Going short: the effects of short travel key switches on typing performance, typing force, forearm muscle activity, and user experience

Abstract

This study examined the effects of four micro-travel keyboards on forearm muscle activity, typing force, typing performance, and self-reported discomfort and difficulty. Twenty participants completed typing tasks on four commercially available devices with different key switch characteristics (dome, scissors, and butterfly) and key travels (0.55, 1.3, and 1.6 mm). The device with short travel (0.55 mm) and a dome type key switch mechanism was associated with higher muscle activities (6–8%,p < 0.01), higher typing force (12%, p < 0.001), slower typing speeds (8%, p < 0.01), and twice as much discomfort (p < 0.05), compared to the three other devices: short travel (0.55 mm) and butterfly switch design and long travel (1.3 and 1.6 mm) with scissor key switches. Participants rated the devices with larger travels (1.3 and 1.6 mm) with least discomfort (p = 0.015) and difficulty (p < 0.001). When stratified by sex/gender, these observed associations were larger and more significant in the female participants compared to male participants. Because the devices with similar travel but different key switch designs had difference in outcomes and devices with different travel were not different, and the results suggest that key travel alone does not predict typing force or muscle activity.

Introduction

Today’s lightweight notebook and tablet computers employ thinner keyboards than earlier models. These designs must sacrifice key travel distance as evidenced by new devices with key travel distances lying outside the current 1.5–6.0 mm standards¹. For example, the Microsoft Surface Pro 4’s keyboard has 1.3 mm travel. Standards such as ISO-9241 (Ergonomics of Human-Computer Interaction) and ANSI/HFES 100 (Human Factors Engineering of Computer Workstations) provide guidelines for the design and development of desktop and laptop computers, but may not be relevant for modern devices. Studies comparing external tablet keyboard attachments to the no-travel, on-screen keyboards have demonstrated better performance with attached keyboard use^2,3. However, the effects of these new short travel key designs on upper extremity muscle activity and typing force are unknown.

Keyboard design characteristics including key travel distance⁴, key size⁵, activation force^6–10, feedback¹¹ key switch mechanism¹², and key location¹³ can affect muscle activity, typing force, and typing performance. However, most of these previous studies were conducted using regular desktop keyboards with travel distances within the recommended range of 1.5–6.0 mm and with dome or linear spring switches.

Key switch force-displacement is an important consideration for musculoskeletal symptoms. Ripat et al.¹⁴ randomly assigned symptomatic typists to use the same commercially available split keyboard¹⁵ in either its normal (0.36 N activation force, 2.8 mm travel) or modified (0.36 N activation force, 0.2 mm travel) design over 24 weeks and found similar improvements from baseline in clinical symptoms and satisfaction in both groups. In addition, symptomatic typists have been shown to type with higher typing forces than those without pain, though typing forces were measured after symptom onset which limits the ability to assign causality^16–18. Higher activation force keyboards are generally associated with higher typing force^8,19 and higher muscle activity^6,7.

The key switch mechanism has also been shown to affect upper extremity biomechanics, user experience, and forearm musculoskeletal symptoms^8,11,20,21. Gerard et al.⁷ found that a buckle spring keyboard with 0.72 N activation force resulted in less typing force and muscle activity than 0.56 N and 0.83 N activation force elastomer dome keyboards, and similar typing force and muscle activity when compared to a 0.28 N activation force elastomer dome keyboard. Participants also preferred the buckle spring keyboard, though this was confounded because it was the keyboard they typically used. In contrast, Bufton et al.¹⁹ found higher typing forces were associated with a 0.86 N buckle spring keyboard compared to a 0.68 N elastomer dome keyboard. In order to achieve thinner designs, many mobile keyboards have abandoned spring and dome keyboards for scissor and butterfly switches^22,23.

