Methods

Subjects

Ten subjects voluntarily participated in this study. The volunteers consisted of eight males (height M=72.6 in.) and two females (height M = 68.25 in.). The participants were college students who were accustomed to using a computer keyboard for typing tasks. All participants were healthy and had no acute or chronic musculoskeletal disorder or pain that could interfere with typing. Nine out of the ten participants normally used the standard straight keyboard and only one subject normally used the split fixed keyboard.


Procedure

The ten subjects were randomly divided into two groups, A and B, each containing five people. Group A was asked to begin typing on the straight keyboard followed by the split fixed keyboard. Group B began typing on the split fixed keyboard followed by the straight keyboard. Each group typed a different passage with a 250 word count and similar difficulty.

The comfort level was evaluated by asking each participant a series of subjective survey questions while the participant typed on each keyboard. Comfort was defined as the absence of perceived discomfort. The subjects were asked questions pertaining to the level of discomfort felt in their wrist, shoulders, neck, lower back, upper back, and chest. They would answer the questions by associating the discomfort they felt with a number from 0 to 4. The numbers corresponded to discomfort levels by the following relationship: 0 - no discomfort, 1 - minimal discomfort, 2 - some discomfort, 3 - uncomfortable, 4 - highly uncomfortable. Subjects related their perceived discomfort to the linguistic queues associated with the rating scale.

Objective measurements were also taken for each individual. These measurements included standing height, sitting knee height, ulnar deviation, extension of the wrist, and glenohumeral abduction. Standing height was measured from the bottom of the foot to the top of the head, as the subject stood erect with both feet flat on the floor. Sitting knee height was measured from the bottom of the foot, perpendicular to the dorsal side of the femur, as the subject sat with his or her back and buttocks against a chair with both feet flat on the floor. A hand-held goniometer was used to measure ulnar deviation, extension of the wrist, and glenohumeral abduction. The vertex of the goniometer was placed on the joint, and angles were measured from neutral position. Neutral position was defined such that the second and third metacarpals were in-line with the radius and ulna, keeping the wrists straight. Extension of the wrist was defined as a positive angle from neutral position. Ulnar deviation of the wrist was defined as a positive angle from neutral position. Glenohumeral abduction was defined as the angle between the upper arm with respect to the saggital plane running through the glenoid fossa. The glenohumeral abduction, ulnar deviation, and extension of the wrist were measured using the goniometer.


Workstation

The workstation was adjusted to the body dimensions of each subject. The chair was adjusted so the subject's feet were flat on the floor and his or her thighs were parallel to the floor. Then, the keyboard tray was adjusted to a height such that the subject's forearm to upper arm angle was within a range of 90 to 110 degrees. The keyboards were centered with respect to the visual display. The height of the visual display table was adjusted so that the top of the visual display screen was level with the subject's horizontal line of sight (Tayyari 1997).

Results

Paired t-test

A paired t-test was performed to determine if the participant's overall comfort level was affected by the design of the keyboard. This test consists of analyzing the differences between comfort levels of each keyboard for each participant. A t value (t₀) and difference score was calculated for each keyboard design. This t value is a measure of how far the average difference score is from zero. The larger the t value, the more likely it is that the difference score is not zero. The null hypothesis states that if there is no difference in comfort level, then the mean of the differences should equal zero. The null hypothesis is rejected when t₀ > t_{a/2, n-1}. The calculated value for t₀ was -0.93553, while the value for t_{0.025, 9} was 2.262. Therefore, the null hypothesis was not rejected. This test revealed that overall comfort level is not significantly affected by keyboard design.

The average comfort level of the neck, shoulders, chest, upper back, lower back, and wrists for each keyboard, are shown in Table 1 and graphed in Graph 1. The steeper the slope of the line on Graph 1, the larger the comfort level difference between the two keyboard designs. Visual inspection of the graph allowed it to be determined that the design of the keyboard does not have a significant effect on the comfort level of the neck, chest, and lower back. Therefore, the comfort levels for the shoulders, upper back, and wrists remained as possibly being affected by keyboard design. Paired t-tests were performed and the t0 values were 2.7136, 1.4056, and 1.9640, respectively. The t_{0.025, 9} was 2.262. Results of the paired t-test demonstrated that keyboard design does have an effect on the comfort level of the shoulder. The t₀ value of the wrist was extremely close to t_{0.025, 9}, and therefore, the comfort level of the wrist was considered in further tests as being affected by the keyboard design.

