Jump to the bottom
Diane Nahl Home Page

CHAPTER 1

INTRODUCTION AND OVERVIEW


1.1 AFFECTIVE ELABORATIONS AND USER PROBLEMS

With the advent in the 1980s of databases accessible through online catalogs and CD-ROM databases such as ERIC, PsycLIT, Sociofile, and many others, there has been a tremendous influx and growth in end-user computer searching in academic libraries (Jacobson, 1991, p.647; Tenopir and Neufang, 1992a, p.25; 1992b, p.54). Research libraries in the U.S. subscribe to many electronic information retrieval systems, including CD-ROM databases (Tenopir and Neufang, 1992a, p.23).

The number of new electronic sources available in libraries will continue to grow because they save physical space in crowded libraries, and because they provide faster and more comprehensive access to information than do print sources. Electronic sources are popular with library users, since users can search several years of an index simultaneously, search the entire abstract as well as many other types of information in the online record, and print-out or download the retrieved records to a disk. There has been an increase in the number of remote users of online catalogs and databases, especially since remote access to electronic sources throughout the country has been facilitated by the Internet (Page, 1993).

Presently, academic library users are faced with many different systems that they must learn to search. Each system may have different user interfaces, search engines, types of search capabilities, and commands. Librarians in academic libraries all over the country are being deluged with demands for on-the-spot instruction. Individualized instruction at the terminal can take on the average 20 minutes per searcher (Stewart, 1993, p.348; Reese and Steffy in Bostian and Robbins, 1990, p.14; Tumlin, 1993, p.417; Whitaker, 1990, p.31). Despite attempts to provide classroom instruction, people still need help during the search. Since it is not feasible to have a librarian at the elbow of each searcher, librarians have developed written point-of-use instructions to give people the information they need at the time they need it. Presently, most remote users of electronic sources must rely on written instructions since few reference librarians are available to searchers while searching online (Stewart, 1993, p.351).

In an informal field study, it was discovered that, despite the fact that most of the end-users were making errors, having difficulty using the various search systems, and expressing frustration, they were consulting neither the written instructions nor the online Help instructions. Although this type of end-user assistance is available in most academic libraries (LOEX, 1992), a search of the literature revealed that there have been few studies on the efficacy of written point-of-use instructions or the online help facilities that are available on many CD-ROM retrieval systems. Bostian and Robbins (1990) studied the effectiveness of different forms and quantities of CD-ROM instruction with 93 students who all searched the same question in a database. One group only received access to the written documentation published by the database producer. Another group was given in addition to that, printed instructions. A third group had in addition, a lecture on search strategy, and a fourth group had all of that, plus a live demo of the database. So that the last group had the most instruction. This group was the only one that differed significantly in satisfaction, search strategy, and use of written instructions.

Bostian and Robbins tallied the number of printed search aids used by each group. On the average, the subjects used fewer than one aid per person. In other words, the typical student did not use any of the available instructions. The groups that received the most instruction were the least likely to use the printed instructions (p.17).

Bucknall and Mangrum (1992) surveyed 1135 library users on their CD-ROM use, including their use of point-of-use instructions. 21% (n=225) of users reported that they had used printed instructions, but only 8% (n=88) had used Help to assist them with their searches. 25% used experimentation, and 5% said they had no problems and required no help during searching. Approximately one out of three users looked at instructions, one out of four experimented on their own, and most of the rest consulted a librarian.

Markey (1984) did a study of online and offline user assistance for online catalog searchers. She found that all of the 33 libraries in the survey provided printed instructions or brochures for the catalog, along with personal assistance from librarians. When she asked end-users in six libraries how they learned to use the catalog, 37% said they had used printed instructions, and 14% reported using the instructions on the screen. In this study, 51% of searchers reported that they were able to learn by using the written instructions.

When Naismith and Stein (1989) tested library users' comprehension of the terminology used in handouts and librarians' speech, they found that their users understood only half of the commonly used terms. These studies show that written instructions do help a significant proportion of the user population, about 1 in 3, but they are undervalued by the majority of end-users, not understood by most, and need to be changed in order to reach more searchers. It is important to test instructional materials to discover whether and why they are effective in making end-users independent searchers. To demonstrate their usefulness to searchers, existing written point-of-use instructions and Help screens need to be explicitly incorporated into bibliographic instruction sessions. This study experimentally compared two types of written point-of-use instructions for one database system. A number of user variables were investigated such as success, satisfaction, search style, frustration during searching, and self-esteem as a searcher.
Back to the top"

Experience with CD-ROM point-of-use instructions shows that students have difficulty using them. Perhaps these user difficulties are due to the fact that system-oriented instructions are written with a focus on the cognitive (or functional) features of operating a CD-ROM information retrieval system. Their brevity and impersonality tend to inhibit language that relates to the affective or personal needs of users. The research question thus arose as to what would happen if one amended the typical point-of-use instructions by adding sentences with language responding to some of the many affective needs users have while searching, such as the need to avoid frustration by being oriented to certain things, the need to feel self-confident by being advised on certain things, and the need to experience enthusiasm or hope by being reassured in some places. Would users still have the same problems with the instructions, or would these affectively elaborated instructions be more helpful and acceptable to users? Are instructions related to success and satisfaction? Do instructions influence searchers' decisions such as selection of search mode or number of search strategies attempted?

1.2 STATEMENT OF THE PROBLEM

A research plan was designed with the following four steps:
Step 1. A sample of point-of-use instructions for CD-ROM databases used in academic libraries was obtained from a national bibliographic instruction clearinghouse (LOEX, 1992). The sample consists of typical examples of point-of-use instructions used in various academic libraries. Given the availability of the H.W. Wilson Company databases at the School of Library and Information Studies, University of Hawaii, and their widespread use in academic libraries, it was decided to select a sub-sample consisting of instructions written for the Wilson databases. Inspection of the instructions gave the impression that they were written in a style that emphasized cognitive facts and definitions, while de-emphasizing affective statements that meet emotional information needs, such as orienting remarks, advice, and reassurances. However it is desirable to show this effect in a systematic way.

Step 2. Each sentence in the obtained sample of instructions for using the Wilson databases was read in context and given a descriptive title with regards to what it was trying to achieve. Some examples include: "Explains Boolean diagram with example;" "Warns that library doesn't subscribe to every journal covered in the index;" "Explains difference between # and : codes for truncation;" etc. As the literature review in Chapter 2 (Section 2.4.3) shows, these descriptive sentences are called instructional speech acts because they describe what the writer intended to communicate to learners in order to help them acquire some particular distinction or skill. Each sentence was first identified as some particular instructional speech act, and then categorized into one of the three behavioral domains (affective, cognitive, and sensorimotor--reviewed in Chapter 2, Section 2.1). The results showed that there was an equal number of cognitive and affective statements. When the number of affective and cognitive speech acts are equal, the A/C ratio is 1. The Affective/Cognitive ratio is a measure of relative emphasis, and thus an A/C ratio of 1 indicates equal emphasis for the two domains in the instructions. What would be the result if the ratio was increased to 2 or 3? In other words, if there were two to three times as many affective speech acts as cognitive, would novices respond better to instructions? Would this type of affectively prominent instructions influence searchers' success, satisfaction, frustration, self-esteem?

It was decided to find a way of inserting affective speech acts in-between the existing cognitive ones. Of course, these affectively elaborated instructions would now be two to three times longer than the original unelaborated ones. Would the disadvantage due to extra length be outweighed by the potential advantages of the user-centered focus? What would be the specific consequences of elaborating the instructions by adding affective speech acts to the 'unelaborated' original version? This question was put to an experimental test.

Step 3. One of the LOEX instructions was randomly chosen from the set of seven that dealt with the Wilson databases. Analysis showed that it contains 25 cognitive and 22 affective speech acts. Every sentence was then read, and one or more affective speech acts were inserted, where it was felt that a novice user would have a problem interpreting the original sentence appropriately due to its conciseness or allusion to previously unexplained facts. The intent of writing the additional sentences was to respond to anticipated user concerns, with the expectation that the added affective speech acts would help reduce or eliminate these concerns. Upon inspection of the sentences that were added, it was discovered that the sentences could be categorized into just three groupings which appeared to be three types of affective speech acts. Their content suggested the following three labels (see Table 2):
1. orienting searchers to necessary background knowledge; this was intended to reduce anger and help maintain reality check;
2. advising searchers regarding strategies and options; this was intended to counteract anxiety and build positive attitudes;
3. reassuring searchers regarding their legitimacy and success; this was intended to help overcome resistance and encourage acceptance.

When the process was completed, the new affectively elaborated text was three times as long, and contained considerably more affective speech acts (82, vs. 22 before, see Table 6). The A/C ratio was now 2.3. At this point, an experiment was designed to test the hypothesis that the affectively elaborated instructions would be more acceptable and helpful to novices.

Step 5. The design is a two-by-two factorial ANOVA (see Figure 6). Freshman and sophomore students at the University of Hawaii served as subjects. They were tested individually on a search workstation at the School of Library and Information Studies, and given four search tasks: half received simple search tasks, and the other half, more complex tasks. The novices were randomly assigned to one of two conditions: one group received the original unelaborated instructions, the other group received the affectively elaborated longer version. Several measures were designed to reflect how users would react to the instructions and how helpful they found them. Areas of user behavior measured include success, satisfaction, search style, frustration or stress, perceived self-efficacy or self-esteem as a searcher, and helpfulness and comprehensibility of instructions. To ensure the objectivity of the data, the researcher trained two undergraduate experimenters who were blind to the design, hypotheses, and research question. Results were expected to show that novices who receive user-centered affectively elaborated CD-ROM instructions are better able to cope with the challenges of the information retrieval environment in comparison with novices who receive system-centered unelaborated instructions.

1.3 EXPERIMENTAL HYPOTHESES

The following null hypotheses were tested:

H1. There will be no significant difference between the affectively elaborated instructions and the unelaborated instructions on interactivity scores.

H2. There will be no significant difference between the affectively elaborated instructions and the unelaborated instructions on frequency of consulting instructions.

H3. There will be no significant difference between the affectively elaborated instructions and the unelaborated instructions on success scores.

H4. There will be no significant difference between the affectively elaborated instructions and the unelaborated instructions on satisfaction scores.

H5. There will be no significant difference between the affectively elaborated instructions and the unelaborated instructions on frustration scores.

H6. There will be no significant difference between the affectively elaborated instructions and the unelaborated instructions on perceived self-efficacy scores.

H7. There will be no significant difference between the affectively elaborated instructions and the unelaborated instructions on helpfulness and comprehensibility scores.