A few studies with micro-travel keyboards (here defined as shorter than the recommended minimum 2.0 mm key travel)^24,25 have shown an association between shorter key travel and reduced muscle activity, typing force, performance, and self-reported typing user experience^11,26–28. Chaparro et al.¹¹ found that participants typed faster and preferred micro-travel mechanical keyboards over a pressure-sensing no-travel keyboard, but they did not measure any typing forces or muscle activity. Similarly, Hoyle et al.²⁷ found better performance and user experience ratings were associated with moderate micro-travel (1.6 and 2.0 mm) keyboards over very short (0.4 mm) and no-travel keyboards. Kim et al.²⁶ measured upper extremity muscle activity, typing force, and performance for no-travel, micro-travel (1.8 mm), and standard-travel (4.0 mm) keyboards, and found higher muscle activity and typing forces were associated with larger key travel. Kia et al.²⁵ recently tested five commercially available micro-travel tablet, notebook, and desktop keyboards (0.5–2.0 mm), and found decreased flexor muscle activity, increased typing forces, decreased ulnar deviation, and increased typing performance was associated with increasing travel distance. However, these differences were relatively small and may be related to unmeasured keyboard factors as the devices did not share the same form factors.

Moreover, few of these studies have examined the effects on specific individual characteristics such as sex/gender, anthropometry, and strength. Past studies have demonstrated higher associations between typing biomechanics²⁹ and technology-related pain and injury³⁰ in women as compared to men. As such, the difference in the effects of mobile computing devices and micro-travel keyboards across sex/gender needs further investigation.

More investigation is necessary to understand micro-travel’s effects on human health and performance in order to provide design guidance. The purpose of the current study was to investigate the effects of four different micro-travel keyboards on forearm muscle activity, typing force, typing performance, and self-reported discomfort and difficulty. In addition, the study sought to investigate if the effects of micro-travel distance keyboards differed between male and female participants. The study tested the null hypotheses that no significant differences will be observed in outcome measures across devices with different key-travel and key-force displacement characteristics and that the same trend will be observed in both male and female participants.

Methods

The study was a repeated measures laboratory experiment comparing four different commercially-available keyboards with small key travel distance with equal number of male and female participants to examine the effect on forearm muscle activity, typing force, typing performance, and self-reported discomfort and difficulty across all participants and within each sex/gender.

Study Population

Twenty participants, balanced for sex/gender, were recruited from the local community. Inclusion criteria for participants were: greater than 21 years of age, perfect or corrected vision, ability to touch type 30 words/minute, English language proficiency, and no history of upper extremity musculoskeletal pain or injury. Female participants ranged in age from 21 to 33 years (average ± standard deviation: 27 ± 3 years), in stature from 153 to 180 cm (168 ± 3 cm), in grip strength from 177 to 314N (245 ±39 N). Male participants ranged in age from 21 to 32 years (average ± standard deviation): (27 ± 4) years, in stature 169 to 186 cm: (179 ± 5) cm, in grip strength 255 to 481 N (373 ± 69 N)). Of the 20 participants, 19 (10 female, 9 male) self-reported to be right-handed. Right hand grip strength was measured using a hand strength dynamometer (Stoelting, Inc., Wooddale, IL, USA) in supinated position with an extended elbow. Participants were coached to grip as hard as possible three times without breaks, and the highest measure was used. All protocols were approved by the local IRB and all participants provided written informed consent before participating.

Experimental Tasks

Participants completed two typing tasks for each of the four devices. The first typing task was a one minute typing test on an online typing test program (Typing Speed Test, TypingMaster, Inc.). For the second task, participants typed for three minutes to transcribe up to three news articles, depending on their typing pace, into a word processor. The chosen articles were on current events and written for an 11–14 year old audience. This was meant to simulate more average typing tasks such as writing an email without inducing fatigue, and participants were asked to type at their normal pace. During the typing tasks, participants sat upright on an armless, backed office task chair with their feet flat on the floor and their thighs parallel to the ground. The monitor was adjusted such that the top of the screen was aligned with the participant’s eyes, and the center was aligned with his/her sternum, and the horizontal distance was the same for all participants. All participants used the same monitor, and brightness and light settings were kept constant for all participants. The devices were placed on a force platform on a desk such that they were centered with the monitor and the gap between the G and H keys was aligned with the participant’s sternum. The desk’s height was adjusted using visual inspection so that the keyboards were approximately at elbow height with approximately ninety-degree elbow flexion and neutral wrists.

