Interface Design for SunSITE statistics

http://sunsite.unc.edu/stats




  • Mark C. McCarthy
  • INLS 257
  • Spring 1997
  • User Analysis Documentation:
    Accessing Statistics on SunSITE

    Those that maintain archives on SunSITE do so with full control of their archives. SunSITE is a digital library and makes no claims of authenticity for content and does not seek to exert editorial control. In contrast, the goal is to facilitate the accessibility of information. The management of SunSITE does not purposefully set out to monitor content or the activities of collection owners. As a result, and for reasons of liability, SunSITE knows little about those persons and organizations that maintain archives on SunSITE.

    The user group for this interface will vary greatly in computational knowledge. This creates an interesting challenge for design. Some of the users will seek very specific information from the Web page while others will come with little knowledge of the system and little idea of what information is available.

    The users maintaining collections on SunSITE are the primary users for this system. These archive owners on SunSITE tend to be experienced users that seek a quick display of statistical information that they can display for sponsors.

    The less experienced users will be visitors to SunSITE or less computer literate users that maintain archives in SunSITE. Because SunSITE encourages diverse collections, some of the managers of archives are not necessarily expert users. The challenge lies in trying to create an interface that allows experienced users to quickly extract the specific information they seek while offering cues for the infrequent user.

    Psychological characteristics

    1. Cognitive style: The diverse array of users cannot be specifically categorized for cognitive styles. Most of the users will be familiar with the basics of Web interfaces. The interface will therefore conform to current Web user expectations.
    2. Attitudes: The users will be coming to the page by choice and will have a very positive attitude toward computing. Some will come with expectations and others will come with simple curiosity. User expectation will be for information about the number of times the pages have been accessed. This information will be provided along with options for table layout or graphs. Some users may seek information about where the pages were viewed from, but this is beyond the scope of the program. The logs used to generate the stats do not include this information.
    3. Motivation to use the system: For some the motivation will be work related but, for most it should be curiosity. We are a non-profit archive; most users do not have sponsors to satisfy.

    Knowledge and experience:

    1. Level of education: (Novice to Expert) Users should have a relatively high level of familiarity with the system. The access to the interface will be through a Web browser and users will need to be able to use a browser to get to it. Those that will be most interested in the statistics will be users with archives on SunSITE and will therefore be familiar with the Web. The people with archives on SunSITE tend to have at least some college, expert computer skills, and moderate ability to communicate in English.

    Physical characteristics:

    Users perceptual abilities and motor skills may vary, but if they are common web users they will have the tools to use the interface. Users will posses the skills necessary for browsing, mouse movement and clicking, and World Wide Web navigation.
  • Mark C. McCarthy
  • INLS 257
  • Description
  • Web Statistics Interface for SunSITE

    The prototype that has been created is for a World Wide Web page. The page is the interface for requesting statistical information about archives on SunSITE. The system allows users to specify the directories that they want information about. It also allows for output as either a table or bar graph.

    The goal of the interface is to create a fast and easy to use system for finding out how many hits specific directories have experienced. The system includes a very brief form to pass the necessary variables to the CGI (common gateway interface) script that carries out the operations. The form includes a text field, radio buttons, and pull down menus. All of these form elements are part of the HTML 3 specification.

    The Uniform Resource Locator (URL) for the stats page has been sent to those that maintain archives on SunSITE through an email list. This is currently the only way that the Web page has been publicized. Owners of the archives will be notified of any updates via the list. While the page in intended primarily for these users, a link to the page may be added to a high level page so the general public can check the number of hits in the future.

  • Mark C. McCarthy
  • INLS 257
  • Spring 1997
  • Rationale for Design Decisions
    Statistical Information Interface on SunSITE

    The Overall Design:

    There are a number of points that were important in the overall design of the Web page. The first of these was the file size of the HTML document. When considering design issues it is important to consider the medium that the system is operating on. While the user will be using a computer, the actual document must travel over the Internet. This can lead to latency problems that will increase as the file size increases. To avoid these problems the page was kept simple and small in size.

    From a creative perspective, there was a desire to incorporate exciting graphics and more interactive functionality with simple programs supported by popular Web browsers. These designs would not make the interface any more efficient and would make the file larger. In an a study of the Web and user interaction, Nancy Lightner, Indrail Bose, and Gavriel Salvendy (1996 *998), found that users' two main concerns were speed of access and searching for specific information. The current system is designed to facilitate these needs as efficiently as possible.

    The Web page has been made as compact as possible while maintaining logical groupings for the tasks. One of the key reasons for keeping the actual size of the page to a minimum is to ensure that the whole system is visible upon download. A major problem for Web designers is the lack of consistency and control over user systems. Varying monitor sizes, processor speeds, bandwidth limitations, and browsers will affect the layout and function of a design. The design for the stats page should enable the whole page to be visible on just about any system without any scrolling necessary. In this way the user can see all the variables available for input. This should eliminate confusion over how much information is needed and how to submit the information.

