What users desire from and how they react to a single search tool is well documented; speed, accuracy, relevant materials first, de-duplication, and an easy to search interface.⁵ However, the needs and desires of librarians, who are also significant users of the product, are less explored. Librarians have not always taken to federated search tools in the same way as their students.⁶ They often consider the unique fields or specialized search terms significant, especially for more advanced research, but also for undergraduate students who need to be able to develop the skills to extract the most relevant information from the datasets retrieved. With the loss of this functionality, students often retrieve huge amounts of information that is not necessarily relevant to their topic. For this and other reasons, many librarians are reluctant to teach the federated search tool that their library has. Lampert and Dabbour surveyed librarians nationally and held small group discussions; they found that two-thirds of the librarians, who were themselves confident in their abilities to search federated tools, did not teach federated search tools. The librarians disliked the lack of certain functionality. Lampert and Dabbour, like others, found librarians had somewhat ambivalent feelings about federated searches.⁷ Marian Hampton, guest contributor to the Serials Review column “The Balance Point,” noted that single interface searching across many sources is a benefit to metasearching but that librarians are “concerned that federated searching presents an illusion that information is simple to find.”⁸ In the same column, Pugh also commented on the tension between wanting to improve students’ information literacy skills and offering them an easy search interface.⁹ In their study of the MetaLib Combined Search (MCS), Tang, Hsieh-Yee, and Zhang found that librarian views were more negative of the usefulness of MCS when compared with student perceptions. The librarians found it “not useful, and hard to figure out.”¹⁰ While librarians have had both positive and negative reactions to federated search for a variety of reasons, specific concerns were the search interface and the results display. McHale pointed to the lack of date and peer-review filters, the lack of controlled vocabularies and indexes, and that most federated search tools do not include all databases.¹¹ Lampert and Dabbour also pointed to the loss of a controlled vocabulary, the inability to limit to peer-reviewed, and the loss of individualized features of specialized search engines as reasons why librarians were not whole-heartedly embracing federated search tools. They also found that users have a hard time distinguishing between the types of resources retrieved.¹²

Designing a search interface is a lengthy, iterative process involving many stakeholders. Research on how users seek information is presented primarily in the information science and information architecture literature. Information science researchers have been studying how users seek information since before World War II, although this area of study flourished beginning in the 1950s.¹³ Early information seeking behavior studies were more concerned with the use of information sources and how users navigated an existing system rather than studying information needs. These studies led to a better understanding of how people seek information and how the design of a system can aid or hinder a user.¹⁴ As information sources went online, librarians began investigating how users interacted with databases, and later, webpages to locate information.¹⁵ These studies showed that an understanding of information seeking behavior can positively influence search interface design.¹⁶ New models of information seeking behavior are being developed as search engines, such as Google, change how people search and retrieve information.

Another important source for the design and layout of search interfaces comes from information technology literature. Usability testing, which software engineers have been conducting since the 1970s, has greatly improved how search interfaces are designed because they are based on actual tasks users perform. Usability studies originated with computer software design firms but were applied to web interface design in the early 1990s.¹⁷ Pioneers in usability testing of webpages include Jakob Nielsen and Jeffrey Rubin whose works are often cited in the library literature as well.¹⁸ While usability testing has focused more on webpages and web interface designs than database design, the principles of user-centered design are central to both. Librarians have taken these techniques, often developed to study consumer behavior, and applied them to library webpages and search interfaces.

