DLI Social Science Team Home Page

Index

Diary

Internal Reports

Completed Papers

Papers in Progress

Conference Presentations

Site Visit and Quarterly Reports

Main DLI Page

Web Client- DeLIver



send comments or questions to: l-neuma1@uiuc.edu

John and Jane Q. Engineer: What About Our Users?


Emily Ignacio

Laura Neumann

Bob Sandusky

1 November 1995


INTRODUCTION

A fundamental component of our DLI project is sociological research and evaluation related to information seeking and use. This piece, synthesized by Laura Neumann, is a joint effort by members of the social science team (Emily Ignacio, Bob Sandusky, and Laura Neumann, specifically) to take a first attempt at integrating, making some sense of, and sharing the large corpus of data that our team has collected over the past year.

The nature of the data and the questions that we were asked to address led us to use a qualitative approach to analyze the data. Each of us, coming at the data from different angles and backgrounds has something different to contribute to the final outcome of this first round of analysis. Additionally, each section of this paper addresses a different level of addressing the issues, each contributing to a multi-dimensional picture.

This large set of data is the result of several separate data collection efforts (see Appendix A). First, there are ten interviews that Laura Neumann conducted with engineering and physics professors at this university in the fall of 1994. They were asked questions about their use of libraries, computers and their general information finding habits. Second, Emily Ignacio and other members of the sociology team cooperated in three focus groups asking professors, graduate students and undergraduates about their journals and computer use and how they found information that they needed. Third, Laura Neumann did a series of observations during the winter of 1994-1995 of the Grainger reference desk and noted the questions that patrons brought to the reference librarian as well as how these questions were answered. Fourth, Laura Neumann and Emily Ignacio did a series of observations and interviews with Uni High students using the computer room and library to find out how they found information. And finally, observations were conducted by Bob Sandusky, Emily Ignacio and Laura Neumann in the Grainger reference area of how patrons were using the different search systems (including Mosaic) to find information.

These data were brought together and were analyzed with several different questions in mind:

  • What search strategies/keys are used and why?
  • What mistakes are made? What are biggest problems/barriers to use?
  • How do users get help? What do they do when they're stuck?
  • What content is sought? (what journals, fields)
  • When and why is full text retrieved?
  • When do people print? Do they print pieces or entire articles?
  • What will people do with the material they've retrieved?
  • How satisfied are users? What do they see as advantages?
  • How much do people understand about the system and its use (how do they conceptualize the system?)
  • What work tasks is the system supporting?
  • Because the purpose of this paper is to provide a multi-dimensional view of the users' information seeking tactics, the results are not organized according to these questions. A separate paper (entitled The Grainger Observations) specifically addresses these queestions.

    This paper is an attempt to share what we have been learning and hearing from our respondents. We are not suggesting that the points brought up are unknown; indeed, we are very aware that our work only supports much of what has already been written. We do think, however, that members of the larger Digital Library project, and specifically the testbed team, would be interested to know what we are finding. In addtion, the statements and suggestions that we mention are those of our respondents; we realize that some of them are not feasible, and others are being worked on already, but again, we thought that the people working on this project would be interested in hearing what our respondents are telling us. We hope that this document will be the starting point in a longer discussion between the Social Science team and other parts of the Digital Library project, especially the testbed team. We want our research to contribute to more fully understand the issues.


    SECTION 1: POINTS OF INTEREST

    In the first section, organized by Emily Ignacio, each question is addressed briefly, pulling out specific quotes and points made in the data on each topic. These are snippets, condensations, and pieces from interviews, focus groups and observations. They were chosen because they are the comments that were repeated time and time again and were found ina variety of sources of data. In addition, people's comments on what they wish a system would do for them and other interesting insights are mixed in.

