ASSESSMENT FOR HCI
1. Human Computer Interaction (HCI) is about designing
computer systems that support people so that they can carry out their activities
productively and safely. HCI has a role in the design and development of
all kinds of systems, ranging from those like air traffic control and nuclear
processing, where safety is extremely important to office systems, where
productivity and job satisfaction are paramount, to computer games, which
must excite and engage users.
OR
HCI studies the communication between man and
machine. In the past few years it advanced to the fastest growing sectors
of the computer industry as the complex systems are at the brink of becoming
unmanageable and use able by normal people. HCI involves both hardware
and software user and system modelling, cognitive and behavioural sciences,
human factors, empirical studies, methodology, techniques and tools.
2. All HCI takes place within a social and organisation context. Different
kinds of applications are required for different purposes and care is needed
to divide tasks between humans and machines, making sure that those activities
that are creative and non-routine are given to people and those that are
repetitive and routine are allocated to machines. Knowledge of human psychological
and physiological abilities and more important still, their limitations
is important. As this figure shows below this involves knowing about such
things as human information processing, language, communication interaction
and ergonomics. Similarly it is essential to know about the range of possibilities
offered by computer hardware and software so that knowledge about humans
can be mapped on to the technology appropriately. The main issues for consideration
on the technology side involve input and output techniques, dialogue techniques,
dialogue genre or style computer graphics and dialogue architecture. Tools
and techniques are needed to realise systems. Evaluation also plays an
important role in this process by enabling designers to check that their
ideas really are what users want.
Insert image
3. The impacts of the following are :-
VDUs :-
One issue in particular, which has been the focus of considerable media
attention is health hazards and the use of VDUs. All sorts of claims have
been made about the potentially dangerous physical effects of prolonged
use of VDUs. The most scare mongering have included the inducement of epileptic
fits, spontaneous abortions and radiation sickness, and the development
of contacts and skin diseases. Those more commonly reported include complaints
of eye strain, headaches and muscular fatigue. Clearly this is very sensitive
issue where the application of HCI knowledge and research can clarify the
nature of the problems. The situation however is by no means clear-cut
and there is considerable conflicting evidence. With regard to radiation
induced effects, it appears that these should be minimal since the emission
level of radiation from VDUs is no higher than natural background levels
(Terrana et al., 1980). But there may still be cause for concern in particular
circumstances, such as long term exposure and low level emissions from
other equipment components. In contrast there is considerable evidence
to suggest that using VDUs can cause a number of visual, skeletal and muscular
strains, the most common symptoms of which are burning, of which are burning,
itching and watering eyes, blurred vision, aching shoulders and backache.
Keyboard inputs:-
The most common method of entering information into the computer is through
a keyboard. Since you have probably used them a lot without perhaps thinking
about the related design issues, thinking about keyboards is a convenient
starting point for considering input design issues. Broadly defined , a
keyboard is a group of on-off push buttons which are used either in combination
or separately. Such a device is a discrete entry device. These devices
involve sensing, essentially one of two or more discrete positions( for
example keys on keyboards, touch-sensitive switches and buttons) which
are either on or off, where as others (for example, pens with digitising,
tablets moving joysticks, roller balls and sliders) involve sensing in
a continuous range. Devices in this second category are therefore, known
as Continuous entry devices.
Pointing devices :-
Pointing devices are input devices that can be used to specify a point
or path in a one, two or three dimensional space and like keyboards, their
characteristics have to be considered in relation to design needs. Example
pointing devices include joysticks, track balls and mice. Pointing devices
are typically continuous entry devices, although a device like a mouse
uses both continuous movement and (with its button) discrete movement.
Graphical interfaces :-
Ideally, it may seem desirable to present information on the screen that
has characteristics. The visual system could then use the same processes
that it uses when perceiving objects in the environment. In particular,
design and manufacturing applications might benefit from the use of realistic
images in helping the users design and create objects. The problem with
this approach however, is the high cost of real-time image generation.