H8. There will be no significant difference between the affectively elaborated instructions and the unelaborated instructions on knowledge scores.

H9. There will be no significant difference between simple tasks and complex tasks on success scores.

H10. There will be no significant difference between simple and complex tasks on satisfaction scores.

H11. There will be no significant difference between simple and complex tasks on the time it takes to complete.

H12. There will be no significant correlation between success scores for the four tasks (simple or complex).

H13. There will be no significant correlation between success and satisfaction.

H14. There will be no significant correlation between perceived self-efficacy as a searcher (self-confidence) and success scores.

H15. There will be no significant correlation between perceived self-efficacy as a searcher (self-confidence) and satisfaction.

H16. There will be no significant correlation between perceived self-efficacy as a searcher (self-confidence) and frustration during searching.

H17. There will be no significant correlation between helpfulness ratings of the instructions and satisfaction.


CHAPTER 2

THEORETICAL APPROACH


The method and theory developed in this dissertation is a synthesis of four approaches found in the literature on human-computer environments:
(1) educational psychology (taxonomy of user behaviors--affective, cognitive, and sensorimotor information needs);
(2) user documentation (principles of user-centered design and technical communication);
(3) information science and bibliographic instruction (models of search behavior);
(4) text analysis (speech act theory, cognitive engineering).


2.1 THE THREE BEHAVIORAL DOMAINS OF THE SEARCHER'S WORLD

The theoretical perspective adopted in this study is grounded in behavioral psychology and education. Largely through the recent emergence of the field of bibliographic instruction (BI), librarians and information specialists have now become acquainted with the notion that information seeking can be categorized, for the purpose of instruction, into skills that can be identified and defined in the language of behavioral objectives (Dusenbury et al., 1991; Jakobovits and Nahl-Jakobovits, 1987). For the purpose of this study, the assumptions that characterize this approach are summarized in the following propositions.

The searcher's world can be examined from the perspective of the three interacting domains of behavior: Affective behaviors (A) involve the searcher's emotional and motivational reactions to the environment. For example, getting through the steps of a search requires goal-directed persistence to continue to exert the effort to stick with the task. Persisting at the search task, wanting to print something, or taking care to avoid typos are examples of affective behaviors.

Cognitive behaviors (C) involve the searcher's knowledge and decision-making throughout the search activity. For example, repeatedly using the F4 function key to print records on Wilsondisc requires having learned (or knowing) which key it is. To limit a set that is too large, a decision must be made on several available alternatives (more specific search terms, date, language, etc.). Interpreting an instruction requires normal knowledge of natural language.

Sensorimotor behaviors (S) involve the searcher's sensory awareness and motor coordination. Seeing where the cursor is currently located on the screen, finding information on screens, finding a needed explanation in the instructions, or operating the keyboard are examples of sensorimotor behaviors.

Any activity people perform must be a composite of the three domains. Thinking cannot proceed without some direction or goal. The simple act of pressing the ENTER key requires some motive or goal-expectation. Looking for the cursor requires both the knowledge of its function and the desire to locate it for the sake of some goal or interest. Thus, the three domains always act together. Despite this fact, educators designing behavioral objectives for students have separated the three domains and treated them as problems in their own right (Nahl-Jakobovits and Jakobovits, 1993; Jakobovits and Nahl-Jakobovits, 1987). A recent example of this separatist orientation in BI serves as an illustration (Kuhlthau, 1991; 1988) (see Section 2.1.3). The research in this dissertation focuses on the effects of affective elaborations of cognitive instructions. Future research will have to attempt to integrate the three domains in the area of instructional text.

2.1.1 Users' Self-Perceptions in the Three Domains

Bandura (1989) has analyzed the concept of "human agency" in task involvements and has summarized a considerable amount of empirical research on how individuals operate in the three behavioral domains. Of interest to this study is his concept of perceived self-efficacy, which is one of the dependent measures used in this experiment (see Chapter 5). There are several aspects of perceived self-efficacy that relate to the psychological or socio-emotional functioning of users in a search environment:

1. Self-Control
The capacity to exercise control over one's own thought processes (C), motivation (A), and action (S) is a distinctly human characteristic. These three modalities of behavior operate as interacting determinants.

2. Self-Beliefs
Self-efficacy beliefs operate on action through motivation such as affective arousal, and through cognitive sequencing of thought patterns.

3. Goal-Directedness
Human behavior is regulated by forethought that includes "cognized goals."

4. Goal Setting
The stronger their perceived self-efficacy, the higher the goals people set for themselves and the firmer their commitment to them.

5. Perseverance
Self-efficacy beliefs influence the amount of effort people are willing to exert to overcome obstacles: the stronger the belief in one's abilities, the greater the persistence in one's task. Self-doubts cause them to abort their attempts prematurely and to settle for mediocre solutions.

6. Frustration and Stress
Threat is not a property of the objective situation. Instead, it is a "relational property" between perceived coping ability and potential problems in the environment. People who believe they can exercise control over potential problems do not have apprehensive thoughts or anxiety in a learning situation. People who have self-efficacy doubts have a lowered capacity to function and suffer impairment to their immune system (health threat).

7. Strengthening Self-Efficacy Beliefs
Self-perceptions of efficacy are enhanced through mastery experiences (successes) and through modeling influences (simulations and rehearsals).

According to those who have used Bandura's approach to measuring self-efficacy beliefs, "there is a lack of any standardized self-efficacy measure because such expectancies are task-specific," and therefore an appropriate questionnaire must be constructed in each measuring situation (Barling and Abel, 1983, p.267). An application of this principle may be found in another empirical study which found that self-efficacy beliefs affected goal-level, task performance, and goal commitment (Locke, et al., 1984). In their study, the task to be performed was to give uses for common objects. The hierarchy of expectancies was presented to subjects in the following format prior to the task (Locke, et al., 1984, p.243):

__________________________________________________________________
Yes/No Certainty
(0 - 100%)

I can list 2 uses in 1 min. _______ _______
I can list 4 uses in 1 min. _______ _______
I can list 6 uses in 1 min. _______ _______
I can list 8 uses in 1 min. _______ _______
etc.
__________________________________________________________________

The level (or magnitude) of self-efficacy is measured by the number of lines checked off as Yes. For example, individuals who say Yes to "I can list 8 uses in 1 min." have a higher level (or greater magnitude) of self-efficacy belief than subjects who say Yes to only "6 uses in 1 min." The strength of the self-efficacy belief is measured by the average percentage entered by the subjects in the second column. Subjects who are 80% certain they can list 6 uses in 1 min. have a stronger self-efficacy belief than those who give a 50% estimate for the same level (6 uses). A similar approach was used by another researcher on self-efficacy (Lee, 1984).

This method was adapted to the present study by presenting searchers with a list of tasks arranged in order from fewest (one task) to more (up to four tasks) (see Appendix D). Bandura used a similar approach with patients suffering from snake phobia (Bandura, Reese and Adams, 1982). In this case, a method of efficacy induction was used by dividing up the task into smaller and easier units. As subjects' successes increased, so did their experience of mastery and their self-efficacy expectations: "As subjects' self-efficacy level was raised, they experienced progressively less anticipatory and performance distress while coping with threats" (Bandura, et al., 1982, p.5). In this research, it was expected that affectively elaborated instructions would facilitate novice users' view of themselves as capable of doing a variety of search tasks, more than novices who are exposed to unelaborated instructions.

2.1.2 The Generality-Specificity Issue in Affective Integration

The point-of-view adopted in this study is that the three domains of behavior always act together in the sense that no human activity can be performed without all three domains. This insight has been known in psychology since Swedenborg (1843; 1977, p.33, #4667), who realized that "thoughts and intellectual things are but forms appearing thus from the affections." He wrote that learning takes place by the infusion of the affective into the cognitive, which together trigger the sensorimotor (1751; 1975, p. 171, #3158.2). Consider the act of walking somewhere: there must be a sensorimotor component (motor movement), a cognitive component (map or knowledge of the geographic layout), and there must be an affective component (a goal to reach or a desire to continue). Consider the act of choosing a database from a screen menu: there must be an affective component (the motive to search or explore a database), there must be a cognitive component (the reasoning involved in matching the description of the database to the information need), and there must be a sensorimotor component (pressing a key, moving a mouse, etc.).

Another point-of-view adopted in this study is that the three behavioral components in any activity may vary in their degree of integration. Depending on the situation, the activity, or the individual's state of mind, the degree of integration of the three behavioral components may be strong or weak. For example, a disinterested student in a classroom may be taking notes on the lecture (cognitive and sensorimotor components) but the affective component is general rather than specific. The student is taking notes out of a feeling of necessity or coercion relating to grades, graduation, and career, i.e., extrinsic motivation. In contrast, a student who is interested in the topic and the lecturer's point-of-view supplies an affective component that is more specifically related to the cognitive, i.e., intrinsic motivation.
Back to the top"

Consider a library user who feels an initial aversion to automation, which is a negative affective state. At the same time, a user can contemplate the convenience of the online catalog for finding a needed book, which is a positive affective state. User's activities in the library can be conceptualized as consisting of a hierarchy of goals or desires. A goal such as going to the library to find a book is more general than a goal dealing with the details of the execution of the necessary steps, such as looking the title up in the catalog, printing the record, and retrieving it from the stacks. Even more specific is noticing whether it is checked out, or figuring out which floor the call number is located on. In Figure 1, the more general goals are represented as more distant (A4 or A5) than the specific goals (A1, A2, or A3). The more distant a goal, the less it is integrated with the cognitive and sensorimotor elements. The positive affect in the motive to find a desired book is general or distant compared to the motive to become proficient in desirable information retrieval skills. The desire to learn the online catalog is a more specific motivation than the desire to retrieve a needed book. It is assumed that the more specific the relation is between the affective motive and the cognitive reasoning, the more integrated the activity, hence it leads to more successful performance. The schema in Figure 1, depicts these theoretical relationships.

This theoretical schema shows the network of relations between the three components. Each squared item represents a skill in one of the domains. The distance between them (length of the arrow) corresponds to the degree of integration of any skill or activity. This distance is defined as a continuum, which means that any degree of proximity is possible between elements of the three domains. The schema depicts some of the connections between the affective and cognitive components as closer, meaning that they are more specifically interrelated, or more affectively integrated (e.g., A1 -- C1 or, A2 -- C2 or, A3 -- C3), while others are farther, more general, and less integrated (e.g., A4 -- C4 or, A5 -- C5).

