Relationship of Hand Size and Keyboard Size to Typing Performance Metrics

Home , Caps lock, Dvorak keyboard layout, Home key, Lock key

A thesis proposal presented to

the faculty of

the Russ College of Engineering and Technology of Ohio University

In partial fulfillment

of the requirements for the degree

Master of Science

Warnaka R. Gunawardena

December 2013

This thesis proposal titled

Relationship of Hand Size and Keyboard Size to Typing Performance Metrics

WARNAKA R. GUNAWARDENA

has been approved for

the Department of Industrial and Systems Engineering

and the Russ College of Engineering and Technology by

Diana J. Schwerha

Associate Professor of Industrial and Systems Engineering

Dennis Irwin

Dean, Russ College of Engineering and Technology 3

ABSTRACT

GUNAWARDENA, WARNAKA R., M.S., December 2013,

Industrial and Systems Engineering

Relationship of Hand Size and Keyboard Size to Typing Performance Metrics

Director of Thesis: Diana J. Schwerha

Touch screen surfaces are increasingly used in mobile phones, tablets, notebook and desktop computers, and other industrial applications as a convenient method of interaction between a user and a device. They offer versatile features such as being space- saving, easy to clean, less noisy for data input, and adjustable when compared to other data input methods. While research has been conducted in the relationship between hand size and mobile phone use, little research has investigated the relationship between touch screen keyboard size, hand anthropometry, and performance metrics. In this study, the relationship between the user’s anthropometric data and the relative size of the touch screen keyboard to typing speed and the typing accuracy was studied.

Thirty participants were recruited based on their hand size (Small n = 10, Medium n = 10, and Large n = 10). They were given 15 sentences to type on three different sizes of keyboards (14x14, 18x18, and 22x22 mm). The speed was measured by characters per minute and accuracy was measured by the correct characters per minute and incorrect ratio. Characters per minute and correct characters per minute data were statistically analyzed using a mixed model ANOVA. Keyboard sizes were significantly different for both characters per minute and correct characters per minute at p < .05. However, hand size and the interaction factor were not significantly different. Future research may focus 4 on further developments on touch screen keyboard sizes to improve the speed and accuracy rather than the hand sizes.

ACKNOWLEDGMENTS

I would like to pay my sincere gratitude to everyone who helped me in many ways to excel in my studies at Ohio University.

I am greatly thankful to Dr. Diana Schwerha, my advisor and research director, for her continuous support and guidance and encouragement given to me during my studies at Ohio University. Also, I wish to convey my heartily appreciation to my thesis research committee members Dr. David Koonce, Dr. Gary Weckman, and Dr. Andrew

Snow for their time, direction, and suggestions given to produce a quality research project, without their support this task would have been a distant reality.

I am thankful to my ever loving wife, Sumali for the never-ending support and patience given to me to complete my research. Further, I am thankful to Mr. Praveen

Gopallawa, who was there to assist me in various ways during research data collection.

My heartiest gratefulness goes to the Department of Industrial and Systems Engineering for the assistance given, especially to Ms. Tonya Seelhorst. Additionally, I would like to give my deepest gratitude to each person who helped me at some point in my studies at

Ohio University.

TABLE OF CONTENTS

Page

Abstract ...... 3

Acknowledgments...... 5

List of Tables ...... 8

List of Figures ...... 9

Chapter 1. Introduction ...... 11

Chapter 2. Literature Review ...... 13

2.1. Touch Screen History of Technologies ...... 13

2.2. Table-Top Touch Screens, Dual Touch Notebooks and Applications ...... 17

2.3. Factors which Hinder the Use of Touch Screens ...... 20

2.4. Improving Touch Screen Keyboards ...... 22

2.5. Lessons Learned from Physical Keyboards ...... 26

2.6. Guidelines for Touch-Screen Keyboard Design ...... 28

2.7. Relationships between Hand Size and Performance on Notebook Computers and

Mobile Phones ...... 30

2.8. Research Hypothesis ...... 32

Chapter 3. Method ...... 34

3.1. Study Population ...... 34

3.2. Research Instrument and Scoring ...... 36

3.3. Participant Testing ...... 42

3.4. Data Compilation ...... 43 7

3.5. Data Analysis ...... 43

Chapter 4. Results ...... 45

4.1. Test Results for Normality ...... 45

4.2. Descriptive Statistics ...... 45

4.3. Results of Repeated Measures Two-Way ANOVA ...... 49

4.4. Non-parametric Tests ...... 52

4.5. Survey Results ...... 53

Chapter 5. Discussion ...... 54

Chapter 6. Conclusion ...... 57

References ...... 59

Appendix A. Study Flyer ...... 64

Appendix B. Consent Form ...... 65

Appendix C. Test Sentences ...... 68

Appendix D. Study Survey ...... 70

Appendix E. Normality Test Results ...... 73

Appendix F. Descriptive Results Characters per Minute ...... 76

Appendix G. Descriptive Results Correct Characters per Minute ...... 77

Appendix H. Descriptive Results Incorrect Ratio ...... 78

LIST OF TABLES

Page

Table 2-1 Comparison between Different Approaches in Touch Screen Keyboard Design

...... 23

Table 2-2 Hand Anthropometric Data ...... 31

Table 2-3 Different Piano Keyboard Sizes...... 32

Table 3-1 Hand Size Measurements ...... 34

Table 3-2 Touch Screen Keyboard Dimensions ...... 36

Table 4-1 ANOVA Tests of Within-Subjects Effects for Characters per Minute ...... 49

Table 4-2 ANOVA Tests of Between-Subjects Effects for Characters per Minute ...... 50

Table 4-3 ANOVA Tests of Within-Subjects Effects for Correct Characters per Minute ...

...... 50

Table 4-4 ANOVA Tests of Between-Subjects Effects for Correct Characters per Minute

...... 51

Table 4-5 ANOVA Pair-Wise Comparison for Keyboard Size ...... 51

Table 4-6 Non Parametric Ranking of Hand Size and Key Size (Kruskal-Wallis Test) .. 52

Table 4-7 Kruskal-Wallis Test Statistics Hand Size and Key Size (Kruskal-Wallis Test)

...... 52

Table 4-8 Descriptive Data of the Survey Responses ...... 53

LIST OF FIGURES

Page

Figure 2-1a Resistive touch screen ...... 14

Figure 2-1b Capacitive touch screen...... 14

Figure 2-2 Acoustic Surface-wave ...... 15

Figure 2-3 Acoustic-Pulse...... 16

Figure 2-4 Infrared internal reflection method ...... 17

Figure 2-5 Table top used at a futuristic school ...... 18

Figure 2-6 Tabletop computer with two users ...... 18

Figure 2-7 Acer Iconia dual touch ...... 19

Figure 2-8 Acer Iconia virtual KB ...... 19

Figure 2-9 Red lines represent key boundary and dotted line shows the anchor region .. 25

Figure 2-10 DVORAK keyboard layout ...... 26

Figure 2-11 Ulnar deviation when typing on a rectangular keyboard ...... 27

Figure 2-12 Key spacing according to OSHA guidelines ...... 28

Figure 3-1 The sketch used to determine participants hand size ...... 35

Figure 3-2 (a) Size 1 keyboard (14*14); (b) Size 2 keyboard (18*18); (c) Size 3 keyboard

(22*22) ...... 37

Figure 3-3 Layout of the keyboard ...... 38

Figure 3-4 Program set-up ...... 39

Figure 3-5 Sample of the touch screen while testing ...... 40 10

Figure 3-6 (a) Fully adjustable ergonomic chair, (b) Height adjustable touch screen

monitor, and (c) Complete work station ...... 41

Figure 3-7 Typing test demonstration ...... 42

Figure 4-1 Descriptive results - Main effect Key size ...... 46

Figure 4-2 Descriptive results - Main effect hand size ...... 46

Figure 4-3 Descriptive results - Characters per second ...... 47

Figure 4-4 Descriptive results - Correct characters per second ...... 48

Figure 4-5 Descriptive results - Error rate ...... 48

CHAPTER 1. INTRODUCTION

Touch screen surfaces have been used by many as a convenient method of interaction between a user and a device. They are also considered to be a space-saving, easy to clean, less noisy interface of data input when compared to other data input methods, such as physical keyboards and mice. Increasingly, they are used in mobile phones, tablets, notebook and desktop computers, and other industrial applications.

Products such as the Samsung SUR40 (Touch Table) [1], Acer Iconica (Dual Touch screen Laptop) [2] & Lenovo IdeaCenter Horizon Table PC [3] give us a glimpse of the future on different ways a touch screen could be used. Almost all touch screens use a software keyboard or a number pad to input data and to interact with the system.

