Home » Articles posted by drannalcox (Page 8)
Author Archives: drannalcox
I thought nothing ever disappeared from the internet!
It seems as though every day we are being warned that once something is on the internet it never really disappears. Although I can find articles written about me 9 years ago http://www.timeshighereducation.co.uk/189972.article unfortunately I can’t find my new webpages (and nor can anyone else!) because the powers that be have changed a single character in the URL! They’ve decided that underscores are out of fashion and hyphens are the new black! So my webpage is now at http://www.ucl.ac.uk/uclic/people/a-cox. The cost of this tiny change? My business cards need re-printing, URLs on posters being displayed at CHI2013 won’t work, and my colleagues and are having to go through all our webpages updating the links.
Expensive slip
My OH bought me a coffee this morning using his chip-and-pin card. The waiter handed him the machine, he entered his PIN, pressed enter, and handed the machine back to the waiter. What’s wrong with this picture? Well, the machine hadn’t been asking for the PIN, it was asking how much tip he would like to give. So now, not only did we have a 4 figure bill for a couple of coffees, he’d told this waiter the pin for his card!
The waiter said that my OH wasn’t even the first person to do this. Last week, someone had made the same error, but had also gone as far as entering their PIN the 2nd time to actually pay. The waiter had chased the customer into the street to explain to him that he’d paid over £2,000 for his lunch!
The true course of scientific research
Brian Romans’ blogpost that he wrote in response to that of Chris Chambers discusses the frequent reality of the course of scientific research. It’s perhaps even an understatement to say that things don’t always work out quite the way you expect them to. Brian argues that there’s a lot to be learned from a piece of research that ‘fails’ and that “stumbling about” can sometimes mean that you stumble into something significant. As Daniel Richardson’s rather honest account makes clear, this way of talking about the process of research is in stark contrast to the way we teach it:
“It generally begins at a conference bar or through an email with a friend when we’ll come up with a half-formed idea and we won’t have any idea what we’re doing. We’ll plunge through the literature, we’ll revise, we’ll completely change the focus, or do one experiment that won’t work but then something strange will happen and we’ll investigate that instead. It’s a collaborative, shoddy, directionless sort of exploration, completely different from the very clean, straightforward way that we teach research.”
It’s not just that we teach research as though it were a very linear process, but it’s also the way we usually report it in papers, theses, talks etc. I remember being a grad student and wondering how on earth I was ever going to create a thesis that hung together, out of a patchwork of studies with unpredicted results. More recently, it felt rather unnatural at first to describe a “failed” experiment in my talk last week at CHI2012. On the one hand it didn’t contribute anything to the theory being put forward in the paper, but, on the other, it was important to demonstrate how our original ideas (about how challenge leads to increased experience of immersion in videogames) were just wrong.
Having spent a few moments myself in a conference bar last week, I’m full of enthusiasm and half-formed ideas, which I hope to investigate over the coming months. Rather than being daunted though about whether experimental results will work out as I expect them to, I’m already looking forward to the unpredictable journey ahead and the challenge of weaving together an interesting story out of whatever happens….perhaps even in time for the CHI2013 deadline.
Safer but slower: incremental number entry interfaces
Drug dosing errors arising from incorrectly entered numbers account for a significant portion of the adverse drug events in hospitals. This study compared two different types of number entry system to see which resulted in fewer errors. While having their eye movements tracked study participants were asked to enter a number using either a numeric keypad (like those on calculators) or an incremental entry system which uses up and down arrows. With the numeric keypad, users must focus on the keypad itself (in order to find the next key to press) and may not notice errors on the display, whereas with the incremental interface, users tend to monitor the display resulting in higher error detection and consequently better entry accuracy. The incremental system produced significantly more accurate entries although entry took longer
Oladimeji, P., Thimbleby, H., COX, A. (2011). Number entry interfaces and their effects on error detection. INTERACT 2011: 13th IFIP TC13 Conference on Human-Computer Interaction. Author URL
Do people become over-reliant on cues?
Previous research by Mike Byrne has demonstrated that aggresive cues are effective at reducing post-completion errors. For such cues to work, they have to appear just-in-time, i.e. at precisely the time the user is supposed to perform the action. Our question was “do people might become over-reliant on such cues?” That is, if they missed seeing the cue, or it wasn’t there for some reason, would their performance being adversely affected? If you’re interested in our results either check out our poster at CogSci 2011 this week or read the paper:
Ament, M., Lai, A. Y. T., COX, A. (2011). The Effect of Repeated Cue Exposure on Post-Completion Errors. CogSci 2011: 33rd annual meeting of the Cognitive Science Society.
Errors in number entry – a taxonomy
Incorrect drug doses are a common medical error. Using a set of number entry errors gathered last year, we have created a taxonomy which lists the varying types of errors collected and groups them by potential cause and at what point they occur during the number entry process. Entry errors can occur when a number is misread from a prescription, when digits get jumbled up in memory before typing, or when actually pressing the buttons on a keypad. From this initial taxonomy we can investigate whether some errors are more likely at certain stages of number entry and begin to understand their causes.
Wiseman, S., Cairns, P., COX, A. (2011). A Taxonomy of Number Entry Errors. HCI2011: The 25th BCS Conference on Human-Computer Interaction. Author URL
Recovering from interruptions
Hospital wards are busy places, and medical professionals conducting routine tasks, such as setting up an infusion pump, are likely to find themselves interrupted at some point in the sequence. We have been investigating ways to reduce cognitive slips caused by interruptions, such as exploring the value of encouraging users to stop and think before resuming a task following an interruption. Findings suggest that the process of retracing previously achieved sub-goals can help with resuming a task. Ways of encouraging people to take time before resuming their task rather than jumping straight back in have been tested. In addition to demonstrating that people make speed/accuracy tradeoffs when resuming after an interruption, we are also investigating whether we can predict from eye-movement data if someone will make an error. Ultimately, as this work matures it will allow us to make predictions about device designs that better support error avoidance.
Back, J., Brumby, D. P., Cox, A. L. (2010). Locked-out: Investigating the effectiveness of system lockouts to reduce errors in routine tasks. Proceedings of the 28th international conference extended abstracts on human factors in computing systems, CHI EA ’10. ( pp.3775-3780). New York, NY: ACM Press. Author URL Publisher URL