An illustration clarifies these relations. The following is a section of an information sheet from Hamilton Library titled "Searching SilverPlatter: ERIC, PsycLIT, Sociofile":

HOW TO PRINT
1. Mark the records you want to print by pressing the ENTER key. (Asterisks will appear to the left of each line.)

The language of this instruction uses a more distant affective justification in relation to the cognitive than the following version of it:

HOW TO PRINT
1. When you find a record you like and you want to print it at the end of the search, you need to mark it by pressing the ENTER key. You will know that it is marked for printing when you see asterisks appearing to the left of each line.

The original wording may be compared to the A4-C1 connection in Figure 1, while the reworded version is like the A1-C1 connection. This may be seen when examining the connections. In the original wording, the goal justification for C1 ("Mark them by pressing the ENTER key") is fairly general: "Mark the records you want to print" (A4). Compare this to A1: "When you find a record you like and you want to print it at the end of the search." Here two affective elements in the searcher's environment are being explicitly stated: (1), the reference to finding a record one likes, or is pleased with; and (2), the reference to wanting to print it later. Note also that the original wording speaks to some 'generalized other' searcher, while the reworded version addresses the individual directly. It can be seen the original wording is cognitively oriented and takes a system-centered perspective. The reworded version is user-centered since it provides affectively relevant functions, as explained.

Here is a second example from a similar information sheet:

"HOW TO SEARCH PsycLIT on CD-ROM
Press F2 (to display FIND command)
Type search term or phrase
Press ENTER key"

Note that system-oriented instructions rely on titles or sub-titles to supply the general affective justification. In this case, the title "How to search PsycLIT" refers to the motive of the user (e.g., "When you want to search PsycLIT" -- which is equivalent). It is possible however to supply a closer motive: "When you come up with some terms or phrase to search and you're ready to type it in" -- this more specifically refers to the searcher's actual situation. Similarly, for the second cognitive line, "Type search term or phrase," it is possible to supply a closer affective component than the current title: "When you have spotted the FIND command prompt on the bottom left of the screen." For the third line, "press ENTER key," the affective elaboration could be, "When you have finished typing in the term and you're satisfied it is correctly typed."

2.1.3 Earlier Studies on the Affective Aspects of Information Seeking

Kuhlthau cites a Master's thesis (Fleming, 1985) and a Doctoral dissertation (Lederman, 1979) reporting research that details how students facing a library research task were "apprehensive or confused about the assignment from the beginning, [yet] they did not seek clarification from the instructor or assistance from the library staff" (Kuhlthau, et al., 1990, p.8). Furthermore, these emotions negatively affected the acquisition of information. Kuhlthau also cites Mellon's findings that three in four undergraduates "described their initial response to library research in terms of fear, frequently termed 'library anxiety'" (Mellon, 1986, p.160). Kuhlthau summarized the evidence as follows:

These studies indicate that uncertainty and anxiety are commonly experienced at the beginning of the search process, and that individual users perceive this to be solely due to their own inadequacy, rather than an experience shared by other users. This perception may limit access to information by inhibiting productive search behavior, actions such as employing expert assistance, tolerating uncertainty, and using the system to form new constructs (Kuhlthau, et al., 1990, p.8).

This statement clearly shows that the earlier conception of how the affective domain influences the searcher is global and separated, in the sense that the affective domain is seen as a problem in its own right. The terms used to refer to affective behavior are general, such as, fear, anxiety, confusion, apprehension, avoidance, and even paranoia against librarians. This characterization of the earlier notion of the affective component is further corroborated by the fact that Kuhlthau uses separate sections to deal with the affective and the cognitive domains (i.e., "Studies of Affective Aspects of Information Seeking" followed by a section titled "Cognitive Models of Information Seeking") (Kuhlthau, et al., 1990, pp.8-9).

These previous observations of a general library malaise are no doubt to be taken seriously. However, the focus in this study is different. The interest here is to examine the way in which individual search skills operate as units composed of the three domains. As pointed out earlier, learning how to print a record on a CD-ROM facility involves acquiring an affective skill unit, which can be stated in an instructional objective as follows:

When you wish to print a record,
press F4 while the record is showing on the screen.

The first part deals with the affective "wish to print," and the second part deals with the cognitive component of the skill (i.e., the meaning of the function to print), as well as the sensorimotor "press the key." One often sees the instruction "Press F4 to print record." Here the idea undoubtedly is that the affective component ("When you want to print it") need not be stated explicitly since it is 'understood.' This may be so, but the purpose of this research is to examine to what extent it may be a disadvantage to rely on the expectation that the affective component will be triggered in the reader without explicit rendition. What has been called here the unelaborated system-centered instructions are written with the usual assumption that much of the affective component need not be overtly expressed, and this is to be contrasted with instructions where the affective component is explicitly spelled out and connected to the cognitive component.

2.1.4 Illustrations of Affectively Elaborated Instructions

Exhibit 1
The following example is a section of a point-of-use instruction sheet. The actual lines from the information sheet are rendered in ALL CAPS and the presupposed affective component is added in lower case.

_________________________________________________________________

WHAT TO DO WHEN THE TERMINALS FREEZE
and you're lost or frustrated because the computer won't do anything

WHEN CURSORS ARE STUCK AT THE TOP OF THE SCREEN
and you're panicked because it won't move down as it is supposed to

IT MEANS THE PRINTER NEEDS TO BE RESET
if you want to fix the problem by yourself (and you can, very easily!)

PUSH THE BLUE BUTTON TO RESET THE PRINTER (REFER TO ACCOMPANYING DIAGRAM)
if you have the confidence to push an electric switch that you normally don't touch (this is a good time to start...)

THE PRINTER WILL BEGIN TO PRINT AND THE CURSOR WILL RETURN TO THE BOTTOM OF THE SCREEN
and you will be relieved to be able to continue your work. You can also congratulate yourself for being able to fix the frozen cursor problem!

_________________________________________________________________

It may be seen from this illustration that when the affective component is rendered explicit, the focus of the instructions change from system-centered (in ALL CAPS) to user-centered (lower case). Instead of presupposing the user's emotions and motives, these are now explicitly referred to, made legitimate, and sympathetically treated with self-esteem building speech acts.

Exhibit 2
Figure 2 shows another example from the same source. Once again, the material in all caps and the diagram are in the original cognitive system-centered instructions, while the affective user-centered component is in lower case. In connection with the content of the affective component, note the use of speech acts that reassure the searcher and echo the satisfaction of goals and intentions ("you can be reassured...", "you can rely...", "you'll be glad"). This illustrates how the language of user-centered instructions is a relational discourse in which the librarians or surrogate instructors relate themselves to the concerns of the end-user. This contrasts with the concise style of system-centered discourse which gives an appearance of objectivity and impersonality. The next illustration further explores this stylistic difference.
Back to the top"

Exhibit 3
The following is a section of an information sheet available at CD-ROM workstations. The new affective component (in smaller font) is added and interwoven with the original (in all caps).
__________________________________________________________________
HOW TO SEARCH
APPLIED SCIENCE AND TECHNOLOGY INDEX
READERS' GUIDE ABSTRACTS
ON CD-ROM
SEARCHING:

1. SELECT THE DATABASE you desire FROM THE "WELCOME TO HAMILTON" SCREEN
(USE ARROW KEYS, THEN PRESS THE ENTER KEY). That action will start-up that database.

2. FOLLOW THE INSTRUCTIONS ON THE SCREEN -- you won't be sorry -- UNTIL YOU COME TO THE "WILSONDISC - DISC SEARCH MENU" SCREEN. That will get you closer to your goal.

3. SELECT "2. WILSEARCH" -- it's what you want. (PRESS THE ARROW KEY and you'll see the highlighted 'selection' line move down, THEN press THE ENTER KEY so the computer gets your message to start the search program.)

4. SELECT "1. ENTER NEW SEARCH." This step brings you to the search screen. (PRESS THE ENTER KEY, or else the computer doesn't get your choice!)

5. ENTER YOUR SEARCH TERM/S -- this is a crucial decision! There may be more than one way of phrasing your question. You might want to do a new search more than once with different words.

6. PRESS THE END KEY, THEN PRESS THE ENTER KEY, and your search begins. Watch the results on the screen. It will only take a few seconds.
__________________________________________________________________
Note that adding the affective user-centered component achieves several relational or communicative speech acts:

-- it creates specific expectations (thus, controls them)
-- it gives reassurances (thus, tends to reduce anxiety)
-- it builds trust in the system

A final illustration establishes these points further. This time, instead of adding explanations after the original cognitive instructions, the affective component appears as prefacing remarks.

Exhibit 4
__________________________________________________________________
HOW TO SEARCH PsycLIT ON CD-ROM
SEARCHING FEATURE
SUBJECT SEARCH USING ONLINE THESAURUS

You'll have a much better chance to find what you're looking for if you use the ondisc Thesaurus of official terms and phrases you can search on, instead of your own personal words which may not be recognized by the program. To get to the online Thesaurus PRESS F9 (THESAURUS).

You need to obtain a translation of your own topic words into the terms recognized by the system. The computer will help you do the translation. To begin, TYPE your own SEARCH TERM OR PHRASE.

You'll notice on the screen a list of terms that are synonyms for your own terms. When you feel that one of the official terms fits the topic you're looking for, HIGHLIGHT DESIRED TERM(S) USING [up and down] ARROW KEYS and PRESS ENTER TO SELECT TERMS. By pressing Enter you tell the computer you're going to use that term for a search, but the search won't start until you have finished highlighting all the official terms that might fit your topic. You can highlight and press Enter as many times as you wish to have a term included in your search.

When you're done selecting terms for this one search, you're ready to tell the computer to start the search. PRESS F (FIND) TO SEARCH FOR TERM(S) (SYSTEM WILL COMBINE TERMS WITH OR).
__________________________________________________________________

The affectively elaborated instructions are obviously more wordy and repetitious, which means that it takes users longer to read them. An experienced end-user may be able to read the cognitive component alone (all caps in the exhibit above) and understand how to proceed. This might require just a few seconds since the total number of words to be read is only 30. However, many end-users are novices at using computers as well as at searching. They may read the 30 words over and over again, with puzzlement, frustration, and mounting panic. For them, the extra wordiness of the affectively elaborated instructions may contribute to remaining in a more rational, less emotional state of operation. In the above example there are 214 words to read, or seven times more than the separated cognitive instructions. However, in the process they will have been exposed to considerably more relational discourse, including the following:
-- a motive for using the available ondisc Thesaurus
-- an interest in comparing own term to system term
-- encouragement to select more than one term if applicable
-- differential expectations when pressing ENTER vs. F (FIND)

The question arises as to how much affective wordiness is minimally required. Undoubtedly there must be a trade-off factor between length of the instructions and the degree of their facilitation effect. Perhaps too much wordiness becomes a liability (takes too long to read; is too cumbersome to consult during the search; reactivity against being patronized). This issue needs to be addressed by future research. For now, the purpose of this study is twofold, one is to assess two treatment effects on novice's search behavior, namely, type of instructions and task complexity. The second purpose is to explore the network of interrelationships that interact dynamically in the searcher's world, such as, success, satisfaction, frustration, self-confidence, helpfulness and comprehensibility of instructions, interactivity, and selection of search mode. The next section presents the theoretical principles by which the instructions in this study were prepared. Beyond its usefulness for the Wilsondisc system, this process could be useful in preparing instructions for novices in any other CD-ROM or online system. This will form a guide and rationale for others interested in preparing such instructions.