Independent Variable: Device

Each participant typed on four commercially-available micro-travel devices (Table 1), each with a different combination of key-travel, short (S) or long (L), and mechanism (dome, butterfly, or scissor). The devices were Tablet S (0.55 mm): Apple iPad Pro (Apple Inc, Cupertino, CA), Notebook S (0.55mm): Apple MacBook Pro (Apple Inc, Cupertino, CA), Tablet L (1.3 mm): Surface Pro 4 (Microsoft Corporation, Redmond, WA), and Notebook L (1.6 mm): Microsoft SurfaceBook (Microsoft Corporation, Redmond, WA). These were chosen for their different key travel distances and mechanisms as well as their current commercial availability. The order of the devices tested was randomized and counterbalanced within sex/gender across the 20 participants. Participants typed on each keyboard for five minutes before the trials in order to become comfortable and acclimated to the devices.

Table 1:

Experimental Devices and Properties

	Tablet S (short)	Notebook S (short)	Tablet L (long)	Notebook L (long)
Brand	Apple Inc.	Apple Inc.	Microsoft Corporation	Microsoft Corporation
Model	iPad® Pro	Macbook® Pro	Surface® Pro 4	Surface Book®
Year	2016	2016	2015	2016
Travel	0.55 mm	0.55 mm	1.30 mm	1.60 mm
Switch Mechanism	Dome	Butterfly	Scissor	Scissor
Force	73–83 g	60–70 g	60–70 g	60–70 g
Key Length × Width	1.5 × 1.5 cm	1.7 × 1.6 cm	1.6 × 1.5 cm	1.5 × 1.5 cm
Space between keys	0.4 cm	0.2 cm	0.3 cm	0.3 cm

Dependent Variables: Muscle Activity

Study outcomes included the normalized median amplitude of electromyography signals of six (6) right forearm muscles: the extensor carpi radialis (ECR), extensor carpi ulnaris (ECU), extensor digitorum (ED), flexor carpi radialis (FCR), flexor carpi ulnaris (FCU), and flexor digitorum superficialis (FDS). Only the right-side was tested because previous research has shown no significant differences in muscle activities between the right and left arm during typing³¹. After cleaning each participant’s right forearm with alcohol and abrading with an exfoliating sponge to reduce impedance, bipolar electromyography sensors were mounted on the muscle bellies as identified through palpation and asking the participant to do demonstrative movements while the researcher provided resistance^32,33. Delsys 8 Bagnoli system (Delsys, Natick, USA) measured, filtered, and amplified the EMG signals prior to their recording on a personal computer at 2000 samples per second using a USB A-to-D backplanes (NI cDAQ-9172; National Instruments, Austin, USA). Post-processing of the signals in Matlab (The Mathworks, Inc., Natick, MA, USA) included rectifying the signal and then second-order low pass filtering at 3 HZ (single pole)^33–35.

Prior to the typing tasks protocols and data collection, participants performed maximum voluntary contractions (MVC) for each muscle resisted by trained experimenters in order to normalize the amplitude of the electromyographic signals. For each contraction, participants were instructed to ramp up to their maximum muscle output, and each MVC was collected for 3 seconds, 3 times with at least 2 minutes of rest between trials. The highest 1-second average amplitude of the processed EMG signal during the three 3-second trials provided the MVC value for each muscle. The median value for each muscle’s EMG signal was calculated for the middle 30 seconds of the typing test and the middle two minutes for the long form typing task to avoid discrepancies caused at the beginning or end of the task.

Dependent Variable: Typing Force

The median typing force was measured by custom force platform under each keyboard^29,34. The platform had three miniature compression load cells (ELFF-B4–10L; Measurement Specialties, Hampton, VA, 9.55 mV*N⁻¹, 44.48 N max) mounted underneath it in a triangular pattern³⁴. A USB backplane (NI cDAQ-9172; National Instruments, Austin, USA) powered, amplified, and sampled the signal from each of the three load cells at 2,000 samples per second. Post processing of the data included band-pass filtering (1–20 Hz, 6th order Butterworth filter) to remove noise and wrist contact using the biomechZoo toolbox in Matlab (The Mathworks, Inc., Natick, MA,USA)³² and custom Matlab code. When unloaded the root mean square value of sensor noise was 0.0063 N. Normalized force was calculated as typing force divided by grip strength. The median value for each signal was calculated for the middle 30 seconds of the typing test and the middle two minutes for the long form typing task to capture sustained typing and avoid discrepancies caused at the beginning or end of the task.