    The overall design includes a form, menus, and buttons as described by Deborah Hix and H. Rex Hartson in Developing User Interfaces Ensuring Usability Through Product and Process (1993 *pg72). The interface includes a number of the guidelines as specified in their book. Some of the key guidelines used included consistency and visually appealing layout. User tasks are set together in logical groupings with visual cues delineating the different areas. Another guideline that is followed is the use of visual cues in various fields. This is an important feature to help users understand what is expected for each task. Two similar guidelines that are also reflected in the design are the use of default values in fields and explanatory messages for expected field inputs. With the implementation of these cues the system should seem logical to the user and help to avoid confusion.

    Below is an explanation of design decisions for each of the tasks for the system.

    Enter the URL:

    A text field or form fill-in is used to enter the address of the archive to query. Since the statistics are being made available primarily for the use of the collection managers users will know the desired URL. There is also a sample entry under the text field to illustrate the proper format. Research has suggested that this type of form fill-in requires little training and simplifies data entry ( Palanivel and Helander *pg 141). In addition, due to the enormity of the collections available it will not be reasonable to offer a menu of choices for all directories.

    An important feature of the text field is the visual cue offered. The field already contains the first portion of a SunSITE URL. This serves two functions; it cues the user about the expected input format, and it saves the user from having to enter this portion of URL. The aim is to make this task as quick and clear as possible.

    Start and End Dates:

    These values will be presented to the user using pull-down menus. The range of dates to be queried will be determined by chosen values from a start date menu that will allow the user to choose a month, day, and year. These values will be compared to an end date menu with similar menu choices. The choices for the month will be English abbreviations, the days will range from 1 to 31, and the years will be given from 1996 to the present. The numbers and months are organized in a reasonable fashion to help improve performance ( Palanivel and Helander *pg 137). The available logs only go back to 1996, hence the date limitation.

    The menu design was chosen to indicate to the user what values are available and to help eliminate confusion. The format that the statistics code needs to process could be confusing if entered in a text field. The different formats for representing dates in different cultures could also lead to many errors. Menus offer a number of advantages over other designs. They can reduce errors, eliminate the need for memorization, and require little training (Hix and Hartson *pg 62).

    There is a lot of information available on the use of menus in interface design. Carey, Mizzi and Lindstrom (1996) did an investigation into user response to pull-down menus and traditional menus. In their work they found that users preferred traditional menus. Despite this, the stats page includes pull-down menus. There are two reasons for this design. The first reason is the focus on compact design and screen real estate. Traditional menus require more screen space (preece, *pg265). The second reason is that most users will be expert users and the study found that they less resistant to pull-down menus. It seems that expert users are more familiar with the pull-down design.

    Display By:

    The system can currently allow the user to choose whether they want daily or monthly statistics. These choices will be presented to the user with radio buttons. Radio buttons have been chosen so the user has cues as to what variables are available. In addition, radio buttons are exhaustive and mutually exclusive (Hix and Harton, *pg 64). This makes clear that the user is to choose only one value and it shows which is currently selected. Another advantage is that it allows for a default selection at startup.

    Output Format:

    The output format choice will be made using radio buttons. The reasons for this design are: The choice of radio buttons allows the user to see what values are available; A default can be set; and users can see current choice. If there were many formats to choose from radio buttons would not be practical.

    Recursion Level:

    The recursion level, or depth to be descended into the directory, will also be chosen with radio buttons. The system currently only supports viewing one directory deep because of concerns over processor limitations on SunSITE The term "recursion level" has been defined on the page to help users understand the task.

    Submit/Reset:

    The submit/reset task is presented using push buttons (Hix and Hartson, *pg 62-63). The buttons are clearly marked and the function of each is clear. In addtion, push buttons are common to forms on Web pages and should be familiar to the intended users.

    Further Consideration:

    The original plans for output choices (line or bar graphs) will not be incorporated in the early prototypes. The focus will be on creating an easy to understand query interface that will deliver the desired results to the user.

    Another consideration is a very fast access design for the expert user. The system will include a separate web access point that will not offer cues or options but will be accessible by simply entering the a specific URL that includes the directory for which the statistics are desired. The instructions for the expert system will be available from the main interface page. In this way the frequent user can learn the faster system if desired. The URL for the main statistics page will be made available through the channels of communication available at SunSITE. The primary channel will be the webgroup list, a collection of email addresses for all collection managers on SunSITE.

  • Mark C. McCarthy
  • INLS 257
  • Task Analysis
  • Usability Specifications:
    (Based on Hix and Hartson, 1993, pg 223)
    Usability Attribute Measuring Instrument  Value to be Measured Worst Acceptable Level  Current Level Planned 

    Target Level 

    Initial performance "Enter directory in text field" 

    tasks per benchmark 1 

    Number of errors on first trial 0 0
    Learnability "Get single recursion table of stats for specific day" 

    tasks per benchmark 6 

    Number of errors on first trial 0 0
    Initial performance "Get single recursion table of stats for specific day" 

    tasks per benchmark 6 

    Length of time to successfully get stats on first try  90 seconds 45 seconds 30 seconds
     
  • Mark C. McCarthy
  • INLS 257
  • Scenarios

    These scenarios are based on the scenarios for design in "Will the Real Scenario please Stand up?," by Robert Cambell

    The frequent user of the statistics

    David is the creator of the Neurosurgery pages in the SunSITE archives. David spends a great deal of time and effort updating the information on the pages. His hope is that these pages will be a resource used by neurosurgeons around the world. Now David wants to dedicate more of his time and funding to this project.