2.2 TAXONOMY OF USER BEHAVIORS

To help teachers respond to increased societal demands for accountability in instruction, educators have developed the notion of behavioral objectives (Bloom et al., 1956; Krathwohl et al. 1964). The teacher's instructional intention is described in specific behavioral terms and worded in terms of a measurable skill. For example, "Student will be able to give correct definition for 80% of the new terms to be acquired." Note that this behavioral objective deals with a cognitive skill--defining terms. One can also have affective behavioral objectives in instruction. For example: "Student will be motivated to persevere at the task until success is achieved" or, "Student will feel confident enough to explore new procedures," are affective skills. Examples of sensorimotor objectives would be, "Student verbalizes support for the activity" or, "Student enters command statement correctly." A distinctive feature of the perspective in this dissertation is to apply the word "skill" to all three 'behavioral' domains.

2.2.1 Taxonomy of Library Skills

Various attempts have been made to integrate the three behavioral domains into a unified taxonomy. In one attempt, a scheme was developed for classifying library skills and common errors (Jakobovits and Nahl-Jakobovits, 1987). It employed the three standard behavioral domains (affective, cognitive, and sensorimotor) at three levels of internalization:
Level 1: orientation (demonstrating willingness to be taught and to practice; acquiring library terminology; carrying out hands-on exploration of the library environment);
Level 2: interaction (maintaining positive library attitudes; learning search sequences; conducting searches);
Level 3: internalization (demonstrating support for the library; understanding how knowledge is organized; promoting the library in one's life).
The research in this dissertation focused on the affective domain and its relation to the cognitive domain, as expressed in the text of instructions written by librarians to help novice users of CD-ROM databases. The sensorimotor domain was not investigated, though it is recognized that future research should address all three domains simultaneously. It is important to point out that the three domains are behavioral in the sense that inner processes such as thinking and feeling are behaviorally defined as skills. Bruner (1986, p.94-95) describes the paradigm shift that occurred with the birth of cognitive psychology:

The new cognitive psychology declared that the choice that guides action is as real as the action that ensues; principles of choice require explanation as a form of mental action.

For instance, drawing a correct conclusion from given facts is a cognitive act that can be learned. Similarly, persevering in the face of difficulty is an affective act that can be taught. In the cognitive behaviorist perspective, cognitive and affective acts are learnable skills, as much as sensorimotor acts.

2.3 USER DOCUMENTATION AND USER-CENTERED PRINCIPLES OF DESIGN

A paradigm shift is taking place in the area of technical writing and user documentation; this change is from a system-centered to a user-centered focus.

2.3.1 What is User-Friendly?

Bradford (1984) relates the recent emphasis on user-processes to the diversification of audiences for microcomputer documentation. As a result, "writers have begun to abandon the traditional posture and to establish themselves as teachers. ... We are often expected to encourage and entertain our audience. We are no longer playing technician; we are being forced to play the more appropriate role of educator" (Bradford, 1984, p.65).

According to Bradford, the elements of the "user-friendly persona" include humor, irony, cartoon graphics, reinforcement, personification, pronoun voice (first vs. second person), and eloquence (analogy and metaphor). Chavarria (1982) warns writers that they need to be friendly without being "patronizing or cute." She refers to the "ease-of-use movement" that stresses "simplifying messages and error recovery, adding procedures, and using headings and other information locators. This movement has been good for technical publications, because it has reminded us writers to think of the needs of the readers" (Chavarria, 1982, p.W-26). She advises writers to consider the characteristics that make dialog friendly, and to apply these to instructional manuals:
1) genuineness (the writer's involvement)
2) self-revelation (or the writer's presence)
3) supportive psychological climate (tone)
4) empathy (addressing readers' perspective)
5) non-judgmental (users are never blamed)
6) equality (vs. the attempt to manipulate).
The key, according to Chavarria, is to communicate a positive attitude in writing, using the following approach:
1) do not include information just to impress readers;
2) don't be "impersonal and machine-like in an attempt to sound
objective" (p.W-27);
3) address users directly ("you");
4) ask them questions within the text;
5) be supportive by making information easy to understand (e.g.,
by the use of graphics and white space);
6) get to know the audience (their demographics, knowledge,
needs, interests);
7) express liking for users: "We must affirm the worth of readers
in our tone" (p.W-28);
8) avoid making novice users feel left out.

Bradford (1984, p.66) asserts that "an important function of user-friendly documentation is to relieve anxiety that accompanies the novice's learning the computer or acquiring the synergisms required by a new system." He cautions against the inappropriate use of humor and does not recommend personifying the computer (as when the computer dialog display addresses itself as "I"), since a machine with an electronic brain or personality (complete with ego and motivation) could be seen by novice users as an adversary, rather than a friend. Neither should writers rely on the traditional conventions in technical writing--the passive voice, denotative diction, the absence of humor or of analogy. According to Bradford, "the alien task and complex hardware should not be described by the disinterested persona if we are to effectively engage the naive audience" (1984, p.68). Once writers abandon the traditional objective persona, "they become a noticeable presence" and "the appropriateness of that revised persona to their audiences is as crucial to their success as is the accuracy of the information" (Bradford, 1984, p.68).

There is an awareness in the literature on documentation that novices and more advanced users are to be addressed in different tones, as in the following example addressed to software designers by a recognized expert in interface design:

In designing a system for novices, every attempt should be made to make the user at ease, without being patronizing or obvious. A message telling users not to be nervous is a bad idea. Users will feel best if the instructions are lucid, expressed in familiar terms, and easy to follow. They should be given simple tasks and gain the confidence that comes with successful use of any tool or machine. Diagnostic messages should be understandable, non threatening, and low-key. If the input is incorrect, avoid blaring phrases such as 'ERROR 435 - NUMBERS ARE ILLEGAL' and merely state what is necessary to make things right 'MONTHS ARE ENTERED BY NAME'. Try to avoid meaningless, condemning messages such as 'SYNTAX ERROR' and give helpful, informative statements such as 'UNMATCHED RIGHT PARENTHESIS'. Constructive messages and positive reinforcement produce faster learning and increase user acceptance" (Shneiderman, 1980, p.226).

Users resent messages that imply that the computer is in charge. Perhaps in recognition of this tendency, the Library of Congress changed one of its prompt messages "from the authoritarian 'ENTER NEXT COMMAND' to the servant-like 'READY FOR NEXT COMMAND" (Shneiderman, 1980, p.226).
2.3.2 Philosophy of User Documentation

One design philosophy that has come to the fore in recent years is called "minimalist," which takes the point of view that, in writing technical materials, less is better (Carroll, 1990). This systems design approach to user documentation is summarized by Brockmann (1990, p.95) in the form of "general minimalist design tips":
1) cut secondary features (overviews, introductions, summaries);
2) focus on what users need to know in order to immediately apply
it to productive work;
3) test repeatedly during design;
4) use pictures of what the screen is supposed to look like to make
it easy for users to coordinate the documentation with the
screen information;
5) link new information continuously to what users already know;
6) encourage active exploration of the system by intentionally
leaving some information incomplete. This incompleteness
forces users to practice on the system instead of just reading
about it.

"Macropunctuation devices" such as tables, indentation, and frequent headings allow the visual structuring of text which contrasts with the traditional methods of writing continuous unmarked prose. Documentation experts employ several known methods of writing, some of which are reviewed in Brockmann (1990, pp.193-215). One is called FOMM (Functionally Oriented Maintenance Manuals) and uses schematics and troubleshooting charts. Another is called STOP (Sequential Thematic Organization of Proposals) in which the page format structures the content (e.g., two page modules; graphics with extended captions; examples; summary tables). A third is called PLAYSCRIPT which is characterized by using the imperative voice, beginning each step with action words in the present tense, specifying the system response that is expected and, linking actions together into a numbered chain sequence.

Style and wording are also important considerations. Brockmann (1990, p.241) refers to seven common practices for language edit:
1) edit contextually;
2) maintain coherence;
3) weed out abstractions;
4) minimize sentence complexity;
5) eliminate nonessential preliminaries;
6) break up dense writing;
7) watch out for fuzzy words (jargon, abbreviations, and
acronyms).
In the words of the designer of the minimalist approach, "the rule of thumb in minimalist design is to try first to cut and condense text and other passive components, but the goal of this is to enrich the training experience. The trick is to give the learner more to think about, but less to overcome" (Carroll, 1990, in Brockmann, 1990, p.95).

The minimalist design philosophy has its problems, as stated by Brockmann (1990, p.100):
1) learning by self-discovery is less predictable, and gaps or lack of
depth in learning may appear;
2) concise writing may become cryptic;
3) the minimalist design assumes a motivated audience, when this
may not be the case;
4) in allowing the free choosing of goals, some users pick
unattainable or ineffective goals for themselves;
5) it's quite different from the way documenters have traditionally
been encouraged to write.

In essence, user documentation and other types of help instructions all have a common purpose: to quicken the progress of novice users from a state of conscious (and sometimes painful) performance to automatic performance. "What makes an expert an expert and how can we filter that down to parrots and novices quickly?" is a question to which Brockmann offers three answers (1990, pp.104-105):
1) Let users build on success. Experts have confidence because they have been successful in the past. "Ten-minute" type guides and point-of-use instructions can give novices a sense of success which might encourage continued use and mastery of the system.
2) Help users develop "a visualization or mental map" of the system, since we know that "experts have a mental map of the material that allows them to easily organize and access information" (p.105).
3) Make users practice, practice, practice so that they may become automatic performers.