According to Barrett & Krueger [4] touch screen keyboards lack the tactile feedback, making the performance lower than a physical keyboard. Surveys conducted by

Benko et al [5] showed that "lack of a decent keyboard" is the number one concern with table top touch screens. On the other hand, due to the inherent software keyboard, touch screens provide a good platform for customization. Given their recent development, however there is room for improvement in touch screen keyboard design.

Improvements to the touch screen keyboard design have been made with the aim of improving the efficiency of the typing as well as the accuracy. These include pattern recognition [6], synchronized physical keyboards [7], moving keys (liquid keyboard) [8] and adaptive typing [9]. Although physical keyboards have evolved to tackle ergonomic issues, proposed improvements have not directly addressed the ergonomic issues which have followed the standard touch screen layout. Also, as the virtual keyboard is 12 customizable, virtual keyboards have the potential to adjust to different hand anthropometry. While research has been conducted in the relationship between hand size and mobile phone use, little research has investigated the relationship between touch screen keyboard size, hand anthropometry, and performance metrics. This gap in the literature is what my research seeks to address.

The proposed research aims to find the possible relationship between the hand anthropometry, typing speed and typing accuracy. The hypotheses are listed below:

1. For touch screen keyboards, typing speed is significantly related to hand size and

keyboard size.

2. For touch screen keyboards, typing accuracy is significantly related to hand size

and keyboard size.

If the hypotheses are supported, a touch screen keyboard could be sized to the user’s hand size, to improve typing metrics. Such a concept can be used in information kiosks, tabletops and even dual touch laptops. This would be useful in places such as airports, where there is a diverse group of users with different physical attributes.

CHAPTER 2. LITERATURE REVIEW

2.1. Touch Screen History of Technologies

In 1965, E. A Johnson [10] described his thoughts on capacitive touch screen displays. His idea was to have thin wires laid across the screen which could be used to recognize the touch input. In 1983, Hewlett Packard released the first commercial touch screen computer, HP 150 [11]. These first generation touch screens were activated by the finger interrupting light beams, which ran parallel to the screen. With advances in the

Graphical User Interfaces (GUI) in the 1980s, input systems, such as touch screens, were in demand. Since then, touch screen concepts have evolved rapidly. They have become ever more popular in the 21st century due to the integration of the technology in products such as mobile phones, tablets and touch monitors. According to the German flat panel display forum [12], the touch panel industry has grown from $4 billion to over $13 billion dollars from 2009 to 2011, tripling in over 2 years. It is expected to be $23 billion by the year 2017. A market survey by Elo Touch Solutions in 2004 [13] has shown that 97% of

Americans between the ages of 18-34 believe touch screens are convenient to use. In the same survey 96% indicated that touch screens make work easier, and nine out of ten believe it saves time.

Touch screens come in different technologies such as resistive, capacitive, acoustic pulse, acoustic wave, and infrared ( IR) [14]. A resistive type touch screen [15] has a resistive coating on the glass and an adjoining conductive coating separated by insulating dots (Figure 2-1a [15]). The conductive coating faces the user, and it is again covered with a flexible hard-coated membrane. When a user applies pressure on the 14 membrane, the conductive coating touches the resistive coating and creat es an electrical contact. This is reflected as a voltage change due to the resistance, and it is converted to an analog reading in the x and y axes of the screen. Since the signal is generated by pressure, the screen could be activated by gloves, finger nails, credit cards and styluses.

Figure 2-1a Resistive touch screen [15]

Figure 2-1b Capacitive touch screen [16]

The capacitive type touch screens have a conductive coating, which is supplied by a low voltage [16]. When the user places his finger on the screen, electrical conductance 15 creates a current that flows through the finger. The current is measured from the corners of the touch screen, and the relative position is calculated, as shown in Figure 2-1b [16].

As an electrical current is required to flow from the screen, non-conducting materials, such as gloves and credit cards, cannot be used to activate capacitive touch screens.

Acoustic surface wave [17] (Figure 2-2) and acoustic pulse [18] (Figure 2-3) technologies determine the location of the point with the use of piezoelectric transducers located at the edges of the touch screen. As the screens themselves do not have any coating, these screen types provide better image clarity than resistive and capacitive types.

Figure 2-2 Acoustic Surface-wave [17]

Figure 2-3 Acoustic - Pulse [18]

IR grid touch screen technology [19], relies on an array of infrared light beams and detectors which are laid across the screen. The working principle is similar to the first versions of touch screens in the 1970s and 1980s. Another type of IR touch screen is

"Frustrated Total Internal Reflection" developed by Jeff Han [20]. In it, IR light is projected to the side of a glass and the light reflects internally (internal reflection). As the user touches the screen, it changes the refraction index, which releases the IR rays from the touched point. As shown in Figure 2-4 [22], IR cameras mounted under a projected surface "see" the points of IR light as the fingers touch the screen.

An advanced version of the IR technology used in touch screens, called diffused illumination, uses cameras to capture IR light which is projected from the top or underneath the touch screen [20]. This technology was introduced by Microsoft as

"PixelSence" [21] in 2007.

Figure 2-4 Infrared internal reflection method [19]

2.2. Table-Top Touch Screens, Dual Touch Notebooks and Applications

In 2012, Samsung partnered with Microsoft to release "Samsung SUR40 with

Microsoft PixelSense" [1], a 40-inch multi-user computer with a full High Definition (full

HD) display (Figure 2-5 [1]). This product type, also known as "Tabletop" or "Touch surfaces," was simply a large multi-touch screen computer laid horizontally. Due to the orientation of the product, multiple users interact in the same screen at the same time. A tabletop, for example, could be used to help in idea generation at a meeting. Images could be dragged, resized and organized with the use of fingers. The product also has the possibility of being implemented in schools (Figure 2-6 [1]) and offices, which would benefit from the touch interaction. In 2013, Lenovo launched their own Tabletop "

Lenovo IdeaCenter Horizon Table PC". It was a 27 inch multi touch computer and was sold at a price of $ 1,700 [3].

Figure 2-5 Table top used at a futuristic school [1]

Touch input eliminated the need of a pointing device such as a mouse. Also, with the use of a virtual keyboard in the touch screen, a physical keyboard was not required to interact with the tabletop. More applications of large area touch screens have been released in the recent past. One such application is a large touch screen Musical

Instrument Digital Interface (MIDI) Controller [22].

Figure 2-6 Tabletop computer with two users [1]

New applications of touch screen technology have surfaced in the computer notebook design as well. Acer Iconia [2] is the world's first dual-touch screen notebook design. It replaces the traditional physical keyboard of the notebook with a touch screen

(Figure 2-7 [2] and 2-8 [2]). When the user touches the bottom screen with ten fingers, the virtual (or soft) keyboard appears. The user can then type into the virtual keyboard.

Figure 2-7 Acer Iconia dual touch [2]

Figure 2-8 Acer Iconia virtual KB [2]

When compared to physical keyboards, there are some advantages in using touch screen keyboards. Physical keyboards are bulky and usually have mechanical moving parts that create noise. Dust and dirt accumulate in-between keys in physical keyboards and needs cleaning. Touch screens, however, are easy to clean and maintain when compared to physical keyboards. Touch screens are also easy to be designed as spill resistant when compared to physical keyboards and mice. Physical keyboards take up space when not in use. On the other hand, virtual keyboards could be put away so that the screen space could be used for another activity. Touch screen keyboards have the ability to be customized. An English keyboard could be easily switched to another language without the need of having physical change to the device. The lighting of the screen or the color of the keyboard can also be changed according to a user’s preference.

2.3. Factors which Hinder the Use of Touch Screens

According to a survey conducted by Benko et al [5], the lack of a proper keyboard was considered to be the biggest drawback in tabletop computers. In their survey, 25 out of 58 respondents answered "a keyboard" to the question "what single feature of standard desktop computers do you miss out when using an interactive tabletop?" Their comments on the greatest frustration related to keyboards are as follows [5].

1. Lack of decent keyboards;

2. Virtual ones don't work well;

3. Physical ones may occlude;

4. Typing, soft keyboard works poorly, no feedback. 21

One of the biggest drawbacks in virtual keyboards is the inability to receive tactile feedback [4], [23]. When a user types on a physical keyboard, the finger senses the physical keys. Also, when the button is pressed, the mechanical sound gives auditory feedback, which helps the typist to continue to press the next key. Mobile phone designs have attempted to provide a solution by giving the option of "haptic feedback", where the phone vibrates when the user presses a key. Automated Teller Machines (ATMs) provide audible feedback to its customers when pressing keys on the touch screen. However, in the case of a tabletop, haptic feedback [24], [25] is questionable due to the concurrent multi-user support feature and possible ad-hoc interaction. Unavailability of a textured surface on a touch screen keyboard makes "touch typing" difficult. Also, attempting to place the fingers on the touch screen to get a feel of the key placement creates ambiguity in the computer. It fails to identify whether the keystroke was intentional or simply adjustments of the fingers.