Dependent Variable: Typing Performance

For the timed typing test task, typing performance was measured in words per minute (WPM). The computer software provided a measure of typing performance in words per minute. An adjusted typing score was also calculated by the software by subtracting the number of errors from the total words typed.

Dependent Variable: Perceived Experience

Participants self-reported discomfort and difficulty with each keyboard after each trial. After each new device, participants received 2 paper 10 cm visual analog scales which asked them to mark between 0 and 10 where 0 was “no discomfort” and 10 was “a lot of discomfort”, and 0 for “no difficulty” and 10 for “a lot of difficulty”.

Statistical Analysis

A repeated measures ANOVA for each dependent study outcome tested the null hypothesis followed by Tukey’s post-hoc comparison for each keyboard design. Individual ANOVA models estimated the 6 muscles’ median %MVC, median typing force, median normalized force, typing speed, adjusted typing speed, discomfort, and difficulty. The muscle activity and typing force models included participants as a random effect, and device and typing task as fixed effects, while the performance and experience models did not include typing task. In addition to the models for all participants, separate models were stratified across the two sex/gender groups to examine the effects of device within sex/gender^36–38. Significance criteria for the F statistic (alpha value) was set at 0.05. When a significant effect was found, a post-hoc analysis with Tukey HSD provided between group comparisons.

Results

Forearm extensor (ECR, ED, ECU) muscle activity did vary significantly across the devices with less median muscle activity (p < 0.01) on Notebook S, Tablet L, and Notebook L compared to Tablet S (Table 2). No significant differences were observed for the forearm flexor muscles (FDS, FCU, FCR). The timed typing test was associated with significantly higher muscle activity across all of the muscles tested than the long form transcription task (p < 0.001). No significant two-way interaction effects were observed between device and task.

Table 2:

Across participant marginal means (and standard errors) for forearm electromyography (%MVC) and typing force. RMANOVA Keyboard Device and Typing Task

Keyboard Device
	P Value1,2	Tablet S	Notebook S	Tablet L	Notebook L
Median EMG %MVC
ECR	0.001	4.5A (0.49)	4.2B (0.49)	4.2B (0.49)	4.4AB (0.5)
ED	<0.001	8.1A (0.77)	7.3B (0.77)	7.5B (0.77)	7.7B (0.8)
ECU	0.006	7.8A (0.84)	7.2B (0.84)	7.2B 0(.84)	7.3B (0.8)
FDS	0.105	3.0 (0.36)	2.7 (0.36)	3.0 (0.36)	2.8 (0.4)
FCU	0.607	2.6 (0.53)	2.6 (0.53)	2.7 (0.53)	2.6 (0.5)
FCR	0.116	2.3 (0.32)	2.1 (0.32)	2.3 (0.32)	2.3 (0.3)
Median Typing Force
Newtons	<0.001	0.47A (0.02)	0.42B (0.02)	0.42B (0.02)	0.42B (0.02)
Normalized (%)	<0.001	3.5A (0.3)	3.2B (0.3)	3.1B (0.3)	3.2B (0.3)

¹Repeated Measures ANOVA with participant as a random variable, Keyboard Device (4 levels), Typing Task (2 levels). Bold values indicate a significant effect (p<0.05).

²For significant main effects, Tukey’s Post-Hoc groupings are ranked such that A>B>C>D. Values with the same superscript letters indicate no significant difference

Typing force varied significantly across devices (Table 2) with Tablet S associated with significantly higher median typing force compared to the other devices (p < 0.001). The timed typing test was associated with significantly higher typing force than the long form transcription task (p < 0.001). No significant two-way interaction effects were observed between device and task.

Typing speed also varied significantly across devices with slower typing speeds for Tablet S compared to Notebook L and Notebook S (p = 0.004) (Table 3). After adjusting for typing errors, the significant difference remained (p = 0.004). Notebook L was also associated with a faster average adjusted typing speed (72 WPM) compared to Tablet S (64 WPM) tablet device (p = 0.004). Adjusted speeds on Tablet L and Notebook S were not significantly different from adjusted speeds on Tablet S and Notebook L.