    Problems with current systems (library/Illinet/IBIS/CD ROMs):

  • system crashes often or is "down"
  • patrons are intimidated by the size and complexity of system
  • patrons do not understand the system: the structure, how to use it
  • in addition, patrons do not understand what information is necessary to successfully search in the system and what difference different types of searches make
  • patrons don't know what is in each database and why it matters, they do not know what they are searching
  • patrons do not understand the results of their searches:
  • example: patron wants to find the journal Physics Today. Gets four options for Physics Today. Doesn't realize that they are different libraries within U of I.
  • "the help is not helpful"... patrons are not getting the help they need with the on line facilities
  • patrons don't have access- there are not enough terminals, the ones with printers are almost always occupied
  • if there is a great deal of information on the screen, patrons will not read it
  • the interface to IO+ and IBIS appears "primitive" and "dated"
  • the lack of time depth of the databases is problematic- people want access to older materials as well as up to the minute things, neither of which is in the databases
  • acronyms always present difficulties: patrons are usure if they can use them to search for information or how to find out what they are, etc.
  • the system is often to slow and seems cumbersome, patrons will not wait for it
  • Other related problems:
  • patrons don't know their libraries; e.g.:
  • what is in the reference section versus the rest of the library
  • what is in "current periodicals" (how current?)
  • what does it mean if the system says something is "snagged" or "missing" or "being bound"
  • patrons sometimes are not aware of the specialization of the library
  • patrons don't know how to access things at other libraries- how inter- library loan works, what is possible
  • inding the correct keywords or controlled vocabulary is almost always a problem
  • patrons almost never use the Library of Congress headings
  • access to CD ROMs is limited to the physical library building. Patrons who don't go to the library want access to these things in their office
  • patrons are not familiar with what is available
  • e.g. the INSPEC Thesaurus system is almost never used unless a patron is pointed there by a librarian
  • Problems finding journals:
  • difficult to find specific issues of a journal
  • journals are not at library when the computer said it should be there
  • get too many hits when searching by title- one person needed to find a journal named "Computer" and got 2000 hits; other patrons who tried to find "Scientific America" and "Nature" had the same problem.
  • all words in title are "insignificant"
  • difficult to find conference proceedings because of the way they are cataloged
  • on-line searches do not give abstracts all of the time. Patrons have complained about having to actually get the articles to see if they are relevant or not.
  • can't get trade publications
  • journals are dispersed all over campus and patrons don't like running after them
  • journals are often stolen or the relevant materials are ripped out.
  • even though reference librarians try to do some instruction on the most efficient way to search for things, for example by ISBN or ISSN numbers, most patrons do not grasp what they are saying or forget journals are out of date too quickly to be useful for some people
  • People want the system to be flexible:

    They want to be able to customize their searches - they want to have a lot of control over the searching process:

  • people want to be able to limit and broaden the search
  • By physical location: many have said that they want to be able to limit the search to Grainger Library
  • others have said that they want to be able to broaden the search: they want to be able to broaden it nationwide and world wide
  • additionally, they want to broaden and limit "on the fly"and to do it while they are in a search
  • People have also said that they want the computer to be able to keep track of what they have searched, in a sort of running search history AND be able to present them with references that they will be interested in based on this search history.

    People want to be able to search "by sight." They worry about the loss of the ability to browse. They have asked if they would be able to visualize the appearance of a book and if there were going to be "virtual shelves." The spatial arrangement of the library is very important. Many professors have stated that they were used to the arrangement of the books in the old engineering library and still have to get a feel for where things are at Grainger.

    Faculty and students mentioned that they would like to be able to search by:

  • color of the book
  • specific graphs: one professor mentioned that she often searches journal articles by looking for specific figures, not by the author
  • page layout: some professors will choose which articles to use by looking at the fonts (if the font looks "dated", the information in the article must be dated (and therefore useless) as well.) Additionally, professors mentioned recognizing an article by the look of the pages and their layout.
  • System should be able to catch spelling errors and offer synonyms or controlled vocabulary terms for key words.

  • Several patrons asked reference librarians for help when they couldn't spell key words or authors' names.
  • Faculty and students want acronym dictionaries, a thesaurus, and a spell checking device in the system.
  • System should be able to allow them to "snowball" search forward and backward - students and faculty want to see how many times a specific article had been cited since its publication, and be able to link to the papers cited in the references.