Also, from an ergonomics point of view, the horizontal placement of the tabletop makes the user bend his neck for long durations. This concern was also highlighted in the survey conducted by Benko et.al. [5]. In their study, some users preferred the tabletop to be adjusted, so that the device could be used single-user and multi-user interchangeably.

According to a study conducted by the Harvard School of Public Health [26], neck flexing while looking down could increase the neck extensor activities.

2.4. Improving Touch Screen Keyboards

There are many ideas that target to improve touch screen keyboards. Some methods aim to replace the lapse of tactile feedback. Other methods aim to customize the touch screen keyboard so that the errors created by the fingers are minimized. Each method has its advantages and disadvantages.

The comparison between the different ideas is shown in Table 2-1. Rashid and

Smith [5] proposed a method to identify patterns created when typing on the touch screen. They found out that on a standard ―qwerty‖ layout, keystroke patterns of 99.5 % of the words extracted from a dictionary were found to be unique. Therefore, a touch typist would be able to type on a touch screen and the pattern created could be interpreted by the computer. Although the method suits touch typists, those who are not expert in touch typing could have difficulty learning this technique.

Table 2-1 Comparison between Different Approaches in Touch Screen Keyboard Design

Idea Author Year Basic principle Advantages Disadvantages

1. A beginner would have Keystrokes are hard time treated as 1. Does not identifying the relative inputs require a visual word on the Relative in a continuous keyboard screen before it Keyboard / Rashid, space. A 2008 2. Uses a has been Pattern Smith dictionary is dictionary to corrected. recognition used to identify correct the 2. Does not the pattern and words describe how to predict the the numerical typed word. row can be used with the setup Using 1. Improved Hartmann, Physical 1. Space on the physical typing speed and Morris, keyboards, mice screen is used keyboards 2009 accuracy due to Benko, used with a up for on table physical Wilson tabletop keyboards. tops keyboard

1. Fingers at 1. Home key natural typing ambiguity in posture practical Moving home 2. Positions of implementation Liquid Sax, Lau, 2011 keys relative to surrounding 2. Have not Keyboard Lawrence hand keys are relative confirmed if the to the system could be orientation of used for a numeric home keys row of keys 1. Advantages are in the long run and not for short-term Improved typing use Keyboard learns speed with Adaptive 2. Assumes Findlater, and adapts to reduced error (learning) 2012 flexibility of the Wobbrock the user’s rate over a long typing fingers is constant typing duration of 3. Prone for typical typing touch screen drawbacks in short- term use

An MIT Technology Review [27] in touch typing describes the fact that schools are not teaching how to touch type, as they believe students already know how to touch type. As most of the students interact with mobile phones before properly using a computer, they lack the ability to touch type. Instead they revert to hunt and peck methods (e.g., a person looks for the key and press with the same finger).

Hartmann et al [7] checked the possibility of using physical keyboards and mice with tabletop displays. Benefits of the idea are high-precision and high-performance input and reduced physical movement compared to conventional touch input. Also, the physical device can be used as point of reference to identify the user. However, using physical keyboards consumes valuable screen space which can be used for other tasks when the keyboard is not used.

An improved method named "Liquid keyboard" by Sax et al [8] uses the natural finger placement to determine the key layout of the two alphabetical key rows in the

―qwerty‖ layout. As in touch typing with physical keyboards, the home keys "A, S, D, F,

J, K, L;" are assigned to the ten fingers. This configuration allows the keyboard to be oriented towards the user’s hand and thus provides personalized key input. Keys which are adjacent to the home keys are activated when the fingers move relative to the home keys. For example, the key "G" will be activated when the index finger of the left hand moves to the right and presses on the screen. Although this technique is a great improvement, it suffers with a "home key issue". The system (computer) fails to distinguish between intentionally pressing a home key and the finger returning back to the home key after pressing an adjacent key. Also, the method still needs to identify a 25 solution to press numeric keys, which are two rows away from the home keys in the

―qwerty‖ layout.

An adaptive method to train the system (computer) to identify and correct the touch input was proposed by Findlater & Wobbrock [9] in 2012. They checked the possibility of having an adaptive system which has a visually dynamic or a static key arrangement. Their study suggested that having a visually static layout improved the typing speed compared to the dynamic model. This finding suggests that even though the adaptive system seems to be customized for a user, the user still needs visual guidance in a standard layout. Research conducted by Gunawardena et al [28] shows that the dynamic changes to the key layout in adaptive typing should be moderated to reduce error (Figure

2-9 [28]).

Figure 2-9 Red lines represent key boundary and dotted line shows the anchor region [28]

Gunawardena et al found that having a smaller anchor area in a dynamic model reduces the error. Their research further demonstrated that having a static layout creates less error than the dynamic model.

2.5. Lessons Learned from Physical Keyboards

There are many areas where the physical keyboard design had a negative impact on the user and typing efficiency. The existing ―qwerty‖ layout is a descendant from the common typewriter layout [2 9]. This inefficient key layout was created due to a mechanical key jam issues faced by typists. The creators of the computer adopted the same layout to the keyboard and it became popular worldwide. There have been attempts to improve the key layout with different styles such as the DVORAK layout [29]; however, the popularity of the new keyboard design is not significant (Figure 2-10 [32]).

Figure 2-10 DVORAK keyboard layout [30]

The standard keyboard layout, used in many computers and laptops, is flat and rectangular. Therefore once the fingers / hands are placed on the keyboard, the wrist is 27 bent in an ulnar direction (Figure 2-11). Tittiranonda et al [31] studied different physical key board designs which aim to reduce awkward wrist movement while keeping the standard keyboard layout as the control. The study showed that over a six month period, a split type (winged) physical keyboard design had reduced pain caused by ulnar deviation.

Figure 2-11 Ulnar deviation when typing on a rectangular keyboard [32]

A physical keyboard also has a fixed key layout. Ergonomics teaches that it is better to fit the machine to the man, instead of changing the man to fit the machine [30].

Therefore, the lack of ability to change the key layout forces users with different hand sizes to adjust to the same size. If the keyboard is used at a school, the keyboard key layout should be smaller than what is preferred by an adult. Due to practicality, the manufacturability and feasibility design of a physical keyboard which changes the button size and the distance with each button is not feasible. However, touch screen keyboards can be customized to accommodate such design improvements. Although 100% implementation of a ―Dvorak‖ layout could take generations due to the learning process, 28 a winged keyboard design implementation could be possible if it is proven to be a standard.

Many touch screen keyboard layouts are based on the typical rectangular shaped

―qwerty‖ keyboard. The size of the keyboard depends on the screen size, resolution and the default size of the keyboard itself. Findlater et al [9] have suggested that such keyboards are not suitable for touch typing, as the user would place the fingers on the keyboard in a natural arc and not in a linear orientation.

2.6. Guidelines for touch-screen keyboard design

The Occupational Safety & Health Administration (OSHA) has provided with guidelines on the horizontal and vertical key spacing in physical keyboards [33] . It is mentioned that the horizontal distance between two keys should be 18-19 mm and the vertical distance between two keys should be 18-21 mm (Figure 2-12 [35]). However,

OSHA does not cite a guideline for the touch screen keyboard size.

Figure 2-12 Key spacing according to OSHA guidelines [33]

The Human Factors and Ergonomics Safety Society (HFES) and American

National Standards Institute (ANSI) have provided guidelines for physical keyboards and as well as touch screen keyboard design. According to ANSI/HFES 100-2007 Human

Factors Engineering of Computer workstations[34], a physical keyboard’s horizontal and vertical distance between two keys should be18 mm to 20 mm. For virtual keyboards, the minimum active width and height of a virtual key should be 9.5 mm. It also shows that touch areas greater than 22 mm do not improve performance. The dead space or the inactive area surrounding the key is recommended to be 3.2 mm.

Since virtual keyboards have a different interface than traditional keyboards, researchers have studied the relationship between muscle exertion, performance metrics and virtual keyboard sizes. Kim et al [35] showed that typing on a virtual keyboard with a Key size of 13 x 13 mm could reduce the typing speed by 15% when compared to 16 x

16 mm keyboard. The research also showed that the participants had increased shoulder muscle activity when typing on a 13 x 13 mm Key sized layout. Interestingly, the smaller

13 x 13 Key size had improved overall accuracy than the larger Key sizes. A research conducted by Mary et al [36] gave evidence of Key size significance when using a virtual keyboard. However in the same research it was shown that the key space between the virtual keys was not significant with the dwell time.

2.7. Relationships between Hand Size and Performance on Notebook Computers and Mobile Phones

Chaparro et al.[37], compared the usability of touch-screen tablets and netbooks.

In their study, subjects’ typing speed was approximately 40% faster in netbooks than

Apple iPad. There was no significant difference in the typing speed between iPad used in landscape and the portrait mode. The netbook has a larger typing area, tactile feedback and familiar setup. Even in table top keyboards, it was found that the standard touch screen layouts perform significantly better than other smaller sized touch screen keyboards such as the radial keyboard and the pinpoint keyboard [38].