More over, when considered against the actual needs of an application,
such a degree of realism is often unnecessary. For example, in a flight
simulator it is less important to deceive pilots into believing that they
are flying through real terrain than it is to provide all the necessary
information in the right form to allow them to function as if they were
in a plane. It is classified in terms of
(i) the various kinds of graphical modelling techniques used to represent
three dimensional objects and scenes, and
(ii) the various forms of graphical coding used to represent different
types of system information at the interface. The focus is on the efficacy
of the different representational forms in terms of the function they are
intended to provide, their perceptual discriminability and ease of recognition.
And more follows,
Graphical user interfaces (GUIs) have brought quantifiable benefits to
users and organisations. Recent studies have shown that users of graphical
user interfaces make fewer mistakes, feel less frustrated, suffer less
fatigue, and are more able to learn for themselves about the operation
of new packages than users of non- graphical or character-based user interfaces.
From a software designer's point of view, however, GUIs are more difficult
to design than character-based interfaces. The user's interaction with
the GUI is more complex because it is based on principles of direct manipulation
and concurrent user access to multiple windows, icons, menus and input
devices. A character-based interface, however, normally only allows the
user sequential access: first view a menu, then make a selection, then
view the next screen, then enter the data. With the character-based interfaces
the designer can (within limits) design the user interface so that the
user will undertake a task in a predefined sequence. With the GUI, the
designer can design the user interface so that certain actions are allowed
on interface objects and the user will decide which actions to take and
(within limits) in what order.
Systems which support groups of users working together are capable of bringing
productivity gains to organisations; for example, by enabling design teams
to collaborate on the development of computer-based designs and to hold
discussions or brainstorming sessions over the network. Consulting companies
can pool their world-wide expertise to prepare bids rapidly for complex
projects. Normally group ware systems support users sharing information
about common objects, such as a report, a bid or a design, in such a way
that each user sees the same version of the object on their screen at the
same time. The user interfaces to group ware systems are normally graphical
but with the added complexity of supporting communication between group
members and the sharing of common objects. Often many different media will
be used to aid communication between group members; video and voice interaction
may be as important as text and graphics. From a software designer's point
of view, designing user interfaces for group ware systems is more difficult
than designing a GUI because the designer must also consider multi-media
interaction and support for communication between people.
The design and implementation of a successful user interface is important,
therefore, not only to the software designer but also to users, groups
and organisations. The next section describes the main activities which
contribute to the design of the user interface and discusses the role of
the software designer in each of these activities.
4. In the user interface design phase the design of the interface is determined
and tested. This involves:-
* the design of user actions
* how to give feedback to the user
* the screen design
* design of possible user activities
* representation of the system’s functionality to the user
* planning of task sequences
* design of access mechanisms
* the design of user interface objects and interaction styles.
Classes of user interface
Interaction between the end-user and the system
is achieved through interactive dialogues. There are a number of different
classes of interactive dialogue and each of these has advantages and disadvantages
depending on the situation in which they are used. The five major classes
are discussed below:
a.Command language
Command language dialogues are those in
which the user types instructions to the computer in a formally-defined
command language. For example, `mv file1 file2', is a UNIX command for
copying file1 into file2. The advantages of this approach is that it is
very flexible, allowing users to create their own commands; it supports
user initiative and it appeals to `power' users, typically to software
and system developers. Command language usually requires a significant
level of training and a high degree of memorisation.
b.Natural language
Natural language interfaces are those in which the user's command language
is a significant, well-defined subset of some natural language such as
English. For example `Which women work in New York City' is a typical user
input to the Intellect system from AI Corp., Cambridge, MA. Natural language
interfaces are typically easy to learn, although they often require considerable
typing skills on the part of the user. They can also be slow to use if
the system is unclear as to the exact meaning of the user request and has
to seek clarification. However, natural language systems are increasing
in sophistication and a great deal of research and development work is
currently being undertaken.