Table 3 :

Performance: Across participant marginal means (and standard errors) during a 1 minute typing test.

Keyboard Devices
	P Value1,2	Tablet S	Notebook S	Tablet L	Notebook L
Total Speed (WPM)	0.004	67B (4)	73A (4)	72AB (4)	75A (4)
Adjusted Speed3 (WPM)	0.004	64B (4)	70AB (4)	69AB (4)	72A (4)

¹Repeated Measures ANOVA with participant as a random variable and keyboard device (4 levels). Bold values indicate a significant effect (p<0.05).

²For significant main effects, Tukey’s Post-Hoc groupings are ranked such that A>B>C>D. Values with the same superscript letters indicate no significant difference.

³Adjusted speed calculated as Total Speed- Errors.

The self-reported questionnaire revealed significant differences in perceived discomfort and difficulty across the devices (Table 4). Participants rated Notebook L with least discomfort, however this was only significantly different (p = 0.0148) from Tablet S. For difficulty, participants rated Notebook L least difficult, followed by Tablet L. These difficulty ratings were significantly lower (p < 0.001) than Tablet S.

Table 4:

Self Report: Across participant marginal means (and standard errors) for self-reported discomfort and difficulty.

Keyboard Devices
	P Value1,2	Tablet S	Notebook S	Tablet L	Notebook L
Discomfort (cm)	0.015	3.05A (0.45)	1.53AB (0.45)	1.44AB (0.45)	1.07B (0.45)
Difficulty (cm)	<0.001	4.80A (0.48)	3.17AB (0.48)	2.05B (0.48)	1.99B (0.48)

¹Repeated Measures ANOVA with participant as a random variable and keyboard device (4 levels). Bold values indicate a significant effect (p<0.05).

²For significant main effects, Tukey’s Post-Hoc groupings are ranked such that A>B>C>D. Values with the same superscript letters indicate no significant difference.

The effects of device on forearm muscle activity showed similar trends with both sex/gender groups; however, statistical significance was present for almost all of the muscles in the female participants and for only one muscle in the male participants (Table 5). Five (ECR, ED, ECU, FDS, FCR) of the six muscles tested showed significant differences (p < 0.05) across the four devices in female participants compared to only one (ED) (p = 0.009) in male participants (Table 5). Tablet S induced greater forearm muscle activity than the other three devices for both female and male participants.

Table 5:

Across participant marginal means (and standard errors) for electromyography (%MVC) and typing force. RMANOVA Keyboard Device and Typing Task

Keyboard Device
	P-Value1,2	Tablet S	Notebook S	Tablet L	Notebook L
Median EMG %MVC
Females
ECR	0.003	4.7A (0.76)	4.3B (0.76)	4.4B (0.76)	4.5AB (0.76)
ED	<0.001	7.6A (1.0)	6.7B (1.0)	6.8B (1.0)	6.9B (1.0)
ECU	0.009	8.4A (1.4)	7.3B (1.4)	7.4B (1.4)	7.5AB (1.4)
FDS	0.033	3.9 (0.61)	3.2 (0.61)	3.8 (0.61)	3.4 (0.61)
FCU	0.388	3.4 (0.90)	3.2 (0.90)	3.6 (0.90)	3.3 (0.90)
FCR	0.005	2.8A (0.40)	2.4B (0.40)	2.6AB (0.40)	2.6AB (0.40)
Males
ECR	0.179	4.3 (0.66)	4.2 (0.66)	4.1 (0.66)	4.2 (0.66)
ED	0.009	8.7A (1.14)	7.9B (1.14)	8.1AB(1.14)	8.5AB (1.14)
ECU	0.588	7.3 (0.97)	7.1 (0.97)	7.1 (0.97)	7.0 (0.97)
FDS	0.853	2.1 (0.27)	2.1 (0.27)	2.1 (0.27)	2.2 (0.27)
FCU	0.991	1.9 (0.51)	1.9 (0.51)	1.9 (0.51)	1.9 (0.51)
FCR	0.419	1.8 (0.49)	1.9 (0.49)	1.9 (0.49)	1.9 (0.49)
Median Typing Force
Females
Newtons	<0.001	0.46A (0.03)	0.41B (0.03)	0.39B (0.03)	0.41B (0.03)
Normalized (%)	<0.001	4.2A (0.35)	3.7B (0.35)	3.7B (0.35)	3.5B (0.35)
Males
Newtons	0.003	0.48A (0.03)	0.43B (0.03)	0.46AB (0.03)	0.44B (0.03)
Normalized (%)	0.019	2.9A (0.35)	2.7B (0.35)	2.7AB (0.35)	2.8AB (0.35)