In terms of large touch screen keyboards, Sears et al. [39] showed in 1993 that as the touch screen keyboard size changed, the typing speed was considerably impacted. For example, in his study there was a gradual increase in the words per minute when the Key size was changed from 0.57 cm to 2.27 cm. A 2.27 cm key is larger than a typical keyboard Key size. However this research does not indicate the point at which the performance level would start to drop. When the keyboard becomes larger the fingers need to travel further and affects the typing speed. Sears conducted his experiments on a

640 x 480 resolution touch screen. Modern technology has produced touch screens which have better resolution, faster response and better viewing angles. Also, in his research there was no consideration of the hand anthropometries effect on the result. Therefore,

Sears’ research provides the path for more research on the size of the touch screen keyboard. 31

Hand anthropometry is different from person to person. For example according to the Georgia Tech Human Systems Engineering Branch (HSEB) the hand breadth of a 5th percentile female is 69 mm and 95th percentile male is 95 mm [40] . The 50th percentile male hand breadth is 15% larger than that of a female. Table 2 shows more anthropometric data related to the hand size.

Table 2-2 Hand Anthropometric Data 95th 5th percentile 50th percentile Dimension Gender percentile (mm) (mm) (mm) Male 173 189 205 Hand length Female 159 174 189 Male 98 107 116 Palm length Female 89 97 105 Male 44 51 58 Thumb length Female 40 47 53 Male 20 23 26 Thumb breadth Female 17 19 21 Index finger Male 64 72 79 length Female 60 67 74 Male 78 87 95 Hand breadth Female 69 76 83

Before the smart phone revolution, the physical keypad on mobile phones had similar hand anthropometry issues. According to Balakrishnan & Yeow [41], female subjects found that it was easier to use the keypad than male subjects. They concluded that the difference in preference was due to the hand size difference. Smaller fingers could access the keys when large fingers had difficulties and errors in typing. In the case of smart phones with touch screens, studies show that even thumb input changes from 32 person to person [42] [43]. Thumb size and reach influences the user experience of the mobile device. Changing the size of a key affects the overall size of a mobile device which could then hamper the market appeal. However, table tops have the space required for customization due to the large touch screen area.

Studies have been conducted with piano players and the piano keyboard size.

Originally pianos were created based on the preference of male European piano players

[44]. With time, as the piano was not just for a specific class, it was determined that the size of the piano keyboard should be reconsidered and made in a way so that smaller sized individuals are not at a disadvantage. Therefore piano keyboards have been made in different sizes to accommodate the different hand sizes. Table 2-3 shows the different standards used in the sizing of piano keyboards.

Table 2-3 Different Piano Keyboard Sizes

Keyboard standard Dimension (mm) Target group Conventional Keyboard 165.1 Large 15/16 - DS Standard™ Keyboard 152.4 Universal 7/8 - DS Standard™ Keyboard 140.7 Small 3/4 - DS Standard™ Keyboard 129.9 Child

2.8. Research Hypothesis

In all attempts to improve touch screens, the initial layout is thought to be the standard. The reason for this phenomenon may be that the key layout and size have not changed from the physical keyboard layout and size. Although physical keyboards follow a standard size due to manufacturability, this research sought to explore the relationship 33 between hand size, keyboard size and typing performance metrics. The hypotheses are listed below:

1. The typing speed in touch screen keyboards has a relationship between the hand size

and the touch screen keyboard size.

2. The typing accuracy in touch screen keyboards has a relationship between the hand

size and the touch screen keyboard size.

If the hypotheses are supported, then future keyboards could be size-adjusted to the users’ anthropometry in order to improve performance metrics. Also, sizing could be used in conjunction with other improvement methods, such as the anchored dynamic key layout [28].

CHAPTER 3. METHOD

3.1. Study Population

After obtaining the IRB approval for the study, the data collection commenced in

October, 2013. Thirty subjects were recruited from Ohio University using a flyer

(Appendix A) distributed in the university premises. Participants were 18 years or older and they did not have any hand related injuries. Recruitment was based on their hand size; small, medium, and large. Hand sizes were determined according to published anthropometric data [42]. The hand size measurements used in the study are shown in the table 3-1.

Table 3-1 Hand Size Measurements Size Hand Length (l mm) Hand Width (w mm) Small 172 > l 93.5 > w Medium 191 > l >172 103.5 > w > 93.5 Large l > 191 w > 103.5

A sketch was drawn using above hand size measurements and used to group the participants. Figure 3-1 shows the sketch used to determine participants' hand sizes.

Participants were asked to place their hand on the box to decide which hand size.

Participants were categorized as Small if their hands fit into the Small box; Medium if their hands fit into the Medium; Large if their hands went beyond the medium box.

Testing time slots were assigned according to the availability and convenience of the participant. This resulted in proportional sampling.

Figure 3-1 The sketch used to determine participants hand size

When the participant came to the testing room, first he/she was given the consent form (Appendix B) to read through and sign. After that the participant’s hand size was checked again to confirm the size. The next step was conducting the typing test using the touch screen and the survey.

3.2. Research Instrument and Scoring

The touch screen used was the Dell ST2220T. It was a multi-touch, high- definition screen and had wide angle viewing (178°). Therefore the screen was visible to the user even when it is horizontally placed flat in front of the typist. The touch keyboard was based on the Key size standards set by HFES/ANSI[34]. First, the keys were plotted in design software to understand the Key sizes and relative positions. The Microsoft

Windows 8[45] virtual keyboard was used to determine the layout (key placement) of the keys. The measurements of three different keyboard sizes are given in the table 3-2 and figure 3-2 shows three different key board sizes.

Table 3-2 Touch Screen Keyboard Dimensions Size Horizontal Key size (mm) Vertical Key size (mm) Size 1 14 14 Size 2 18 18 Size 3 22 22

(a)

(b)

Once the relative co-ordinates of the keys were decided, Microsoft Visual Studio

C# and XAML was used to create the program for the test. The program was broken down to four main windows. The participant number entry screen was used to associate a number assigned to the participant with the collected data. Once the participant number was entered the program showed the next screen with the keyboard. The sentence was picked randomly from a file which has 15 unique sentences [46] (Appendix C). The keyboard size was picked from an internal array code which has 3 randomized keyboard 38 sizes. The "Qwerty" layout was used in the touch screen keyboard. In order to minimize the participants accidently skipping sentences by pressing the enter key, the enter key was placed away from the QWERTY layout (Figure 3-3).

Figure 3-3 Layout of the keyboard

Once a key was pressed, the pressed key, corresponding sentence character and the time were stored in a comma delimited file (Comma separated Values/CSV). Once the next button was pressed, the data was directed to the next file (Figure 3-4). The folder structure and naming was maintained in order to eliminate the possible risk of confusion.

Once the participant had typed 5 consecutive sentences, the program automatically prompted a countdown of 30 seconds to provide the 30 second break. At the end of the 30 seconds, the participant had to press the "Ok" button to move to the next sentence. At the end of the 15th sentence the participant was prompted with a thank you screen to acknowledge the end of the typing exercise. 39

Figure 3-4 Program set-up

The distance between the keys was kept at a constant 3.2 mm as recommended by

ANSI [34] for all three keyboard sizes. The keyboards did not have the "backspace",

"arrow" or "delete" key. Also, the typist could not see the letters typed as each letter was replaced by an asterisk symbol (Figure 3-5). This was to monitor the number of errors that the participant made during the typing.

Figure 3-5 Sample of the touch screen while testing

The workplace was designed to accommodate the participant’s comfort, which made them focus on typing. The workstation comprised a fully adjustable ergonomic chair, height adjustable touch screen monitor, and also proper environment, such as lighting and air quality. Figure 3-6 demonstrates the work station that was used in the study. The work station was setup to give a complete silent environment to minimize the outside disturbances. The work station was in front of a large whiteboard screen to minimize any visual distractions.

(a) (b) (c) Figure 3-6 (a) Fully adjustable ergonomic chair, (b) Height adjustable touch screen monitor, and (c) Complete work station

At the end of the typing session, subjects were asked to fill out a survey of three different questions to find out the comfortableness of typing to each participant and the

System Usability Scale (SUS) questionnaire[47] to check their impression on typing on the touch screen keyboard and (Appendix D). The feedback ―1‖ indicated that the participant is ―Strongly Disagree‖ and ―5‖ as ―Strongly Agree‖.

The survey data was keyed into an excel sheet and the SUS score was calculated.

One was subtracted from the response of the question 1, 3, 5, 7, and 9. For question 2, 4,

6, 8, and 10, the participant’s response was subtracted from 5. By doing above subtraction it converted each response to a scale of 0 to 4 (0 – the most negative response 42 and 4 – the positive response). Then all converted responses for each participant were added up and multiplied by 2.5 to provide the SUS score out of a maximum score of 100.

