7. Further Work and Conclusions



This thesis has shown how a set of tools have been developed for the elicitation of information about feelings, people and places. Despite the difficulties encountered, requiring modifications to some of the modules, the tools that have been produced have delivered some interesting results. This chapter will start by suggesting new areas for the application of the computer package, and will also address ideas for possible adaptations to the modules of the package as greater computing power becomes available. The chapter finishes by examining some of the conclusions drawn during the package development.


7.1 Planned Changes to the Package


We now examine some of the proposals for changes to existing modules in the package, and at the reasoning behind the changes. Ideas for augmenting the existing modules in the package are also examined.

7.1.1 Customising a building.
It is intended that once a basic building-shape has been chosen, it will be possible to customise it by adding further items, such as different windows etc. There will also be a facility for repositioning items on the basic house-shape. A possible interface for this is now being designed, based on the slide-bar, and a further tool which fills the slide-bar with other objects of the same type as an object chosen from a palette. This will address a problem that was apparent in early tests with the Scrapbook: that children will quite often not accept a picture as being a representation of their house because they will point out that it does not look like their house. By adding an ability to customise the picture it may be possible to present the child with sufficient extras that they will accept the house representation. It is also likely that if a child has customised a house in a certain way, and it turns out to be a true representation, that other things that the child describes are more likely to be believed.

7.1.2 Free-Form Text-Input
This would add to the versatility of the package. A child could at any time click on a button - whereupon a text box would appear which allowed the child to type a message. This message would be stored with the results file. This feature could be used in whatever way was thought useful - whenever a child wants to comment on anything. Such a system would allow the child's comments to become part of the recorded "script" of the interview session, and give them the validity of the rest of the child's recorded actions.

7.1.3 New Interface for the Emotions and People Module
There are still some problems with the current interface for the emotions and people module, and so some ideas have been put together for a possible new interface. The current interface suffers from the following drawbacks:

  1. If there are more than seven people the whole family must be reduced in size to fit all the individuals into the window. This has the effect of drawing the names in a smaller font too. When the font becomes smaller it is more difficult to read, until the point at which it becomes impossible to read.

  2. Only the top half of the display is used to hold the family to choose from, and selected individuals move from there to the bottom. In effect, just under half of the display is empty at any one time.


A new interface may include the following features:

  1. The names could be attached to the figures such that they fit better into a small space. (See figure 7.1 for an example of the current interface, and figure 7.2 for a suggested change).

  2. The display could be divided by a vertical line which marks the boundary between selected and unselected individuals. This boundary line could move to accommodate more individuals as they are selected and move from one side to the other.

    Fig. 7.1 On the left a group of people with their names in boxes underneath them, on the right the same group, but shrunken slightly in size. Note the names are more difficult to read.


Fig. 7.2 Although the people are smaller, the names are still easy to read since they are not reduced in size. They are repositioned to make way for each other.

7.1.4 Setting up the Package in Advance
The SAGE approach depended on the interviewer being aware of the child's "universe" before the interview took place. Carrying this over into the computer package means creating settings, each complete with a global set of people that can be chosen by the child to populate the setting. An interface for this set-up module (or possibly a separate program) has been discussed. Figure 7.3 shows a mock-up of a dialog box for defining care contexts, or settings. The interviewer uses this to create settings for a child to talk about. For each setting the type of context is recorded from a pop-up menu of choices (e.g. full time residential, secondary residential, visited non-residential, Out, Education, Leisure, Other), and a name or description can be associated with it. The interviewer also assigns dates to definition of the context (if these are known). Dates are defined either as historical (having a start and end value), or current (starting at a particular value and still ongoing).

Each care context may have appropriate speech associated with it. For example, when asking about a current full time residential setting, questions would probably refer to "the place you live now". The default speech type proposed by the package would match with the context type, and the type of dates associated with it. The interviewer may override the default if he wishes.


fig. 7.3 Dialog for defining a care context.


7.1.5 Customising the Package
In order for the package to have wide applicability, it is envisaged that there will be a vast array of artwork which may be selected for presentation at a particular interview. This artwork would include specific scenes for the emotions module, and specific individuals for the people module. A technique for presenting this artwork in a manner that makes it easy to select the required items has to be developed. Some work has already been done on an interface that will permit a user to choose from a set of items using buttons which define the properties of the item (figure 7.4). Consider the selection of scenes for the emotions module. The proposed interface will present a set of pop-up menus which describe aspects of the scene, and in doing so will narrow down the choices available. The dialog box which is used to select the scene, and which contains the pop-up menus, will also display a thumbnail view of any scene which matches the specifications, with an option to preview a selected scene at the full size.


fig. 7.4 Customising the available scenes for the Emotions module.






fig. 7.5 clicking the buttons in the "people in scenes" section of the dialog results in a filter being applied to the scenes being offered for selection.

The "people in scenes" section acts as a filter on the number of different scenes available for selection (figure 7.5). There are three buttons:

Child only , which excludes all scenes which feature other people and disables the four buttons in the lower-right of the section.

Anyone , which includes all scenes which feature any other person and disables the four buttons in the lower-right of the section.

Include , which enables the four buttons in the lower-right of the section, thus allowing the user to select the type of other person who is permitted to appear in the scenes.

