"Arts, Faculty of"@en . "Psychology, Department of"@en . "DSpace"@en . "UBCV"@en . "Li, Hiroe"@en . "2011-03-02T03:04:49Z"@en . "2007"@en . "Master of Arts - MA"@en . "University of British Columbia"@en . "Personal Digital Assistants (PDAs) are mobile devices that offer a range of applications and are used in various environments (e.g., finding contact information driving while). These environments and applications vary on the level of attentional demands. The key interests of the present study were to explore the interference that occurs when both the PDA task and environment are highly attention demanding. The main goal of the present study was to investigate the attentional demands of two types of PDA tasks: Navigation and data entry. Using a dual-task methodology, I conducted two experiments that explored the amount of attention (Experiment 1 and 2), and two experiments that investigated the types of attention (Experiment 3 and 4), required by the two PDA task types. For the first two experiments, a tone discrimination task was chosen as the secondary task as it has been shown to require general attentional resources. Participants first completed the tone discrimination task alone in order to assess performance in the baseline condition. In the test phase, participants completed a set of PDA tasks concurrently with a tone discrimination task. To assess the type of attention required by PDA tasks, a method used to reveal the types of attention was first validated in Experiment 3. The validated method was used in Experiment 4. Participants completed a task that either drew on visuo-spatial resources or articulatory/auditory resources concurrently with either a PDA navigation or data entry task. The two main findings of the 4 experiments were: Navigation requires more attention than data entry; data entry requires more articulatory/auditory resources while navigation requires both articulatory/auditory and visuo-spatial resources, but more of the latter."@en . "https://circle.library.ubc.ca/rest/handle/2429/31888?expand=metadata"@en . "A T T E N T I O N A L D E M A N D S O F D I F F E R E N T T Y P E S O F P D A T A S K S by H I R O E LI B.A. , University of British Co lumb ia , 2005 A T H E S I S S U B M I T T E D IN P A R T I A L F U L F I L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F M A S T E R O F A R T S in T H E F A C U L T Y O F G R A D U A T E S T U D I E S (Psychology) T H E U N I V E R S I T Y O F BRIT ISH C O L U M B I A August 2007 \u00C2\u00A9 Hiroe L i , 2007 Abstract Persona l Digital Ass is tants (PDAs ) are mobile dev ices that offer a range of appl icat ions and are used in var ious environments (e.g., f inding contact information driving while). T h e s e environments and appl icat ions vary on the level of attentional demands . The key interests of the present study were to explore the interference that occurs when both the P D A task and environment are highly attention demanding. The main goal of the present study was to investigate the attentional demands of two types of P D A tasks: Navigat ion and data entry. Us ing a dual- task methodology, I conducted two exper iments that explored the amount of attention (Experiment 1 and 2), and two exper iments that investigated the types of attention (Exper iment 3 and 4), required by the two P D A task types. For the first two exper iments, a tone discrimination task was chosen as the secondary task as it has been shown to require general attentional resources. Part ic ipants first completed the tone discrimination task alone in order to a s s e s s performance in the basel ine condit ion. In the test phase , participants completed a set of P D A tasks concurrently with a tone discrimination task. To a s s e s s the type of attention required by P D A tasks, a method used to reveal the types of attention w a s first val idated in Exper iment 3. The val idated method was used in Exper iment 4. Part icipants completed a task that either drew on visuo-spatial resources or articulatory/auditory resources concurrently with either a P D A navigation or data entry task. The two main f indings of the 4 experiments were: Navigat ion requires more attention than data entry; data entry requires more articulatory/auditory resources while navigation requires both articulatory/auditory and visuo-spat ial resources, but more of the latter. ii Table of Content Abstract ii Table of Content iii List of Tab les v List of Figures vi Acknowledgements vii Introduction 1 Experiment 1 10 Method 10 Part icipants 10 Apparatus 10 Materials 11 Tone Discrimination Task 11 P D A T a s k s 12 Neuropsycholog ica l Tests 13 Design 13 Procedures 14 Resul ts 15 Data Preparat ion 15 P D A T a s k s 17 Tone Discrimination Task 18 Discuss ion 20 Exper iment 2 23 Method 24 Part icipants 24 Apparatus 24 Tone Discrimination Task 24 P D A T a s k s 25 Design 26 Procedures 26 Resul ts 27 Data Preparat ion 27 P D A T a s k s 28 Tone Discrimination Task 30 Discuss ion 33 Experiment 3 36 Methods 37 Part icipants 37 Materials 37 Input T a s k s 38 Letter Trac ing Task '. 38 Sen tence Dec is ion Task 39 Output T a s k s ; 40 Point ing 40 Tapp ing 40 Say ing 40 Neuropsycholog ica l Tests 40 Design 41 iii Procedures 41 Resul ts 4 2 Data Preparat ion 4 2 Input and Output T a s k s 4 3 Discuss ion 4 4 Experiment 4 4 7 Part ic ipants 4 7 Apparatus 4 7 Input T a s k s 4 7 Output T a s k s 4 7 P D A T a s k s 4 8 Design 4 ^ Procedures 4 ^ Resul ts 5 0 Data Preparat ion 50 P D A T a s k s 5 1 Secondary T a s k s 55 Discuss ion 56 Genera l D iscuss ion 59 Conc lus ion 67 References 8 7 iv List of Tab les Table 1 The number of Opt imal S teps /P resses for each P D A Task as an Indicator of Task Difficulty 68 Table 2 Percentage of Correct , Incorrect, and Adjusted Identifications in the Tone Discrimination Task ac ross Condi t ions 69 Table 3 The Number of S t e p s / P r e s s e s Required by each P D A Task across Tone Discrimination Task Difficulty 70 Table 4 Percentage of Correct , Incorrect, and Adjusted Identifications in the Tone Discrimination Task across Task Condit ions and Tone Discrimination Difficulty 71 Table 5 The number of S t e p s / P r e s s e s Required by each P D A Task A c r o s s Condi t ions 72 v List of Figures Figure 1. Picture of the dev ice used for the experiment 73 Figure 2. R e s p o n s e t ime (mean of median) for correct identification of target tones across condit ions 74 Figure 3. Percentage of errors in data entry and navigation tasks as a function of tone discrimination difficulty 75 Figure 4. R e s p o n s e time (mean of medians) of correct identifications in the tone discrimination task as a function of tone discrimination difficulty and task condi t ions. . . 76 Figure 5. Stimuli for the practice trials of the letter tracing task 77 Figure 6. Stimuli for the test trials of the letter tracing task 78 Figure 7. Illustration of a trial of the letter tracing task with top/bottom instructions 79 Figure 8. Illustration of a trial of the letter tracing task with outside instructions 80 Figure 9. Image of s taggered Y and N of the output task (pointing) 81 Figure 10. M e a n complet ion time as a function of input task and output task 82 Figure 11. Percentage of P D A task errors as a function of input task, output task, and P D A task type 83 Figure 12. Typing speed (words per minute) in data entry tasks as a function of input task, and output task 84 Figure 13. Complet ion time in navigation tasks as a function of input task, and output task 85 Figure 14. Complet ion t ime of the concurrent tasks as a function of input task, output task, and P D A task type 86 vi Acknowledgements Many people have helped me to accompl ish this work. I a m deeply grateful to my supervisor, Peter Graf, for his time, pat ience, gu idance throughout the entire process. I would like to thank his valuable comments on my writing, constant encouragement to present my work at var ious conferences, and mentorship in guiding me in my work. He has taught me, and cont inues to teach me, how to use words carefully, how to express myself clearly, and how to organize my thoughts logically. I would like to thank my committee members , J o a n n a McGrene re and Ron Rensink, for their thoughtful considerat ions of my thesis and for lending their time and expert ise to help me improve it. I'd like to thank Gery Fung and Magg ie Sekhon for their careful and dedicated data entry. I a lso want to thank Ralph Hakst ian and Jeremy B iesanz for teaching me the statistics knowledge that I needed in order to perform the relevant ana lyses for this thesis work. I would to thank my friend and lab mate, Carr ie Cuttler, for helping me with my Engl ish, providing valuable feedback, and sharing her exper iences about academic life with me. I want to thank my lab mate and friend, J e s s G a o , for shar ing her knowledge about conduct ing exper iments with me, and providing me adv ice in handling living related matters. I would a lso like to thank Daniel S i u , my lab mate and friend, for cheering me up when I was feeling upset, for continually urging me to persist, for showing constant care for my wel l-being, and for teaching me how to become a better person. I would like to thank Lynn Fontanil la for l istening to me rambling on about this project for two years . I a lso want to thank all of my family in Hong Kong and Vancouver , for providing generous help to me. I a lso want to thank everyone in my family for their love and support. I especia l ly thank my mom and dad , for their uncondit ional support during vii these two years, for their understanding when I didn't regularly spend time with them. Finally, I would like to thank my sister, Yuk ie L i , for bringing me count less moments of joy when we spent t ime together, and for assist ing me with the household chores when I got too busy. I a lso would like to thank the El izabeth Young Lacey foundat ion, and my supervisor, Peter Graf, for believing in my potential to f inancial ly support me. 1 Introduction Persona l Digital Ass is tants ( P D A s ) are mobile dev ices that are equipped with various appl icat ions and used in a wide range of environments. For example , these devices are used to look up route information to navigate around a city (Goodman , Gray, Khammampad & Brewster, 2004), to look up and a c c e s s medical reference information (Embi , 2001), to write and send emai l messages , and to store contact information (e.g., a phone number). S o m e examples of environments in which these appl icat ions are used include a busy sidewalk, a quiet room, and a bright, sunny day on the beach. The nature of the environments in which P D A s are used and the applications for which P D A s are used differ widely. O n e of the most important var iables that distinguish between the var ious environments and appl icat ions is the attentional demands . Environments and appl icat ions differ with respect to their attentional demands . For example, a busy street is more attention demanding than a quiet room. Looking up a stored document in a directory is more attention demanding than pressing a button to turn off the dev ice. The overall goal of this thesis was to learn about the interactions of the attentional demands between various environments and appl icat ions. There are severa l possib le combinat ions of environments and appl icat ions in relation to their level of attentional demands . First, both the environment and application can require little or no attention. One such example can be recording a s imple voice message on a dev ice while walking on a quiet street. The second combinat ion occurs when either