3.3. Participant Testing

After the participant completed the consent form documents, they were given a subject number to protect their confidentiality. The subject number was entered on the start screen of the program and described the sections of the screen before typing.

Participants were then asked to type as accurately and quickly as possible for each keyboard. After typing 5 sentences, the program prompted the interval screen before the next keyboard trial. They were informed that they are allowed to take even more than 30 seconds rest if required. All participants were given the same introduction to prevent bias from the speed and accuracy. Figure 3-7 shows how a participant would type in a testing session.

Figure 3-7 Typing test demonstration 43

After the typing test, each participant was asked to complete the survey. All the survey data were securely stored in a sealed envelope with their subject number.

Immediately after successful completion of the test, the participant was given compensation of USD 20. Their signature was obtained to show that they received the compensation money. Total test duration was approximately 30 minutes.

3.4. Data Compilation

Once the testing was completed, an excel template was used to summarize the data from each CSV file generated by the program. Due to the complicated nature in typing errors, each sentence was checked manually in the excel template and highlighted so that the errors are captured by the template. Total of 450 CSV files were checked and the summary data was compiled together. In each process, "check-sum" calculations were performed to minimize errors in data handling. Survey data was entered to the same table as separate columns.

3.5. Data Analysis

In the analysis phase of the research, data accumulated from the touch screen program was checked for typing accuracy and speed. Data was analyzed using two way repeated measures Analysis of Variance (ANOVA) in SPSS Statistics V17.0 for a statistical significance of p=0.05. The three different sizes of keyboards were compared for the participants' performances. Therefore the significance of accuracy and speed on different groups on each keyboard were tested. 44

The two independent variables were hand size and keyboard size. The two dependent variables were speed and accuracy. The speed was measured by total number of characters that a participant typed per minute for each sentence. Accuracy of the typing was measured by the correct characters per minute and incorrect character percentage for every sentence. Before conducting the ANOVA test, the data was tested for normality. The between factor was hand size and the within factor was keyboard size.

The five repetitions were averaged within each keyboard.

CHAPTER 4. RESULTS

Data was obtained from 30 participants (17 male and 13 female) from Ohio

University. Ten participants with small hand size, 10 participants with medium hand size, and 10 participants with large hand size participated in the study.

4.1. Test results for Normality

All three dependent variables; characters per minute, correct characters per minute, and incorrect ratio were tested whether they are normally distributed. The test results from SPSS statistics are shown in the Appendix E. According to Shapiro-Wilk method, only characters per minute and correct characters per minute data were normally distributed but not the incorrect ratio. Therefore, the repeated measures ANOVA test was conducted only for character per minute and correct character per minute data.

As the incorrect ratio was a non-parametric variable, Kruskal-Wallis test was performed. In order to satisfy the assumptions in the test the data was tested for homogeneity before performing the Kruskal-Wallis test.

4.2. Descriptive Statistics

The average measurements of total characters typed per minute per sentence and correct number of characters per minute per sentence are the dependent variables.

Descriptive data for each main effect is shown in Figure 4-1 and 4-2.

180.0 8% 175.0 7% 170.0 6% 5% 165.0 4% 160.0 3% 155.0 2% 150.0 1% 145.0 0% 14 x 14 18 x 18 22 x 22 CPM 176.2 176.9 170.1 CCPM 163.4 167.2 158.1 ER 7% 5% 7%

Figure 4-1 Descriptive results – Main effect key size (Characters per minute (CPM), correct characters per minute (CCPM), error rate (ER))

185.0 9% 180.0 8% 175.0 7% 170.0 6% 5% 165.0 4% 160.0 3% 155.0 2% 150.0 1% 145.0 0% Small Medium Large CPM 179.5 172.5 171.1 CCPM 169.8 157.9 161.0 ER 5% 8% 6%

Figure 4-2 Descriptive results – Main effect Hand size (Characters per minute (CPM), correct characters per minute (CCPM), error rate (ER))

Detailed breakdown of the performance in each keysize and hand size are shown for Characters per minute (Figure 4-3, Appendix F), Correct Characters per minute

(Figure 4-4, Appendix G) and Error rate ( Figure 4-5, Appendix H).

200

150

per minute per 100

50 Characters Characters

0 14 x 14 18 x 18 22 x 22 Small 180.47 180.89 177.24 Medium 177.24 175.43 164.93 Large 170.92 174.25 168.04

Figure 4-3 Descriptive results –Characters per minute

200

150

100 minute per per 50 Correct Characters Characters Correct

0 14 x 14 18 x 18 22 x 22 Small 169.84 172.03 167.45 Medium 159.12 163.8 150.88 Large 161.23 165.73 155.97

Figure 4-4 Descriptive results – Correct characters per minute

12.0%

10.0%

8.0%

6.0%

4.0% Incorrect Ratio Incorrect 2.0%

0.0% 14 x 14 18 x 18 22 x 22 Small 6.0% 5.0% 5.0% Medium 10.0% 6.0% 8.0% Large 5.0% 5.0% 7.0%

Figure 4-5 Descriptive results – Incorrect ratio 49

4.3. Results of Repeated Measures Two-Way ANOVA

Using repeated measures Two-Way ANOVA, the effect of hand size and keyboard size was tested for to determine if they were significant factors for characters per minute and correct characters per minute. The level of significance for the test was set at p=0.05.

Table 4-1 shows the within-subjects effects for the fixed variable characters per minute. The Key size was significant (p<0.05), however there was no significance for the interaction factor of Key size * hand size.

Table 4-1 ANOVA Tests of Within-Subjects Effects for Characters per Minute Type III df Mean F Sig. Partial Source Sum of Square Eta Squares Squared Sphericity Assumed 842.25 2 421.13 4.64 0.014 0.147 Key Greenhouse-Geisser 842.25 1.82 461.4 4.64 0.017 0.147 size Huynh-Feldt 842.25 2 421.13 4.64 0.014 0.147 Lower-bound 842.25 1 842.25 4.64 0.040 0.147 Key Sphericity Assumed 314.9 4 78.73 0.87 0.489 0.060 size * Greenhouse-Geisser 314.9 3.65 86.25 0.87 0.482 0.060 Hand Huynh-Feldt 314.9 4 78.724 0.87 0.489 0.060 Size Lower-bound 314.9 2 157.45 0.87 0.431 0.060 Sphericity Assumed 4900.54 54 90.75 Error Greenhouse-Geisser 4900.54 49.3 99.43 (Key Huynh-Feldt 4900.54 54 90.75 size) Lower-bound 4900.54 27 181.50

There was no statistical significance for within-subjects effect of hand size. Table

4-2 shows the test result for hand size.

Table 4-2 ANOVA Tests of Between-Subjects Effects for Characters per Minute Type III Sum Mean Partial Eta Source df F Sig. of Squares Square Squared Intercept 2736626.1 1 2736626.1 566.35 1.2E-19 0.954 Hand Size 1227.2 2 613.6 0.13 0.881 0.009 Error 130464.8 27 4832

As with Characters per minute, the correct characters per minute was also statistically significant with the Key size at p<0.05. As shown in Table 4-3, the interaction factor Key size * Hand size was not statistically significant as well.

Table 4-3 ANOVA Tests of Within-Subjects Effects for Correct Characters per Minute Type III Partial Mean Source Sum of df F Sig. Eta Square Squares Squared Sphericity 1249.72 2 624.86 4.67 0.013 0.148 Assumed Key Greenhouse- 1249.72 1.81 690.12 4.67 0.016 0.148 size Geisser Huynh-Feldt 1249.72 2 624.86 4.67 0.013 0.148 Lower-bound 1249.72 1 1249.72 4.67 0.040 0.148 Sphericity 187.73 4 46.93 0.35 0.842 0.025 Key Assumed size * Greenhouse- 187.73 3.62 51.83 0.35 0.824 0.025 Hand Geisser Size Huynh-Feldt 187.73 4 46.93 0.35 0.842 0.025 Lower-bound 187.73 2 93.86 0.35 0.707 0.025 Sphericity 7221.18 54 133.73 Assumed Error Greenhouse- 7221.18 48.89 147.7 (Key Geisser size) Huynh-Feldt 7221.18 54 133.73

Lower-bound 7221.18 27 267.45

Hand size was also not statistically significant with the correct characters per minute as well. Table 4-4 shows the between-subjects test result for correct characters per minute.

Table 4-4 ANOVA Tests of Between-Subjects Effects for Correct Characters per Minute Type III Sum Mean Partial Eta Source df F Sig. of Squares Square Squared Intercept 2388103.1 1 2388103.1 633.37 2.8E-20 0.959 Hand size 2268.4 2 1134.2 0.3 0.743 0.022 Error 101802.8 27 3770.5

A pair-wise comparison was made to check for significance between the Key sizes. Table 4-5 shows the pair-wise comparison with the significance for each Key size.

