Usability testing of proximity detection

Abstract

A study with 18 test persons was done to get a better understanding of the usability of proximity detection and the evaluation of the minimum size of control elements. The test persons did two different tests on a smartphone with integrated proximity detection and the associated test application. Several tasks with changing states of the proximity detection were made. Differences of several parameters were determined on the basis of the collected data. It could be shown that the task completion had been simplified.

Motivation

Smartphones and other mobile devices have small screens and are operated with fingers. This may lead to a fat finger problem (obscuration of the control elements) or the fingers define an imprecise area of contact. A current solution is the definition of minimum sizes for control elements by the manufacturers. Can the error rate of the interactions be minimized even with small buttons by means of proximity detection? Would the speed of the interaction increase by means of proximity detection? Can proximity recognition help to find certain entries in table lists more quickly?

Methodology

18 test persons of both gender (balanced) between 20-35 years (med: 23, Ø: 25.6) were tested. The environment of the test person is predominantly characterized by student or university. The test persons usually have good experiences with laptops, smartphones and similar devices. The test environment consists of a test application (2 tests with 2 or 3 test variants) and a test device. The questionnaires consist of a personal questionnaire, system usability scale and an additional questionnaire (freely formulated answers). In the first test (list selection), the aim was to check the effect of the highlighting of the list entries while increasing speed. The second test (precision test) was to check the effect of the strokes of controls while increasing speed and accuracy. For this purpose, a test room was created during the entire study to create unchanged conditions. The testers were welcomed, they got all relevant study information and questionnaires. A brief interactive demonstration of proximity recognition was offered. Finally, the test series and the software evaluation were started. The test series ended with the additional questionnaire.

	Test 1: list selection	Test 2: precision test
Dependent variable	proximity detection	proximity detection, button size
Conditions of proximity detection	switched off, switched on	switched off, switched on, switched on + selection support
Independent variable	Duration (ms)	Total duration (ms), Duration (ms), Errors, Aborts

Results

The first test (list selection) was not significant and an improvement due to the proximity recognition could not be confirmed. In the second test (precision test) the differences for button sizes from 10 mm to 1.4 mm were not significant. Significant differences had been measured in sizes between 2.4 mm and 1.4 mm.

Discussion

Can the error rate of the interactions be minimized even with small buttons by means of the proximity detection? Significantly more test persons could correctly select the buttons between 2.4 mm and 1.4 mm with activated proximity detection. 14 of 18 test persons indicated that the recognition of the approach facilitated the task. For large buttons, a user-specific implementation is useful. Would the speed of the interaction increase by means of proximity detection? The speed increased with an activated proximity detection in both tests. In the second test, the slope of the regression line with an activated proximity detection was significantly greater than without proximity detection. Can proximity recognition help find certain entries in table lists more quickly? The measured values and the system usability score did not indicate any improvements after the evaluation. Poor performance may have been strongly affected by the test setup. The highlighting of list items were irritating, as an interaction with the items had not been possible.

Conclusion

The proximity detection has potential. An improved precision could only be achieved with very small button sizes. With an activated proximity detection, the speed worsened within the scope of the study. The majority of users have adopted and preferred the proximity detection. Further research would be desirable: An evaluation of the remaining data sets could be done. The selection algorithm or the hardware can be optimized. There are many research perspectives in this field.