Due date: 15 December, 1997, 3pm GMT.
Worth 25% of the marks for this module. Your name should be on every page, and each page should contain information about only one question.
I was again, overall, pleased with the results. Nearly everyone has 2 to 3 tools to think about how interfaces work now, and most people have them substantially mastered. Late papers had 5% (raw) taken off for each day late.
The University of Nottingham is considering changing its computer addresses from the abbreviated form (e.g. psychology.nottingham and cs.nott) to a full form for clarity (e.g. psychology.nottingham). What are the potential costs and benefits of this approach. (Leave whether it is a good idea to the person balancing the costs and benefits!)
What I was looking for:
A statement of further, necessary assumptions about, for example, tasks and typing speed.
A computation of how long each takes based on the KLM. Not including mental operators was bad as was including them in the wrong places. Including them and noting that there are some problems in knowing where to do this was ok. This was enough to pass, but for a better mark, you needed to show that you understood what this meant. If you were really keen, you could graph the time differences vs. typing speed (CM&N do this in their book in a few places).
A discussion noting many, but not necessarily all of: the possible increased error rate with the longer name due to keystrokes, the possible increased error rate on the alias because of now knowing what to type; how KLM does not predict errors or novices very well; the role of memorability and regularity; other limitations and incomplete aspects of the KLM; the culture and use of email addresses and changing them (e.g. bookmarks), and where they are stored.
A concluding summary noting the relative role of the features mentioned above in changing the results. The answer should have cited Card, Moran & Newell, at least.
Where people appeared to misunderstand the question, and answered usually far too short, I tried to give the benefit of the doubt, but here, just doing the KLM analysis, was about half the work because there are known problems and omissions on the KLM that must be noted with respect to this change.
No penalties for spelling or very awkward writing per se, but points were lost for all questions where these were bad enough to truly interfere with expression. Other aspects, too numerous to mention here, also affected the marks of individual papers.
Evaluate the effect of the software described below. You can (and will have to) make assumptions, but be sure to explain and justify them. Comment on learning time, time to use, and amount of knowledge necessary. (If this is not a complete enough description for your analysis, make reasonable assumptions, and note them clearly.)
Things I looked for (starting with 75%):
* description of what the arrow looked like and any other assumptions (missing, -2) missing references or support for arguments (-2)
* enumeration of the possible tasks, and how their frequency could influence the answer (nec. for A) (+ 5 if you included this)
* the role of Fitts' law in judging if the two-headed arrows were faster to get to, and perhaps the increase in errors with a smaller target (if each arrow was smaller). Better answers used Fitts' law to compute the time difference rather than just mentioning it. (nec for B) (- 5 if not mentioned)
* a full specification of the goals, operators, methods and selection rules for the two versions (nec. for C) The altCdef should have more selection rules and more methods (or a different method to go to the nearest arrow set). (-10 if not included) (-5 for leaving out or if put into methods not mentioning selection rules)
* a final, summary answer that noted the role of the factors above. Generally, if your GOMS analysis was efficient, the number of methods did not increase much so learning time can be compared with GOMS, but is negligible in this case, and the new method was a good deal if you can assume that the mouse is nearby the new smaller arrow. (- 5 to 10 bad implications or incomplete) Whether it was a good idea or not depended on your assumptions. No one computed total time savings, but most people's analyses suggested that it would save some time. I don't think I'll bother to load it (I have a lot of extensions already), but it looks like it should be included in future scrolling windows.
Certain designs (e.g, both arrow moved together, or assuming pointing time remained the same), lead to trivial analyses of bad designs. These were not heavily penalised if they were explained. Other aspects, too numerous to mention here, also affected the marks of individual papers.
Prepare a description of the activities involved in sending a single email message. Try to include as much information as necessary, and choose and justify an appropriate representation. Explain your assumptions. (about 1 page)
Imagine basing a manual on the description. What else would a novice (who does not about email and the internet) need to know? (about 0.5 page)
What I was looking for:
You should have started with further assumption, like type of machine and mailer (else -5). The choice of task analysis must be justified (else up to -20), better answers also noted why other methods were not chosen. Plain text was a relatively bad method (- 5), HTA is good, GOMS is ok, KLM is a bit weak (-2 or 3). The TA should be fairly complete in any case for a single message (up to -10 for a bad or incomplete analysis). Other books (such as Dix, Finlay, Aboud & Beale) were good references (no refs lead to -2).
The manual should be based on the task analysis (not answered, -10), with the areas that the TA does not cover a ripe area for inclusion in the manual. Most people didn't quite notice this or did but didn't mention it explicitly. Those that mentioned explicitly how to use the TA as the basis for the manual got up to +5. Other aspects, too numerous to mention here, also affected the marks of individual papers.