For my last year of university, I chose to take a course on human and computer interaction (HCI) in order to learn more about usability and how to conduct proper experiments in HCI. In this course, I performed a couple of interesting experiments, one of which I will be summarizing in this post.
The experiment that I will be discussing is regarding text input on smartwatch devices. As smartwatches are becoming more prevalent, the need to come up with an input design that is able to perform efficiently and accurately becomes increasingly important. Smartwatch users want to be able to accurately input into their device without too much hassle. Some of the current input methods on the Apple Watch include a curated list of responses you can select from, or an area where you can draw characters onto the screen using your finger to send messages. These methods of input can be restrictive or even difficult as users are not given as much control as they have in a keyboard, and manually writing out each letter can be cumbersome.
The goal of my usability study was the test two potential methods of inputting text on the smartwatch. The first method consists of a QWERTY keyboard, the same as the keyboard found on the iPhone or your laptop, shrunken to the size of a smartwatch. I will be referring to this method as the “normal keyboard”. The second method was a zoom keyboard, where the user taps once to enlarge a certain area of the keyboard, and then again to select a character. My study focused on how accurate and fast the inputs were depending on what type of activity the user was doing (i.e. walking around or sitting down). Activity is an extremely important factor to consider, as many people use their smartwatches on the go.
Four participants were chosen to complete the experiment. The experiment consisted of 4 different trials – Using the normal keyboard while sitting down (sitting + normal), using the normal keyboard while walking (walking + normal), using the zoom keyboard while sitting down (sitting + zoom), and using the zoom keyboard while walking (walking + zoom). Each trial consisted of typing five phrases. The trials were put in different orders for each participant in order to assure that the results were not affected by the learning curve (i.e. they perform better on later experiments due to having been ‘trained’ by the previous experiments). The experiment was conducted on a mobile device that was strapped to the user’s wrist to simulate a smartwatch.
My predictions for the experiment were as follows, and they were all supported by the results of my experiment.
- The zoom keyboard will have a higher average accuracy rate compared to the normal keyboard, as the zoom keyboard provides a bigger interface and more flexibility for errors.
- The zoom keyboard will have a larger time per character compared to the normal keyboard on average, due to the fact that the user must click on the zoom keyboard twice for one input.
- There will be greater change in accuracy between zoom and normal when walking than when sitting, due to the zoom keyboard providing more precision.
After performing the experiments, accuracy and speed (measured in seconds it took per character) were compared between each of the different trials. The zoom keyboard had an average accuracy of 77.5%, and normal keyboard had an average accuracy of 44.6%, indicating that the users were able to type more accurately using the zoom keyboard. Many participants expressed frustration with the normal keyboard setup, especially during the walking + normal keyboard trial due to the fact that it was extremely easy to make typos. This indicates that on average, the zoom keyboard is able to perform better.
The difference in accuracy between the trials is most noticeable between the walking trials, with the average walking + normal accuracy rate being 27.6%, and the average walking + zoom accuracy rate being 67.5%. There was also a greater accuracy rate in the sitting trials as well, although not as apparent, with sitting + normal having an accuracy rate of 67.9% and sitting + zoom with 87.5%. Both accuracies for either keyboard dropped when comparing external activity (sitting vs. walking), which highlights the importance of having an input method that is resilient to any external activity.
The change in accuracy between sitting and walking for the normal keyboard was 40.3%, whereas the change for the zoom keyboard was only 20%. This indicates that the external activity has an effect on how accurately the user is able to input the given phrases, with the normal keyboard being more affected than the zoom keyboard. This could be due to the shakiness when walking affecting the normal keyboard more, as the zoom keyboard provides more precision.
On average, the zoom keyboard (with an average of 1.06 seconds per character) took more than double the time of the normal keyboard (with an average of 0.486 seconds per character. This increase in time can be attributed to the fact that users must click twice on the zoom keyboard to input the desired character, compared to the normal keyboard, where they only need to click once.
Overall, the conclusions that can be drawn from this experiment is that keyboard type is significant in terms of activity, accuracy rate, and time it takes to input characters. The experiment also indicates that there may be a trade-off between time and accuracy, since the users were able to type faster on the normal keyboard, but more accurately on the zoom keyboard. Users were much more satisfied with their experience with the zoom keyboard due to the ability to accurately select their desired characters, regardless of what the activity was being done.
Personally, as the owner of an Apple Watch, I would really like to see some iteration of a zoom keyboard implemented, as the current methods of input are very limited. This was my first time conducting a formal usability test. I am happy to answer any questions if you have any!
Top comments (0)