RGI Project – Exploratory User Study
The global blind population exceeds 45 million (Pascolini & Mariotti, 2011), two million of which resides in the US (American Foundation for the Blind, 2008).The blind comprise a significant user group that interacts with information retrieval (IR) systems, including digital libraries (DLs), in entirely different ways from sighted users. In this study, a “blind user” refers to an individual who lacks the functional sight to see information presented on a computer screen. For these users, interacting with an IR system is a listening activity. They predominantly rely on text-to-speech software called screen-reader (SR) to interact with computers and the Internet (Lazar et al., 2007). An SR identifies and interprets textual content on the screen and presents this aurally through a synthetic voice (Di Blas et al., 2004).
The previous literature reveals only glimpses of the problem. It does not provide an in-depth discussion of how and why help-seeking situations arise for blind users in IR systems interaction, nor does it provide insight into their unique cognitions, perceptions and actions. These shortcomings make it difficult for blind users to understand the information conveyed by SR. They need help mechanisms designed specifically to accommodate their unique DL interaction needs, and to assist them in effective IR.
The literature gap demands a closer examination of their cognition and behavior in DL interactions.
This proposed study represents the initial phase of a multi-phase project. We will develop an understanding of various help-seeking situations blind users encounter in performing specific search tasks, and the types of help needed to deal with these situations. In addition, we will investigate the factors that lead to the help-seeking situations of blind users. Three questions will be investigated: (1) What are the typical types of help-seeking situations blind users face in using DL?; (2) What types of help do blind users need in order to solve each type of help-seeking situation?; and (3) What are the factors that lead to the typical types of help-seeking situations experienced by blind DL users?
To answer these questions, we will conduct a user study with 30 blind subjects from across the United States, and employ multiple methods to collect data. The objective is to recruit subjects with a basic information search skill and at least three years of experience using a search engine through a screen reader to retrieve information from a DL. Potential subjects will be prescreened via a pre-questionnaire and will represent blind users with diverse demographic characterization. Each subject will receive $100 as a participation incentive. Since the blind comprise a low-incidence user population, recruiting a sufficient number of participants can pose unique challenges. To address this challenge, we will work closely with our blind-serving partner organizations in recruitment. In collaboration with project partners, this user study will be conducted in using real DLs that blind users might be interested to use.
In order to identify diverse types of interactions, three typical types of search tasks will be employed in the study: known-item search, specific information search, and subject-oriented search. Known-item searching refers to finding an item when a user knows particular information about that item, such as author or title. Specific information searching is to look for exact data or facts. Subject-oriented searching indicates looking for items with common characteristics (Xie, 2008). These tasks will help researchers investigate the different types of help-seeking situations blind users experience in accomplishing different search tasks. These tasks are comparable to the researchers’ previous study funded by a grant from the Institute of Museum and Library Services (IMLS). The findings will enable researchers to compare and contrast the results of this study and sighted users of previous IMLS study, and to further identify the unique help-seeking situations and help needs of blind users.
Multiple data collection methods were applied to explore blind users’ search behavior and help-seeking situations. In addition to serving as a pre-screening tool, pre-questionnaires will be used to solicit demographic information, and search and system knowledge, including information search skills, cognitive style, help-seeking habits, help-seeking style, etc. The pre-search interview includes questions seeking perceptions about help mechanisms and help-seeking behavior in Internet use. Think-aloud protocols provide detailed information about blind users’ perception of their help-seeking problems, and associated help features needed. Transaction logs show the unique information seeking patterns of blind users. In post-search interviews, researchers asked participants to identify: the typical problems in fulfilling their search tasks, the types of help features they used and whether they are useful, the types of desired help features they require, etc. Due to the mobility restrictions of blind users, the researchers traveled to different partner sites that are convenient to blind subjects.
Data were analyzed both qualitatively and quantitatively. Based on open coding and content analysis, types of help-seeking situations that blind users were encountered during the search process. While investigating blind users’ help-seeking situations and associated help needs, the researchers also explored the factors that led to different types of help-seeking situations. First, qualitative data were analyzed by using open coding (Strauss & Corbin, 1990), which is the process of breaking down, examining, comparing, conceptualizing, and categorizing unstructured data. From the open coding we will identify unique types of help-seeking situations blind users face. Specifically, taxonomies of help-seeking situations and associated help needs were identified, based on the individual help seeking-situations and types of help needs revealed by the search processes of the thirty research subjects. Second, factors associated with blind users’ help-seeking experiences were analyzed by correlation analysis.
Blind subjects encountered physical and cognitive situations in interacting with digital libraries. 872 unique situations that blind subjects encountered in interacting with digital libraries were identified from the analysis of ninety search sessions. Physical situations refer to problems in accessing, identifying, operating, or perceiving during the search process. Cognitive situations represent problems with understanding, sense-making, or reasoning during the search process. Physical situations consist of four types: (1) difficulty accessing information, (2) difficulty locating information; (3) difficulty refining and limiting collections or results; and (4) difficulty identifying current status or path. Cognitive situations fall into five main types: (1) difficulty evaluating information; (2) difficulty with help; (3) confusion about multiple programs or structures; (4) avoidance of format, approach, or input fields; and (5) difficulty constructing searches.
This paper discussed the factors associated with the top physical and cognitive situations, which were selected based on their frequency of occurrence and their uniqueness in relation to blind subjects. The top physical situations include: 1) difficulty accessing information, 2) difficulty locating information, and 3) difficulty identifying current status or path. The top cognitive situations consist of: 1) difficulty evaluating information, 2) difficulty with help, and 3) confusion about multiple programs or structures.
Associated factors that led to the top situations were also identified. 2,102 factors were observed across all 872 unique situations. In many situations, more than one factor was observed. Thirty-three types of factors that represent four categories emerged: User, System, Task, and Interaction. However, findings indicated that the top physical and cognitive situations were primarily associated with User, System, and Interaction factors. User factors refer to attributes that define a user (e.g., domain, system, assistive technology knowledge). System factors refer to different aspects of digital library design. Task factors include task attributes such as task type or requirements. Interaction factors involve the quantity or type of search results derived from the user-digital library interaction.
Design implications provided suggestions for reducing help-seeking situations associated User, System, and Interaction factors for both physical and cognitive situations. For example, to reduce situations associated with User factors such as Inadequate System Knowledge, it is recommended that DL’s provide instructional and contextual features, such as an inclusive demo or tutorial about the DL, or by providing context-sensitive help. To reduce situations associated with System factors, such as Unclear Labeling, it is recommended that descriptive unambiguous link labels need to be prominently placed for blind users to easily identify pathways. For Interaction factors related to No results or Confusing results, DL designers could provide an intuitive pathway to assist blind users in modifying their query to eliminate the cognitively demanding task of starting again.
In-depth exploratory investigation of blind users’ interaction with the American Memory Digital Collections
Funded by Research Growth Initiative, University of Wisconsin-Milwaukee
Designing Help Features to Support Blind Users’ Interactions with Digital Libraries
Funded by OCLC/ALISE
Creating digital library (DL) design guidelines on accessibility, usability, and utility for blind and visually impaired (BVI) users
Funded by Institute of Museum and Library Services (IMLS)