2.3.3 The Issue of Length

How long should manuals and instructions be? Of course, it depends on who the readers of the text will be and the purpose for which they read it. There is an additional issue involved:

Technical writers who must produce a manual or some other instructional text are often caught between conflicting goals. On the one hand, as advocates of the readers who must understand and use the text, writers worry about leaving out any bit of information that might be important or useful. On the other hand, as employees who are accountable for producing a text at the least possible cost, they must use text sparingly: The longer the text, the higher the cost. In short, writers must constantly judge whether the importance of an explanation or some other piece of information is worth the cost of printing it. In some sense, all writers face the same fundamental question: what information should a text contain and to what extent should that information be elaborated?" (Charney et al., 1988, p.47).

The pressure to cut length may be due to cost, as is the case with manuals, but it may also be due to space, as is the case with signs, templates, point-of-use instructions, billboards, cards, and, computer screens. Whether due to cost, space, or some other consideration such as people's willingness to read on, the length of instructions is a crucial and practical factor. As the question is posed in the above quotation, to what extent should user information be elaborated?

2.3.3.1 Experiments on Elaboration

In an attempt to give an experimental answer, Charney, et al. (1988, pp.50-56) report a study comparing the effects of varying instructions for performing file saving and other PC-DOS operations by novices. Starting with the IBM PC-DOS manual they wrote four types of elaborations, each administered to a different group: Rich Conceptual & Rich Procedural Elaborations; Rich Conceptual & Sparse Procedural Elaborations; Sparse Conceptual & Rich Procedural Elaborations; Sparse Conceptual & Sparse Procedural Elaborations. In general, "rich procedural" meant adding several examples, and "sparse conceptual" meant that less information was provided about related concepts (e.g., discussing when the root directory is automatically designated in connection with a change directory command--CHDR). The dependent measure consisted of performance efficiency (e.g., time to complete an operation and number of commands issued).

The results showed that the elaborated version (which was about three times as long) produced significantly more effective performances than the standard, unelaborated version. It would seem from this experiment that standard user documentation tends to be too short. Yet some elaborations may be better than others. Charney, et al. (1988, p.63) summarize the following conclusions from their research:
1) No benefit from elaborations of general concepts (e.g., what is a
disk drive).
2) No benefit from general elaborations on when to apply specific
procedures.
3) Significant benefits from elaborations on how to apply
procedures (e.g., well-chosen situational examples).
Thus, readers with specific tasks in mind need less elaboration in the text. While examples are generally seen as useful, those who received working exercises which they had to solve on their own performed better. Again the question is raised regarding the extent to which user information should be elaborated and what should be the nature of the elaborations. To answer this question it will be necessary to examine in some detail the relation between systems and instructions to users. This is because the task of instructions is to introduce the system to users. Unknown system features must be made known to users through the language of the instructions. How elaborated should this language be? At what level of detail must instructions be to give users the ability to perform a task? What can be left out? It may be that space limitations (e.g., instructions on wall signs) and time pressure (handouts can't take too long to read), are two factors that have led librarians to write point-of-use instructions for novice CD-ROM users which are so crisp, technical, and impersonal that they are have limited effectiveness in dealing with the psychological needs of novice users.

The experiment in this study tested the hypothesis that certain types of elaborations of user instructions, despite their greater length, are more effective in achieving user acceptance and understanding. These types of elaborations are called affective since they deal with novices' feelings and reactions while attempting to perform a CD-ROM search.
Back to the top"

2.3.4 System Design and User Documentation

Prototype testing of new designs and systems has been a common practice for generations. For example, the typewriter underwent over twenty-five experimental stages before it was released to the public (Mares, 1909, in Brockmann, 1990, p.178). Prototype testing of documentation is similar to field testing for systems. Its purpose is to test the readability and usability of instructions using an iterative process to refine the document. In the computer age, document prototyping is even more important given the complexity of electronic systems, where documenters have come to realize that "they can only discover the proper design for documents from actually users using their documents" (Brockmann, 1990, p.178).

At first, it was not easy to know how to write documentation for computer software, as the following story shows:

Joseph D. Chapline, history's first computer user documenter in the late 1940s and early 50s with the Eckert-Machuly Computer Corporation, tells the story that he was stumped as to how to write the BINAC operations and instruction manual in 1949 until he had to teach the BINAC to a group of actual users in a class. It was their questions and their understanding or confusion with the class notes and lectures that convinced Chapline what to write and how to write it. (Brockmann, 1990, p.178).

Protoyping testing reveals an actual reader's reactions to some proposed text. One professional in the field puts it this way:

I can observe patterns of usage, learn what works and what doesn't, and polish my skills based on real information about how customers use my manuals, and what they like and don't like. I can generalize from what I learn. Testing lets me avoid mind-reading. I don't have to guess what readers want or anticipate their needs on an uncanny sixth sense" (Margolis, 1989, p.7 in Brockmann, 1990, p.179).

Document prototyping attempts to capture users' experiences. Three methods discussed by Brockmann (1990, pp.180-182) are in common use. One is to observe users and debrief them after the test. The observation phase may include the verbal exchanges users have with each other when several of them sit in the same room while undergoing the test. A second method is to obtain a "reading protocol" in which users tape record their thinking-aloud verbalizations while working on a task. A third method of document prototyping is the "playback technique" which is the recorded sequence of acts of the user, a concept similar to transaction logs in database searching.

In the debriefing phase, users are interviewed regarding both their cognitive and affective experience. One expert practitioner describes the questions to be asked:

In debriefing sessions with users testing an on-line HyperCard presentation, Barbara Harvie of Software Publishing Corporation asks these questions [to get at reader attitude]: did you feel stupid or frustrated at any point; do you feel that you learned anything; were you logical; were you bored at any point; how do you feel about the level of the information--too much--too little--not relevant--useless; how do you feel about the overall metaphor" (Margolis, 1989, p.6, in Brockmann, 1990, p.182).

It is clear from this that the practice of prototype testing of user documentation creates a climate of acceptance of the user-centered approach and helps develop better techniques for analyzing user behaviors.

2.3.5 Writing Style and Types of Users

2.3.5.1 Types of Readers

There are two basic purposes for reading instructional text. One is 'to read to learn', which occurs when one studies a subject. The other purpose is 'to read to do,' which occurs when one wants to use an appliance or computer application. A document may be read by following one of these four reading strategies, according to Brockmann, 1990, pp.103-106; 191-193):
1) critical reading (e.g., evaluating a document);
2) receptive reading (for thorough comprehension);
3) search reading or scanning (e.g., looking up some specific item);
4) skimming (to get a general drift).
Along with reading strategies, the writer must also consider levels of users (parrot, novice, intermediate, expert) and types of users (regular, intermittent, transfer). Parrot users require a ten-minute task-oriented guide. They want to practice immediately on a practical task and need the feeling of success quickly. Novice users need task-oriented tutorials and exercises that give them a mental model of the whole system. Intermediate users need a product capability reference manual with practice, as well as a means to visualize the system. Over time and use, intermediate users become expert users. Transfer users need explanations of how the new system differs from what they are used to.

2.3.5.2 User Alienation

Intermittent users provide a special problem for documentation and instructions since they work with the system so infrequently that they do not go through the learning stages the others do. In fact, "these users may not be willing to advance from one learning level to the next and demands upon them to do so generally only breed resentment" (Mozeico, 1982, in Brockmann, 1990, p.105). As a result, the following recommendations are offered:
1) documentation should be very brief -- half an hour of reading is
considered maximum);
2) an easy method of visualization should be presented to allow
intermittent users to explore the system and to discover
additional details as they need them;
3) an obvious or visible means of navigation must be provided to
guide them along their chosen procedural path with menus,
messages, and commands, always letting them know what
can be done next.
4) the existence of computerphobia must be recognized and various
tactics should be used to overcome resistance to change.
These include "hand holding, more reassurances, and more
explicit linkage to what the audience already knows and feels
comfortable with" (Brockmann, 1990, p.107).

Writers of user documentation are thus clearly aware of what they call "user alienation" (Mirel, 1988, p.289; Norman, 1987, p.327; Brockmann, 1990, p.109). The tone of writing is seen as a method for countering computer alienation by evoking greater receptivity to technology and greater involvement. The tone of the document reflects the relationship between designers and users. A cross-cultural example is cited by Brockmann (1990, p.113) involving software used both in the United Sates and elsewhere. "ENTER THE CORRECT AMOUNT" is appropriate in the U.S. but it is not acceptable in Singapore due to its "failure to project an adequate 'span of discretion' for the audience." Again, "FIRST, YOU SHOULD MAKE A BACKUP OF YOUR PROGRAM DISK" is permissible in the U.S. but in Great Britain it is deemed too informal, and is rendered instead as, "FIRST, THE USER SHOULD MAKE A BACKUP OF THEIR PROGRAM DISK." The position adopted in this dissertation is that affectively elaborated instructions may be an effective way to deal with 'user alienation' and other user-centered needs. Future research will have to show how the cultural problem just mentioned can be handled with affective elaborations.

2.3.5.3 Norman's Model of the Human-Computer
Interface

Understandability and usability are two principles that apply equally to design and documentation issues. According to proponents of "cognitive engineering" (Norman, 1988; Norman and Draper, 1986) understandibility of instructions depends on providing a good conceptual model of the task or system. Good models are those that allow users to predict the effects of their actions. Users will invent false models when the appearance of some system features (the visible setup) does not match the functions desired. These principles are diagrammed in Figure 3 (based on Norman, 1988, p.16). The designer's model is formalized by the design process itself. The user's model evolves through interaction with the system. The system image is made up of the visible structure or appearance of the interface (controls, keys, instructions, labels, visualization aids or interface metaphor, etc.). All communication between designer and user takes place through the system image. If the system image does not make the design model clear, users end up with a wrong mental model as evidenced by symptoms such as inability to predict, difficulty in using, having a shallow understanding, and experiencing troubles. One of the key factors in helping users develop an accurate conceptual model is to provide immediate feedback for every action or choice made. Another good strategy is to provide a metaphor or analogy which helps users map the relationship between controls or commands and their consequences in the world. Physical analogies and cultural standards have served as sources of metaphors. As Norman points out, the desk top analogy of the Macintosh computer has been very successful in helping users visualize the system features for managing directories, folders, and files. The position taken in this dissertation is that affectively elaborated instructions is an effective method for improving the accuracy of the mental models users construct for themselves while interacting with the computer system.