    How they like to do searches (not in any particular order):

  • by author
  • the problem is, they can't get a list of the articles the author has published through the IO+ system
  • by topic
  • problem: they can't search using acronyms. This seems to be very important to them.
  • problem: they have a terrible time finding the correct keywords or controlled vocabulary terms
  • Most people who are not novices, and even some novices simply need to locate something that they already have a reference for. They are not often doing general searches, and this is particularly true of graduate students and professors.

    Steps that people follow when doing reading journal articles:

    1. Read the abstract or first paragraph

  • they want to see if the article contains information they want to know. They use this to weed out irrelevant articles.
  • If you're just looking for data, go directly to the figures and the results section. ("These speak a thousand words.")
  • Several faculty members spoke of a list serve which sends out abstracts. They skim the abstracts and when they find something that they like, they send out a request, and the full article is sent in 10 seconds. They then print the articles out so they can read it at their leisure. In fields where very recent data are important, services such as these are very popular and greatly relied upon.

    2. Read the introduction and the conclusions to see if the author is writing about anything interesting.

    3. Read the bibliography - to see if the author is citing any important people or articles. Also, they want to make sure that they know what's out there and that they are not repeating an experiment that has already been done.

    4. If any of the above things are helpful, then read the entire article.

    Why access to a printer and copier is necessary:

    They like to print out the full-text of articles because they can highlight them and read them at their own leisure - they're not bound by the placement of the computer. Even students printed out lecture notes that were available on-line because they liked to bring the notes to class so they could follow the lectures better.

  • One faculty member said: "I like printouts. You can go home, put your feet up, turn the lights down and write on the paper."
  • Another professor mentioned: "Its horrible reading papers on the screen...you need trifocals."
  • a graduate student mentioned that "the letters in the IEEE system are too small - you can't scan the article, much less read it on screen. I always end up printing on that thing."
  • People like to flip back and forth between sections of the paper.

  • "I have journals at my office on-line and they are awkward. I can't browse them like I can a paper copy. How do you flip through them? You can be more casual [with a paper copy] than you can with a computer."
  • They also like to highlight and mark important quotes. Some students mentioned that they keep two copies of the same article: then they can have one "clean" version of the article and mark the other article up.

    The quality of the printout matters as well: one student gave his reason for downloading his files while using the INSPEC system instead of printing them: "because of the quality and speed of the printer (it is a crappy little dot-matrix printer)", but notes the time is probably not an issue, seeing as it is taking quite a while for the disk to do the downloading. He mentions that he will take this disk to his office where he has a laser printer.

    Additionally, he explains that "most people use the IEEE system for the free printouts...The worst thing about the Grainger is the photocopy machines. Often you end up walking up and down the whole building looking for one with paper and no paper jam. But when you go and tell the people at the reference or the circulation desk that there is a problem with all the printers in the building, half the time they won't even go and fix them. That is why the IEEE system is really appealing."

    Why they thought we should we have access to full-text retrieval:

  • "I hate having to run all over campus just to find some articles only to find out that the journal isn't there!"
  • the necessary pages are often torn out
  • Inter-library loans take too long.
  • Some journals the students need are being bound or are snagged, so they can't get to the article.
  • Why they use journals:

  • Need to see what others are doing (keep on top of current research)
  • Learn more about what a certain author is doing.
  • See how theories are applied (undergraduates said this often)
  • textbooks are too general - journals have more in-depth, detailed information
  • helps people choose research topics (undergraduates said this often)
  • "I use journals just to learn material, like a textbook but there is no text book for what I am doing." (graduate student)
  • Relevant parts of journals (in no particular order):

  • abstract
  • bibliography
  • calculations
  • raw data
  • data
  • numerics
  • figures
  • indexes
  • Other important things besides journals (in no particular order):

  • class notes
  • dissertation/theses
  • conference proceedings
  • equipment specification sheets
  • technical reports
  • books on programming
  • course books
  • software information
  • lab assignments
  • reference sheets
  • information on publicly available programs
  • specific programming commands
  • Note: everything about doing research is based on the socialization process of becoming a professional:

    What topics to use:

  • instructors often give students very broad topics and encourage them to narrow the topic on their own.
  • Use of the system:

  • people need very basic instruction to learn the system. In the beginning of the semester, reference librarians make an effort to sit them down and give them very basic instruction. In addition, students mentioned being exposed to library resources through introductory classes, residence hall programs, and peers.
  • How to find topics: (see above for narrowing topic)