c.Menu systems
Menu systems allow the user to issue commands
by selecting choices from a menu of displayed alternatives. Menu systems
are popular since they reduce learning time, reduce the number of keystrokes
necessary and structure decision making. Most of the currently available
fourth generation environments provide screen design tools which support
the development of menu-based interfaces.
d.Form filling dialogues
Form filling dialogues are those in which
the user enters data by filling in fields in one or more forms displayed
on the screen. The use of forms on the screen considerably simplifies data
entry and requires very little training to use. Forms management tools,
similar to those available for menus, can be found within fourth generation
environments.
e.Direct manipulation interfaces
Direct manipulation interfaces are those in which the user manipulates,
through button pushes and movements of a pointing device such as a mouse,
a graphic or iconic representation of the underlying data. An icon is a
graphical symbol or pictogram used instead of words. Most direct manipulation
interfaces use window systems or environments, in which the user's screen
is divided into a number of possibly overlapping rectangular areas, each
of which handles a specific function. Direct manipulation interfaces represent
task concepts visually, are easy to learn and use, encourage exploration
or experimentation with the system features and generally result in a high
level of user satisfaction. Such interfaces are traditionally difficult
to design and to program. However, most of the user interface design standards
currently being put forward are based on direct manipulation interfaces.
Choosing the most appropriate class of user interface to match the needs
and expectations of the users is an important aspect of good user interface
design. For any given class of user interface a number of design decisions
must be made by the interface designer, particularly in terms of what information
should appear on the screen, how much information, in what order, what
type of error messages, where should error messages be displayed on the
screen and so on.
5. There are several recognised styles of interaction that can be
utilised.
Rapid or Throwaway prototyping :-
is also used to collect information on requirements and on the adequacy
of possible designs. In rapid prototyping, the prototype is thrown-away,
in the sense that it is not developed into the final product, although
it is an important resource during the project’s development. Recognises
that requirements are likely to be inaccurate when first specified. The
emphasis is an evaluating the prototype before discarding it in favour
of some other implementation. Incremental
prototyping :- allows large systems to
be installed in phases to avoid delays between specification and delivery.
The customer and supplier agree on core features and the implementation
is phased to enable an installation of a skeleton system to occur as soon
as possible. This allows requirements to be checked in the field so that
changes to core features are possible extra, less important, features are
then added later. The system is built incrementally, one section at a time.
Incrementally prototyping is based on one overall design.
Evolutionary prototyping :-
is the most extensive form of prototyping, it is a compromise between production
and prototyping. The initial prototype is constructed, evaluated and evolved
continually until it forms the final system. Some designers believe that
more acceptable systems would result if evolutionary prototyping were interspersed
with periods of requirements animation or rapid prototyping. The only problem
is that evolutionary prototyping tends to encourage designers to “fix”
on a particular solution too soon rather than exploring the alternatives
more fully. Compromise between production and prototyping. The system can
cope with change during and after development. Helps overcome the traditional
gap between specification and implementation.
A full prototype
as the name suggests, contains complete functionality although with lower
performance.
A horizontal prototype
shows the user interface but has no functionality behind the buttons.
A vertical prototyping
contains all of the high level and low level functionality for a restricted
part of a system.
High fidelity prototyping
refers to prototyping through a medium, such as video, which resembles
as closely as possible the final interface. High fidelity video prototypes
tend to be popular with commercial organisations because they make the
product appear very polished and aesthetically pleasing.
Low fidelity prototyping
involves the use of materials that are further away from the final version
and that tend to be cheaper and faster to develop. For example, a software
version of the interface with cut down functionality would be higher fidelity
than story boards.
Chauffeured prototyping
involves the user watching while another person, usually a member of the
development team, ‘drives’ the system. It is away to test whether the interface
meets the user’s needs without the user actually having to carry out low
level actions with the system. This may appear to contradict the intentions
behind involving the user, but it can be useful for confirming, for example,
the sequence of actions needed to perform a task.
Wizard of Oz prototyping
also involves a third party, but in this case, the user is unaware of it.