2.3.5.4 User Errors

By studying the psychology of everyday actions, cognitive engineers have discovered many reasons why people make errors. First, one should note that people blame themselves for errors: "It is as if they take perverse pride in thinking of themselves as mechanically [or technologically] incompetent" (Norman, 1988, p.35). Second, experience has shown that good system interfaces can be designed that make errors easy to detect and easy to reverse or, if not, have minimal consequences. Designers should heed the motto: If an error is possible, someone will make it. One common source of errors is inadequate mapping where users develop misconceptions in their mental models of the system. The existence of conflictual habits is another source of error (e.g., frequently hitting the wrong key for a particular operation). Users need to be reminded that making errors is routine in everyday automated activities such as conversations and keyboarding. Natural situations allow for routine correction of errors, such as retyping or abandoning a sentence midway and beginning a new sentence. Error recovery procedures have become essential in designing computer systems, at least in theory.

Norman (1988, p.107) has identified several types of user errors. Capture errors occur when a frequently performed activity suddenly takes charge instead of the intended one (e.g., hitting the wrong Function key; hitting return instead of an F key; waiting while staring at the screen without having entered the command; etc.). Description errors result from performing a correct action but on the wrong object, which occurs when we are distracted, bored or fatigued. Data driven errors occur when the sight of some object or information intrudes into an action sequence (e.g., typing the wrong information on a line; choosing an unintended menu item). Associated activation errors occur when an association of ideas or thoughts trigger a non-relevant action (e.g., typing an incorrect subject heading phrase). Loss of activation errors occur when the goal of an action is momentarily forgotten or when part of a sequence is temporarily inhibited. Mode errors occur when users forget what mode they are in when there are several (e.g., typing all caps when not intending to). By studying the nature of user errors during searching, librarians will be able to write affectively elaborated instructions that orient novice users to recover from common errors, give them advice on how to proceed. and reassure them that it is normal to make mistakes (see Table 2).

2.3.5.5 A Theory of Action

Norman's theory of the structure of actions (1988, pp.46-51) is adapted here to reflect the focus on the affective, cognitive, and sensorimotor domains of behavior (Bloom et al., 1956; Krathwohl, et al., 1964; Merrill, 1972). The diagram in Figure 4 depicts the six stages identified by Norman, and their relation to the three behavioral domains as viewed in this dissertation. The cycle starts with users' intentions which are specific goals or motives in a particular situation. This activity occurs in the affective (A) domain of feelings, needs, and values (preferences, priorities). Intentions are translated into a specific plan which occurs in the cognitive (C) domain of thoughts, reasoning, models, and inferences. When intentions and plans are joined together, the overt action that results is called execution, which occurs in the sensorimotor (S) domain of afferent/efferent connections involving sensory organs and motor fibers.

These first three stages are called stages of execution. The last three are called stages of evaluation. After the execution of an intention through motor action, users observe the consequences that ensue in the world. This process is called perception and occurs in the sensorimotor (S) domain of seeing, hearing, and touching. Perceptions need to be interpreted in the cognitive (C) domain in terms of one's expectations based on understanding. Finally, interpreted perceptions must be evaluated in the affective (A) domain of values, preferences, and priorities.

From the point of view of searchers, there are going to be two problems in their actions. The first is called the gulf of execution and refers to the difference between intentions (Affective) and the allowable actions in the system (Norman, 1988). The second user problem is called the gulf of evaluation and refers to the amount of effort it takes to get feedback on a specific action. The purpose of user documentation or instructions is to reduce these two problems to a minimum. Learning allowable actions--the gulf of execution--is facilitated through user documentation principles that provide a good mental map, as reviewed above in connection with the minimalist writing philosophy. The amount of effort users must exert to get feedback--the gulf of evaluation--is to be bridged by visibility of system features and, once again, appropriate documentation and instructions. The position is taken in this dissertation that affectively elaborated user-centered instructions may be effective in reducing both the gulf of execution and the gulf of evaluation. By providing users with orientations, advice, and reassurances, allowable actions are interconnected with intentions, while it takes them less effort to receive feedback.

2.4 THE ANALYSIS OF INSTRUCTIONS

The structure of text has been an active area of research in cognitive psychology, as evident from some reviews of that literature (Bransford, 1979; Anderson and Reder, 1979; Reder et al., 1986). As reviewed by Reder (1982), it was cognitive psychology's attempt to teach reading that raised the issues of what constitutes comprehension and how text is processed. What types of inferencing goes on during reading with meaning? How is information encoded in memory and how is it retrieved when needed? The idea arose that "long term memory is a network of interconnected propositions" (Reder, 1982, p.213). The act of reading adds new propositions to the network in long term memory. These newly encoded propositions act as linkages which aid retrieval from memory "at test" time, when the information needs to be used.

One of the sub-areas of research on text processing has been the study of the effects of writing various types of elaborations of statements upon memory retrieval and skill learning tasks. For example, Reder, et al. (1986) compared the effects of two versions of a user's manual for novices learning file management with the IBM-DOS system. The longer version was almost three times more wordy and was called the elaborated version. The shorter version was called unelaborated. The results showed that novices who were given elaborated texts operated in a more efficient way than novices who read the unelaborated version. How is this effect explained?

2.4.1 Text Elaboration Theory

The construct of elaborations has been defined as "any information that supports, clarifies, or further specifies the main points of a text. Elaborations can take many forms, including examples, details, analogies, restatements, and deductions" (Reder, et al., 1986, p.64). The text itself is called TBR (to be remembered text), while the embellishments are referred to as ET (elaborated text). In other words, writing instructions takes two steps:
Step 1: write to-be-remembered text
Step 2: insert elaborated text
There are three reasons given for the potentially facilitative effect of elaborated text upon later recall of to-be-remembered text:
(1) the embellishments in elaborated text act as "cued recall" since
more linkages and hence, more channels of access, are laid
down in the tracks of memory;
(2) elaborated text helps users "generate additional concepts from
which to spread activation" (Reder, 1982, p.213);
(3) elaborated texts "allow for inferential reconstruction of the to be
remembered text item" (Reder, 1982, p.213).
In other words, writing elaborations of instructions should facilitate recall of pertinent information when needed by users during task performance.

Researchers in this area are concerned about the differential length between elaborated texts and unelaborated texts. In fact, experimental studies have shown that in the case of 'declarative knowledge,' such as that a school text book ordinarily provides, readers score higher on quizzes when they are given substantially reduced summaries of the text content (Reder, 1982). This effect appears to be the opposite of what one would expect from the theory of elaborations outlined above which would predict that a textbook, being more elaborated than a summary, would provide better knowledge scores. Yet studies show that students provided with summaries do better. Reder and associates comment on this contradiction:

The studying of summaries facilitates performance on tests of factual knowledge, because it allows the reader to devote full attention to the essential facts, exactly those that must be retrieved at test. In other words, studying elaborated texts impedes learning the main points of the text because reading the elaborations reduces the amount of time subjects can devote to the main points. This Total Time Law is a well established verbal learning phenomenon (e.g., Bugelski, 1962; Cooper and Pantle, 1967) (Reder et al., 1986, p.65).

However, the Total Time Law is not the only factor operating here, as shown by studies that allow readers more time for elaborated texts, but the subjects who used summary texts still performed better. Another factor is the "interference effect" of elaborations discussed by Reder et al.: "interference affects retrieval rather than encoding" (Reder et al., 1986, p.65).

The contradiction was resolved when Reder and associates demonstrated that elaborations behave the opposite way in the skill learning domain: rather than elaborations being a source of interference in memory retrieval (as is the case in learning of declarative knowledge), they now become a source of facilitation and greater efficiency (Reder, et al., 1986). In a situation where some skill has to be acquired, as in the case of novices learning file management in PC-DOS, elaborations of sentences in the user's manual may prove to be facilitative and beneficial due to the uncertainty of the situation and the complexity of the task. Thus, the extra length of the elaborated text outweighs the liability of time.

2.4.2 Types of Elaborations

Three types of elaborations are recognized:
(1) Helping novices "appreciate the meaning of new concepts," such
as, "what conditions affect the usefulness of a procedure."
This reflects an affective orientation, as adopted in this
dissertation.
(2) Helping novices "remember and compare procedures" such as
pointing out "effective combinations" and "deepening
understanding." In other words, the writer adds content that
concerns the cognitive behavior of users.
(3) Helping novices "to apply procedures in a specific situation."
This refers to providing specific examples that show "what
their own command must look like." In other words, the
writer adds content that concerns the sensorimotor
behavior of users.
Although Reder and associates do not explicitly refer to the three domains of behavior (A/C/S), it is clear that their conceptualization of what novices need in a computer learning situation matches the theoretical perspective adopted in this research.

The results of two experiments reported by Reder, et al. (1986) clarify the specific characteristics of elaborations that are beneficial to the performance of novice computer users. They used two versions of a user's manual for the IBM-PC DOS. The elaborated version was written first, and amounted to about 11,000 words. The unelaborated version was then constructed by deleting portions of the text until it was about 3,500 words long. Reder and associates were satisfied that "none of the elaborations contained new information necessary...for criterion tasks" (1986, p.67). Novices sat by the computer and read the instructions (either the elaborated or unelaborated version--randomly assigned), and stopped when they felt ready. They had been told that they would not be allowed to consult the instructions once they started operating the computer. The group that received the elaborated version took an average of 50 minutes to read and study them, while the group receiving the unelaborated version took an average of 40 minutes. The dependent variables were four measures of performance:
(1) proportion of tasks correctly completed;
(2) average time of completion of tasks;
(3) average number of commands issued;
(4) proportion of commands issued per minimum steps
required (measures efficiency and takes into account
unnecessary actions).
Except for the first measure, the general pattern of significance favored the group that received the elaborated version. This conclusion was confirmed when they examined what type of user behaviors were significantly better for the elaborated group namely, execution errors, goal specific errors, and error recovery moves. The novices who had received the unelaborated instructions were able to accomplish the same tasks but they did them less efficiently, with more errors.
In the second experiment, Reder and colleagues investigated the relative benefits of two types of elaborations, "syntactic" and "conceptual." Syntactic elaborations dealt with the syntax of commands with a view to teaching better procedures for specific file management actions. The version that contained syntactic elaborations was called "syntax rich" and contained examples of correct commands and descriptively rich format statements. Conceptual elaborations dealt with the "need to determine the appropriateness of information to task constraints" (Reder et al., 1986, p.71). The version that contained these elaborations was called "concept rich" and dealt with the "purpose of the commands and when it was a good idea to use them." They also contained "basic topics" on drives, paths, and subdirectories. "Syntactic elaborations," in the language of this research, appear to focus on the user's cognitive behaviors since they relate to the functions of commands and procedures. Since "concept rich elaborations" relate to the appropriateness of using certain commands given certain goals, they deal with the user's affective behaviors.
Back to the top"

The results generally showed that the syntax-rich version was more beneficial. The authors believe that this outcome would change if conceptual elaborations were to be more carefully constructed to
facilitate retention. Their paper concludes with the following expectation:

Conceivably, if the mapping between the concept examples and the task requirements were as close as the map between the syntax examples and the task requirements, performance would also benefit from the conceptual elaborations. ... The benefit or liability of elaborations is a function of their quality and relevance to the main points that they support." (Reder et al., 1986, p.73-4).