  • Upper level students and graduate students have to learn how to find new research ideas. Professors often teach students how to do this. Professors see their role as trainers of graduate students, they pass their information finding habits on to their students: "there is an educational aspect to this and the student really does have to learn how to do some of these things. When you first take on a grad student, there is really not a great deal of work that you can get out of them. I don't mean to be insulting, but usually, in my case, the students have to be here for about 3 years before they are really contributing. They are learning until then and so, you can give them things that will help them learn, but you can't give them everything to learn.... my students, of course, are real computer users, and if they can get the important information from the computer.... they are probably being guided by me to go and look someplace else."
  • We have not yet obtained any information concerning who trained the faculty and staff to use the systems. We hope to gain more knowledge concerning this topic by conducting more observations and interviews with these users.
  • Other things people mentioned they would like:

  • "Make it idiot proof."
  • keep track of personal histories of searches
  • Visual Aids- "We need more visual aids." "There are too many technical terms - the figures help a lot."
  • Acronym dictionary
  • should be reference links - should be able to click on a citation and be able to look at that paper
  • ability to click between sections of paper (somewhat simulates "flipping" between parts of a paper.)
  • better ways to narrow the search:
  • ability to pick your own time frame to narrow or broaden search
  • should be able to pick the time frame ("I'd should be able to narrow the time frame to 6 months, 2 years, or between 1963-1973 if I needed to.")
  • there should be an "except" button: some people want to exclude certain authors from their search
  • geographical options: prompts of local, national, or worldwide
  • ability to download article to a disk
  • ability to e-mail things
  • ability to cut and paste from articles
  • ability to digitize graphs to get at the original data
  • have conference presentations on-line

  • SECTION 2: FRAMING THE ISSUES

    This section, by Laura Neumann, attempts to offer a more "meta-level" or a slightly more broad view of what seems to be falling out of the data. This section is not only about finding useful quotes or points; instead, quotes are contextualized, which offers some broader ideas on how to think about search strategies. Like Section 1, this analysis is based on all data collected.

    Search strategies: How do people find things?

  • as would be logical, this seems to vary on the level of experience a person has in their field, with information systems, and in general (roughly broken into novice, intermediate, and expert).
  • although it seems that experience in a field is the most important (i.e. a person may have zero experience with many library systems but still can find information in her/ his own ways; s/he has had time to develop her/his own strategies.)
  • this also seems to vary on what the task, why the person is searching for information, is the person looking for a specific piece of information (a number, a formula, a reference), and a general area (what has been written on X topic, what has Y author written this year, how do articles in Z journal look?).
  • in addition, the medium of the information the person is searching for will change how they search- electronic or paper; book or journal; conference proceeding or technical report.
  • finally, the method of reading a document is also connected somehow. There seems to be two levels of searching: searching on a higher level for a document and then searching that document for a piece of information. Thus the question of how people go through documents is relevant here.
  • Novices

    Undergraduates and the Uni students are lumped together in this category. Most of their searching is done by keyword and topic. They also want to search by location, they want to be able to search on the collection by locality, i.e. search only through the holdings of the undergraduate library. The Uni students actually can do this, search only the Uni library. They focus less on the author and the journal, and seem less discriminating about the quality of what they retrieve and rather focus on the amount of effort it takes to obtain the document. They do some limiting by year. They do some browsing, and particularly for the Uni students, the search on the Internet in non-directed and creative ways, remembering paths, contexts and the look of information rather than the traditional specifics that items are usually searched for by.

    It seems that these people are looking for information on a general area rather than specific pieces of information. They do not have a deep awareness of the formal search strategies and systems that are available. In their inexperience, they do not have a deep awareness of the field of possible information that they are searching and seem to move rather randomly about in it.