    In order to get the permission from his superiors, David must prove that the pages are being utilized by a wide audience. The supervisors ask that David get statistics each week and make the information available for them. They state that they don't want anything fancy, just the numbers. David goes to http://sunsite.unc.edu/stats where he can get the statistics he needs.

    David enters the directory for his research, dates for the last 6 months, the monthly output, and chooses a table format. The data comes back with number of hits for each page within his directory. He prints out this information and has it sent to his supervisor. The word comes down that these statistics are too difficult to interpret. Now David must find a way to display the information in a more usable format. David goes back to the stats page, but this time he sets the output format for bar graphs.

    The infrequent user of the statistics

    Carl has just recently been given an account on SunSITE and has created a resource for educators in correctional institutions. . He has no funding for his work and does it on his free time. He got a message about the new statistics page that is now available on SunSITE. Out of curiosity he checks it out. When he gets to the page he sees the first instructions to enter the directory that he would to get statistics from. He does this and then moves down the page making choices for the presentation of the material. He chooses the statistics for the last week and has them displayed as bar graphs.

    General description of task characteristics

    How frequently do users perform the task?

    The users' frequency of use will vary greatly with each user. Those that are getting funding to maintain archives will have a greater incentive to check statistics than those that have creative collections they maintain in their spare time.

    What are the time constraints on the task?

    The task will take very little time to use when the system is running properly. It will allow those that have frequent use to get very quick results while providing cues for the infrequent user.

    Is system use mandatory or discretionary?

    The use of the system is discretionary.

    Questions related to the environment or context

    How important is the task?

    For many the task will be one of satisfying curiosity. In contrast, some users will depend on the statistics as an indication of how useful their archives are and if it is a good investment of time and energy.

    Where is the task performed?

    The task can be performed from anywhere with an Internet connection and a web browser. Most users will use the system from where they normally work with the Web. For most archive owners that will be here in the United States, but for some collection owners it will be from other continents.

    What is the relationship between the users and the data?

    In some cases the data will reflect the number of http requests for their pages or in other cases the information will show activity on a page that the user doesn't have a part in creating.

    What other tools does the user have?

    Expert users can try to put into place alternative statistical gathering agents but this is difficult and not recommended. These agents require extra processor power that can degrade SunSITE performance if there are many such programs running.

    What is the turnover rate among users?

    After the initial implementation the user turnover rate will be low. The exception is the curious visitor to sunSITE that decides to check the stats.

    Questions related to the complexity/difficulty of the task

    How complex is the task? How difficult is the task?

    The task is very simple for those familiar with the World Wide Web. All the parts of the interface are common to Web pages.

    How structured is the task?

    The task is well designed and structured for ease of use. The interface is designed to reduce errors while offering as many options as the current software can support.

    What training will be provided? How is the task learned?

    There will be no training besides the information made available on the system. There are a number of cues and examples to support users.

    What happens when things go wrong?

  • When things go wrong the user will get information other than he/she was looking for. The name of the directory that is being queried appears at the left of the table. The names of lower level directories and files also appear in a cell at the left. If an invalid directory is entered a table will display with the name of the incorrect data at the left and all other variable showing zero. The user can simply go back to the page and re-enter the data.
  • GOMS Task Description
  • Goals:

    Tasks:

    Actions:

    Hierarchical task analysis

    GOMS Task Description

    Method for accomplishing goal of obtaining Web page access statistics on SunSITE.

    Step 1 Start browser.

    Step 2 Accomplish goal of getting to the GUI stats page.

    Step 3 Enter top level directory to be queried in text field.

    Step 4 Choose start date from pull down menu.

    Step 5 Choose end date from pull down menu.

    Step 6 Choose display format from radio buttons.

    Step 7 Choose output format from radio buttons.

    Step 8 Choose recursion level from radio buttons.

    Step 9 Submit query.

    Step 10 Return with goal accomplished.

    Bibliography

    Hix and Hartson, pg 223 for the interaction chart.

    Preece for Hierarchal task analysis chart, pg 415

    Preece for GOMS pg 420.

  • Savage, Pamela
  • user interface evaluation in an iteratve design process..
  • Robert Cambell Wll the Real Scenario please Stand up

  • Human Factors Issues in dialog design
  • Thiagarajan Palanivel and Martin helander
  •  

    NJ Lightner Problems it te WWW

    JM CArey

    Robert Cambell, Will the real scenarios please stand up.