Paper Reading #10 - Designing Adaptive Feedback

Comments:
Comment 1
Comment 2

References:
Title: Designing Adaptive Feedback for Improving Data Entry Accuracy
Authors: Kuang Chen, Joseph M. Hellerstein, and Tapan S. Parikh
Venue: UIST 2010, Oct 3-6, 2010
Summary:
In this paper, the authors describe a new method for data entry that uses prediction to minimize error rates. The method is called USHER and works by building a probabilistic model that figures out the likelihood of potential entries. It then offers defaults, highlights likely options, and gives warnings. By using this system, the authors hope to decrease entry error rates while also avoiding using the technique of double entry.

They then test the effectiveness of their system by building a java frontend for USHER and having African data entry clerks enter a series of medical entries into a database. These entries were premade for the experiment and were checked for validity afterwards. They discovered that while defaults made the process quicker, they did not decrease the error rates. However, the highlighting system made a significant difference.

Discussion:
There are a lot of reasons why I like this paper. First off, they did a good job of using hard scientific data to back up their claims. Many of the papers so far did not do this. Second, they used some concepts from cognitive psychology that are similar to the ones we learned from The Design of Everyday Things. Finally, they showed practical applications for the technology that could be used later.

As far as the data entry is concerned, I definitely support this method. I have done data entry in the past and it's very boring; I often let my mind wander when I was doing the entries. In the medical world, where data entry is common, this is a very bad thing. Hopefully by adding these entries in, less errors will be made.

(Image courtesy of: anecdote.com)