A common theme among all these studies is that search behavior changes. Searchers will move between browsing and searching depending on the results of their search.¹⁹ Web interface designs also need to take into consideration that users bring their own experiences and expertise to a search.²⁰ Researchers in information behavior emphasize that search is an iterative process so designs need to support users in refining and reformulating their search.²¹ The fact that different user behaviors may require different designs and layouts makes designing webpages a challenge. Over time, some de facto standards and best practices have developed to help designers create useful search interfaces. Kalbach summarizes different search behaviors and suggests design strategies that support each of these.²² Rose argues that search interfaces need to support the following information seeking behaviors:

• Different interfaces (or at least different forms of interaction) should be available to match different search goals.
• The interface should facilitate the selection of appropriate contexts for the search.
• The interface should support the iterative nature of the search task. In particular, it should invite refinement and exploration.²³

Layout and design of a page has a major impact on the users’ success in locating.²⁴ In their book Search Patterns, Morville and Callendar lay out some of these de facto standards for the design of features and functionality that make for a good user interface.²⁵ Another useful source for design guidelines is the Research-Based Web Design and Usability Guidelines from the U.S. Department of Health and Human Services, which distills much of the research on good web design into a single document.²⁶ Russell-Rose describes the use of design patterns as a way to design a set of webpages that support different user search strategies and information seeking behaviors.

Although each website needs to be designed in context, there are a number of features and layouts that work well overall. Research supports the use of progressive disclosure so that as the user moves toward their desired target, the information provided by the system becomes more relevant and specific.²⁷ Most design guidelines also strongly recommend the use of hierarchical faceted metadata which Morville and Callendar, corroborated by other researchers, refer to as “arguably the most significant search innovation of the past decades.”²⁸ Other recommendations include the use of filter and branching to help users refine their search; content type and format are the most useful filters.²⁹ This includes features like auto-complete or auto-suggest that provide immediate feedback to users. The system should also allow alternate views such as sorting by date or author.³⁰ Hearst et al. recommend providing users with help and immediate responses. Layout too can influence user success.³¹ For example, placing the primary navigation menu in the left panel or along the top and placing navigation bars at the top of page layouts promote success.³²

While information seeking behavior research and usability studies have had a strong influence on how search interfaces that meet different user needs are designed, the results of this research and these studies are not always systematically applied to the actual design of an interface. Search interfaces, Rose states, “often reflects the attributes of the tools used for searching, rather than about how people look for information.”³³ Ferrara notes that search engines are often built for very narrow search strategies that basically support just keyword searching.³⁴ This kind of design overlooks the basics of human search behavior.³⁵

Participatory design is a user-centered design method for involving users in a design process of an end product such as a webpage or a search engine interface. Participants act as co-designers at the design's formulation stage with a goal of creating a usable product incorporating new technology to meet their needs. Participatory design originated in Sweden as computers were being introduced into the work setting; key concepts included “democracy, power and control at the workplace.”³⁶ Indeed, a participatory design process asks workers to voice how best to integrate technology into their work practices based on their knowledge, experience, and expertise. Bowler et al. highlight the iterative and circular aspects of participatory design and note the vital contextualization it brings to the design process—meaning that can be overlooked in usability tests.³⁷ Bowler also notes that as “new approaches have been added to the collage,” the definition of participatory design has changed.³⁸ Moreover, as the method has gained traction, its strengths and weaknesses have become apparent. Spinuzzi recommends it to “rapidly and incrementally improve” the product in question, to focus on “democracy in the workplace,” or to tap into users’ strengths. However, it is not the best method for securing user input when “radical change” is sought because of its focus on using an existing design, which can limit big picture thinking.³⁹

Participatory design is a growing area of interest as evidenced by a conference devoted to the practice and numerous publications detailing practitioners’ experiences.⁴⁰ Interest in and use of this method has also grown among librarians and libraries. The publication Studying Students: The Undergraduate Research Project at the University of Rochester,⁴¹ popularized this method in the library profession, generating enough interest that the editor, Foster, now offers workshops through The Council on Library and Information Resources for librarians wishing to learn more.⁴² Rochester is one of several libraries that has employed participatory design methods to improve library services; here we focus on libraries use to improve websites and users’ ability to access library provided content.⁴³