2.4.3 Definition of Speech Acts

One aspect of the research on elaborations that needs improving is the development of more systematic procedures for describing the specific differences between regular, unelaborated writing and elaborated versions thereof. The research in this dissertation attempts to give a systematic answer to this specificity problem by designing a procedure that permits the classification of the content of elaborations. The approach borrows from content analysis and speech act theory as described in a recent review and application of the technique to behavioral objectives for bibliographic instruction (Nahl-Jakobovits and Jakobovits, 1992). The main idea in content analysis is to read portions or units of the text (e.g., sentences or assertions) and to categorize each unit into a predefined system of classification. For example, students were asked to keep notes on how they were doing their library research, then they wrote a summary report. The sentences of the report were categorized into one of the three behavioral domains (affective, cognitive, sensorimotor). The following are some illustrations (Nahl-Jakobovits and Jakobovits, 1992, p.9-11):

Report: "As always, it's much easier to study in the library because it is quiet and there are no distractions like the television."
Speech act category: +A (positive affective skill)--feeling inner calm in the library environment.
Report: "So now I know what journals look like!"
Speech act category: +C (positive cognitive skill)--demonstrates the acquisition of a library relevant distinction.
Report: "But one day when I was looking through the Psychology shelves I saw General Journal of Psychology. ... I noticed that there were lots of journals like the Journal of Social Psychology..."
Speech act category: +S (positive sensorimotor skill)--expresses and describes noticings and browsing sequence during a search.

In this approach, the three domains and levels of behavior provide a theoretical basis for the content analysis. The concept of speech acts comes from linguistics, philosophy of language, and communications (Austin, 1965; Carney, 1979; Coulthard, 1977; Habermas, 1984; Pool, 1959; Searle, 1969; Labov and Fanshel, 1977; Vendler, 1967). Sociolinguists have shown that ordinary speakers of English operate on the basis of speech acts, as attested by their use of ordinary expressions that identify what someone is doing, such as, they are greeting one another, they are proposing a settlement, he is asking a question, he is denying it, she is saying something friendly, and so on. These are all speech acts, i.e., they are verbal behaviors that have interpersonal significance. In this research, the sentences of CD-ROM instructions written by librarians for novice users were analyzed into speech acts. For example, a sentence might be: "... You can search Wilsondisc in two ways. ..." This is a speech act since it is recognizably written to inform users of some specific fact. This type of speech act is common in instructional language. The sentence, "Some of these articles may be relevant to your subject" is an affective speech act since it is intended to point out the value of a search result, in an attempt by the writer to reassure users and insure their satisfaction. Speech act theory, especially the perspective of Habermas (1984), is further discussed before the presentation of the proposed taxonomy of speech acts (see section 4.4.2 and Table 2).
CHAPTER 3

MEASURING USER BEHAVIOR IN INFORMATION SEARCHING


3.1 HISTORICAL PARADIGM SHIFT

Dervin (1991, p.3) argues that the technologies of information systems, far from solving societal problems, are rather "reifying and exasperating social inequities world wide." She states that it is an "illusion that these technologies ... allow us to be better informed simply because they move something we call information about in ways that we think are more powerful" (p.3). She argues for a paradigm shift that would reconceptualize information from "a thing which is either transmitted, manufactured, or processed" to the position that "in order to make the term more useful in the arena of human affairs, the term must be reconceptualized in communication terms" (p.3-4). After reviewing the causes of the paradigm shift from "information as an observer construct versus user construct" (Dervin, 1983, p.160), she makes the following recommendations regarding desirable practices of a user-based information system:

* frequent use of question-answer formats, as in the so-
called "Dear Abby" format;
* frequent use of reporting values, motives, and
frameworks for observing along with results of
observing;
* frequent reporting of conflicting observations resulting
from different frameworks;
* frequent use of case studies in presenting information;
* multiple-category systems for organizing information,
with a substantial number being user-based, as in
lists of user questions for situation facing;
* frequent use of random samples of users detailing their
information needs to determine the content of
messages and the nature of systems;
* use of situational analysis of audiences rather than
"personality demographic analysis" (p.175).

It is plain from this list that the weaknesses of the system-centered paradigm could be remedied in a fundamental way by including the affective component of the searcher's world in several ways:
* question-answering
* inclusion of values and motives
* case history
* explicit references to conflicts
* varieties of samples of users with a focus on the
dynamics of the situation rather than on individual
personalities.

The need for change in information system design and management is necessary since database searching is increasingly performed by untrained people. This is recognized by Jacobson in a recent statement:

The increased power achieved by information retrieval systems in recent years is making highly complex software interfaces commonplace in many professional and public venues. The proliferation of these systems is accompanied by increased numbers of untrained or minimally trained end users, and expectations for continued growth are widespread. Given these developments, the user services area of library and information science must now contend with the evolution of entirely new information searching and retrieval behaviors (Jacobson, 1991, p.647).

Thus, new search aims will have to be addressed by system designers and researchers. The implications for library policy and information science, according to Jacobson, is that "measures of search session success should rely less on expert evaluations that is now common, and more on success as users themselves define success" (pp.654-5).

3.2 RESEARCH ON THE USER'S PERSPECTIVE

Jacobson reviews four types of theories that have been proposed for explaining the search process viewed from the user point-of-view. One approach is to treat the activity of inquiry as the result of "anomalous states of knowledge" (Belkin, 1980). A different approach consists in studying students longitudinally and identifying the "stages of evolution of a research activity" (Kuhlthau, et al., 1990). A third approach attempts to understand searchers' decisions by specifying their "mental models" of the search problem and its solution (Borgman, 1986). A fourth approach is Dervin's "sense-making" concept which is defined as the "movement of thoughts and questions through cognitive time-space" as explained by Jacobson:

In this sense, one may be said to move through a series of thoughts and experiences, to encounter 'barriers' to one's progress, to 'lose' one's way, etc. In either case, any given movement can be interrupted when an individual is confronted by the need for some form of guidance; that is, when they need to know something. Based on the work of Carter (1980), such conditions are considered cognitive 'gaps,' and may be exemplified in their most general form by the need for street directions or instructions on the use of Boolean connectors, depending on the behavioral context. Information seeking is defined within this framework as 'gap bridging.' Gap bridging can be accomplished by thinking up an answer, asking for help, looking for useful information, or by any other functional method that enables the individual to continue moving. These gap-bridging attempts are operationalized as 'questioning,' whether spoken by individuals or unspoken (Jacobson, 1991, p.649).

To understand the user's perspective, researchers (e.g., Fidel, 1991a, b, c) have studied how users make decisions by observing their 'moves' within a search activity. Examples of moves made by searchers include any activity that allows the searcher to continue progressing, as for instance the following:

-- using help files
-- reading the screen
-- issuing commands
-- guessing
-- re-reading the screen
-- asking questions of themselves or others
-- consulting documentation
-- trying to get to a menu
-- trying to get out of a menu
-- trying to figure out where they are
-- modifying a search string
-- displaying a record

Marchionini claims that, at this stage, designers of information systems have now adopted a "user-centered design philosophy" and have begun implementing interfaces that take into account "end-user information seeking needs" (1992, p.156). He makes three assumptions about the characteristics of end-users in electronic environments. One is that end-users are not interested in system characteristics such as design, elegance, or data structure. Instead they want to get to the end of their work; they want answers, not pointers; they want document delivery, not information retrieval. A second assumption is that "end-users want to achieve their goals with a minimum of cognitive load and a maximum of enjoyment" (p.156). Part of this assumption includes the law of least effort:

Moreover humans seek the path of least cognitive resistance and prefer recognition tasks to recall tasks; most people will trade time to minimize complexity. Finally, humans will perform better and continue to use systems that are pleasureful or interesting (Marchionini, 1992, p.157-8).

Computer-augmented information seeking can create a more relaxed, less stressful, more pleasant search environment. In other words, if it is user-centered, it is affectively more benign. In the past, attempts at designing more user-friendly information systems have led to software interfaces known as "metaphors," such as the Desk Top metaphor for the operating system on Macintosh computers. This approach has been applied more recently to developing alternatives to traditional command language interfaces. For example, Borgman and colleagues have developed the "library metaphor" for an OPAC interface in elementary school that uses the Dewey system and allows students to browse hierarchical graphic bookshelves (Borgman et al., 1990, p.55-68). Some of the more recent search support innovations include, according to Marchionini's review, the following:

* query-by-example (Zloof, 1977);
* spatial database representations (Herot, 1980);
* online thesauri tied into the search command
sequence (Chen and Dhar, 1990);
* interfaces that offer suggestions to users, including
dynamic queries that evolve as part of the
interaction with the system (Meadow, 1988);
* problem articulation as pruning (where users begin with
all items in the database selected and each query
results in a subset, e.g., Computer Library
CD-ROM);
* problem articulation "on the fly" through active
browsing and judicious navigation in hypertext
systems (Frisse and Cousins, 1989).

Other developments include the addition of alternate input devices such as datagloves, gesture recognizers, speech recognizers, and eye trackers (Marchionini, 1992, p.160). He quotes Gauch, saying that "In electronic environments, the IR problem is not finding information, it is filtering information" (p.161). This comment reflects the fundamental dilemma of end-users in today's information environment. Marchionini concludes that user-centered interfaces should accommodate individual differences, cultural diversity of individuals, and affective and cognitive characteristics of individuals. The proposed study focuses on an important feature of the movement to design more user-friendly information systems, though restricted to the addition of affective components in point-of-use instructions. When system interfaces fail users, point-of-use instructions could be important in helping users overcome those weaknesses.

3.3 THEORETICAL MODELS OF SEARCHING BEHAVIOR

3.3.1 Reformulations

According to Dalrymple, "the user-centered approach derives its questions and methodology from the behavioral sciences" (1992, p.272). In her study, subjects were assigned search problems and their cognitive processes were studied by having them think-aloud while searching. Of special interest were their "reformulations," an expression which refers to the attempt to refine the terms of a query. Reformulations are assumed to be cognitive processes that interact with searchers' long term memory retrieval operations, which may be conceptually similar to database access and retrieval operations. "Presearch reformulations" transform the language of the information need into terminology that initiates the search. "Search reformulations" are modifications of search terms during the search. For example, the statement, "I'm looking for books about women writers in 20th century German literature" needs to be translated into the form, "Feminist writers." This presearch reformulation may be modified later during the search by adding, "Feminist writers German literature." Dalrymple found that a significantly greater number of reformulations were made when subjects search the online catalog as opposed to the card catalog. She calls for additional investigations on reformulation to determine whether it can be taught, how it is affected by the urgency of the information need or by individual differences in search style, and whether it acts as a request for feedback from the system. Reformulation has been adopted as one of the measures for the present study.