Table 4-5 ANOVA Pair-wise Comparison for Keyboard Size 95% Confidence (I) Key (J) Key Mean Interval for Std. a Measure size size Difference Sig.a Difference Error (mm) (mm) (I-J) Lower Upper Bound Bound 18 x 18 - 0.64 2.12 0.763 -5 3.71 14 x 14 22 x22 6.14* 2.43 0.018 1.15 11.14 Characters 14 x 14 0.64 2.12 0.763 -3.71 5 18 x 18 per Minute 22 x 22 6.79* 2.78 0.021 1.08 12.5 14 x 14 -6.14* 2.43 0.018 -11.14 -1.15 22 x 22 18 x 18 -6.79* 2.78 0.021 -12.5 -1.08 18 x 18 -3.79 2.6 0.156 -9.11 1.54 14 x 14 22 x 22 5.3 2.9 0.078 -0.64 11.24 Correct 14 x 14 3.79 2.6 0.156 -1.54 9.11 Characters 18 x 18 22 x 22 9.09* 3.41 0.013 2.09 16.08 per Minute 14 x 14 -5.3 2.9 0.078 -11.24 0.64 22 x 22 18 x 18 -9.09* 3.41 0.013 -16.08 -2.09 52

4.4. Non-parametric tests

As the Incorrect ratio was not normally distributed, the data was tested with a non-parametric Kruskal-Wallis test. The group ranking is shown in Table 4-8 and test statistics is shown in Table 4-6 and Table 4-7 shows the Kruskal Wallis test results performed on Hand size and keyboard size and two independent variables. Results did not indicate that keyboard size or the Hand size was a significant factor in each Key size performance.

Table 4-6 Non Parametric Ranking of hand size and Key size (Kruskal-Wallis Test) N Mean Rank

Average Incorrect Small 30 43.17 Character percentage Medium 30 52.37 with Hand size Large 30 40.97 Average Incorrect 14 x 14 mm 30 47.87 Character percentage 18 x 18 mm 30 42.27 with Key size 22 x 22 mm 30 46.37

Table 4-7 Kruskal-Wallis Test statistics Hand size and Key size (Kruskal-Wallis Test) Average Incorrect Character Average Incorrect Character percentage for Hand size percentage for Key size Chi-Square 3.215 .739 df 2 2 Asymp. Sig. 0.200 0.691

4.5. Survey Results

The SUS score was calculated as reported in the method. An average SUS score of 71.67/100 was recorded. According to Usability.gov, a score of 71.67 is an above average score for usability.

Following are the three questions participants answered in the survey.

Q1. ―Before this study, I usually look at the keypad when I type using a physical keyboard.‖

Q2. ―Before this study, I usually look at the keypad when I type using a touch screen keyboard as with mobile phones and other devices (iPad etc.).‖

Q3. ―Before this study, I would only use my index finger to type most of the keys in the physical keyboard.‖

Table 4-8 shows the descriptive of SUS data and the survey data. From the mean of the different values obtained for Question 1 to 3 shows that majority of the participants looked at the virtual keyboard than the physical keyboard. Also, the majority of the participants have a tendency of using all their fingers to type in physical keyboards than using index finger to type ( hunt and peck method).

Table 4-8 Descriptive data of the survey responses Measurement Minimum Maximum Mean Std. Deviation SUS score 37.5 90 71.67 12.25 Q1 1 5 3.17 1.4 Q2 2 5 4.23 0.89 Q3 1 3 1.77 0.96 54

CHAPTER 5. DISCUSSION

For this experiment, the main effect of keyboard size was statistically significant at p <0.05 for both dependent variables, which were Characters per minute and Correct

Characters per minute. The effect of the Key size on the two dependent variables shared a similar pattern across the three different Key sizes. In terms of mean values, all three groups typed their best in the 18 x18mm keyboard followed by 14 x14 mm keyboard.

The 22 x 22 mm Key size showed lower performance than the other two Key sizes in both instances. A pair-wise comparison between the Key sizes showed that characters per minute on the 14 x 14 mm Key size was 6.14 characters per minute better than the 22 x

22 mm Key size, which was statistically significant at p = 0.018. Also, the 18x 18 mm

Key size was 6.78 characters per minute better than 22 x 22 mm Key size which was statistically significant at p = 0.021).

Characters per minute only consisted of the time factor. On the other hand, correct characters per minute took into account both time and accuracy. The 18 x 18 mm Key size showed improved results followed by the smaller 14 x 14 mm Key size and then the

22 x 22 mm Key size. The results show a significant (p=0.013) drop of mean 9.1 correct characters per minute from 18 x 18 mm size to the 22 x 22 mm. The 14 x 14 mm Key size had a slightly lower correct characters speed than the 18 x 18 mm Key size.

Characters per minute and correct characters per minute results for each keyboard suggest that the study obtained results comparable to similar studies and the specifications set by OSHA and ANSI for keyboards. In other studies, the typing speed dropped significantly when the Key size was reduced to 13 x 13 mm [35] from the 55 recommended sizes (19 +/- 1 mm) and after 22 x 22 mm there was no significant improvement in the performance [34]. A possible reason for the phenomenon might be due to the increased distance a hand needs to travel to hit the larger key. The smaller Key sizes means less distance to travel, however as the size is smaller, more time is needed to target smaller keys.

There was no statistical significance of the hand size with the result. Among three hand size groups, the small hand size group had the highest average characters per minute in all three keyboards while the medium and the large size hand groups had lower values.

The medium hand size group was performing at an average of 7 characters per minute better than the larger hand size group in the 14 x 14 mm keys, but quickly dropped to the same level as the Key size increased subsequently from 18 x 18 mm to 22 x 22 mm. The small hands may be more adaptable in moving around the keyboard variations than the other two sizes. However, the result was not statistically significant (p > 0.05) between hand sizes.

Medium hand size group seemed to have an increased error rate when they were typing in all the Key sizes than the other two groups. As the data was not normally distributed, the variable was not tested in ANOVA. Kruskal-Wallis test result showed that with respect to mean rank, the medium hand size had a higher rank for error ratio than the rest. However, the difference was not statistically significant with each Key size.

One of the critical observations in the study was that all the participants raised their hands from the keyboard when they started typing in the touch screen. I assume this effect had the medium sized hand group at a disadvantage. As physical keyboards have 56 standard sizes, the medium hand size group might have been used to keeping their palms on the keyboard or table where the small and large hand groups might have been already used to such conditions in physical keyboards. The assumption was made, as many of the medium hand size participants questioned if they can keep their hands resting on the touch screen surface during the typing test and almost immediately changed the way of typing by lifting the hands.

The interaction factor between the hand size and the keyboard size was not significant for characters per minute and the correct characters per minute. We hypothesized that if the interaction factor was significant, that keyboards could be sized for the anthropometry of the individual. Our study indicated that this was not the case and that certain Key sizes were better for all hand anthropometries.

The SUS scores did indicate that the participants had a more than average usability rating for typing on the touch screen. Also, the survey revealed that the participants felt the requirement to constantly look at the screen to find the keys in touch screen typing.

One of the limitations that I faced during this study was having limited number of participants. I had only 10 participants for each hand size. The study results may have been different if I had at least 30 participants from each hand size since that will be a better representation of each population. Also, once the limits for the hand sizes were decided, it was noted that for large hand sizes and small hand sizes the groups comprised of either male or female participants. There might be an influence from gender on the study as typing could be related to dexterity/flexibility of hands. 57

CHAPTER 6. CONCLUSION

Participants with 3 different hand sizes were asked to type on keyboards with 3 different Key sizes. The hand sizes small, medium and large were based on anthropometrical data. The three Key sizes (14 x 14 mm, 18 x 18 mm and 22 x 22 mm) were based on the previous studies and ANSI recommendations. The gap between the keys was kept the same at 3.2 mm. After typing 15 sentences on the three keyboards (5 sentences each) the characters per minute, correct characters per minute and the error ratio per character was calculated. Each participant was given a survey after the testing which included the SUS questionnaire.

Characters per minute and correct characters per minute demonstrated a normal distribution and was further tested in two-way repeated measures ANOVA and found statistical significance (p<0.05) for Key size. The typing speed on the 18 x 18 mm Key size was higher than the 14 x 14 mm Key size followed by the 22 x22 mm Key size. The effect of Key size resulted in patterns which were similar to related studies. Incorrect character ratio failed the normality test and was not tested for analysis of variance.

However, a Kruskal-Wallis test on the non-parametric data set indicated a higher ranking for medium hand size for error rate and approximately equal ranking for the small and large hand sizes. However, both hand size and Key size were not statistically significant

(p > 0.5).

