Older Adults Disproportionately Hindered by Touch Screen Interfaces in Driving Tasks

Introduction

Since their initial development, touch screens have become ubiquitous among smartphones, laptops, and tablets. Their dynamically reconfigurable interface provides numerous advantages for users and developers over static interface layout of keyboards, computer mice, and classic button and display screen cell phones. Touch screens also enable a user to tap directly on what they want to interact with on a device, rather than navigating to that element with arrows or mouse input. However, a drawback of these newer designs is increased difficulty with reading user input into these capacitive-based touch pads. Older adults especially struggle with touch screen based input, although it is unclear to what extent this is due to confusion about the input method versus physiological changes (e.g., Favilla & Pedell, 2013). This predicament can lead to older adults feeling less capable and being less willing to engage with and adopt touch screen technology, creating a situation of disuse (Vaportzis et al., 2017). Experimental results are mixed about what effect age has on the electrodermal response and skin conductance levels, making this an area that could benefit from expanded study (Catania et al., 1980; Eisdorfer et al., 1980; Venables & Mitchell, 1996). In this article we will look at the rates of hits versus misses in a tablet task performed by older adults and younger adults completing a simulated driving trial. A touch screen interface was chosen for the secondary task due to the increasing popularity of touch screen interfaces in personal vehicles.

Methods

Procedure

After reviewing and signing informed consent, participants were given a short practice driving scenario to become acquainted with the controls, adjust the driver’s seat to their preferred distance from the steering wheel, and to get a feel for the driving environment. Following this introductory simulation, participants filled out a simulator sickness questionnaire, provided their basic demographic information, and older adults took the MoCA. Visual acuity was measured using a Snellen chart (Hetherington, 1954). Participants were instructed to maintain a speed of 45mph (20.1 m/s) and lane keeping while performing an IVT, hitting the ‘Next’ button on a tablet attached to the center console and reading a set of numbers out loud, whenever they saw a stimulus appear on their windshield.

Participants practiced the secondary task a minimum of five times or until they were comfortable with it and consistently able to get the numbers to change by pressing the ‘Next’ button (see Figure 1), then drove a two-minute-long scenario where they were asked to maintain 45mph while performing the secondary task every 25 seconds plus or minus two seconds to get a baseline measure for secondary task performance.

OPEN IN VIEWER

Figure 1.

Experimental task layout. Participants pressed the ‘Next’ button to change the numbers. Experimenters changed the scenario buttons at the bottom to match which scenario was being run. P - practice, B - equal (both) task priority, D - driving task priority, T – IVT (tablet) task priority.

During these practice sessions, experimenters often had to coach older adults on how to get the touch screen to respond to their touches (e.g., duration of the touch, force of the touch). Even with this coaching, it quickly became evident that the touch screen simply did not respond well to older adult touches and a few of our older participants quickly became frustrated with the IVT portion of the experiment.

During experimental trials, the task prompts appeared no sooner than 19 seconds after the start of the scenario. Task bandwidth was varied with high frequency set as every 8 seconds plus or minus one or two seconds and low frequency was every 11 seconds plus or minus one or two seconds for the task prompt. Participants drove twelve four-minute driving trials, six with low prompt frequency, six with high prompt frequency. We asked participants to change task priority (driving, IVT, equal) every four trials. There was also simulated wind to make the driving task more difficult, which also varied in frequency.

Results

The ratio of hits (i.e. ‘Next’ button pressed correctly on the tablet) to misses (i.e. attempt to hit ‘Next’ was made but the numbers didn’t change) on the tablet was found to be significant for age t(217) = -7.129 such that older adults had significantly more misses relative to the number of hits than younger adults (see Figure 2). The percentage of hits relative to the number of expected hits was also significant for age t(217) = -7.129, showing that older adults had lower percentages of hits compared to expected than younger adults. Note that hits beyond the expected value were excluded from this data so that 100% was the highest score any participant could receive. One older adult trial and one younger adult trial were excluded from analysis due to the inability to pull hit/miss data from those videos. No other significant main effects were found for hit/miss data.

OPEN IN VIEWER

Figure 2.

Ratio of hits to misses by age group.