    Intermediates

    Graduate students and newer professors fall into this category. These people are getting a feel for their field and thus have different search tasks and strategies than either the novices or the experts in addition to sharing the search strategies of both. These people have the peculiar task of placing themselves and their work in their field, finding their niche and the niche of others. They do keyword or topic searches on occasion, but, since they have a better feel for the information they need and want, are more discriminating in the items they will accept and retrieve. They are more concerned with finding the correct keyword and express a strong desire for a synonym finder, a thesaurus and an acronym finder to be readily available. In their searching of a general area, they do more browsing of a directed sort, for example reading through all the recent issues of one particular journal. In their searching for a particular piece of information and in their awareness of quality issues, they do more snowball searching, following the references back from one article. They express a desire first, to have live reference links- to be able to only click on a reference and be brought to that paper, and secondly, to be able to follow references forward in time, to see who has cited this paper since it was written. As more experienced people, intermediates seem to be more familiar with the formal search systems available to them and use them fairly heavily. [why are these most useful to this group?... the particular pitch of the contents] They also mention using the "find" functions of software to aid them in getting to where they want to be. They are able to pick out what information is most useful to search on and where to go to find what they are looking for (a greater familiarity with the contents of databases.) Intermediates are in the early formation stages of their own personal information finding infrastructure. At this point they are beginning to develop workarounds and rely more heavily on the formal structures available to them.

    Intermediates seem to also be more apt to ask others for information, for pointers or suggestions as to how to go about finding things or search strategies. As graduate students, they are in a unique position to do so, as professors point out, a large part of their job is guidance of the graduate students in getting to know the field.

    Experts

    Experts are those people with a great deal of experience in their field. I define them as people who have built up their own information infrastructures in which they have refined their workarounds to the point that information comes to them. These people rarely do keyword or subject searches, they know their field well enough that such a general search is probably not useful. They instead focus on finding specific pieces of information. They have a "feel" for their field, but they do browse on occasion just to see what is out there. They do less going out and finding information, as I said, they have established their personal infrastructure so that the information they need comes to them in one way or another. Colleague and friend networks are very important in getting the information that they need.

    Their personal information infrastructure can be understood as located along several different continua:

  • they vary from using formal (IO+, current contents services, conference proceedings, subscriptions, etc) infrastructure to informal (colleague networks, bulletin boards, being an editor, browsing.) It is interesting to note that these vary and crossover in interesting ways. What is the line?
  • they vary in relation to the established channels, within established channels (IO+, current contents, being an editor, etc.) to outside established channels (using the news groups or the Internet).
  • they might be institutionally supported or provided (IO+, Illinet) or might rely only on personal initiative (browsing, journal subscriptions, being a reviewer)
  • each of these may be used formally (doing an author search in IO+, reading conference proceedings, or subscribing to a list server) or informally (using IBIS to find an L.C. heading to assign to an article that is being written, going to conferences to meet with friend
  • colleagues to chat about the latest research, contacting a librarian at an institution that you used to be employed at because she can find what you need.)
  • For experts, the office as an information universe becomes an "archaeological dig." It can range from fully ordered (everything carefully files and recorded) to what looks like utter chaos (piling, shuffling, lack of formal easily recognizable system). The process of finding things in the office is about discovery... there is serendipity, the "methods" of piling as cataloging and indexing, friends with whom information is exchanged very informally, colleagues with whom more guarded exchanges occur. The office has various inputs of information: the phone, the door, the mail slot, the Internet connection. There are information flows within the space of the office, information traverses a path from entry to placement in different areas, to replacement, and over again. People navigate their offices and find things in a process that includes hunches vague recall about the "look", the "feel or flavor," the color, the density, the length, the accessibility, the source, and the type of the information, among other things.

    What is done with retrieved information? What work is the system supporting?

    Why do people go and get things?

    This question also seems as if it would vary on

  • the level of experience a person has in their field, with information systems, and in general (roughly broken into novice, intermediate, and expert)
  • although it seems that experience in a field is the most important, a person may have zero experience with many library systems but still can find information in her/ his own ways; in other words, s/he has had time to develop her/his own search strategies.
  • the method of reading a document is also connected. There seems to be two levels of searching: searching on a higher level for a document and then searching that document for a piece of information. Thus the question of how people go through documents is relevant here.
  • finally, this issue is closely linked to the type of information that is retrieved. Technical reports will be used for a different task than will journal articles, than will books, than will conference proceedings. Of course, each of these has some commonalty of use, but each also has its own flavor of utility.
  • Novices

    As people finding their place into and in a field, the novices' tasks differ greatly from others'. The undergraduate and high school students are doing work for their courses, first find and narrowing down a topic for research, then finding references and writing the paper. They also use articles as models for their own work, they are trying to find out what a paper should look like, or what their research should look like. Occasionally they will use a retrieved piece of information to snowball search for other information, but it does not seem that this is common. Most of the information that they need for their courses is given to them in their textbooks or course notes, but sometimes they will go to journals to get a more detailed piece of information. These people, again, are working in a more general searching mode.