The user interacts with a screen, but instead of a piece of software responding
to user’s requests, a developer is sitting at another screen answering
the queries and responding to the real user. This kind of prototyping is
likely to be conducted early in development to gain an understanding of
the user’s expectations. There is an added advantage for the development
team is using Chauffeured or Wizard of Oz prototyping in that extra understanding
can be achieved through being involved so closely with the users.
The
use of different kinds of prototype in different stages of a design results
in a two-phase view of iterative design. In the first phase, prototypes
are developed to gather different forms of information, and radically different
alternatives may be tested in parallel. At some point, this prototyping
ends with a proposal for a single full initial design. One solution is
then iterated through design, code and tested cycles. Any further radical
changes are unlikely, as production standards will now be in force, and
major changes will be expensive. This phase can be regarded as a convergent
fine-tuning stage with a slow cycle time. The earlier prototyping phase,
by contrast, is a divergent, exploratory and bold stage, with fast cycle
times and the preservation of alternative designs.
6.
(a) The expert evaluation
will be as follows, because the expert will evaluate it by looking at :-
* the kind of information required,
* the nature of the system or specification that is being evaluated,
* the stage in the life cycle being evaluated
* whether or not statistical validity is needed (for example, if you are doing a formal experiment or large-scale survey)
* the resources available.
Another answer:-
* the characteristics of the users (or the predicted users) of the product who take part in the evaluation (for example, experience, age, gender, psychological and physical characteristics)
* the types of activities or predicted activities that the users will do. These may range from tightly specified tasks, which are defined and controlled by an evaluator, to activities decided by the users.
* the environment of the study which may range from a controlled laboratory situation to a natural work setting. If it is the latter, the study is known as a field study.
* the nature of artefact being evaluated, which may be anything from a series of sketches to a working software prototype or fully developed product.
Even in predictive evaluations, in which experts attempt to predict the usability of a system without directly involving users, evaluators take these same four aspects into account. They do this by drawing on their knowledge of the kinds of things that typical users would do and the sorts of circumstances in which they know typical users of the system will work.
User’s evaluation :-
As well as examining user’s performance, it is important to find out what
they think about using the technology. However good user’s performance
scores are when using technology, if they do not actually like using it
for some reason it will not be used. Sometimes, for example a quite small
and seemingly trivial feature (to designers) may be extremely annoying
to users. Surveys using questionnaires and interviews provide ways of collecting
users’ attitudes to the system.
(b). Heuristic evaluation:-
Molich and Nielsen (1990) devised a method known as heuristic evaluation
in response to need for cheap, cost effective methods that could be used
by small companies who could not afford or did not have the facilities,
time or expertise necessary to do usability engineering. In heuristic evaluation
reviewers examine the system or prototype as in a general review or usage
simulation, but their inspection is guided by a set of high-level heuristics
(Nielsen, 1992), which guide them to focus on key usability issues of concern.
The following list is typical of the kind of heuristics that can be used.
* use simple and natural dialogue
* speak the users’ language
* minimise user memory load
* be consistent
* provide feedback
* provide clearly marked exits provide shortcuts
* provide good error messages
* prevent errors
According to Nielsen (1993) each reviewer normally does two or more passes through the interface in order to :
* inspect the flow of the in the interface from screen to screen,
* inspect each screen one at a time against the heuristics, examining features such as system messages, dialogue and so on.
Cognitive Walkthroughs :-
the third kind of expert review is a walkthrough. As in software engineering,
the goal of a walkthrough in HCI design is to detect problems very early
on so that they may be removed. Walkthroughs involve constructing
carefully defined tasks from a system specification or screen mock- ups.
A typical example would be to walk through the activities (cognitive and
operational) that are required to get from one screen to another. Before
doing the walk through experts determine the exact task that will be done,
the context in which it will be done and their assumptions about the user
population. They then walk through the task, reviewing the actions that
are necessary to achieve the task, and attempt to predict how the user
population would most likely behave and the problems that they would encounter.
In many respects this is similar to a review, except that it requires a
more detailed prediction of user behaviour.