¹Repeated Measures ANOVA with participant as a random variable, Keyboard Device (4 levels), Typing Task (2 levels). Bold values indicate a significant effect (p<0.05).

²For significant main effects, Tukey’s Post-Hoc groupings are ranked such that A>B>C>D. Values with the same superscript letters indicate no significant difference

The device’s effect on typing force showed similar trends and significance in both groups (Table 5). Tablet S was associated with more absolute and normalized typing force than the other devices in both female and male participants, though a more significant result was observed for female normalized force than male normalized force (p < 0.001 vs. p = 0.019).

Both groups performed similarly across the four devices with slowest performance achieved on Tablet S (p < 0.05) (Table 6). Female participants typed fastest on Notebook L in terms of both total speed (p = 0.021) and adjusted speed (p = 0.034), whereas male participants typed fastest on Notebook S for total speed (p = 0.023) and adjusted speed (p = 0.013).

Table 6:

Performance: Across participant marginal means (and standard errors) for adjusted speed (calculated as words typed minus errors) during a 1 minute typing test.

Keyboard Devices
	P Value1,2	Tablet S	Notebook S	Tablet L	Notebook L
Total Speed (WPM)
Females	0.021	68B (6)	70AB (6)	69AB (6)	76A (6)
Males	0.023	67B (7)	76A (7)	75AB (7)	74AB (7)
Adjusted Speed (WPM)
Females	0.034	66A (6)	67A (6)	66A (6)	73A (6)
Males	0.013	62B (6)	72A (6)	72A (6)	72A (6)

¹Repeated Measures ANOVA with participant as a random variable and keyboard device (4 levels). Bold values indicate a significant effect (p<0.05).

²For significant main effects, Tukey’s Post-Hoc groupings are ranked such that A>B>C>D. Values with the same superscript letters indicate no significant difference

After stratifying by sex/gender, fewer differences across devices were observed for self-reported discomfort and difficulty (Table 7). Only difficulty in male participants showed significant differences (p = 0.01) with Tablet S rated significantly more difficult than Tablet L and Notebook L.

Table 7:

Across participant marginal means (and standard errors) for self reported discomfort and difficulty.

Keyboard Devices
	P Value1,2	Tablet S	Notebook S	Tablet L	Notebook L
Female
Discomfort	0.143	2.88 (0.64)	1.09 (0.64)	1.03 (0.64)	1.16 (0.64)
Difficulty	0.054	5.08 (0.75)	3.54 (0.75)	2.58 (0.75)	2.24 (0.75)
Male
Discomfort	0.288	3.17 (0.63)	1.82 (0.63)	1.71 (0.63)	1.01 (0.63)
Difficulty	0.009	4.61A (0.65)	2.93AB (0.65)	1.70B (0.65)	1.82B (0.65)

¹Repeated Measures ANOVA with participant as a random variable and keyboard device (4 levels). Bold values indicate a significant effect (p<0.05).

²For significant main effects, Tukey’s Post-Hoc groupings are ranked such that A>B>C>D. Values with the same superscript letters indicate no significant difference.

Discussion

In response to the widespread availability and use of thinner keyboards, the goal of this study was to examine the effects of different key travels and key switch mechanisms currently available in thinner keyboards on forearm muscle activity, typing force, typing performance, and self-reported discomfort and difficulty. Device significantly affected biomechanical outcomes, and the differences across devices were similar for all four outcomes: Tablet S with short travel and a dome type key switch mechanism was associated with higher muscle activity, higher typing force, slower typing speeds, and worse self-reported discomfort and difficulty.