    Intermediates

    Intermediates, again, graduate students and new professors, due to their lack of experience in the field are using information in yet a different way. They are using it to place themselves and others within and between fields. They are gaining and understanding of what is out there and what is relevant. They are developing standards of quality with which to evaluate information. They use some documents to point them to other documents through the reference sections. They also use journal articles in particular to model their own work: to model their methods and techniques in their work and to model their writing. They look to articles as examples of publishable material which they strive to follow. They do literature searches for papers they write, for thesis research, for background information on experiments that they are doing. Additionally, intermediates will look to journals to retrieve single bits of information, a formula that they do not want to personally derive, a specification, a graph. These people are in the general search arena as well as looking for specific pieces of information.

    Experts

    Experts, again, due to their knowledge of the field and experience do less general searching to find all about something and do more retrieving of specific bits of information or specific documents. They are also using these documents to keep up with their field, or to retrieve a specific number or fact that they need for comparison with their own work.

  • the difference between research work and teaching (teaching requires more thoroughness?)
  • There is a difference between research on immediate topics versus historical topics. Each requires a different search strategy which covers different places and types of information.
  • when finding stuff for research:
  • They model own work, compare what they are doing to what is out there.
  • See what else has been done by other people, see if they are doing something new, find out how their findings compare to others'.
  • when finding stuff for paper writing:
  • Again, they model their own work against others'.
  • They are doing a literature review.
  • Again, they are seeing what others have done along the same lines.
  • when finding stuff for proposal writing
  • Again, they model their own work against others'.
  • They are doing a literature review (marshaling their allies).
  • Again, they are seeing what others have done along the same lines.
  • finding specific bit of information (formula, #, name, another useful reference, etc.)
  • use documents to lead to other documents by:
  • full reference
  • call number
  • author
  • journal
  • keywords
  • professional status maintenance for experts ("field mapping" for intermediates)
  • keep up with new work of colleagues
  • see what's out there if they are embarking on something new
  • These points are all linked back. What you are looking for will determine how you search and vice versa.


    SECTION 3: MORE QUESTIONS TO BE ASKED

    This section, by Bob Sandusky, is an attempt to bring in other related topics and other questions that might be explored and new angles on how to investigate the issues in question. Rather than making individual points or trying to take a meta-level view, this section refers to a more personal experience of exploration and ideas that were drawn in as the exploration went on.

    The Search Context

    This is about the contexts within which people do on-line searches at Grainger.

    I want to focus on the person's motives, the references they have in hand when they sit down at the terminal, what their quest is, what motivates them, what sort of help they have had in the past (training, experience) or are using this time, which systems they are using and why. What results do they obtain, in which format, and what do they do with them. Maybe a term to use is the externalities of the search process. I'm not looking at the details of how the system works, the specific search strategies used by the user, etc.

    Another way I make sense of this is by using the black box metaphor. I'm trying to focus on the inputs and outputs of the black box. The black box is the particular system(s) and commands and responses. I'd like to look at what's offered to the black box as input / user knowledge / starting points and what comes out when they logout or stand up and walk away.

    I can't make any claims for causality here: there are too few subjects per system to make it a valid sample, and we aren't putting in controls to make this an experiment.

    Primary data for this is the observations done during this summer in the Grainger reference area. The interview and focus group transcriptions are also informing this little bit of analysis.