University of Rochester Libraries (UR) 2007 ethnographic study of student research behaviors reported on their successful use of participatory design workshops to discover what students want from library space for research, how they research and write papers, and what they want from a library website.⁴⁴ Though several methods were employed to gather student input, the participatory design workshops are most pertinent to this article. During the workshops students were asked to design a website from scratch and to redesign the libraries existing site based on paper mock ups. This study typifies their design process and the value of integrated and repeated use of participatory design. They continuously employ user centered methods to learn about their users to improve services. For example, they turned to participatory design workshops to learn about faculty and graduate students use of the library website and what their ideal site would include.⁴⁵ The workshops helped “create an informational basis” for identifying key tasks when using the library website.⁴⁶

To gain ground on solving seemingly intractable barriers users encounter when navigating libraries’ siloed content, Rochester partnered with research teams at Cornell University, Ohio State University and Yale University in the eXtensible Catalog project.⁴⁷ The teams studied student and faculty research processes to learn more about these barriers and facilitated “a participatory, collaborative process of co-creation” to design a new software suite that will search across multiple content repositories.⁴⁸ Their focus on work processes broadened their research horizon beyond a focus on tweaking features to understanding how students and faculty conduct scholarly work in an information environment where library provided resources share space with many other providers.

Nikolova-Houston used participatory design techniques to revise her websites featuring Slavic medieval manuscripts.⁴⁹ She used three stages of participatory design: discovery, site evaluation, and site prototyping. Discovery included semistructured interviews, written surveys, classroom observation, and review of documents such as syllabi. Then, she asked her participants to evaluate her existing sites using common usability criteria and, last, she held one prototyping session. Participatory design “created a more intimate social atmosphere” between her and her participants and gave her confidence in her understanding of their tacit knowledge and future needs.⁵⁰

Participatory design is a successful method for involving users in a design process to learn about their needs, thoughts and opinions about what they would like in a site and for demonstrating to them “that the online sources they design can be used for academic teaching and research.”⁵¹ However, additional research demonstrating the effectiveness of user created and inspired designs would be welcome.⁵²

Even though we asked the groups to design the results page rather than the initial search page, all groups followed good practices by including a basic search feature on their designs.⁵³ Some added features that reduce users’ cognitive load. For example, two groups included an additional place where the previous search displayed so users can see what was searched.⁵⁴ It is possible that the other groups assumed that the search entered would remain in the search box. One group wrote the search phrase “Mark Twain” into the search box probably with this intention.

Surprisingly, only three groups offered a link to an “advanced search,” although they did not specify which fields they would have included. Our initial assumption was that this would be a very important feature since librarians have often commented on the lack of precision in federated search tools.⁵⁵

All of the groups had options for refining the search; this is where we found the most creativity in the designs and it was clearly very important. The most common way of refining a search was through the use of facets usually displayed on the left hand side, a standard recommended by Morville and Callender.⁵⁶ The most common facet was content type which Morville and Callender consider one of the most important refinement options.⁵⁷ The most prevalent content types were book (n = 3), media (n = 3), and article (n = 2), but in some cases, the groups simply included a button labeled “content type” (n = 2) which leads us to believe that once the user clicked that button, he or she would be presented with the appropriate content types. One group added websites to the results, possibly as a way to direct users to related materials on the web. Only three groups included an option for limiting the search to peer-reviewed, which was probably related to time constraints but was a little surprising to us since it is a feature librarians have found lacking in some federated search tools. Two groups allowed users to refine searches to just full-text or what is “at my library.” Moreville and Callender consider format the other most significant way to refine a search.⁵⁸

What we found most interesting was the number of groups (n = 5) that added a rating or voting system to indicate they wanted “more like this.” This widget was included in the packet, so we cannot say how many groups would have thought of this without a prompt, but it was clearly popular and is recommended by search interface designers as one way to direct users to the best content.⁵⁹ One group used this feature exclusively, with no other options for refining search results. In this design, the user would refine results by marking them as “Hot” and “Not.” The other groups used terminology such as “more like this.” One group included a relatively sophisticated method to let the user rate a number of articles as “Thumbs up” or “Thumbs down” and then rerun the search using a “refine search using my likes.” One group included a “trash basket,” which we interpreted as a similar rating feature.