7.1.6 Identikit
This proposed tool will allow a user to customise the appearance of people that they have selected using the People module. Customising, in this context, means the addition of features that will allow a user to specify a person more exactly. A number of features of an individual may be changed with this tool: hair colour and style, colour of skin, addition of glasses, addition of facial hair. Without such a tool it would be impossible to present the user with sufficient breadth of individuals for them to choose from. It is obviously important in this package to give the user enough scope to be able to identify people. Customising will take place once all the people have been selected, though the module will permit the user to modify their choice of individuals in case the act of customising brings to light some problems in their choice of people. This particular module has not yet been written, though a number of ideas have been mooted for it, and prototypes have been designed using HyperCard (see Appendix D for further information about HyperCard). One idea for a possible interface is to use the slide bar (see section 4.4.6) to present a series of items that may be placed on the pictures of the individuals. In this respect it may borrow from the interface for customising a building.


7.2 Potential Enhancements Under Consideration


7.2.1 External Control of the Package
In order for the package to be versatile it is necessary to be able to have some lee-way to navigate as required amongst the various modules. If the interviewer has to do this while the child is using the system it could become off-putting to the child. One possible answer to this would utilise the networking facilities built into every Macintosh, and control the package externally from either another Macintosh, or an Apple Newton:

Macintosh - A package called Timbuktu allows a user on one Macintosh to observe, and to take control of another. This would permit someone in a different location to watch what is taking place on the interviewer's Macintosh, and to change the course of the package through the modules as new information comes to light.

Apple Newton - A slightly simpler technique would involve connecting an Apple Newton and a Macintosh via a network. The Newton and the Macintosh share network protocols for the transfer of data, and a simple program could be written for the Newton which would allow it to send navigation commands to a package running on the Macintosh. Such a system would be likely to be less intrusive than having to reach across the child to use the mouse to select a different module.

7.2.2 Extended Use of Colour and QuickTime Video
As the level of sophistication of computer games used by children has increased, so children's expectations for all computer packages have risen. When the modules of this project were first designed, it was necessary to avoid the use of colour, since colour Macintosh systems were too expensive to be generally available. This is no longer the case, and colour has been incorporated into the package where it is reasonable to do so, and where it does not complicate the interface.
Now that powerful Macintosh systems are generally available, one place in which colour and QuickTime could be used is the upgrading of the bird agent figure into a high-quality cartoon, or a video of a person or figure. This could lead to improved speech quality, and the ability of the agent to bring to the child's attention, things that they should be aware of. One important aspect of the package is that it should captivate the child so that they enjoy using it, and the possibility of making the package more exciting by having better quality visual images could help achieve this.


7.3 Conclusions


A number of conclusions about the development of complex computer systems can be gathered by examining the history of the project so far:

7.3.1 User-Testing and Prototyping
Interfaces developed away from users may be doomed to failure. It is important to try interfaces on the target group of users as soon as possible in the development phase. Even though some of the user interfaces in this project seemed to be perfectly reasonable, when tried on the target group of users they failed to work satisfactorily. Once it becomes obvious that an interface is not working it is important that it must be refined or replaced as appropriate.

The simplest method of designing an interface turns out to be drawing it on a piece of paper. Ideas for a number of different dialog boxes and other interface items within the package were fleshed out using designs produced in this manner. Although this technique can only show the interface very roughly, once a design has been accepted it can then be prototyped in HyperCard, and these prototypes can be tried out on a group of target users.

7.3.2 Participative Development
The involvement of designers from the fields of psychiatry, forensic science, psychology and computer science has been very valuable in creating usable designs in this project. By being able to share information from all the pertinent viewpoints, it has often been possible to see shortcomings in procedures before they are tested and demonstrated to fail on real users. One of the main problems with this attempt at participative development, however, has been informing the non computer-literate members of a team of just what may easily be implemented, and what it may be very difficult to implement. Having a team member with skills in both psychology and computing helped to translate the "psychology speak" into computing and vice versa, and helped reduce this problem.

7.3.3 Barriers to Communication
Easy-to-use computer systems can help break down communication barriers. Results from using modules of the package have already shown indications that children have reported their experiences more fully and more accurately, than they do when traditional interview techniques are used. This seems to back up the research in this area.

7.3.4 Modelling Real-Life Situations
It would be preferable to allow the child complete freedom of choice when using the package, however, this is simply not possible. Real-life situations are too complex, there are too many possible options, and attempting to offer all the possibilities leads to a system that is too unwieldy and difficult to use. What is really vital is that there are enough alternatives, so that the user's choice is not too limited. What constitutes "enough" is different for each situation, and it is only by testing the range of choices on the target group that it is possible to be sure that the range offered is sufficient. This package is also able to offer the ability to modify the alternatives offered, depending on decisions taken by the interviewer in advance of the interview. This means that the package can be targeted more specifically at an individual user, while also being more applicable to a larger target audience.

7.3.5 Usefulness of Function Libraries

By building up libraries of often-used functions, and where possible, making very specific functions more generally applicable, it has been possible to reduce the complexity of the code in individual modules, and to pass on changes and bug-fixes to every module that could take advantage of them very simply.


7.4 Overview


The package documented in this thesis has so far been used only in pilot and small-scale studies, but the results have shown great promise. The team at Liverpool are very satisfied with the results to date, and plan to distribute copies of the package to other psychologists. This will form the alpha testing phase, and it is anticipated that the feedback from their observations will provide the data for further changes to the package, and alternative uses for existing modules. From the history of the development of the package to date, and the responses of the people who have seen it, it seems clear that it will prove a useful tool for anyone who wishes to talk to children about places, people and feelings. The design of the system has made it versatile enough to be of use in many different sorts of "interview", whilst being easy to set-up by non computer-experts. It is also clear that the effect of such a simple to use, friendly computer system, is to make people feel more at ease, and more likely to answer questions truthfully. Although the package will continue to undergo changes, it seems that even in its present state it will be of much use.
back to Chapter 6. forward to Appendix A