Overall the results show that given similar activation force, larger key travel distances were associated with better performance and experience. The fastest typing speed and least discomfort and difficulty were observed for the device with the largest travel distance. Participants’ self-reported more discomfort and difficulty with the devices with the shortest travel distance. These results are consistent with most previous research^26,27. Kia et al.²⁵ also found better typing speed and accuracy was associated with larger key travel.

The present results suggest that key travel alone does not predict biomechanical outcomes and that key mechanism and activation force are also important factors in key-switch design. Specifically, the two devices with the same short travel (0.55 mm) had the largest differences across most muscles, though this difference was relatively small (< 1.0%MVC). These two devices differed in activation force and mechanism: Tablet S had a dome switch mechanism and a higher activation force than Notebook S which had a butterfly switch mechanism. Similarly, this study found that key travel distance was not strictly associated with typing force, typing performance, or perceived experience, as Tablet S was associated with the worst results across these measures compared to the other three devices. These results align with studies that have tested different activation forces^8,20,26,28 and mechanisms^7,12,19.

In terms of trends observed within sex/gender stratified groups, the results suggest that the effect of the 4 devices was different in female and male participants similar to what has been observed in general occupational health and safety studies^36,39,40. More statistically significant differences across devices were observed in female participants compared to male participants. This result suggests that women may be more affected by micro-travel keyboard designs than men, similar to Won et al (2009) which suggested differences in anthropometry increased the awkward postures and higher muscle activity for female participants. Though the differences observed were small, it is possible that over prolonged use, these differences could correspond to increased risk^41,42. Moreover, the kinematics of the users was not measured as it is suspected that the differences in posture would be small across these devices; however, perhaps reaching some of the non-home row keys magnified the effects of the different devices. In this study, differences in strength probably played a role as all the metrics were normalized by strength or maximum voluntary contractions, meaning that female participants exerted more of their maximum effort than male participants.

As the demand for thinner keyboards and devices increases, it is important for mobile technology designers to consider both the user’s experience and sex/gender when deciding on key force-displacement tradeoffs. Others before have made these recommendations for general concepts surrounding occupational ergonomics concerning sex/gender in both practice and research^36,39. These results support the importance of considering gender in usability testing to ensure that the designs mitigate differences observed within different populations.

This study investigated the effect of four commercially available devices, and as such it is not possible to disentangle the various travel, force-displacement, and mechanism characteristics in a full factorial designed study. The comparisons made above were done so within this context; however, they are supported by other studies that have shown similar patterns. The study did not collect wrist posture which could account for the increased muscle activity in Tablet 1, which did not have any wrist supports; however, participants were set up to maintain similar wrist postures across devices, and the other dependent variables corresponded with the EMG results.

Furthermore, the study was conducted in a laboratory setting, and the short duration of typing may not accurately reflect real-world, long term device use, and it is possible that computing has changed so much that long form transcription is also not reflective of the modern typing experience. In addition, the self-reported experience measures were collected after only a short experience with each device and may not be valid measures of long term discomfort and difficulty. However, consumers would likely only type on devices for a short period in the store, and thus these results may be important to manufacturers. The electromyography effect sizes were relatively small (< 1.0 % MVC) across all devices and may not corresponded to an increased risk clinically; however these subtle differences could matter over long term exposure. We were unable to blind the device brands, and participants’ prior experiences with them could have affected their self-reported results and behavior. Gender is a proxy for multiple biological and psycho-social characteristics, and stratifying by a male/female binary may not explain the root cause of the differences observed.

Key travel is only one design feature which may not fully account for the differences in typing force, muscle activity, typing performance, and participant experience. Future studies should explore key spacing, over-travel, and activation force in addition to key travel. In addition, future studies should account for individual differences such as anthropometry, strength, and sex/gender.

This study found differences in upper extremity biomechanics, typing performance, and self-reported experience across 4 micro-travel keyboard devices with 0.5mm (Tablet S & Notebook S), 1.3 mm (Tablet L), and 1.6 mm (Notebook L) travel distances during short laboratory-based typing tasks. The results further support the importance of other design features such as activation force and mechanism, as well as keyboard form factors. These considerations will be important as the market continues to demand thinner mobile devices.