the environment or the P D A application is attention demanding. One of the examples is trying to type a formal email (i.e., high in attention demand) in a quiet room (i.e., low attention demand) . The focus of this thesis is to explore the third combinat ion, 1 when both the environment and application are highly attention demanding, for example, typing a formal emai l while walking on a busy sidewalk. O n e of the factors that determine the level of attentional demands of an activity is the extent to which cognit ive control is required to execute that activity. Schne ider and Shriffin (1977) dist inguished between two types of cognit ive p rocesses , automatic and control led, that var ies on the amount of attentional demands . Automat ic p rocesses require little or no attention. Th is is because the execut ion of the activity that relies on automatic p rocesses draws on previously learnt responses . A s a result, automatic p rocesses can operate in combinat ion with other activities that are attention demanding. Activit ies that rely on automatic processing require extensive training and practice. In contrast, controlled p rocesses are initiated through effort and cognit ive control. Due to the fact that these p rocesses require substantial attention, they will interfere with other activities which a lso require controlled process ing. The interference observed while combining activities that are high in attentional demands points to a general limited capacity in attentional resources. Two different c lasses of attention theories have been proposed to explain var iance in interference during multi-tasking: S ing le resource theories (Kahneman, 1973) and multiple resource theories (Parasuraman & Dav ies , 1984). Single resource theories propose that there is only one general reserve of attention that can be al located amongst various concurrent tasks or activities. The extent of interference will depend in part on the load which each of the activities imposes and the allocation of attention among the activities (Moray, 1967). A s the primary task demands more of these resources (i.e., becomes more difficult) fewer are avai lable for a concurrent or secondary task, and performance on the latter task deteriorates accordingly. Primary-task workload would be inversely reflected 2 in secondary- task performance, depending on the al location of attention toward the primary activity (Moray, 1967). The multiple resource theory suggests that there is more than one resource for processing information. O n e of the most wel l-known multiple resource theories was formulated by Badde ley and Hitch (1974). They proposed a tripartite system that involves an attentional controller (central executive) ass is ted by two s lave systems: The visuo-spatial sketchpad and the phonological loop. The central execut ive is responsible for coordinating information from the s lave sys tems. The central execut ive is assumed to function like an attentional sys tem that selects and operate control p rocesses and strategies. The visuo-spat ia l sketchpad is used to hold and manipulate visual images. The phonological loop is used to hold and manipulate speech-re lated information. Accord ing to this theory, two activities that require the s a m e types of attention would show task interference. That is, there should be observable decrements in performance on an activity executed singly compared to tending that activity with a second activity that a lso requires the s a m e resources. Interference between the environment and appl icat ions can a lso arise when there is competit ion for the s a m e sensory/perceptual p rocesses . A n example includes trying to find a phone number (i.e., application) while driving (i.e., environment). Both activities demand the visual modality and thus it is necessary to switch between the two activities in order to handle both tasks concurrently. In other words, the efficiency of performing both activities is limited at the speed in which vision can be switched between the two activit ies. The limits of attending to both activities are restricted at the visual level of p rocess ing. There is a large area of research that investigates the interference that occurs due to competit ion for the same sensory/perceptual p rocesses 3 (Broadbent 1957, 1958). However, this area is not of interest to the present thesis and will not be d i scussed further. Previous Research Prior research on attention and technology use has general ly focused on exploring the effects of individual dif ferences in attention on technology usability. For example, des ign guidel ines are general ly targeted towards accommodat ing individuals with reduced attentional capacity, such as older adults (Connel ly & Hasher , 1993; Kotary & Hoyer, 1995). Morris and Venkatesh (2000) noted that large amounts of information are usual ly presented on small d isplays (e.g., mobi le devices) , and that could be problematic for older adults who have difficulties in handling a lot of information at the same time and sorting out task relevant information. They suggested that interfaces with reduced information content make it eas ier to focus attention on relevant information and reduce the time spent on information search . A id in focusing attention can be provided by structuring the information, providing spatial and temporal cues , and manipulating the screen layout. Guide l ines of interface des ign have a lso included recommendat ions to exc lude graphic details that may be decorat ive to prevent distraction (Hawthorne, 2000). Individual di f ferences in certain cognitive abilities have been shown to be related to information search performance on a wide range of technologies. Severa l studies have found that individuals with lower spatial ability and poorer vocabulary skills take longer to retrieve information from a hierarchical da tabase on a computer (Freudenthal, 2001; V incente , H a y e s & Wi l l iges, 1987). This effect due to spatial ability was prominent even after taking into account prior exper ience with technology (Vincente, Hayes & Wil l iges, 1987). Individuals with poorer spatial ability performed less efficient searches on a cell phone (Ziefle & Bay, 2004). On searching for information from the Wor ld Wide 4 W e b , individuals with poorer spatial ability took longer to find the relevant information (Dahlback, Hook & Sjol inder, 1996). Lower mental rotation, verbal and visual memory performance was l inked to greater time spent on the task in virtual reality navigation (Moffat, Zonderman & Resn ick , 2001). Data entry performance on a ful l-sized keyboard has been linked with basic cognitive abilities in a study by Cza ja and Sharit (1998). A g e , process ing speed , motor skil ls, v isuo-spat ial ski l ls and prior computer exper ience have been found to have an impact on entering data into records or search f ields. A m o n g these factors, visuo-motor skills and memory predicted the number of typing errors above and beyond prior exper ience with computers. In a recent study (Li and Graf, 2007), certain cognit ive abilit ies were found to predict data entry performance on a P D A . A m o n g sensory abilit ies, episodic memory, perceptuo-motor ski l ls, and verbal intell igence, we found that sensory abilities and episodic memory were the stronger predictors for different types of data entry errors on a P D A . The finding that verbal intell igence had no predictive power is not surprising in view of the fact that none of the data entry tasks were des igned to chal lenge verbal skills or to require extensive language processing. Objectives and Motivations The motivation for this research was to learn more about the factors that influence the usability of P D A s in order to increase the usability of these dev ices. The majority of research and design guidel ines have focused on providing support for individuals with poorer cognit ive abilities. However, the usability of technologies can be compromised even for individuals with relatively better cognit ive abilities in certain situations. If the appl icat ion is inherently very attention demand ing , then usability is reduced when these appl icat ions are used in environments that are a lso attention 5 demanding, for example , recording a complex voice m e s s a g e while driving in busy traffic. Moreover , if both the environment and application demand the same type of attention, then usability can also be compromised. A s a first step to understanding the interference that might be produced in these situations, one of the objectives of this thesis is to learn more about whether certain P D A tasks are more attention demanding than others, and about the type of attention that they require. The second motivation for this research was to examine the relationship between attention and P D A usability in order to guide design speci f icat ions for these devices to be more suitable for the var ious environmental demands . Whi le it has certainly been inferred that attention plays a crucial role in the usability of technology (Hawthorn, 2000), there have been few studies that empirically demonstrated a link between attention and usability. In addit ion, while there is a plethora of des ign guidel ines on designing websi tes, cell phone menus, and interfaces of var ious software appl icat ions and computer sys tems, the guidel ines may not be appl icable to P D A s . P D A s generally have different s ized sc reens and rely on a different interaction technique (i.e., touchscreen using stylus). Thus , I bel ieve that is a lso important to generate empirical data that serves as a start for basing design guidel ines for P D A s . Contributions This type of research serves to identify whether future research needs to be focused on reducing the attentional demands of certain types of P D A tasks in order to maximize the effective and efficient use of these dev ices . O n e of the limitations of previous studies is that they often only report either the complet ion time (an index of efficiency), or number of errors (an index of efficacy). Thus , it is difficult to ascertain whether the individuals exper ienced a speed-accuracy trade-off. In addit ion, the studies that reported the number of committed errors are expressed as absolute values. A s a 6 result, it is difficult to interpret whether the error p resses were cons idered high or low in relation to the total number of p resses . Moreover, the criteria used to classify errors were unclear. The exper iments in this thesis addressed the three limitations mentioned earlier. The exper iments used clearly developed criteria to categor ized errors, used standardized methods to calculate the data, and reported indexes of both efficiency and effect iveness. This research deve loped and validated a dual-task methodology to a s s e s s the attentional demands of technology use. Whi le this methodology has been used in a variety of human factor studies (see Wickens , 1992 for a review) and memory studies (Craik, Govon i , Naveh-Ben jamin & Anderson , 1996; Fe rnandes & Moscov i tch , 2000; Naveh-Ben jamin , Craik, G u e z , Dori, 1998), this method has not yet been applied to specif ical ly a s s e s s the attentional demands of P D A tasks. The des ign of the secondary task must meet severa l criteria. First, the secondary task must be des igned to a s s e s s interference produced by the attentional demands of the P D A tasks, and not interference produced by competit ion of the s a m e perceptual p rocesses of the P D A tasks. S e c o n d , the secondary task must include measurable performance var iables. O n