An interaction between age and wind frequency was found for the ratio of hits to misses such that younger adults had slightly higher hit:miss ratios for high wind frequency (M =_0.88) than low (M = 0.86) while older adults had lower hit:miss ratios for high wind frequency (M = 0.45) than low (M = 0.48).

Another interaction for hit-to-miss ratios was found between age and priority instructions such that hit:miss ratios increased for younger adults as IVT priority increased (driving M = 0.84, equal M = 0.87, IVT M = 0.89) but for older adults driving priority (M = 0.50) had the highest hit:miss ratio and equal priority (M = 0.44) had the lowest.

IVT priority ratio (M = 0.47) was between driving and equal priority. No other significant interactions were found for the ratio of hits to misses, and no significant interactions were found for percentage of hits relative to expected hits.

Discussion

We observed older adults disproportionately struggle using a tablet interface compared to younger adults during our driving simulator study. While not surprising taken alone, what was surprising was that even with a very simple task (hit one button once) with detailed coaching from the experimenters on the duration and force of the touch required to get a response from the screen older adults still had trouble. Some of the misses did come from the occasional miss of the ‘Next’ button target, but even when an older adult appeared to hit the button in the middle of the target with appropriate force and duration the tablet would sometimes fail to recognize the touch. This suggests that there may be a physiological reason contributing to older adults having trouble with touch screen interfaces.

Touch screens work because they are covered in a thin layer of an electrically conducting material (IOP, 2022). When a person’s finger contacts a touch screen surface, the electrical conductivity of that finger creates a measurable change in the electric charge occurring on a grid of transparent capacitors which are used to indicate the input touch location. When someone’s fingers are dry, this may cause problems because a layer of moisture on the tips of fingers helps facilitate this change in electric charge. Dry or leathery skin affects 30% to 60% of all older adults (Hurlow & Bliss, 2011), giving them a physiologically disadvantaged starting point when attempting to use touch screens. This is an important aspect to keep in mind when designing electronics that may be used by older adults and potentially adds to the list of impediments contributing to their successful adoption of new technologies.

This aspect combined with increased dual task difficulties experienced by older adults, like using an interface on the center console while driving, may further exacerbate this issue. Younger adults had the lowest IVT performance scores when the tablet task was the lowest priority, but for older adults it was the equal priority condition where driving and the IVT were both high priorities that had the lowest IVT performance scores. This suggests that older adults were more affected by the dual-task cost than younger adults, which is consistent with previous literature (Neider et al., 2011; Verhaeghen et al., 2003).

Older adults experienced more trouble with the tablet than the younger adults and this might have made the secondary task itself more difficult for older adults. The more attentionally demanding and difficult task may have confounded the dualtask cost for older adults in similar ways that talking on a cell phone did compared to less demanding invehicle tasks in Strayer and Johnston’s study (2001). The added difficulty of trying to equally prioritize the driving and IVT tasks relative to the single priority conditions may have overwhelmed older adults and they had even more trouble trying to perform the IVT task than they were when they prioritized driving. This flustering effect may have important implications for older adult driver safety while using in-vehicle technologies. If in-vehicle technologies require too many attentional resources to use, and this is especially true for older adults who are already more vulnerable road users, we may see an increase in accidents caused by distraction or a pattern of disuse of these systems in older adults. To make sure older drivers aren’t disproportionately negatively impacted by invehicle technologies intended to help them, to the authors agree with the National Highway Traffic Safety Administration (NHTSA) guideline to include them in the usability research done on new and existing vehicle HMIs (nhtsa.gov). Using younger adults to test new IVT HMI designs is generally more convenient for developers but may lead to a design that is not well suited for older adults. Using older adults during user testing can help make products that are good for both older and younger adults since designs that work well for older adults will likely also work well for younger adults. This highlights the importance of age diversity inclusion in the design process. This is consistent with the findings of Cooper and colleagues (2020) who found that when older adults use in-vehicle systems they were slower to react to cues to perform a task, took longer to complete the tasks, and reported the tasks as more demanding. Both our and Cooper and colleagues’ findings suggest that, while older adults may benefit from invehicle technologies, they could struggle to use them which may cause frustration and disuse.