).
The ability of the WWW to provide three-dimensional worlds is provided through Virtual Reality Modelling Language (VRML) `plug-ins' to current web browsers. The viewing of VRML through appropriate browsers provides excellent tools for the modelling and internal representation of three dimensional environments which can be transformed into fully immersive systems. As VRML continues to be developed anticipated advances include greater interactivity, behaviour and functionality in shared and multi-user environments. As virtual reality type technologies evolve the applications become almost unlimited. The use and development of virtual worlds is likely to reshape the interface between people and information technology. New ways of communicating with information, the visualisation of information and methods of exchanging information will enhance the current methods and processes of collecting and using the information which can be interchanged between the client and server.
A virtual environment can represent any three-dimensional world that is either real or abstract. This includes real systems like buildings, landscapes, archaeological excavation sites, crime scene reconstruction's and so on. Of special interest is the visual and sensual representation of abstract systems like population densities, information flows, and any other conceivable system including artistic and creative work of abstract nature. These virtual worlds can be animated, interactive, shared, and can expose behaviour and functionality which has the potential to be of great benefit for social science research.
The National Center for Supercomputing Applications (NCSA) is at the forefront of VRML development. Much of the development work and applications of the VRML technology by NCSA provide examples of the future potential of this technology. Examples of the work can be view at the dedicated NCSA VRML web site (http://www.ncsa.uiuc.edu/General/VRML/VRMLHome.html).
Graphics Multimedia Virtual Environments Visualisation Contents