ETH Zurich :
Computer Science :
Pervasive Computing :
Distributed Systems :
HCI HS2007 : Fitts' Law
Exercise: Fitts' Law
due 23 Oct. 2007
The purpose of this exercise is to familiarize yourself with Fitts' Law, one of the
most commonly used predictive models in HCI research.
Try to read at least MacKenzie's 1992 paper on Fitts' Law. Optionally, read Fitts' original 1954 article. If you want to know what this is good for (you know you want to!), see Bruce Toganazzi's 1999 AskTok-column on Fitts' Law.
Try to run the experiments by at least two people (i.e., yourselves), preferably also by your friends and family members. The more the better!
- Begin with Fitts' original tapping task, using the tapping task form. Compute ID from the values on the form, using MacKenzie's formula.
- Download the IDTest software (repackaged from Barton A. Smith's original IDTest23.zip). Run either the
.BAT or the
.SH-file, using the defaults (which are being read from
IDTest.ini). Use different people and different pointing devices. Make sure to appropriately name each test run, in oder to correctly analyze the data later.
Use a standard spreadsheet software to analyze your data. Import the datafile(s) into the software (using tab-separated text-file import) and use the Summary data at the end of each run to chart time vs DI (make sure to use an x-y scatterplot). All spreadsheet tools should allow you to combine multiple series into a single chart, allowing you to compare different input devices.
You can also add a trendline to your data points:
- In Excel 2003: Select the chart/series, then click Add Trendline in the Chart menu. Choose Linear in the Type tab.
- In Excel 2007: Select the chart/series, then click on the Chart Tools/Layout Ribbon. Click on Trendline in the Analysis section and choose Linear Trendline.
- In OpenOffice: Select the chart and choose Insert > Statistics > pick out trend.
- In Gnumeric: Select the series to which you want to add a trend line (in the Customize Chart dialog), next click on
the Add button and choose the trend line type you need.