The SUS questionnaire showed that the touch screen typing usability level was above average, which might suggest further improvement is needed for the maturity of the touch screen keyboards. However, in the process of improving the touch screen 58 keyboard, this research found that Key size is a significant main effect and hand size is not. Future research could benefit from this result and focus on factors which emphasize more on the Key size rather than the hand size.

REFERENCES

[1] ―Features - Touch Screen Displays SUR40 | Samsung Signage Solutions.‖ [Online]. Available: http://www.samsung.com/us/business/commercial-display- solutions/LH40SFWTGC/ZA-features [Accessed: 29-Nov-2012].

[2] ―Acer ICONIA 6120 Dual-Screen Touchbook Review - Watch CNET’s Video & Read Our Review.‖ [Online]. Available: http://www.cnet.com/laptops/acer-iconia- 6120-dual/4505-3121_7-34440624.html [Accessed: 29-Nov-2012].

[3] ―IdeaCentre Horizon 27" Table PC (US).‖ [Online]. Available: http://www.lenovo.com/products/us/desktop/ideacentre/horizon/ [Accessed: 28- Apr-2013].

[4] J. Barrett and H. Krueger, ―Performance effects of reduced proprioceptive feedback on touch typists and casual users in a typing task.,‖ Behav. Inf. Technol., vol. 13, no. 6, pp. 373–381, 1994.

[5] H. Benko, M. R. Morris, A. J. B. Brush, and A. D. Wilson, ―Insights on Interactive Tabletops : A Survey of Researchers and Developers,‖ 2009.

[6] D. R. Rashid and N. a. Smith, ―Relative keyboard input system,‖ Proc. 13th Int. Conf. Intell. user interfaces - IUI ’08, p. 397, 2008.

[7] B. Hartmann, M. R. Morris, H. Benko, and A. D. Wilson, ―Augmenting interactive tables with mice & keyboards,‖ Proc. 22nd Annu. ACM Symp. User interface Softw. Technol. - UIST ’09, p. 149, 2009.

[8] C. Sax and E. Lawrence, ―LiquidKeyboard : An Ergonomic , Adaptive QWERTY Keyboard for Touchscreens and Surfaces,‖ no. c, pp. 117–122, 2011.

[9] L. Findlater and J. Wobbrock, ―Personalized Input : Improving Ten-Finger Touchscreen Typing through Automatic Adaptation,‖ in Proceedings of the 2012 ACM annual conference on Human Factors in Computing Systems CHI 12, 2012, pp. 815–824.

[10] E. A. Johnson, ―TOUCH DISPLAY - A NOVEL INPUT/OUTPUT DEVICE FOR COMPUTERS,‖ Electron. Lett., vol. Vol. 1, no. No. 8, pp. 219–220, 1965.

[11] ―HP Virtual Museum: Hewlett-Packard-150 Touchscreen Personal Computer with Hewlett-Packard 9121 Dual Drives, 1983.‖ [Online]. Available: http://www.hp.com/hpinfo/abouthp/histnfacts/museum/personalsystems/0031/ [Accessed: 29-Nov-2012]. 60

[12] S. Bieller, ―42 nd DFF Working Group Meeting of the German Flat Panel Display Forum,‖ Solothurn , Switzerland, 2012.

[13] ―Young Adults Increasingly Favor Touch Screen Technology - Elo TouchSystems - Tyco Electronics.‖ [Online]. Available: http://www.elotouch.com/Solutions/CaseStudies/psbsurvey.asp [Accessed: 29- Nov-2012].

[14] ―Compare all Elo touch technologies - Elo TouchSystems - Tyco Electronics.‖ [Online]. Available: http://www.elotouch.com/Technologies/compare_all.asp

[15] ―Elo Touch Solutions — AccuTouch five-wire resistive touch technology.‖ [Online]. Available: http://www.elotouch.com/Technologies/AccuTouch/default.asp [Accessed: 29- Nov-2012].

[16] ―How surface capacitive touch technology works - Elo TouchSystems - Tyco Electronics.‖ [Online]. Available: http://www.elotouch.com/Technologies/SurfaceCapacitive/howitworks.asp [Accessed: 29-Nov-2012].

[17] ―How IntelliTouch/SecureTouch surface wave touch technology works - Elo Touch Solutions.‖ [Online]. Available: http://www.elotouch.com/Technologies/IntelliTouch/howitworks.asp [Accessed: 29-Nov-2012].

[18] ―How acoustic pulse recognition touch technology works - Elo TouchSystems - Tyco Electronics.‖ [Online]. Available: http://www.elotouch.com/Technologies/AcousticPulseRecognition/howitworks.as p [Accessed: 29-Nov-2012].

[19] ―How CarrollTouch infrared touch technology works - Elo Touch Solutions.‖ [Online]. Available: http://www.elotouch.com/Technologies/CarrollTouch/howitworks.asp [Accessed: 29-Nov-2012].

[20] T. R. Alex Teiche, Ashish Kumar Rai, Christian Moore, Donovan Solms, Görkem Çetin, Jimi Hertz, Laurance Müller, Lynn Marentette, Nolan Ramseyer, Paul D’intino, Ramsin Khoshabeb, Rishi Bedi, Seth Sandler, Taha Bintahir, Thomas Hansen, Multitouch Technologies, 1.0 ed. 2009, pp. 1–47.

[21] ―The Power of PixelSenseTM.‖ [Online]. Available: http://www.microsoft.com/en- us/pixelsense/pixelsense.aspx [Accessed: 29-Nov-2012]. 61

[22] ―Emulator | SmithsonMartin Inc.‖ [Online]. Available: http://www.smithsonmartin.com/products/emulator/ [Accessed: 29-Nov-2012].

[23] U. Hinrichs, M. Hancock, S. Carpendale, and C. Collins, ―Examination of Text- Entry Methods for Tabletop Displays,‖ Second Annu. IEEE Int. Work. Horiz. Interact. Human-Computer Syst., pp. 105–112, Oct. 2007.

[24] H. Richter, S. Loehmann, F. Weinhart, and A. Butz, ―Comparing Direct and Remote Tactile Feedback on Interactive Surfaces,‖ in EuroHaptics’12 Proceedings of the 2012 international conference on Haptics: perception, devices, mobility, and communication, 2012, pp. 301–313.

[25] C. McAdam and S. Brewster, ―Distal tactile feedback for text entry on tabletop computers,‖ in Proceedings of the 23rd British HCI Group Annual Conference on People and Computers: Celebrating People and Technology, 2009, pp. 504–511.

[26] J. G. Young, M. Trudeau, D. Odell, K. Marinelli, and J. T. Dennerlein, ―Touch- screen tablet user configurations and case-supported tilt affect head and neck flexion angles.,‖ Work, vol. 41, no. 1, pp. 81–91, Jan. 2012.

[27] ―Out of Touch with Typing | MIT Technology Review.‖ [Online]. Available: http://www.technologyreview.com/view/425018/out-of-touch-with-typing/ [Accessed: 29-Nov-2012].

[28] A. Gunawardana, T. Paek, and C. Meek, ―Usability guided key-target resizing for soft keyboards,‖ Proc. 15th Int. Conf. Intell. user interfaces - IUI ’10 , p. 111, 2010.

[29] ―Early Typewriter History.‖ [Online]. Available: http://www.mit.edu/~jcb/Dvorak/history.html [Accessed: 29-Nov-2012].

[30] D. M. Licht, D. J. Polsella, and K. R. Boff, ―Human Factors, Ergonomics, and Human Factors Engineering: An Analysis of Definitions,‖ no. 513.

[31] P. Tittiranonda, D. Rempel, T. Armstrong, and S. Burastero, ―Effect of four computer keyboards in computer users with upper extremity musculoskeletal disorders.,‖ Am. J. Ind. Med., vol. 35, no. 6, pp. 647–61, Jun. 1999.

[32] ―Enabling Access : Blog posts labelled: split keyboards.‖ [Online]. Available: http://enablingaccess.ca/blog.html/categories/split+keyboards

[33] ―OSHA Ergonomic Solutions: Computer Workstations eTool - Components - Keyboards.‖ [Online]. Available: http://www.osha.gov/SLTC/etools/computerworkstations/components_keyboards. html#design [Accessed: 28-Apr-2013].

[34] ―Human Factors and Ergonomics Society: Product Detail.‖ [Online]. Available: http://www.hfes.org/Publications/ProductDetail.aspx?ProductID=69

[35] J. H. Kim, L. Aulck, O. Thamsuwan, M. C. Bartha, and P. W. Johnson, ―The Effects of Virtual Keyboard Key sizes on Typing Productivity and Physical Exposures,‖ Proc. Hum. Factors Ergon. Soc. Annu. Meet., vol. 57, no. 1, pp. 887– 891, Sep. 2013.