    The first parameter of the search context is the motive for the quest. When a user comes to the library or logs onto an on-line information system, there is motivation. The quest that the user is embarking on is not likely to be trivial. Even the most straightforward of these systems (probably IO+/LCS) is not easy to use. I think that the typical user isn't going to begin using these systems casually or for entertainment (vs. e-mail or WWW). So, the question is what motivates people to expend the energy to log on or come to the library to use these systems? So far, the Grainger observations show a variety of motives for searching, from job searching, to experimenting with unfamiliar systems (just sort of messing around with them to see what they can do), to directed browsing, to looking for very specific subject matter, to looking for specific information objects (like an article) [see details below]. In contrast, some of the interviews and focus group results I've looked at seem to talk more about normative motives like writing articles and proposals (faculty and graduate students) or completing assignments (undergraduates). The real life observations are useful in identifying a wider range of motivations for usage. This is interesting to me also because many of the Grainger observations were done without asking the users directly what their motives were. Their actions suggest some of the motives listed above while others told us specifically in response to our questions.

    The next issue is the references people have in hand when they begin a search. It seems to me that most searches that are directed at all would begin with some specific starting point: a URL, a search term, a vaguely or specifically conceptualized topic or subject, and author, a citation, an article with citations, a title, a journal, etc. Several of the Grainger observations were of people who did not seem to have any concrete references (on paper). Some of those without paper seemed to be fairly well focused (perhaps because the user had the subject /author/etc. in his or her memory) but a few seemed very scattered, almost random. The latter seem to be more likely to be experimenting with the systems to get a sense of what they are like or what they can do. How strong of a relationship is there between motive and having a reference at hand; a starting point? The interviews seem to get at this at least a little, I will look there, also.

    Training or experience in the past and getting help is next. So far, we haven't asked anyone about training or experience they have had with specific systems. In a few observations, we have seen the users talking to librarians or other users about the systems. In a couple of cases they have interacted with the observers! Also of interest is the use of L.C. subject heading lists (on IO+) and use of thesauri: generally, I call this getting help with words. I also think in terms of help from on-line sources and help from off-line sources. Librarians, other users, books, pamphlets are examples of off-line help. Help screens (and maybe BI) are examples of on-line help. Perhaps the interviews and focus groups will provide more data here. It seems to me that the on-line help and information at the terminals is extremely thin, and not of much use. If someone really wants to know how a system works and / or what information it covers, how do they do that?

    It's interesting to look at which systems people choose to use and why they use them. Is their choice based upon familiarity and comfort? Capabilities (a lot of people like printing the full text for free from the IEEE system)? Knowledge and awareness of the coverage of various systems (I doubt it because I have a hard time figuring this out)? Don't have a lot of concrete information about the why here, we can observe pretty easily what they choose to use.

    What they take away from the search session is important, too. This is usually easy to observe, but we have only asked some of the users we've observed exactly what they plan to do with the results. The interviews and focus groups might provide some normative types of responses. I'll look there, too.

    Sometimes people seem to take the citations from something like I.E. or IEEE and go to IO+ to check holdings. A lot of people just walk away. Do they follow up? How?


    CONCLUSION

    This work is only a first attempt at organizing, making sense of, and sharing the information that the Social Science team has been collecting. It has been generated through the use of qualitative methods, specifically grounded theory. The goal of this theory is not to generate statistics but rather to reveal trends, commonalties and important points in the data and then to situate these in a larger analytical framework. Each member of the sociology team comes from a slightly different background, as should be obvious from reading this paper, but we see these different views as helpful in revealing multiple facets of the issues. We hope this document will lead to discussion between the Social Science team and interested parties on the rest of the Digital Library project, especially the testbed team. We not only want to discuss the findings in this document, but also the method that we used to analyze the data, how it is or is not useful, and what is to be done in the future.

    This paper is an attempt to share what we have been learning and hearing from our respondents. We are not suggesting that the points revealed are novel; indeed, we are very aware that our work only supports much of what has already been written. We do think, however, that members of the larger Digital Library project, and specifically the testbed team, would be interested to know what we are finding. Secondly, the statements and suggestions that we mention are those of our respondents. We realize that some of them are not currently technologically feasible, and that others are being worked on already, but again, we thought that the people working on this project would be interested in hearing what our respondents are telling us. We find their comments and actions fascinating and wish to use what we are finding to help build a useful digital library.