c e val idated, this method can a lso be used to a s s e s s the attentional demands of other technological dev ices, such as cell phones, mp3 players, and appl icat ions on personal computers. Overview The overal l goal of this thesis is to understand the attentional demands of P D A tasks. Whi le there are many appl icat ions that can be used on a P D A , the majority can be classif ied in one of two types of tasks: Navigation and data entry. Navigation refers to locating a certain file, folder or device information through the menu options. Data 7 entry refers to entering data (i.e., a specif ic phrase or numbers) via a touch-screen keyboard using a stylus. The dual-task methodology was used as the general set-up for measur ing the attentional demands of performing P D A tasks. The methodology involves having an individual perform two tasks concurrently (i.e., divided attention condition), and to measure task interference relative to single-task performance (i.e., full attention condition). In my exper iments, the P D A tasks were a lways designated as the primary task. Therefore, changes in secondary task performance were attributed to the reduced availability of attention due to al location of attentional resources to the P D A tasks. For exper iment 1 and 2, I chose tone discrimination as the secondary task s ince this task has been shown to involve the central execut ive (Klauer & Stegmaier , 1997). The second reason is that this task was des igned to avoid the s a m e perceptual p rocesses used by the P D A tasks (i.e., auditory and vocal for tone discrimination; motor and vision for P D A tasks). Th is arrangement ensured that the tone discrimination task a s s e s s e d the attentional demands of P D A tasks, as opposed to interference produced by competit ion for the s a m e modalit ies. Third, the tone discrimination task offered observable performance var iables which could be measured objectively, which al lowed compar isons of any changes in performance when tone discrimination was performed with the P D A tasks. I des igned and conducted two experiments to explore whether data entry and navigation tasks require different amounts of attention. In Exper iment 1, the secondary objective was to adapt the dual-task methodology to explore the attentional demands of the P D A tasks. Part ic ipants were asked to first perform the tone discrimination task alone to get a basel ine assessmen t , followed by performing a ser ies of navigation or data entry tasks concurrently with the tone discrimination task. 8 Fol lowing up on the f indings from Experiment 1, Exper iment 2 was designed to determine whether the results obtained in Exper iment 1 were due to attentional demands or due to the difficulty of switching between the tone discrimination task and P D A tasks. Part ic ipants first completed an easy and hard vers ion of the tone discrimination task in the basel ine condit ion. In the test condit ions, participants completed the easy and hard version of the tone discrimination task with either a set of data entry and navigation tasks on the P D A . Exper iment 3 and 4 were des igned and conducted to explore whether performing the data entry and navigation tasks require different types of attention. The primary objective of Exper iment 3 was to confirm that a method deve loped by Brooks (1968) is valid and reliable for revealing the types of attention, specif ical ly v isuo-spat ial and auditory/articulatory attention. The same tasks, a task that draws on visuo-spatial resources and a task that requires articulatory/auditory resources, descr ibed in the original study were used . For Exper iment 4, the method validated in Exper iment 3 was employed to a s s e s s the types of attention required by the P D A tasks. Part icipants performed concurrently either a navigation or data entry task with one of the two tasks that require different types of attention descr ibed in Exper iment 3. 9 Exper iment 1 The primary objective of this study was to explore the attentional demands of two P D A task types: Navigat ion and data entry. A dual-task methodology was used to a s s e s s the attentional demands of each type of task. Whi le the dual-task methodology has been used successfu l ly in other areas of research, it has been never been employed for assess ing the attentional demands of P D A tasks. Thus , an additional objective of the present study was to adapt the dual-task methodology to the exploration of the attentional demands of P D A tasks. Method Participants Twenty-six undergraduate students were recruited through the subject pool in the psychology department at the University of British Co lumb ia . They were compensated with one course credit in return for their participation. The experiment was conducted with the approval of the University of British Co lumb ia behavioral ethics review board. Apparatus A n unmodif ied Hewlet t -Packard i P A Q rx3715 handheld computer was used for this experiment [see Figure 1]. Th is device has a color sc reen which is 2.26 inches wide and 3.02 inches high. A s shown in Figure 1, five hardware buttons are posit ioned below the screen. The button in the middle is a navigation key. The four other buttons are designed for access ing different appl icat ions and P D A status information. To interact with the dev ice, users either press these buttons or use a stylus to select icons or menu options on the sc reen . Data entry is done via a touch screen Q W E R T Y keyboard using a stylus. On this sc reen , each letter, digit or symbol has a 'target area ' (i.e. the area for select ing each letter) that measures 4 mm in width and 3 mm in height. 10 Figure 1 Materials The stimuli for the tone discrimination task were pure tones, each exactly 100 ms in duration. O n e tone, herein cal led the standard tone, had a f requency of 4000 Hz. The other tones required for this task, cal led odd or target tones, had f requencies of 4020 Hz , 4040 Hz , 4060 Hz , 4080 Hz , 4100 Hz , 4120 Hz , 4140 Hz , 4160 Hz or 4180 Hz. Al l tones were created using Audaci ty v.1.2.6, a freeware Cross-P la t fo rm Sound Editor (Mazzoni & Dannenberg , 2000). E a c h tone was stored as a .wav file, with single channel 16-bit P C M coding at a sampl ing rate of 44.1 kHz. Tone Discrimination Task Stimulus presentation and response recording for the tone discrimination task were controlled by a P C , using the EPr ime v. 1.0 software (Psycho logy Software Tools Inc., Pittsburg, PA) . Part ic ipants wore headphones to ensure controlled presentation of the tones. The vo lume of the tones was set at a level which each individual considered to be \"comfortable\". For the tone discrimination task, participants l istened to a ser ies of standard tones interspersed with target tones. I instructed participants to say the word 'fruit' into the microphone each time they heard a target tone (i.e., a tone higher in pitch than the standard tone). Th is response word (i.e., fruit) was chosen to ensure that the microphone captured the onset of the vocal response. Part ic ipants were instructed to make their responses as accurately and as quickly as possib le. The tone discrimination task consisted of nine b locks, each with 150 trials. The target tones varied across blocks while the standard tone remained constant. During each trial, a tone was presented fol lowed by a random inter-stimulus interval (ISI) of 500 ms, 1000 ms, 1500 ms, 2000 ms, 2500 ms or 3000 ms (the ISI following a target tone 11 was always 1500 ms in duration to ensure an adequate amount of t ime for responding). The first three trials in each block always involved the presentation of a standard tone in order to \"habituate\" participants to this sound. In the remaining 147 trials, participants were presented with either a standard or target tone. O n each set of thirteen trials, two trials were randomly selected to present a target tone. PDA Tasks Six common P D A tasks were se lected. Three tasks required searching through the menu layers to find information (check the battery, retrieve appointments and find a picture), and 3 tasks required entering data (enter contact information, enter expense information, and make an appointment). E a c h of these tasks can be completed in a number of different ways . Tab le 1 shows the number of s teps required for complet ing each task in the most efficient or optimal manner. Table 1 Check the battery. The instructions for this task directed participants to find the current status of the P D A battery and to report the remaining battery charge. Retrieve appointments. For this task, participants were required to find and report to the exper imenter all appointments scheduled for 7 weeks away from the current date. Find a picture. Part icipants were instructed to find a picture of a green door in the personal folder. Enter contact information. For this task, participants were required to find the contacts function, and to enter contact information for a laboratory, including the name, the complete address , as well as the phone number and the emai l address of that laboratory. 12 Enter expense information. Participants were required to find the Exce l workbook and enter a ser ies of numbers into designated cel ls. Make an appointment. This task required finding the date 6 weeks away from the current date in the calendar, entering a restaurant name for an appointment at noontime, and setting a reminder to go off one hour prior to this event. Neuropsychological Tests A neuropsychological test battery was employed to a s s e s s participants' cognitive abilities. The battery was compr ised of four standardized tests: the Digit Symbo l Substitution Test (Wechsler , 1981), the North Amer ican Adult Read ing Test (Blair & Spreen , 1989), the Reverse Digit S p a n Test (Wechsler , 1981), and the Trail Making Test (Reitan, 1992). I administered each neuropsychological test according to the instructions in the publ ished manuals . The results of these tests are not directly pertinent to the object ives of the present project and thus will not be reported or d iscussed here. Design This exper iment included a basel ine condition and two critical test condit ions identified, respectively, as the data entry and navigation condit ion. In the basel ine condit ion, participants completed the tone discrimination task a lone, while in the critical test condit ions, they completed this task concurrently with a P D A task. The basel ine condition was required in order to find a difficulty level where tone discrimination accuracy was approximately 8 0 % for each participant. Th is accuracy level was selected because it reflects performance that is off the cei l ing, while leaving sufficient down-side room for revealing the addit ional resource demands of the concurrent P D A tasks which had to be completed in the data entry and navigation condit ion. E a c h participant completed the s a m e set of P D A tasks, which are listed in Tab le 1. 