Alternate/Qualitative Methods

Assignment 4, by Zach Tomaszewski

for CIS 703, Fall 2003, taught by Dr. Elizabeth Davidson

In Assignment 2, I laid out an experimental design to test the extent to which conceptual metaphors shape our use of the World Wide Web. However, it would be possible to use other, more qualitative methods to investigate this question.

Archival Study

A study of the effects of metaphor on Web use requires that we know what these metaphors are before we can investigate their effects. One way to determine what metaphors are used to describe using the World Wide Web would be through an archival study.

Such a study would involve searching through a selected collection of written works describing the Web to determine 1) which metaphors occur when describing Web use and 2) the relative frequency of each metaphor.

Metaphors may vary depending on the author or the intended audience. Also, there may be a difference when writing formally verses informally, or when instructing rather than simply describing. Thus, I would recommend selecting material from a number of different publication media: books, journal articles, Web pages, and listserv/newsgroup postings. In each of these media groups, there should be documents intended for beginning Web users, advanced users/beginning developers, and experienced site designers.

Part 1 would require one or more human readers to go through a small number of documents selected from each of these publication media/intended audience categories. (Documents could be randomly selected, but I would recommend first clustering possible selections into well-known/widely-read documents and those not-so-well-known. With only a small handful of documents in each category, random sampling could produce a skewed selection. Yet we do want to sample from the full range of writing on Web use.) From this small number of documents, researchers can identify examples of different metaphors to describe Web use. This is a qualitative study--finding a single example of a metaphor means that it is used.

(A major limitation of this sort of exploratory search is the small number of documents that can be coded for metaphors by an individual researcher. If the wrong representative documents are chosen, the existence of any metaphors that do not happen to occur in those particular documents will be missed.)

Once the metaphors have been identified, they can be searched for in a larger body of literature, which is part 2 of our study. Using electronic keyword searches, instances of metaphor use can be counted. Their frequency of occurrence can be compared between the different publication media and between intended audiences. Given sufficient funding and research time, these intersections of media and intended audience could be further dissected as to whether the document intends to instruct, or whether it is primarily a review or description of use. It could be checked whether some authors use certain metaphors more frequently than other authors.

Such a study would serve well to explore what metaphors have been used to describe Web use, as well as provide some basic quantitative data on each metaphor's relative frequency in different contexts. Yet while this tells us people use these metaphors when writing about Web use, they do not tell us if they rely on them when actually using the Web. (The study described in Assignment 2 aims to do just this.) Also, an archival study looks only at the works of authors, who probably are more familiar with the Web than beginning users (since they're actually writing about it). People may not use the same metaphors as frequently in writing as they do in speech. Also, determining metaphor frequency by relying on electronic searching is plagued with the technical problems of correctly deriving example utterances of a metaphor to search for, as well as accurately matching all such sought utterances only when they are used to metaphorically refer to the Web. (Such a perfectly precise search with perfect recall is practically impossible.) Yet it is the most cost-effective way to examine a large body of literature.

Participant Observation

Another possible study of the use of Web metaphors is to observe different groups while they actually use and discuss the Web. Since different groups probably think about the Web differently, we would want to observe different sorts of people. For example, we could attend a class that teaches people how to use the Web, a class on basic Web site construction, and a Web designer's conference. Within these different contexts, we could record what sort of metaphors are used and the contexts in which they seem most prevalent. Observing people in their natural environment of using the Web (and explaining to others how they use it) may provide us additional clues as to why some metaphors are more prevalent than others.

A study like this has a number of strengths. We are not restricted to written texts or artificial lab environments. We can observe the full context of metaphor use, as well as ask questions of subjects to further clarify our understanding of their actions. We can get an intuitive feel for how different people relate to the Web, while keeping an ear open for the metaphors they use. Openly watching behavior in this ethnological (rather than systematic) observation mode is more likely to give us some insights into what factors might influence the use of both the Web and the metaphors that describe it.

The weakness of such a study is that its findings rely heavily on the observer's interpretation of the situation. (Though this is really only a weakness from the positivist's perspective.) Besides the disposition of the observer affecting the findings in general, different periods of observation could be affected depending on the observer's mood or attention on that particular day (a question of instrumentation validity). Also, because an observer can only follow a small number of people, her findings may not represent the population at large (poor external validity).

As we've seen, different methods have different disadvantages, but allow us to examine different aspects of the problem at hand. An archival study can discover the variety of different metaphors use to describe the Web. An experimental design can quantitatively describe the extent to which the reliance on these different Web metaphors affects use of the Web. And a participant observation study can give us better clues as to why certain metaphors are preferred in certain contexts or by different groups.