A Review of:
Sahib, N. G., Tombros, A., & Stockman, T. (2012). A comparative analysis of the information-seeking behavior of visually impaired and sighted searchers. Journal of the American Society for Information Science and Technology, 63(2), 377–391. doi: 10.1002/asi.21696
Objective – To determine how the behaviour of visually impaired persons significantly differs from that of sighted persons in the carrying out of complex search tasks on the internet.
Design – A comparative observational user study, plus semi-structured interviews.
Setting – Not specified.
Subjects – 15 sighted and 15 visually impaired persons, all of them experienced and frequent Internet search engine users, of both sexes and varying in age from early
twenties to mid-fifties.
Methods – The subjects carried out self-selected complex search tasks on their own equipment and in their own familiar environments. The investigators observed this activity to some extent directly, but for the most part via video camera, through use of a screen-sharing facility, or with screen-capture software. They distinguished four stages of search task activity: query formulation, search results exploration, query reformulation, and search results management. The visually impaired participants, of whom 13 were totally blind and two had only marginal vision, were all working with text-to-speech screen readers and depended exclusively for all their observed activity on those applications’ auditory output. For data analysis, the investigators devised a grounded-theory-based coding scheme. They employed a search log format for deriving further quantitative data which they later controlled for statistical significance (two-tailed unpaired t-test; p < 0.05). The interviews allowed them to document, in particular, how the visually impaired subjects themselves subsequently accounted for, interpreted, and vindicated various observed aspects of their searching behaviour.
Main Results – The investigators found significant differences between the sighted participants’ search behaviour and that of the visually impaired searchers. The latter displayed a clearly less “orienteering” (O'Day & Jeffries, 1993) disposition and style, more often starting out with already relatively long and comprehensive combinations of relatively precise search terms; “their queries were more expressive” (p. 386). They submitted fewer follow-up queries, and were considerably less inclined to attempt query reformulation. They were aiming to achieve a satisfactory search outcome in a single step. Nevertheless, they rarely employed advanced operators, and made far less use (in only 4 instances) of their search engine’s query-support features than did the sighted searchers (37 instances). Fewer of them (13%) ventured beyond the first page of the results returned for their query by the search engine than was the case among the sighted searchers (43%). They viewed fewer (a mean of 4.27, as opposed to 13.40) retrieved pages, and they visited fewer external links (6 visits by 4 visually impaired searchers, compared with 34 visits by 11 sighted searchers). The visually impaired participants more frequently engaged in note taking than did the sighted participants.
The visually impaired searchers were in some cases, the investigators discovered, unaware of search engine facilities or searching tactics which might have improved their search outcomes. Yet even when they were aware of these, they very often chose not to employ them because doing so via their screen readers would have cost them more time and effort than they were willing to expend. In general, they were more diffident and less resourceful than the sighted searchers, and had more trust in the innate capacity and reliability of their search engine to return in an efficient manner the best available results.
Conclusion – Despite certain inherent limitations of the present study (the relatively small sample sizes and the non-randomness of the purposive sighted-searcher sample, the possible presence of extraneous variables, the impossibility of entirely ruling out familiarity bias), its findings strongly support the conclusion that working with today’s search engine user interfaces through the intermediation of currently available assistive technologies necessarily imposes severe limits on the degree to which visually impaired persons can efficiently search the web for information relevant to their needs. The findings furthermore suggest that there are various measures that it would be possible to take toward alleviating the situation, in the form of further improvements to retrieval systems, to search interfaces, and to text-to-speech screen readers. Such improvements would include:
• more accessible system hints to support a better, and less cognitively intensive, query formulation;
• web page layouts which are more suitable to screen-reader intermediation;
• a results presentation which more readily facilitates browsing and exploratory behaviour, preferably including auditory previews and overviews;
• presentation formats which allow for quicker and more accurate relevance judgments;
• mechanisms for (a better) monitoring of search progress.
In any event, further information behaviour studies ought now to be conducted, with the specific aim of more closely informing the development of user interfaces which will offer the kind of support that visually impaired Internet searchers are most in need of. Success in this undertaking will ultimately contribute to the further empowerment of visually disabled persons and thereby facilitate efforts to combat social exclusion.
Recommending High Utility Queries via Query-Reformulation Graph