3.3.2 The Affective/Cognitive Gap

In a series of studies on information seeking from the user's perspective, Kuhlthau (1991) explored the cognitive and affective aspects of users' experiences in information seeking situations. She points to a gap between how searchers think and feel, on the one hand, and on the other, how information providers think and organize information. On the one hand, the system's pattern is based on certainty and order; on the other, the user's pattern is based on uncertainty and confusion. Information thus has a mechanistic as well as a human dimension. Users assimilate information from various sources and transform it into meaning through sense-making attempts that result in a personal point-of-view creating new knowledge that is shareable with others (Kuhlthau, 1991, p.361). Her review of the research on information seeking from the user's perspective contains the following highlights:
(1) Most studies conducted in the past are from the perspective of the system (Dervin and Nilan, 1986).
(2) There is a new interest in moving away from system oriented research towards a focus on users of information, especially on their problem-solving and decision-making behaviors (Borgman, 1984).
(3) A fresh approach brings in a cognitive science point-of-view that takes into account the user's thinking processes and evaluations of the information retrieved (Ingwersen, 1982).
(4) There is an acknowledgment now that users interpret information in terms of personal life characteristics and these are likely to be as important as system considerations in information retrieval (James, 1983; Hollnagel and Woods, 1983).
(5) To understand how people find and use information, we need to combine their cognitive map of the searching environment with a model of the system's design features (Belkin, 1984).
(6) Users' ability to articulate requests to the information system depends on their level of understanding of their problem. Since their level of awareness of their information need may be initially low, they may have little success in specifying what is needed (Belkin, 1984).
(7) Users' queries contain at least four levels of awareness of information need: "visceral" (below the threshold of awareness), "conscious" (aware of the need but not yet formulated in specific terms), "formalized" (explicitly expressed), and "compromised" (translated into the terms of the system) (Taylor, 1968).

Kuhlthau emphasizes that an information search "is a process of construction which involves the whole experience of the person, feelings as well as thoughts and actions" (Kuhlthau, 1991, p.362). To put it more specifically,

A model representing the user's sense-making process of information seeking ought to incorporate three realms of activity: physical, actual actions taken; affective, feelings experienced; and cognitive, thoughts concerning both process and content.... While purely cognitive conceptions of information need are adequate for some research purposes, consideration of the affective dimension of users' problems is necessary to address a wider, holistic view of information use (Kuhlthau, 1991, p.362).

Kuhlthau argues for a new "dynamic" perspective on the searcher's world. An information need becomes "the gap between the user's knowledge about the problem or topic and what the user needs to know to solve the problem" (Kuhlthau, 1991, p.362). She argues her perspective through "personal construct theory" originated by Kelly (1963). This cognitive approach describes how new information is assimilated, beginning with "confusion" brought about by "inconsistencies and incompatibilities" between the new information and the currently held constructs. Doubting the validity of the information may follow confusion, and experienced threat may lead one to discard the new information. Or, alternately, users may attempt to incorporate the new information by constructing new hypotheses that allow a more inclusive construct system.

3.3.3 Longitudinal Study of Information Seekers

This represents an interest in the dynamic evolution of an information seeking problem within the personality parameters of a user. To obtain data on this private arena of the searcher's world, Kuhlthau's subjects recorded in writing, their feelings, thoughts, and actions related to their library research project. They also kept logs in which they recorded sources they used, procedures they went through for finding sources, and their evaluation of their usefulness. They also filled out a questionnaire that asked questions on their perceptions of six areas of library activity: topic selection, research assignments, focus formulation, procedures for gathering information, frequency of library use, and role of mediators. Taped interviews were conducted in which students drew flowcharts describing the process they followed.

With this intensive and longitudinal data on 20 college students, Kuhlthau developed a six-stage model of the research process. The model was verified with 385 college students at 11 universities with three data points in time: initiation of the research assignment, its midpoint, and its closure. Table 1 is a summary of her findings (Kuhlthau, 1991, p.367). These observations draw a progression from initial states of uncertainty with vagueness to later stages of optimism, confidence, and satisfaction with more narrowed, focused, and synthesized information. Kuhlthau notes three broad psychological effects: (1) interest in a topic increases as a search progresses; (2) the topic changes as information is gathered; and (3) a central theme evolves as information is gathered. Her conclusion on the importance of the affective component is explicit:

By neglecting to address affective aspects, information specialists are overlooking one of the main elements driving information use (Kuhlthau, 1991, p.370).

One purpose of the study in this dissertation is to infuse point-of-use instructions with the affective component and to demonstrate its facilitating effect on understanding, performance style, and satisfaction.

3.3.4 Search Style of Expert Intermediaries

Fidel has studied several aspects of online searching behavior, including, (1) choosing databases, (2) choosing search terms, (3) conducting the search, (4) reviewing results (feedback review), (5) making new decisions in response to the evaluation of results and, (6) terminating search and getting printout (Fidel, 1991a, b, c). Her approach consists in identifying incidents where a search key was selected, and then fitting each incident into a decision tree, resulting in a catalog of criteria for decisions. For instance, during the process of choosing search terms, the first decision point is to determine whether a "single-meaning term" (which is good for free-text searching) is wanted or a "common term (broad and fuzzy meaning with too many contextual variations)" (1991a, p.493). In case it is a single-meaning term, the next decision point is to map the term to a descriptor. For example, the topic "Anxiety about using computers" is matched to descriptors such as TECHNOPHOBIA (narrower) or ANXIETY (broader). This matching process involves a semantic content (or the concept) and the system language (controlled vocabulary). If no exact match is found, then a partial match may work in conjunction with textwords (key terms) for an inclusive search. If no match is found, textwords can be used to probe indexing further. This case history approach demonstrates that searchers use intuitive as well as explicit "rules" for decisions. Descriptive efforts on how users make decisions in an information retrieval situation produce knowledge engineering trees that can help in the design of intermediary systems that assist the searcher through 'dialog box' inquiries about the user's purpose and scope of the search as well as requesting evaluation of sample retrievals. According to Fidel,
Back to the top"

Understanding how searchers of all types look for information, and how they interact with existing systems, can provide guidelines for searchers' training and assistance (1991b, p.501).

It is evident that decisions during searching are guided by users' perceptions. These can be identified by asking them to state the reason for a particular decision. These reasons or perceptions can often be validated by objectively observable evidence such as transaction logs which record what keys have actually been pressed. Another method of validation consists in examining frequency distributions and correlations across several searchers for a single search, or across several searches for one individual. Other variations include the number of databases, different subject areas, type of information venue (research, industry, other), and number of moves required (task complexity). Having conducted research with these variations, in her conclusions, Fidel emphasizes the finding that search decisions and strategy are heavily determined by system features, especially the structure of controlled vocabulary and the availability of effective online thesauri:

Thus, research should be carried out to discover which features of databases and their thesauri can be standardized without affecting retrieval quality. The role of intermediary expert systems will then be to bridge the necessary differences, employing switching languages and other terminological and semantic networks (Fidel, 1991b, p.514).

Fidel analyzed the verbal and search protocols of 47 professionals in the hope that she might discover "what characteristics of searching behavior constitute a searching style" and "in what way one individual searcher is different from another, all external conditions being equal" (1991c, p.515). Fidel, citing Fenichel (1981), estimates that research on online search behavior has declined recently because prior experiments have failed to provide conclusive results on individual differences such as experience, cognitive attributes (Woelfle, 1984), personality traits (Bellardo, 1985), and type of request (Saracevic and Cantor, 1988).

3.3.5 Types of Moves in Searching

A common explanation given for these partly negative results is that "individual search styles override most measured attributes of searching behavior" (Fidel, 1991c, p.515). She thinks that this attitude is premature since we don't yet understand the specifics of search styles. She analyzed the data on verbal protocols of thought processes while searching, and the data from interviews with searchers to determine reasons for their search selection. She built a two-layered model to represent what she found. The essential features of this model are summarized in the schema shown in Figure 5. The derivation of six types of moves is depicted here. A move is defined as any modification in search strategy aimed at improving results. Operational moves do not change the meaning of a request, while conceptual moves do. Thus, type 2 moves, for instance, are operational moves aimed at increasing precision. Examples of moves include the following:

Type 1
-- Intersect free-text terms to occur in a predetermined field.
-- Limit time (by date).
Type 3
-- Add synonyms and variant spellings.
-- Eliminate restrictions previously imposed.
Type 4
-- Intersect a set with a set representing another query
component.
-- Select a narrower concept.
Type 6
-- Enter a broader descriptor or term.
-- Group together search terms to broaden the meaning of a
set.

Of the 1,244 moves made by the 47 professionals in their 281 searches, 60% were operational and 40% were conceptual. The number of moves to increase recall was about double the number of moves to increase precision. Fidel takes this as a sign of "the difficulty in achieving satisfactory recall in the databases currently available" (Fidel, 1991c, p.518). Operational moves are used to improve precision or to reduce a set, while conceptual moves are used to improve both precision and recall. The average number of moves per search was 5, with a range of 1 to 18. It is of interest that these professional searchers only used 25% of the moves that were available, indicating that "search systems should remind searchers of the complete array of moves possible in online searching" (Fidel, 1991c, p.519). Interactivity (number of moves) was not related to purpose, concern with recall, or subject area. It thus appears to be an element of search style.

In conclusion, Fidel uncovered three search style characteristics of professionals which were not related to results or success:
(1) Level of interaction or number of moves;
(2) Preference for operational versus conceptual moves;
(3) Preference for using textwords versus descriptors in moves.
Two of these three searching characteristics are among the 11 dependent measures in this dissertation because it is believed that these search style measures can be influenced by the speech act content of instructions as well as by other searchers' characteristics such as success, satisfaction, self-confidence, and frustration. Since both cognitive and affective domains of the user's world must be addressed, it is clear that the new user-centered perspective needs to focus on a broader spectrum of the user's involvement in the search process. The next chapter attempts to specify in more detail users' affective and cognitive search behavior.

(end of Chapter 3) Back to the top Forward to Chapter 4