HCI
researchers tend to adapt methods to fit their needs: Polson et al. (1992),
for example, take a strongly cognitive stance. They describe their brand
of cognitive walkthrough as a ‘hand simulation of the cognitive activities
of user’. The focus is cognitive and the underlying aim is to identify
potential usability problems. This is a fairly fine-grained approach, which
relates closely to cognitive task analysis. By structuring the review process
around specific questions, psychological theory can be embedded into it,
which helps to illuminate how well the interface fulfils the cognitive
needs of the intended users. The method is still being developed and more
effective versions that can be adopted by companies may emerge.
Following is example of a checklist for doing a cognitive walkthrough.
Cognitive walkthrough start-up sheet Interface _____________________________________________________
Task _________________________________________________________ Evaluator(s)
_______________________________ Date _______________
Task description:
Describe the task from the point of view of the first-time user. Include
any special assumptions about the state of the system assumed when the
user begins work. Action sequence: Make a numbered list of the atomic actions
that the user should perform to accomplish the task.
Anticipated users:
Briefly describe the class of users who will use this system. Note what
experience they are expected to have with systems similar to this one,
or with earlier versions of this system.
User’s initial goals:
list the goals the user is likely to form when starting a task. If there
are other likely goal structures list them, and estimate for each what
percentage of users are likely to have them.
Next action #:_______ Description:_______________________________________
1. Correct goals
II. Problems forming correct goals
A. Failure to add goals. ____%
B. Failure to drop goals. ____%
C. Addition of spurious goals. ____%
No-progress impasse. ____%
D. Premature loss of goals. ____%
Supergoal kill-off. ____%
III. Problems identifying the action
A. Correct action doesn’t match goal. ____%
B. Incorrect actions match goals. ____%
IV. Problems performing the action
A. Physical difficulties. ___%
B. Time-outs. ___%
Evaluating guidelines :-
Guidelines inevitably contain some overlapping and contradictory advice.
Consistency might, for example, be important for learning to use a particular
system but troublesome when a user becomes experienced. Many of these contradictions
appear early and then disappear during the design process as high level
principles become refined into lower level design rules. Often the constraints
imposed by the characteristics of the users, their work and the environment
will remove the need to choose between contradictory items of advice, because
one refinement of a higher level principle will clearly apply when others
do not. The skill that must be acquired in order to accumulate and apply
guidelines wisely is the ability to evaluate them critically. So how does
one evaluate guidelines?
There
is no mechanical technique. Principles have different forms, which affect
how one goes about evaluating them. Generalisations such as ‘users will
treat computer systems as if the latter were human’ are based on inductive
reasoning from experience. We can see people doing this. Prescriptions
such as ‘develop a consistent intuitive conceptual model when designing
a system’ are based on deduction from the effects of not doing this. To
evaluate guidelines based on inductive reasoning the data on which the
principles are based be available. If, however, these data consist of the
undocumented collective experiences of design community, then inexperienced
designers cannot access them easily. For guidelines based on deduction,
specialist knowledge may be the basis for many of the inferences that lead
to the formulation of a principle. If the authors of a guideline do not
present an argument for a guideline, it is easy to reject it, even if it
is good. Authors may also present a false argument. A reasonable guideline
such as keeping colour coding simple can appear unreasonable if tit is
deduced from the wrong evidence and if it is made too specific (for example,
allowing no more than seven colours at once).
Observing (laboratory and field) :-
in one way or another, several different kinds of evaluation depend on
some form of observation or monitoring of the way that users interact with
a product or prototype. Observation or monitoring may take place informally
in the field or in a laboratory as part of more formal usability testing.
Alternatively, it may be done from a participative or ethnographic perspective
with the aim of really trying to understand how users themselves interact
with technology in natural settings. There are a number of techniques for
collecting and analysing data. Data may be collected using direct observation
with the observer making notes or some other form of recording such as
video may be used. Keystroke logging and interaction logging can also be
done and often they are synchronised with video recording. The way data
is analysed will depend on the question that the evaluators want to answer.