Permitting the user an “alternate view” is another feature web designers recommend as “the optimal view differs by user and task.”⁶⁰ Most of the groups allowed a variety of ways to sort results. Five out of the seven designs used the prefabricated sort widget. In most cases, it was placed on the right or at the top of the screen, possibly indicating a need for high visibility. Another possibility is that this feature is often located at the top right in other databases and the groups followed this layout.

All the groups except one devised ways for the user to access or acquire the actual article, book, media item, or whatever else might be displayed in the results list. Although it was not always explicit, we think that for most of the examples, clicking on the article title would retrieve the item, as almost all the results pages had some way to indicate if the full text was available. The participants used differing terminology, including “ours,” “full-text,” and “get it now,” to indicate availability. Some of the groups used the predesigned star icons or images of an article or a combination of image and text to indicate the content is readily available implementing two important design concepts, that of the meaningful label and the use of attention getting features.⁶¹ All groups except two had links to interlibrary loan services on the results page that indicated to us that an easy way to acquire an item was important. One group placed the item on the right side inside a menu along with items like “my lists,” “my library account,” “home to my library,” and “home to my university.” One group allowed users to share their results with others.

Most groups had the results of the search, meaning in this case the citation or information about the article, book, or media item, displayed in the middle column of the page, which is consistent with most databases and other search systems. In most cases, the groups just indicated with a box—here is where the citation will display—and did not include the specific fields displayed, which we think is due to time and a general sense that we would want the standard information (title, author, journal, etc.) displayed. Only two groups specifically indicated that an abstract would be available which surprised us, but as with other features, we do not want to read too much into the absence of this feature. Certainly having “snippets” that preview the content is a useful feature.⁶²

The groups designed a variety of navigation features to get the user to the next page, a specific set of results, to the library homepage, or to other sites. Several groups included a prefabricated “results x of x” in their design to let users know how many items their search had retrieved. The groups also found that getting the user to the university or library homepage was important. They also felt a link to “my library account” was an important navigational feature.

Access to help was a commonly included and desirable feature.⁶³ Four of the seven groups used a prefabricated question mark on their results page. We assumed the authors intended this to mean that clicking on it would take a user to search engine help pages, but we also realized it could link to a webpage featuring other help options like e-mail, text, phone, etc. Five of the seven groups indicated the ability to get immediate help was important by adding a chat widget. One chat box even had a slightly desperate “Help Me!” label attached. Another group supplied a way to report a technical problem.

Tools and techniques to help reduce cognitive load can help enhance the user experience.⁶⁴ Keeping the search terms visible in a search box is an example of such a technique and one employed by the participants. Other techniques might include ways to save a search or item. Every group except one offered the user the option to print, save, and e-mail results, so this was clearly an important feature, as was an option to mark certain items. This feature was very visible on most designs. One group had a “drag and drop” where users could drag their selections to a folder rather than mark them with a checkbox.

Surprisingly, we did not see the ability to manage citations on many of the designs. Most groups allowed for printing, e-mailing and saving of items, and it is possible the designers meant to allow users to format the citations they had selected, but no group stated this explicitly. We did, however, see a number of groups who included the “citation” label widget on their designs, which could be read in different ways, so our interpretation can only be speculative.

We were interested in how the groups would use visual cues on their designs. The Research-Based Web Design and Usability Guidelines recommend using attention getting features when appropriate to direct users to the right item.⁶⁵ In the results section of the display, nearly every group included some text, image or combination of the two to help identify the content type of the citation retrieved. They also frequently included some kind of symbol, such as the star (n = 4) or an image of an article or journal (n = 3), to show that the full text was available.

When we examined the features that librarians had rated as essential or useful in our preconference survey, we discovered some interesting dichotomies (table 1). Of those features, only the ability to restrict by content type actually appeared on the drawings. The other features, such as refine by subject heading, ability to restrict by peer-review, date or location, which the survey respondents rated highly, showed up less than half the time on the drawings. This is inconclusive, but it did make us wonder what impact the act of drawing their ideal search tool rather than choosing from a predefined list had on the participants’ decision to include or not include something.