Linear Model

After concluding that the comfort levels of the shoulder and wrist were possibly affected by keyboard design, a linear model was generated using SAS software to determine which angles (ulnar deviation, extension of the wrist, and glenohumeral abduction) contributed to the perceived level of comfort. The average of the angle measurements can be found in Table 2. The standard error in the raw data collected for ulnar deviation, extension of the wrist, and glenohumeral angle was 2.06, 1.04, and 1.63, respectively. The comfort level of the shoulder and the wrist were assumed to be functions of the following variables: keyboard design, glenohumeral angle, extension angle of the wrist, and ulnar deviation of the wrist. A variable is significant if its P-value < 0.05. The P-value is the smallest level of significance that would lead to rejection of the null hypothesis. A small P-value indicates how unlikely it is that a test statistic as extreme or more extreme than the one given by the data will be observed given that the null hypothesis is true. Upon analyzing the SAS output, it was concluded that the keyboard design and ulnar deviation of the wrist affect the comfort level of the shoulder. Furthermore, SAS output indicated that none of the variables significantly affect the comfort level of the wrist. The P-values associated with the variables are in Table 3.

Discussion

The initial objective of the experiment was to determine if keyboard design had a significant effect on the user's comfort level. The results of the study expressed that overall comfort level is not affected by the design of the keyboard.

The results of this study showed a significant reduction in ulnar deviation while typing on the split fixed keyboard. These findings did agree with the results of a study conducted by Zecevic, Miller, & Harburn (2000), which showed that there was a decrease in ulnar deviation when a subject used the split fixed keyboard. By reducing ulnar deviation while typing, the wrists are closer to a neutral posture, thereby reducing the risk of work-related musculoskeletal disorders (Marklin 1999).

The split fixed keyboard reduced the extension of the wrist in this study, but not to a statistically significant level. These results agree with the study conducted by Marklin, Simoneau, & Monroe (1999), which stated that the split fixed keyboard did not substantially reduce wrist extension. Rempel, Kier, Smutz, & Hargen (1997) found that wrist extension had a greater effect than ulnar deviation on carpal tunnel pressure. This finding indicates a need for more studies of the upper extremities while typing.

The study did not show a significant difference in glenohumeral abduction. This result is partially supported by Simoneau, Marklin, & Monroe (1999) study that reported that the wider the person is, the greater the ulnar deviation of the wrists while typing. However, the fact that this relationship was low (approximately, r = .20) demonstrates that typing style, as well as position of the elbows has a more significant effect on wrist posture than the width of the person.

Standard goniometer protocol was used, but goniometry appears to be inadequate. A possible reason that there was no significant difference in the glenohumeral abduction in this study is because there was no standard point of reference for the measurement of this angle. There was no clear reference point for the gleno-humeral joint; therefore, its location had to be estimated. The estimated location of the gleno-humeral joint was used as the reference point for the angle measurement. Because of this crude estimation, the measurements had a wide range of variance, and it was difficult to get repeatable data from each subject. A better measurement system, such as electrogoniometry or motion capture, is needed to fully test this hypothesis.

Although this study concluded that overall comfort level while typing is not affected by the design of the keyboard, the study did show that the comfort level of the shoulders was significantly affected by keyboard design and the amount of ulnar deviation of the wrist. A study conducted by Simoneau et al. (1999) reported that subjects would tend to move their elbows anteriorly to rest them on their abdomen, effectively reducing the elbow-to-elbow distance (decreased glenohumeral abduction) and ulnar deviation of the wrists. This finding indicates a need for the individualization of alternative keyboard designs based on the particular typing position of each user and not based exclusively on his or her size. In an effort to maximize shoulder comfort, the glenohumeral angle and ulnar deviation must be reduced. The split keyboard design induces smaller glenohumeral angles and less ulnar deviation and, therefore, increases comfort in the shoulders and reduces the risk of work-related musculoskeletal disorders.

Because seventy percent of the subjects prefer the straight to the split fixed keyboard, and ninety percent of the subjects normally use a straight keyboard for typing tasks, it is believed that the subjects' responses to the survey questions based on initial perception were biased. This could cause the results to be somewhat skewed. These results were not supported by the study conducted by Zecevic et al. (2000) in which they found the preference to one keyboard over another was inconclusive. The results might be more conclusive if a larger, more random sample were taken.

Intuitively, the results of the experiment would indicate that there is a difference in comfort levels between the split and the straight keyboards. From an ergonomic standpoint, the split keyboard provides a higher level of comfort because it allows the user to maintain a more neutral body position than a straight keyboard.

Acknowledgements

We thank Mr. Patrick L. Parker for his direction throughout this research project. We would like to express our thanks and appreciation to Mr. Clay Walden for his help in the statistical analysis of the results of this experiment. We would also like to thank the Mississippi State University students who participated in this study.