a) Please provide BRIEF answers to the following questions:
- What does the acronym HTML stand for?
HTML stands for Hypertext Markup Language.
- Describe two problems with 'think aloud' evaluations.
Think aloud evaluations can yield unreliable results if the process of 'introspection' and talking
about using a system actually interferes with the users interaction. Finite cognitive resources can
be used in maintaining a 'conversation'. Alternatively, this can encourage users to think more
closely about what they are doing than they would under normal interaction. Further problems
stem from the subjective responses that this can elicit.
- Describe two situations in which keyboard input might be preferred over speech for data entry.
Noisy environments can reduce the recognition rates. Speaker independent interaction can be
difficult to achieve, for instance in a walk-up and use cash point machine. Speech impediments
and other forms of disability can create difficulties etc.
- What is a user interface widget.
A widget is a component of a user interface; examples include a scroll bar, an icon, a menu etc.
They can also provide a high-level abstraction for interface development toolkits.
- What is the significance of 7 (+ or - 2) for human-computer interaction?
7 is popularly believed to be the number of unrelated items that people can conveniently hold in
short term memory. The + or 2 refer to the impact of performance shaping factors such as
noise, fatigue and distraction on that capacity. Interface designers can use knowledge of short-
term memory to inform the development of an interactive system. For example; by limiting the
number of menu items or form fields on an interface. Techniques such as chunking can be used
to extend this capacity.
[2 marks per answer, 10 marks in total]
b) Why do Internet retrieval delays continue to be a problem even for those users who have high-
performance computers and good network connections?
[3 marks]
Retrieval delays continue to be a problem even for users with relatively good performance computers and
network connections because:
- The servers to which they are connecting cannot always service the demands that are placed on
them. It can be difficult for companies to predict the computational power required by a server
given the elastic demand of many Internet sites. Further problems arise because servers may only
be overloaded during short periods of the day.
- Users will often tailor their interaction so that it always begins to exceed the performance
characteristics of the available technology. They will run more simultaneous applications;
download more resources etc until delays begin to affect their tasks.
- The developers of many web sites continue to invest in computationally expensive presentation
techniques. The desire to innovate and to attract more users to a site may lead people to use
techniques that stretch even the best available technology.
- etc.
c) Explain why relatively short but unpredictable delays create greater usability problems than longer
delays that users can anticipate.
[5 marks]
If users know that a particular retrieval task will take a long time then they can schedule it appropriately.
For example, a software download for an OS upgrade is likely to consume significant amounts of time and
hence may be done overnight. If the download is unpredictable then it can be extremely irritating. This is
typically more of a problem if the task takes longer than anticipated but can also be an issue if a longer task
completes or aborts long before the user expects. With unpredictable delays there are a number of
techniques that designers can exploit:
- take action to increase the predictability of a download task. Installing additional server capacity
will reduce the maximum delay during peak loading towards the more usual time. Alternatively,
tasks can be refused if the server reaches a particular saturation. It can be argued that this makes
the retrieval delay more predictable even if completion is now less predictable.
- give a worst case estimate. This places a bound on delay.
- more from interactive to batch mode download techniques so that users do not have to sit waiting
to respond to series of questions for unpredictable retrieval tasks. Users can then go away even if
the delay is prolonged.
- etc.
d) You have been asked been asked to design the user interface to an e-commerce web site. The site will
sell expensive, luxury cars to members of the public. Write a brief technical report describing how you
would test the site to determine whether retrieval delays will create significant usability problems for the
users of this site.
[7 marks]
The key point is that this is a high-value product. The site may, therefore, exploit a range of more
advanced marketing techniques and high-end audio-visual presentations. These resources may incur high
download times. Hence the testing must consider:
- Retrieval times from different types of end-user connections. It may be better to provide a more
basic form of the web-site to users who are accessing the site from relatively slow modem
connections. In this circumstance, however, this style of connection may be relatively rare for
potential customers. Market research is required.
- Retrieval times from different geographical regions. It might be appropriate to deploy regional
servers to mirror the main site and provide tailored local information. This approach can help so
that the capacity of the infrastructure is tailored to the local demand at a regional level.
- Retireval times at different times of the day. Any evaluation should not only consider the logical
and physical location of the end-user/customer. It should also consider changes in response times
at different times of the day/week. These will be influenced by the loading on routers etc by other
network users however, periods of peak Internet usage need not coincide with periods of peak
interest in this particular website.
- User expectations and performance characteristics of resources compared to download times.
Retrieval delays and periods of waiting can be justified if the end-user experience justifies their
investment. Subjective satisfaction and purchase decision evaluations can help to assess the cost-
benefit trade-off here.
Each of these issues can be evaluated in different ways. I'm hoping that the best answers will map
from these different characteristics down to the specific techniques that we have introduced in the
course. For example, cooperative evaluation might help with point 4 but may not yield great insights
for 2 given that the evaluator is typically in the same location as the user. Other forms of automated
web-based evaluation might be mentioned by first class answers.