13 Procedures I tested participants individually in a sess ion that lasted approximately 60 minutes. Upon obtaining their written consent, I administered the tasks in the order descr ibed below. E a c h participant first completed the tone discrimination task alone in the basel ine condition in order to find a f requency difference between the standard and target tone at which a level of accuracy of approximately 8 0 % would be ach ieved . In order to find this level, I used a calibration procedure in which the f requency of the target tone was reduced by 20 H z (i.e., the difference between the target tone and the standard tone was decreased) ac ross success i ve blocks of trials. Speci f ical ly, for the first block of trials, the target tone f requency was 4100 Hz. If performance accuracy was above 8 0 % after 150 trials, I reduced the target tone frequency by 20 H z for the next block of 150 trials. I cont inued with this calibration procedure until the participant's performance was approximately at 80%. In the next phase of the experiment, participants completed the tone discrimination task together with the P D A tasks in the order listed in Tab le 1.1 instructed participants to focus on complet ing the P D A tasks accurately and quickly, but to respond to the target tones whenever possible. For each of the P D A tasks, I explained to participants about the goal of the task. For the navigation tasks, participants were given written instructions about the goal of the task. For the data entry tasks, participants were provided with the to-be entered information in written form. Part icipants were free to refer to the instruction sheet at any time during the task. At any time in the course of any of the tasks, participants were permitted to ask for help, for hints or information about how to proceed. 14 The f requency difference between the standard and target tone which was obtained for each subject in the basel ine condition was used for the tone discrimination task when it had to be performed in conjunction with one of the data entry or navigation tasks in the critical test condit ions. I started the tone discrimination task when participants began each P D A task and terminated the task as soon as the P D A task was completed. A Hitachi D Z - M V 3 8 0 A digital v ideo camera was used to create a complete record of participants' but ton-presses and stylus interactions with the P D A from the start (i.e., turning the P D A on) to the end (i.e., turning the P D A off) of each P D A task for offline coding. I started the v ideo recording immediately before participants turned the P D A on and stopped the v ideo recording immediately after participants completed the task and turned the P D A off. Fol lowing the complet ion of the P D A tasks, I administered the battery of neuropsychological tests. After the last neuropsychological test was completed, participants were given a verbal debriefing, as well as a debrief ing form, and course credit. Results Data Preparation For the P D A tasks, I developed a detailed manual with step-by-step instructions for scor ing each interaction (e.g., button-press, stylus-cl ick) with the P D A , as well as for scoring part icipant-experimenter interactions (e.g., requests for help). The manual was developed based on the guidel ines and definitions from a manual in a previous usability study (Graf & L i , 2007). For the manual in the previous study, an iterative method was used for developing the scor ing manual , alternating between writing scoring instructions and applying those instructions, until the manual could be used reliably by one other 15 coder. Two independent coders scored the complete video record of 10 different subjects. The reliability was .80 and above across the 9 different P D A tasks. Part ic ipants' performance of the P D A tasks was coded using the video records. I watched the v ideos that captured participants' interaction with the P D A , and for each task I counted the number of p resses , as well as recorded the amount of time to complete each task. For navigation tasks, a press was scored as incorrect when it d isplayed an undesired screen (i.e., did not lead towards the complet ion of the task) or when it produced no change in the sc reen. Correct p resses yielded the display of a desired sc reen, one that was required to progress towards complet ing the task. Complet ion time for navigation tasks was calculated by the amount of t ime e lapsed from the start to the end of the task. The start of a navigation task was defined by the first time the stylus touched the sc reen; the end was defined by the display of the target screen (i.e., the information to be found). I calculated data entry errors using a method developed by Wobbrock and Myers (2006). Data entry error rate was expressed as a function of uncorrected and corrected errors by the final entered data. The standardized metric words per minute ( W P M ) was used as an indication of data entry speed . I calculated data entry speed by dividing the total number of p resses by the completion time in minutes to compute the characters per minute ( C P M ) metric. Then , I computed W P M by dividing C P M by five. T h e s e methods were chosen as it al lowed compar isons across studies using the s a m e metrics. The results from the tone discrimination task and P D A tasks were screened for outliers, def ined as falling more than three standard deviat ions away from the sample mean. One outlier was found. The outlier was replaced with a non-outlying value, a number that was three standard deviat ions above the sample mean . 16 For each participant, I calculated the mean percentage of correct and incorrect identifications of target tones, adjusted identification of target tones and the median response time (RT) for the correct and incorrect responses in each condition of the experiment (basel ine, data entry, navigation) for the tone discrimination task. A correct identification was scored when a participant made a response to a target tone, while an incorrect identification w a s scored when a participant made a response to a standard tone. The adjusted identification score was computed by subtracting the percentage of incorrect identifications by the percentage of correct identif ications. Three participants did not make any correct responses on the tone discrimination task while they were concurrently performing either the data entry or navigation tasks. The data for these three participants were exc luded from the analysis. The analys is was conducted with data from the remaining 23 participants. PDA Tasks The dependent measures for the P D A tasks were the percentage of correct p resses , the percentage of errors in the data entry and navigation tasks, speed of data entry, and the amount of t ime to complete navigation tasks. The average percentage of correct p resses in navigation ac ross the tasks was 56 .58% ( S D = 10.90). Overa l l , 38 .83% (SD = 14.16) of the total p resses was categor ized as incorrect in navigation. Among the navigation tasks, participants made the greatest percentage of errors while trying to find the page for entering contact information (M = 58.31, SD = 21.53), fol lowed by trying to find the battery (M = 47.27, S D = 24.90), trying to find the correct date to make an appointment (M = 44 .71 , S D = 28.62), trying to retrieve the appointment dates (M = 29.87, S D = 25.58) and finding a picture {M = 36.97, S D = 30.01). Finally, participants made the least percentage of errors trying to find the Exce l program to enter information (M = 15.83, S D = 22.19). 17 The average percentage of correct p resses in navigation across the tasks was 77 .71% ( S D = 5.71). Of the total p resses made during data entry, 6 .60% (SD = .03) were errors. In genera l , the percentage of errors for each data entry task was low. Part icipants made the smal lest percentage of errors while entering information in the Exce l sheet (M = 4 .80, SD = 5.73), fol lowed by entering contact information (M = 7.21, SD = 3.07), and entering appointment information (M = 7.80, SD = 3.00). A paired samp les f-test was conducted using the average percentage of navigation and data entry tasks errors. The results revealed that participants made a significantly greater percentage of errors when performing navigation tasks compared to data entry tasks, f (22) = 11.39, p < .001. Average complet ion time for the navigation tasks was 32.15 s (SD = 14.43). Finding the battery (M = 48 .30 , SD = 27.16) took the longest, fol lowed by finding the contacts page for entering information (M = 41.05, S D = 33.09) and finding the correct date for entering appointment information (M = 41 .65 , S D = 29.09). Finding the appointment date took 30.61 seconds (SD = 29.78), fol lowed by finding the picture (M = 23.04, S D = 21.40), and finding the Exce l sheet (M = 8.22, S D = 6.84). M e a n data entry speed for the data entry tasks was 11.31 W P M ( S D = 2.28). Data entry speed w a s the quickest for entering appointment information (M = 14.93, S D = 4.43), fol lowed by entering contact information (M = 10.08, S D = 4.85), and entering information in an Exce l sheet (M = 8.91, S D = 1.60). Tone Discrimination Task The purpose of this experiment was to explore how performance changed when the tone discrimination task was performed alone versus in conjunction with a P D A task. The dependent measures for the tone discrimination task, in each task condit ion, were 18 the percentage of correct, incorrect, and adjusted identifications of target tones, and the time required to make correct identifications. Tab le 2 shows the mean percentage of correct, incorrect, and adjusted identifications of target tones, as well as the 9 5 % conf idence intervals, for each condit ion. Compared to the Base l ine condit ion, performance was lower in both the Data Entry and Navigat ion condit ion, but this effect was greater in the Navigat ion condit ion. Part icipants made a similar percentage of incorrect identifications in the Basel ine and Navigation condit ion. Part icipants made approximately 1/3 fewer incorrect identifications in the Data Entry condit ion. Table 2 In order to take into account the effect of di f ferences in incorrect identifications on the pattern of correct identif ications, the adjusted identifications were explored in detail. The average va lues of adjusted identifications for the basel ine, data entry, and navigation condit ions fol lowed the same pattern as correct identif ications. The mean value in the data entry and navigation condition was not e n c o m p a s s e d by the 9 5 % conf idence interval of the basel ine condit ion, indicating that performance accuracy was significantly lower in both the data entry and navigation condit ions compared to the basel ine condit ion. Similarly, the 9 5 % conf idence interval of the data entry condition does not include the mean value of the navigation condit ion, indicating that accuracy was significantly lower than performance accuracy in the data entry condit ion. A large effect s ize was found for the percentage of correct identif ications across condit ions, / = 2.31. S ince the pattern of results did not change even when taking into account the incorrect identifications data, only the median response t imes for correct identifications were explored. Figure 2 shows the response time (mean of medians) for correct 19 identification of target tones, as well as the 9 5 % conf idence intervals, in each condit ion. The summar ized results revealed increased response t imes for both the data entry and navigation condit ions compared to the basel ine condit ion. The mean response t imes in the data entry and navigation condit ions were not e n c o m p a s s e d by the 9 5 % conf idence interval of the basel ine condit ion, indicating that response time w a s significantly greater in the test condit ions. R e s p o n s e t imes were not significantly different ac ross the Data Entry and Navigat ion condit ions. A large effect s ize was found for the response t imes across condit ions, / = 0.96. Figure 2 Discussion For this experiment, I developed a version of the dual- task methodology to explore the attentional demands of navigation and data entry tasks. I had participants perform a tone discrimination task in the basel ine condit ion in order to find a difficulty level that would yield accuracy performance at approximately 8 0 % . Next, participants performed the tone discrimination task with either a data entry or navigation task. Per formance for both the data entry and navigation tasks was poorer than what was found in previous studies. In particular, data entry errors were slightly greater and typing speed was s lower compared to studies where participants performed data entry using a Q W E R T Y keyboard on a P D A . Zha and S e a r s (2001) reported 4 % (12.62 W P M ) and 5 % error rate (6.98 W P M ) for two data entry tasks. F leetwood et al . (2002) found that experts enter uncorrected text (i.e., where eras ing errors were not permitted) at a rate of 17.91 W P M and novices enter at a rate of 15.38 W P M with a 2 % error rate for both groups. For navigation tasks, participants took longer to find the information and made a greater percentage of errors compared to the f indings in a recent usability study with a P D A (Graf & L i , 2007). 