[36] M. E. Sesto, C. B. Irwin, K. B. Chen, a. O. Chourasia, and D. a. Wiegmann, ―Effect of Touch Screen Button Size and Spacing on Touch Characteristics of Users With and Without Disabilities,‖ Hum. Factors J. Hum. Factors Ergon. Soc., vol. 54, no. 3, pp. 425–436, Feb. 2012.

[37] B. Chaparro, B. Nguyen, M. Phan, A. Smith, and J. Teves, ―Keyboard Performance: iPad versus Netbook,‖ Usability News, vol. 12, pp. 1–9, 2010.

[38] S. Ko, K. Kim, T. Kulkarni, and N. Elmqvist, ―Applying Mobile Device Soft Keyboards to Collaborative Multitouch Tabletop Displays: Design and Evaluation,‖ ITS’11, Novemb. 13-16, Kobe, Japan, 2011.

[39] A. Sears, D. Revis, J. Swatski, R. T. Crit, and B. Shneiderman, ―Investigating Touchscreen Typing : The effect of keyboard size on typing speed,‖ Behav. Inf. Technol., vol. 12, pp. 17–22, 1993.

[40] S. Pheasant and christine m. Haslegrave, BODYSPACE, 3rd ed. Taylor and francis, 2006, p. 134.

[41] V. Balakrishnan and P. H. P. Yeow, ―Hand anthropometry and SMS satisfaction,‖ J. Appl. Sci., vol. 8, pp. 816–822, 2008.

[42] A. K. Karlson and B. B. Bederson, ―ThumbSpace : Generalized One-Handed Input for Touchscreen-Based Mobile Devices,‖ Ifip Int. Fed. Inf. Process., pp. 324–338, 2007.

[43] M. S. M. Sun and X. R. X. Ren, ―An Evaluation of Thumb Interface for Menu Selection on Mobile Equipment,‖ Innov. Comput. Inf. Control ICICIC 2009 Fourth Int. Conf., pp. 805–809, 2009. 63

[44] ―Our Research.‖ [Online]. Available: http://www.steinbuhler.com/html/our_research.html [Accessed: 06-Jul-2013].

[45] ―Shop Windows | Microsoft Store.‖ [Online]. Available: http://www.microsoftstore.com/store/msusa/en_US/cat/Windows/categoryID.6268 4800?tid=sJhVysIYV_dc&cid=5250&pcrid=25485035573&pkw=microsoft windows 8&pmt=e&WT.srch=1&WT.mc_id=pointitsem_Microsoft+US_google_5+- +Windows+8&WT.term=microsoft windows 8&WT.campaign=5+- +Windows+8&WT.content=JhVysIYV&WT.source=google&WT.medium=cpc

[46] ―Grimm’s Household Tales.‖ [Online]. Available: http://www.sacred- texts.com/neu/grimm/index.htm

[47] B. W. Patrick W. Jordan, B. Thomas, Ian Lyall McClelland, ―Usability Evaluation In Industry,‖ in in Usability Evaluation In Industry, 1 Edition., CRC Press, 1996, pp. 189–194.

APPENDIX A. STUDY FLYER TOUCH SCREEN TYPING STUDY!

If you are

 Age between 18-30  Proficient in keyboard typing  Not suffering from hand or wrist injuries

You will be rewarded with a $20 for just 30 minutes of your time!

Get registered! contact:

Warnaka @ 281.691.**** (******@gmail.com) 65

APPENDIX B. CONSENT FORM

OHIO UNIVERSITY CONSENT FORM

Title of Research: Touch Screen Typing: Use of Hand Anthropometry to Determine the preferred Keyboard Size.

Researchers: Warnaka Rajeew Gunawardena ( Masters Student) Dr. Diana Schwerha ( Advisor)

You are being asked to participate in research. For you to be able to decide whether you want to participate in this project, you should understand what the project is about, as well as the possible risks and benefits in order to make an informed decision. This process is known as informed consent. This form describes the purpose, procedures, possible benefits, and risks. It also explains how your personal information will be used and protected. Once you have read this form and your questions about the study are answered, you will be asked to sign it. This will allow your participation in this study. You should receive a copy of this document to take with you.

Explanation of Study

The outcome of the project could improve tabletop touch screen keyboards. A tabletop is a large touch screen laid horizontally which resembles a table. The virtual keyboards on these devices lack tactile feedback and make it impossible to touch type. This research aims to find a relationship between the user hand size and the typing speed on the touch screen. This research will help to design better touch screen keyboards.

The description of the project is listed below:

1. You will be seated and the height of the table top touch screen and it will be adjusted so that you are in a neutral (ergonomic; wrists straight) posture.

2. You will be asked to type on 3 different touch screen keyboards (different sizes).

3. You will be asked to type 15 different passages (3 keyboard sizes, with 5 passages per size). We anticipate that the each passage will be 20-30 words. You will get a 30 second break in between each passage.

4. We anticipate that the total testing time would be 30/40 minutes after the initial documentation session.

5. After the 15 passages are typed, you will be asked to complete a brief survey about each size of keyboard. 66

6. You will be given $20 at the end of successful test completion. You may elect to stop the experiment anytime, however the reward amount would be prorated.

You should not participate in this study if you have any medical conditions ( Such as Carpel Tunnel) related to hand and wrists as you would be at risk of increased discomfort and higher risk of injury.

Your participation in the study will last approximately 30 minutes after the initial documentation session.

Risks and Discomforts

You might feel discomfort in your hands and wrists due to typing.

Benefits

If the hypothesis is proven, better touch screen keyboard design for table top computers would be possible.

Confidentiality and Records

Your study information will be kept confidential by removing any links between you and the test results.

Additionally, while every effort will be made to keep your study-related information confidential, there may be circumstances where this information must be shared with: * Federal agencies, for example the Office of Human Research Protections, whose responsibility is to protect human subjects in research; * Representatives of Ohio University (OU), including the Institutional Review Board, a committee that oversees the research at OU;

Compensation

As compensation for your time/effort, you will receive USD 20 at the end of the session. Contact Information

If you have any questions regarding this study, please contact below mentioned

Dr. Diana Schwerha Associate Professor Phone: (xxx) xxx-xxxx [email protected]

Warnaka Gunawardena Phone: (xxx) xxx-xxxx [email protected] 67

If you have any questions regarding your rights as a research participant, please contact Jo Ellen Sherow, Director of Research Compliance, Ohio University, (740)593-0664.

By signing below, you are agreeing that:  you have read this consent form (or it has been read to you) and have been given the opportunity to ask questions and have them answered  you have been informed of potential risks and they have been explained to your satisfaction.  you understand Ohio University has no funds set aside for any injuries you might receive as a result of participating in this study  you are 18 years of age or older  your participation in this research is completely voluntary  you may leave the study at any time. If you decide to stop participating in the study, there will be no penalty to you and you will not lose any benefits to which you are otherwise entitled.

Signature Date

Printed Name

Version Date: [2/2/13]

APPENDIX C. TEST SENTENCES

Sentences used in the touch typing testing

1. It happened that the eldest wanted to go into the forest to hew wood, and before

he went his mother gave him a beautiful sweet cake and a bottle of wine.

2. They were once sitting thus when the little grandson of four years old began to

gather together some bits of wood upon the ground.

3. There was once upon a time an old goat who had seven little kids, and loved them

with all the love of a mother for her children.

4. The king ordered all the goldsmiths to be brought to him, and they had to work

night and day until at last the most splendid things were prepared.

5. There were once upon a time a king and a queen who lived happily together and

had twelve children, but they were all boys.

6. The king rejoiced when he heard that she was innocent, and they all lived in great

unity until their death.

7. The two children had also not been able to sleep for hunger, and had heard what

their step-mother had said to their father.

8. Then she seized Hansel with her shriveled hand, carried him into a little stable,

and locked him in behind a grated door.

9. Do you think that could be anything to a man who has struck down seven at one

blow. I leapt over the tree because the huntsmen are shooting down there in the

thicket. 69

10. When she knew the answer to the riddle she wanted to steal away, but he held her

mantle so fast that she was forced to leave it behind her.

11. The maiden took the drumstick, wrapped it carefully in a cloth, and went onwards

again until she came to the glass mountain.

12. The next morning the wolf sent the boar to challenge the dog to come out into the

forest so that they might settle the affair.

13. They traveled up and down, and at last they came into a kingdom where and old

king reigned who had a single but wonderfully beautiful daughter.

14. When they were quite out of sight the poor man got down from the tree, and was

curious to know what was secretly hidden in the mountain.

15. There was once upon a time a shepherd boy whose fame spread far and wide

because of the wise answers which he gave to every question.

APPENDIX D. STUDY SURVEY

TOUCH SCREEN TYPING STUDY SURVEY

Following questionnaire will help the research determine overall usability of touch screen keyboards.

Please check a box to indicate your rating for each question.

1. Before this study, I usually look at the keypad when I type using a physical keyboard.