University of Moratuwa
Question
Asked 11th Jun, 2016
How test a new text input method for usability?
I developed a new text input mechanism for Sinhala (A popular language used in Sri Lanka). The prototype was built for an android touch screen device. I would like to know the options I have for usability testing. Any tests I can carry out based on the cognitive system is preferred.
Most recent answer
@Carl, why do you think people who take online surveys are upset?
Popular answers (1)
RWTH Aachen University
Hello Tulakshana,
if you need a quick and dirty”, reliable tool for measuring the usability and the learnability of your new mechanism on the android you can use the "System Usability Scale" (Brooke). From my experience, it is a short and effective measurement scale; you may only have to rephrase a bit.
If you want to see a short description, here is a link:
Kind regards,
Wiktoria
3 Recommendations
All Answers (9)
Oil Design&Construction Company (ODCC)
Hi dear
What is the meaning of new mechanism? Do you mean, you develop new layout for keyboard? Or implement new character map for Unicode of Sinhala?
If you mean a new layout, a good layout should be designed based on frequency of character using. e.g. in English language, characters such as a e i o u are very frequent, so the place of these characters in keyboard designed to be easy for use. You could process some general books context in Sinhala language and count frequency of characters, then based on the frequency of characters, design your keyboard layout.
With Regards
1 Recommendation
RWTH Aachen University
Hello Tulakshana,
if you need a quick and dirty”, reliable tool for measuring the usability and the learnability of your new mechanism on the android you can use the "System Usability Scale" (Brooke). From my experience, it is a short and effective measurement scale; you may only have to rephrase a bit.
If you want to see a short description, here is a link:
Kind regards,
Wiktoria
3 Recommendations
Yonsei University
Dear Tylakshana,
I understood your question that you want to know about any types of possible empirical test. If so, you can test your prototype in two perspectives. Firstly, you can measure the task performance of the users. For example, for the given text entry, you can measure the time for typing given text with your prototype, and then you can compare the result with the conventional keyboards. During such experiment, you can also measure the typographical error rate for each participant. Secondly, you also can measure some emotional and ergonomic factors of the users. For example, ease of use and physical fatigue may be measured with a questionnaire after the experiment.
3 Recommendations
KDU University College
You can use LUND usability testing,
Lund, A.M. (2001) Measuring Usability with the USE Questionnaire. STC Usability SIG Newsletter, 8:2
1 Recommendation
It is good practice to measure BOTH user performance AND satisfaction. If you measure only satisfaction, you are likely to overlook performance issues. As Wickens and Andre (1995) say, users often like design features that are not supportive of performance, such as overuse of color.
1 Recommendation
Agricultural Research Institute Cyprus
Hi
You can prepare a quick test and ask participants to input a paragraph using an existing (standard) text input mechanism and the same paragraph using your proposed input mechanism and after each task, ask this question Overall, this task was? see http://www.measuringu.com/blog/single-question.php for more details
1 Recommendation
Texas A&M - Central Texas
Good question.
As Elizabeth said, the best practice is to measure both satisfaction and performance. It also matters how these measures are evaluated and the demographics of the subjects. The issue with subjective questionnaires and interviews is that people lie. In some cases it is considered being polite, since you honored them by including them in the survey they will say only positive. On-line surveys tend to produce negative results because the most people taking the survey are upset. Another issue with the subjective approach is the population necessary for the study. Typically a subjective approach requires 40 or more subjects before the data begins to approach the actual mean.
Another approach is performing a comparative test and use statistical results for performance comparision. It is a simple test to construct, but the primary issue is the math. Measuring human performance is an exercise in measuring learning. Human learning is most accurately modeled by either an exponential decay curve or a power law decay curve. Meaning that it is necessary to use parametric and non-parametric statistics. The good news is that it only requires 7 to 10 cases with 14 to 20 subjects (7 to 10 for each data entry method). The details of this approach are documented in An Economical Approach to Usability Testing”, Proceedings of 33rd Annual IEEE International Computer Software and Applications Conference (Compsac 2009), Seattle, WA.
Good Luck.
University of Moratuwa
Link to what @Ahamad was referring to https://www.researchgate.net/publication/230786746_Measuring_Usability_with_the_USE_Questionnaire
Similar questions and discussions
Recommendations
Through research on software testing model selection, seeking the most appropriate testing method to achieve most reasonable testing volume and optimal testing result. Described in detail about the core content of Yin-Yang software testing theory; According to research data of Software Productivity Research, U.S.A, with overall consideration about...
Software systems are often complex in the number of fea- tures that are available through the user interface and con- sequently, the number of interactions that can occur. Such systems are prone to errors when interactions do not work as anticipated. This research introduces a combinatorial method for setting up task-based usability tests. The meth...
The assessment of software systems often requires to consider together qualitative and quantitative aspects. Because of the different nature, measures belong to different domains. The main problem is to aggregate such information into a derived measure able to provide an overall estimation. This problem has been traditionally solved trough the tran...