20 The poorer P D A task performance found in this exper iment is likely due to distraction of concurrently attending to the tone discrimination task. Attending to a second task may have distracted participants from performing the P D A tasks as well as they could when they performed these P D A tasks a lone. However , the results obtained in this experiment were only slightly poorer compared to f indings obtained in previous studies (Graf & L i , 2007; F leetwood et al . , 2002; Zha & S e a r s , 2001) when full attention was avai lable (i.e., when the P D A tasks were completed alone). The findings from this experiment suggest that participants focused on complet ing the P D A tasks as well as possible. The results from the tone discrimination task are consistent with findings from previous studies that a lso found secondary task costs when participants were required to perform two tasks concurrently (e.g., Troyer, Winocur , Craik & Moscov i tch , 1999; Anderson , l idaka, C a b e z a , Kapur, Mcintosh & Craik, 2000). The comparable findings indicate that a dual- task methodology using tone discrimination is valid for assess ing the attentional demands of P D A tasks. Secondary task costs obtained in this study were slightly larger than what was found in previous research that used auditory discrimination as a secondary task. In a study involving the effects of divided attention on encoding and retrieval, l idaka, Anderson , Kapur , C a b e z a and Craik (2000) reported the percentage of words recalled when encoding performed alone was 79%. The percentage of correct words recalled dropped to 5 8 % when encoding was done with a tone discrimination task (magnitude change of 26.58%). Th is finding suggested that performing tone discrimination is attention demanding. Kl ingberg and Roland (1997) reported a 15.66% change in response time to detect a pitch change in a ser ies of presented tones when combined with a visual detect ion task. The greater secondary task costs obtained in this experiment compared to previous studies ( l idaka, Ande rson , Kapur, C a b e z a & Craik, 2000; Kl ingberg & Ro land , 1997) is likely due to the greater attentional demands required by P D A tasks. The finding that tone discrimination performance was affected to a greater extent by navigation tasks than by data entry tasks suggest that navigation is more attention demanding than data entry. R e s p o n s e t imes for adjusted identifications did not significantly differ for the Data Entry and Navigation condit ions. The null difference suggests that the difference in the percentage of adjusted identifications in the data entry and navigation condit ions was not due to simply placing more attention while performing data entry tasks compared to navigation tasks. O n e of the common crit icisms of using a dual-task methodology to a s s e s s attention is that the results can be explained either by attention or task switching. In task switching, individuals who perform two concurrent tasks al locate their attention to only one task at a t ime. To attend to the second task, individuals need to d isengage from the current task and re-al locate their attention the second task. In order to attend to the first task again, individuals need to re-allocate their attention back to the first task. Accord ing to task switching, it is possib le that participants find it more difficult to switch between navigation and tone discrimination compared to data entry and tone discrimination. Thus lower accuracy performance in the Navigation condit ion could a lso be explained by task switching. The second study was carried out to investigate whether the findings in Exper iment 1 were due to task switching or attention. 22 Experiment 2 Exper iment 2 was des igned to investigate whether the dual-task methodology used in Exper iment 1 a s s e s s e d the attentional demands of different kinds of P D A tasks, the difficulty of switching between the P D A tasks and tone discrimination task, or a combinat ion of the two factors. The general design of Exper iment 2 was the same as for Experiment 1. A n addit ional factor of the tone discrimination task, easy and hard was included in this experiment. The additional factor permitted manipulat ion of the required attentional resources to perform the tone discrimination task; more attention is required for the hard version compared to the easy vers ion. I expected that, as in Exper iment 1, adjusted identif ications in the tone discrimination task would be lower when performed together with navigation tasks than data entry tasks. If navigation requires more attention than data entry, then the differences in the percentages of adjusted identifications between the hard and easy tone discrimination task would be greater among navigation tasks than among data entry tasks. Concurrent ly performing a navigation task and hard tone discrimination task should be the most attention demanding, and the percentage of adjusted identifications should be the lowest in this condit ion. S ince an easy tone discrimination task requires less attention, the percentage of adjusted identifications should a lso be lower, but to a lesser extent compared to a hard version of the tone discrimination task. In contrast, s ince data entry tasks require less attention, there is more leftover attention to perform the tone discrimination task. The percentage of adjusted identifications should be similar while performing either an easy or hard tone discrimination task. O n the other hand, if the dual-task methodology which uses tone discrimination is sensit ive to task switching, then switching between a navigation task and tone discrimination should be equal ly difficult should be more difficult than switching between 23 a data entry task and tone discrimination. In addit ion, switching between a tone discrimination task (easy or hard) and a P D A task (data entry or navigation) would be equally difficult. Thus , I expected that, similar to Exper iment 1, the overall the percentage of adjusted identifications would be lower when the tone discrimination task was combined with navigation tasks compared to data entry tasks. In addit ion, I expected that the difference in the percentage of adjusted identifications would be similar between an easy and hard tone discrimination task when combined with navigation tasks. Last ly, I expected that he difference in the percentage of adjusted identifications would be similar between an easy and hard tone discrimination task when combined with data entry tasks. Method Participants Thirty undergraduate students were recruited through the subject pool at the Psycho logy department in the University of British Co lumb ia . They participated individually in this one sess ion study lasting approximately 60 minutes. They were compensated with one course credit in return for their participation. The experiment was conducted with the approval of the University of British Co lumb ia behavioral ethical review board. Apparatus The s a m e dev ice as in Exper iment 1 was used . Tone Discrimination Task The materials, des ign and instructions for the tone discrimination task were the s a m e as for Exper iment 1, except for the duration of the IS I. The interval from the onset of one tone to the onset of the next was narrowed to 600ms, 1100ms, and 1600ms. 24 From the results Exper iment 1, the mean of median response t ime w a s no greater than 1200 ms. Thus , the ISI was reduced s ince the extra time was not needed. PDA Tasks Twelve tasks commonly performed on a P D A were selected for this study. A greater number of tasks were included for this experiment so that there were three P D A tasks in each condit ion that involved a P D A task. This number of P D A tasks per condition was chosen to obtain an appropriate sample of P D A task performance. Six tasks required participants to navigate through the menus: Find a picture, retrieve appointments, check the battery, find a voice recording, find an emai l address , and find owner information. The other 6 tasks required participant to enter text on a Q W E R T Y keyboard using a stylus: Enter sentences, name new folder, enter contact information, enter email m e s s a g e , enter appointment information, and enter expense information. Six of the twelve tasks were identical to the ones used in Exper iment 1. The newly added tasks in Exper iment 2 will be descr ibed below. E a c h of these tasks can be completed in a number of different ways. Tab le 3 shows the number of number of steps required for complet ing each task in the most efficient or optimal manner. Tab le 3 Find a voice recording. For this task, participants were required to locate the folder named 'Persona l ' and to play the voice m e s s a g e named 'recording'. Find an email address. Part icipants were instructed to f ind, and display on the sc reen, the emai l address for the contact 'Aphas ia project'. Find the owner. For this task, participants were required to search through several menus to find the screen that showed the name of the dev ice owner. Enter sentences. Part icipants were instructed to enter two sentences into a Word document. 25 Enter an email message. This task required participants to enter a brief message into the emai l composer in the device. Name a new folder. Part icipants were required to create a new folder and name it 'Exper iment results'. Design The des ign of this experiment was a 2 x 2 factorial with tone discrimination difficulty (easy, hard) and task condition (basel ine, data entry, navigation) manipulated as within subjects factors. The basel ine condition was included in order to find two difficulty levels where accuracy was approximately 8 0 % for the easy tone discrimination condition and approximately 7 0 % for the hard tone discrimination condit ion for each participant. T h e s e accuracy levels were selected because they reflect performance that is off the cei l ing, while leaving sufficient down-side room for reveal ing the additional resource demands of the concurrent P D A tasks which had to be completed in the data entry and navigation condit ion. Procedures I tested participants individually in a sess ion that lasted approximately 60 minutes. Upon obtaining their written consent, I administered the tasks in the order descr ibed below. E a c h participant first completed the tone discrimination task alone in the basel ine condition in order to find a f requency difference between the standard and target tone at which a level of accuracy was approximately 8 0 % and 7 0 % . T h e s e performance levels were identified by means of the same calibration procedure as in Exper iment 1. In the next phase of the experiment, participants completed either an easy or hard tone discrimination task together with the P D A tasks listed in Tab le 3. E a c h participant completed the s a m e set of P D A tasks in combinat ion with a tone 26 discrimination task. Presentat ion of the four condit ions ( P D A task type by tone discrimination task difficulty) was counterbalanced by means of a Latin Square des ign. The instructions for performing a P D A task concurrently with a tone discrimination task were the s a m e as in Exper iment 1. The procedures for starting and ending the tone discrimination task, as well as procedures for recording P D A task performance, were the s a m e as in Experiment 1. After the last P D A task was completed, participants were given a verbal debriefing, as well as a debriefing form, and course credit. Results Data Preparation The coding manual for the P D A tasks was deve loped in the same way as in Experiment 1. The methods for coding P D A task performance and calculat ing tone discrimination task performance were same as in Exper iment 1. Al l data were checked and corrected for transcription and coding errors until accuracy was greater than 99%. The results from the tone discrimination task and P D A tasks were sc reened for outl iers, defined as falling more than three standard deviat ions away from the sample mean . There were two outliers d iscovered in the tone discrimination task and 1 outlier in the P D A tasks. E a c h outlier was replaced with a number either 3 standard deviat ions above or below the sample mean , respectively. S ix participants did not make any correct identification on the tone discrimination task while concurrently complet ing a P D A task. Data from these participants were excluded from the ana lyses . The data from the remaining 24 participants were included for the ana lyses . There were five missing values in the P D A tasks. E a c h value was replaced with the average value from that variable. 27 PDA Tasks The dependent measures for the P D A tasks were the percentage of correct presses, the percentage of errors in the data entry and navigation tasks, speed of data entry, and the amount of time to complete navigation tasks. Overa l l , two-thirds of the presses (M = 66.23) in the navigation tasks were correct. A n in depth examinat ion of the navigation errors by each task reveals that the percentage of errors ranged from 8 .15% to 49 .15%. Nearly half of the total presses while finding the battery were errors (M = 49.15, SD = 23.51). Other tasks had slightly fewer errors, such as finding the page for renaming the folder (M = 42.67, SD - 19.36), followed by finding a vo ice recording (M = 43.38, SD = 20.00), finding appointments (M = 32.52, SD = 21.85), creating a new document in Word (M = 28.00, S D = 31.27), creating a new emai l m e s s a g e (M = 32.94, S D = 29.79), finding the appropriate date for entering information (M = 31.46, S D = 19.80), creating a new contact (M = 23.46, S D = 28.70), finding an emai l m e s s a g e (M = 23.29, S D = 27.58), and finding the page that displays the owner of the device (M = 22.46, S D = 23.17). The fewest errors were made for finding the Exce l sheet to enter information {M = 8.15, S D = 15.22). The percentage of errors for finding the Exce l sheet was substantially lower than the rest of the navigation tasks, and was probably not representative of navigation. This task was not included in further ana lyses . A lmost all the p resses made during data entry were correct. Of the total p resses, 6 .5% were errors. The percentage of data entry errors across tasks ranged from 2.67% to 8.83%. The greatest percentage of errors were made when re-naming a folder (M = 8.83, S D = 9.76), fol lowed by entering appointment information (M = 7.84, S D = 6.19), entering contact information (M - 7.52, S D = 4.22), typing an emai l message (M = 7.37, 28 SD = 6.68), and entering sentences into Word (M = 5.07, SD = 4.11). The least percentage of errors w a s made when entering data in the Exce l sheet. The percentage of errors of each P D A task type ac ross tone discrimination difficulty, as well as the standard errors, is d isplayed in Figure 3. Part icipants made a slightly greater percentage of navigation errors in the easy tone discrimination condition compared to hard tone discrimination condit ion. Simi lar to navigation tasks, participants made a slightly greater percentage of data entry errors in the easy tone discrimination compared to the hard tone discrimination condit ion. Figure 3 A two-way A N O V A was conducted on the error data with the P D A task types (data entry, navigation) and tone discrimination difficulty (easy, hard) as within-subjects factors. The analys is conf irmed the observat ion that participants made more errors while performing navigation tasks than data entry tasks, F (1, 23) = 188.71, MSE -78.54, p < .001, f =2 .19 . No other effects were significant. Typing rate was slightly s lower in the hard tone discrimination condition (M = 12.57, SD = 2.26) compared to the easy tone discrimination condit ion (M = 14.59, SD = 2.91). Typing rate w a s the quickest for renaming a folder (M = 17.50, SD = 4.39), followed by entering sen tences in Word {M = 15.59, SD = 3.86), typing an email message {M = 15.12, SD = 4.77), entering appointment information (M = 13.65, S D = 3.63), entering contact information (M= 10.68, S D = 2.02), and entering information in an Exce l sheet (M = 9.48, S D = 1.85). The results using a paired samp les f-test 1 Cohen's f (1988) is the appropriate effect size measurement used in the context of an F-test. Cohen's f and eta-squared (a power parameter commonly reported with the F-test statistics) are related in the following manner: f - e ta 2 / (1-eta2). 29 revealed that participants entered data at a significantly quicker rate while performing an easy compared to hard tone discrimination task, t (23) = 4 .91 , p < .001, d 2 = 0.78. Average complet ion time for navigation tasks while complet ing an easy tone discrimination task w a s slightly greater {M = 31.85, SD = 9.30) compared to while complet ing a hard tone discrimination task (M = 24.45, S D = 7.21). The complet ion t imes ranged from an average of 55.46 s (SD = 31.48) to 11.86 s ( S D = 12.18). Part icipants took the longest while finding the appropriate screen to rename a folder, followed by finding a vo ice recording (M = 52.67, S D = 29.11), finding the battery (M = 40.83, S D = 20.86), finding the appointment dates (M = 36.29, S D = 27.08), finding the appropriate date for entering appointment information (M = 31.25, S D = 20.02), finding a picture (M = 30 .71 , S D = 28.57), composing a new emai l m e s s a g e (M = 28.13, S D = 24.16), finding the screen that d isplays the owner of the dev ice (M = 17.58, S D = 14.62), creating a new Word document (M = 15.59, S D = 16.02), creating a new contact (M = 15.59, S D = 16.09), and finding an email message . The results using a paired samples f-test conf irmed that participants found the required information significantly faster while performing a hard compared to an easy tone discrimination task, f (23) = 3.24, p < .01, d= 1.40. Tone Discrimination Task The dependent measures for the tone discrimination task, in each condit ion, were the percentage of correct, incorrect, and adjusted identif ications of target tones, and the time required to make correct identifications. shows the mean percentage of correct, incorrect, and adjusted identifications of target tones for each condit ion. A s expected, participants were less accurate while performing a hard compared to easy tone discrimination task in the Bbsel ine condition. 2 Cohen's d (1988) is the appropriate effect size measure to use in the context of a Mest on means. 30 Among the data entry condit ions, participants made a greater percentage of correct identifications while complet ing an easy compared to hard tone discrimination task. Similarly, in the navigation condit ions participants made a greater percentage of correct identifications while complet ing an easy compared to a hard tone discrimination task. Overal l , the percentage of incorrect identifications was the greatest in the Basel ine condit ions (M = 2.43), fol lowed by the navigation (M = 1.82), and data entry condit ions {M = 1.35). In the basel ine condit ion, participants made approximately twice the percentage of incorrect identifications for the easy tone discrimination task compared to the hard tone discrimination task [see ]. In the navigation condit ions, participants made approximately the same percentage of incorrect identifications for an easy and for a hard tone discrimination task. Whi le participants were completing data entry tasks, they made a slightly greater percentage of incorrect identifications while performing a hard versus an easy Tone Discrimination Task . In order to take into account the different percentage of incorrect identifications across condit ions, the adjusted identifications were explored in detail Q. The pattern of results for adjusted identifications was the same as for the pattern of correct identifications. A greater percentage of adjusted identifications w a s found when participants completed an easy version of the Tone Discrimination Task than the hard version. W h e n performed concurrently with data entry tasks, participants made a similar percentage of adjusted identifications in the easy and hard vers ion of the tone discrimination task. The percentage of adjusted identifications was lower while participants performed a hard Tone Discrimination Task compared to an easier version. A two-way A N O V A conducted on the adjusted identif ications data with the task condition (basel ine, data entry, navigation) and tone discrimination difficulty (easy, hard) 31 as the within-subjects factors confirmed the observat ions. Part icipants made a significantly greater percentage of adjusted identifications while performing an easy tone discrimination task, F (1, 23) = 33.49, MSE = 394.70, p < .001, / = 1.16. A main effect of Ttsk condit ion was also found, F (2, 46) = 56.99, MSE = 309.87, p < .001, / = 2.16. A post-hoc test conducted with Fisher 's L S D revealed that the percentage of adjusted identif ications in the basel ine condition was significantly greater than the data entry condit ion, and that the percentage of correct identif ications in the navigation condition was significantly less compared to the data entry condit ion. The critical ana lyses were to examine whether the percentage of adjusted identifications varied between P D A task types across tone discrimination difficulty. The mean difference of adjusted identifications among navigation condit ions (M = 17.14, SD = 22.7'4) was greater than among data entry condit ions (M = 9.51, SD = 21.65). A paired samples f-test conducted on the mean differences between the two condit ions revealed that the difference approached signif icance, f (23) = 1.75, p = .09. S ince the pattern of results did not change even when taking into account the incorrect identifications data, only the response t imes for correct identifications were explored. The summar ized results for response t imes (mean of medians) of correct identifications are d isp layed in Figure 4. Overal l , participants took longer to respond to target tones while performing data entry tasks (M = 861.04) and navigation tasks (M = 947.00) compared to the Base l ine condition (M = 787.85). T ime required for responding to target tones while complet ing data entry tasks was approximately the same when participants were concurrently performing an easy or a hard Tone Discrimination Task . However, response time while performing navigation tasks was greater while attending to a hard compared to easy Tone Discrimination Task . 32 Figure 4 A 2 x 3 A N O V A was conducted on the response time data with the task condition (basel ine, data entry, navigation) and tone discrimination difficulty (easy, hard) as the within-subjects factors. The analysis confirmed that participants took significantly longer to correctly identify target tones when performing a hard compared to an easy tone discrimination task, F (1, 23) = 6.66, MSE = 29073.56, p < .001, / = 0.49. A significant main effect of task condit ion was found F (2, 46) = 19.21, MSE = 15852.62, p < .001, / = 1.23. The interaction effect was not significant. Pos t -hoc tests using Fisher 's LSD revealed that response time was significantly greater while participants were concurrently complet ing a P D A task. In addit ion, response time was greater while performing navigation tasks compared to data entry tasks. The interaction was not significant. Discussion The primary objective of Experiment 2 was to explore whether the findings obtained in Exper iment 1 reflected measur ing attention or task switching. The second objective was to investigate whether the developed dual- task methodology could be reliably used . W e required participants to perform an easy or hard version of the tone discrimination task while complet ing a set of P D A tasks that either focused on navigation or data entry. Overal l , the navigation task error rate was greater than the data entry error rate. The error rates for both P D A tasks were slightly better when combined with a hard version of a Tone Discrimination Task compared to an easy vers ion, but the difference was not significant. Performing hard tone discrimination s lowed down information search ; performing easy tone discrimination s lowed down data entry speed . 33 The complet ion time and data entry speed should be interpreted with caution. The time data reported a lso reflects the exchange of instructions between the participants and experimenter, and when participants asked for help. One possible interpretation of the finding is that participants asked for more help in one condit ion, thus slowing down data entry. Another possible interpretation for the findings is that participants was more relaxed when they realized that entering data and attending to easy tone discrimination was not difficult, and therefore took more time to enter the data. In genera l , P D A task performance was comparab le to f indings in studies that had participants complete these tasks with full attention (Graf & L i , 2007), indicating that participants were paying attention to performing these tasks. The percentage of adjusted identifications was significantly lower and the median response time was significantly s lower while performing the hard tone discrimination task compared to the easy version in the basel ine condit ion. The findings suggest that the difficulty manipulat ion was success fu l , and that the hard version of the task required more attention than the easy vers ion. A s expected, the difference in adjusted identifications was greater between navigation condit ions compared to the data entry condit ions. R e s p o n s e time was the greatest when participants were completing navigation tasks with a hard tone discrimination task. W e were able to replicate the majority of f indings from Exper iment 1. Consistent with Exper iment 1, participants made more errors in navigation tasks compared to data entry tasks. Overa l l , the percentage of navigation errors was slightly lower and overall time to complete navigation tasks was slightly quicker compared to Exper iment 1. The overall data entry error rate was similar to Exper iment 1. However , data entry speed was slightly s lower than what was found in Exper iment 1. 34 Overal l , the pattern of adjusted identifications was similar to what was found in Experiment 1. The percentage of adjusted identifications was lower when performing a P D A task and a tone discrimination task compared to performing only the tone discrimination task a lone. The effect s ize of adjusted identif ications for data entry and navigation condit ions were comparab le to the results from Exper iment 1. R e s p o n s e time for correct identifications while complet ing data entry tasks was quicker than what was found in Exper iment 1. The reduced response time compared to Exper iment 1 is most likely due to the reduced inter-stimulus interval. There is ev idence that response time varies as a function of inter-stimulus interval (France et a l . , 2002). Auditory trace decays as a function of t ime. In longer ISI's, participants take longer to retrieve the compar ison tone prior to making a dec is ion. In summary, the findings from Experiment 2 provide support that a dual-task methodology using tone discrimination is a valid set-up for assess ing the attentional demands of P D A tasks. In addit ion, the replicated f indings from Exper iment 2 also suggest that this method can be used reliably in the context of measur ing attention required by P D A tasks. Finally, Exper iments 1 and 2 provide converging ev idence that navigation tasks require more attention to perform than text entry tasks. Whi le there have been many researchers that study attention as a single resource, a second account of attention argues that attention consis ts of distinct reserves where each resource is responsible for process ing different types of information (Baddeley, 1986). The following two exper iments were des igned to explore whether different types of attention are required to perform navigation and data entry tasks. 35 Exper iment 3 The purpose of this study was to validate a methodology for exploring the different types of attentional resources required for navigation and data entry tasks. I chose an existing method, descr ibed by Brooks (1968), for reveal ing different types of attention: V isuo-spat ia l and articulatory/auditory. This method required participants to perform two tasks concurrently. Part icipants completed a combinat ion of one input task (letter tracing task, sen tence decis ion task) and one output task (pointing, tapping, saying). O n e input task required visuo-spatial resources (letter tracing task), while the other task required articulatory/auditory resources (sentence decis ion task). The letter tracing task involved presenting an image of an Engl ish block letter (e.g., see Figure 5) and making categorizat ion judgements. For this task, participants were asked to decide whether each corner w a s the topmost/bottom-most corner in the letter, or whether each corner was on the outside edge of the letter. The sentence dec is ion task required participants to categor ize each word in a sentence presented over the speakers . They were required to dec ide whether each word was a noun/non-noun or article/non-article. For both tasks, participants were required to provide 'yes ' or 'no' answers . Figure 5 The three output tasks required participants to indicate their answers from the Input T a s k s in one of three ways . Part icipants made their responses by a method that required either v isuo-spat ial resources (pointing), articulatory/auditory resources (saying), or motor abilities (tapping). A combinat ion of these tasks (i.e., one input task and one output task) would be then used in conjunction with the P D A tasks to explore the type of attention required by the P D A tasks. 36 I expect to find the s a m e pattern of results reported in the original study (Brooks, 1968). Complet ion t imes would be greater when both tasks draw on the same type of resources compared to when the two tasks require different attentional resources. Specif ical ly, complet ion time would be greater for the letter tracing task when responding in a manner that requires visuo-spatial attention compared to responding by a method that requires articulatory/auditory attention (i.e., say ing the responses) . In addit ion, I predicted that complet ion time would be greater for the sentence decis ion task when verbal ly indicating the responses compared to indicating answers by a method that requires visuo-spat ial processing. Methods Participants Twenty-seven undergraduate students were recruited through the subject pool in the psychology department at the University of British Co lumb ia . They were compensated with one course credit in return for their participation. The experiment was conducted with the approval of the University of British Co lumb ia behavioral ethics review board. Materials The stimuli for the letter tracing task consisted of an outline of six block letters. Two letters were for practice trials and the remaining four letters were for the test trials. Each letter for the practice trials [shown in Figure 5] had 12 corners: E, H. Each letter for the test trials [shown in Figure 6] had ten corners: F, N, G , and Z . E a c h letter was drawn using M S Paint and stored as a bitmap file. Figure 6 The stimuli for the sentence decis ion task consisted of six famous Engl ish aphor isms (e.g., a bird in the hand is not in the bush). Two sentences were used for the 37 practice trials, and the remaining four sentences were used for the test trials. Each sentence in the practice trial had a length of twelve words: You' l l never plow a field by turning it over in your mind; the grass is always greener on the other s ide of the fence. Each sentence in the test trial had a length o f ten words: Rivers from the hills bring fresh water to the cit ies; a bird in the hand is not in the bush; there is the low fiend who stole the child's candy; no man who has a wife is still a bachelor. E a c h sentence was recorded by speak ing through a microphone, and stored as a .wav file, with single channel 16-bit P C M coding at a sampl ing rate of 44.1 kHz . Input Tasks Letter Tracing Task. St imulus presentation and response recording for the letter tracing task were control led by a P C , using the EPr ime v.1.0 software (Psychology Software Too ls Inc., Pit tsburg, PA) . The stimuli, descr ibed in the Materials sect ion, were presented on a 15\" monitor. The letter tracing task required participants to categor ize each corner of a block letter. The letter tracing task consisted of two practice trials and four test trials (one trial for each letter). In each trial of this task, a letter was presented on the monitor on a white background. Part ic ipants were al lowed to view the letter as long as they needed. After participants indicated that they could remember the shape of the letter, the stimulus was taken off the monitor. Participants were instructed to imagine the letter to ensure that they could generate a mental image of the letter. Part ic ipants were shown the letter again if they indicated that they could not remember. W h e n participants indicated that they could remember the shape of the letter, the image was taken off the monitor. Next, one of the categorizat ion instructions was presented on the screen. Part icipants were required to begin the task starting at the bottom left corner of the letter and to categor ize the corners in a c lockwise direction until all corners were categor ized. 38 In the top/bottom instruction, participants were asked to categor ize the corners that were the highest point or the lowest point in the letter as 'yes ' and all other corners as 'no' (see Figure 7). For example , the correct sequence of responses for the ' F ' stimulus would be \"yes, yes , yes , no, no, no, no, no, no, yes\" . The outside instruction required participant to categor ize each corner that touched an imaginary box surrounding the letter as 'yes ' and all other corners as 'no' (see Figure 8). The correct sequence of responses for the ' F \" st imulus would be \"yes, yes , yes , yes , no, no, no, no, no, yes\". Figure 7 Figure 8 Sentence Decision Task. St imulus presentation and response recording for the sentence dec is ion task were controlled by a P C , using the E P r i m e v. 1.0 software (Psychology Software Too ls Inc., Pittsburg, PA) . The sen tences were presented over two speakers . The sentence dec is ion task consisted of two practice trials and four trials (one trial for each sentence) . The sentence decis ion task required participants to categorize each word immediately after a presented sentence. In each trial of this task, a sentence was presented auditorily through two speakers . After the sentence was presented, participants were asked to repeat the sentence to ensure that they were able to recite the sentence perfectly. The sentence was repeated if participants recited the sentence incorrectly. After correcting the participants, they were asked to repeat the complete sentence again. W h e n participants were able to rehearse the sentence perfectly, one of the two categorizat ion instructions was presented. In the noun instruction, participants were required to categor ize each word (in serial order) that was a noun as 'yes' and all 39 other words as 'no'. In the article instruction, participants were required to categorize each word that was an article as 'yes' and all other words as 'no'. Output Tasks Pointing. A list of Y and N was displayed in a staggered manner on a computer monitor [see Figure 9]. The Y ' s and N's were staggered to force c lose visual monitoring of pointing. E a c h line d isplayed one Y and one N. A total of 12 l ines of Y ' s and N's were presented on the screen for a practice trial, s ince there were 12 corners on each stimulus of the practice trial. Ten lines were presented for a test trial, s ince there were ten corners on each st imulus of the test trial. Part icipants c l icked on either a Y or N on each line to make one response, starting from the top line. Part icipants used a mouse to point and click at the Y (for yes) and N (for no). Figure 9 Tapping. In the tapping output condit ion, participants p ressed a button on a keyboard that cor responded to a 'yes' with their left index f inger and another button for a 'no' response with their right index finger. Saying. In the saying output condit ion, participants responded by saying 'yes' and 'no' into a microphone. Neuropsychological Tests A neuropsychologica l test battery was employed to a s s e s s participants' cognitive abilities. The battery consis ted of four standardized tests: the Forward Cors i Block tapping task (Cors i , 1972; Kesse l s , Zandvoort, Pos tma , Kappel le and Haan , 2000), the Reverse Cors i B lock tapping task (Corsi , 1972; K e s s e l s , Zandvoort , Pos tma , Kappel le & Haan , 2000), the Forwards Digit span test (Wechsler , 1981), and the Reverse Digit S p a n Test (Wechsler , 1981). The tests were administered accord ing to the standardized 40 instructions publ ished in each manual . The results of these measures are not pertinent to the objectives of the present thesis and thus will not be included. Design The des ign of this experiment consisted of a 2 x 3 factorial des ign that had input task (letter tracing task, sentence decis ion task) and output task (pointing, tapping, saying) as within subject factors. Procedures I tested participants individually in a sess ion that lasted 60 minutes. Upon obtaining their written consent , each participant completed the tasks descr ibed below. There were a total of six condit ions in this experiment. In each condit ion, participants completed either a letter tracing task or sentence decis ion task, combined with one method of responding: pointing, tapping, say ing. Presentat ion of the six condit ions was counterbalanced across participants by means of a Latin Square des ign. In each condit ion, participants were given two practice trials to familiarize them with the task, the categorizat ion instructions, and the method of response. Prior to beginning the practice trials, I explained to participants the goal of the task. For condit ions that involved the sentence decis ion task, I expla ined to participants that they were to listen to a ser ies of short sentence and make categorizat ion judgments for each word. I then expla ined each categorization instruction (i.e., noun, article). O n c e participants indicated that they understood the task, I p roceeded to descr ibe the method of response for that condit ion. After the practice trials were completed, a block of four sentences was run as the test trials. The stimuli and categorizat ion instructions were randomly presented ac ross trials. For the letter tracing task, I explained to participants that they were required to look at a block letter presented on the monitor and to categor ize each corner, in a 41 clockwise direction, starting from the bottom left corner. For each letter, I told participants they would be asked to make categorization judgments for each corner. Next, I explained each categorizat ion instruction. O n c e participants indicated that they understood the task, I p roceeded to descr ibe the method of response for that condit ion. These instructions were the same as for the sentence dec is ion task. After the practice trials were comple ted, a block of four letters were run as the test trials. I instructed participants to make their responses immediately when they made a categorizat ion. Th is instruction was given to prevent participants from first collecting several answers and then indicating the responses, which would not reflect completion time produced by the interference. In addition, I told participants to be as accurate as possible in making their responses . The timing for each trial began after the categorization instruction was presented and ended when participants made the last response. Fol lowing the complet ion of the last condit ion, I administered the battery of neuropsychological tests. After the last neuropsychological test was completed, participants were given a verbal debriefing, as well as a written debriefing form, and a course credit. Results Data Preparation The results from the tasks were screened for outl iers, def ined as falling more than three standard deviat ions away from the sample mean . O n e outlier that was three standard deviat ions above the mean was found. Th is outlier w a s replaced with a non-outlying number, a number that was 3 standard deviat ions above the sample mean. 42 One participant did not complete the four test trials of the sentence decis ion task while making verbal responses . The data from this participant were exc luded from the analysis. Data from the remaining 26 participants were used for the ana lyses. The dependent measure for the letter tracing task and sentence decis ion task was the time it took to complete one trial. For each participant, I computed the mean complet ion time for each condit ion. Input and Output Tasks M e a n complet ion time is shown in Figure 10. Overa l l , complet ion time for the letter tracing task was 17.73 s ( S D = .85) and 16.04 s ( S D = .74) for the sentence decis ion task, suggest ing that the tasks were of approximately the s a m e difficulty. The figure revealed that the two condit ions that required over lapping resources had greater complet ion t imes compared to other conditions that had an input and output tasks that required different types of resources. W h e n participants responded by pointing to answers, complet ion time for the letter tracing task (M = 26.87, S D = 7.34) was approximately twice as long compared to categorizing words in a sentence (M = 15.27, S D = 4.45). Similarly, complet ion time was greater when participants made vocal responses for the sentence decis ion task (M = 21.83, S D = 6.01) compared to when the performing the letter tracing task (M = 13.66, S D = 3.57). Complet ion time in the tapping output condition was approximately the same when performing the letter tracing task (M = 12.85, S D = 3.57) as in the sentence decis ion task (M = 11.64, S D = 3.54). Figure 10 A two-way A N O V A was conducted on the complet ion time[s] with the input task (letter tracing task, sentence decis ion task) and output task (pointing, tapping, saying) as the within-subjects factors. A main effect of output task was found, F (2, 50) = 79.24, MSE = 13.05, p < .001, / = 0.86. The interaction between input and output task was 43 also significant, F (2, 50) = 115.38, MSE = 11.03, p < .001, / = 2.97. Complet ion t imes between the input tasks were not significantly different, p > .05. Two fol low-up paired samples f-tests were conducted on the complet ion t imes to explore whether complet ion time varied across input and output tasks. Compar ison between complet ion t imes confirmed the observat ions. W h e n participants pointed to the answers, they took significantly longer to categorize corners of a letter than to categorize words in a sentence, t (26) = 9.04, p < .001. W h e n participants indicated their answers verbally, they took longer to complete the sentence decis ion task compared to the letter tracing task, t (26) = -6.78, p < .001. Discussion The main purpose of the experiment was to investigate whether the method descr ibed by Brooks (1968) provided a reliable and valid means for revealing the different types of attention. Part icipants were required to perform two tasks concurrently. They were engaged in a task that required primarily v isuo-spat ia l (categorizing corners of a block letter) or verbal resources (categorizing words in a sentence) , and they indicated their answers in a manner that required verbally, motor or visuo-spatial resources. The general pattern of results found in this experiment was consistent with my hypotheses. Complet ion time was the greatest when participants handled two tasks that required the s a m e resources. The results showed that responding in a verbal manner was s lowest while categoriz ing words in a sentence, and pointing to the answers on a monitor was s lowest for categorizing corners of a letter. Complet ion time for tapping the answers on a keyboard was approximately the same while combined with either input task, suggest ing that tapping answers required neither visuo-spat ial nor verbal resources. 44 However , there are several dif ferences between the results obtained from this experiment and the results reported by Brooks (1968). In this experiment, participants took 15 s to complete the sentence decis ion task when pointing to the answers, while the complet ion time was slightly shorter (10 s) in the original study. The greater complet ion time found in this experiment can be expla ined by the different technique used record to the responses . Whi le Brooks (1968) required participants to circle the answers on a piece of paper, I required participants to use a mouse and select their answers on a computer. Part icipants may have been less famil iar with the mouse and computer method than the typical pencil and paper method, and this may explain why the complet ion time was greater in the present experiment. In the original study, participants took almost twice as long to tap the answers for the letter tracing task (M = 14.1 s, SD = 5.4) compared to the sentence decis ion task (M = 7.8 s, SD = 2.1) whi le I found that complet ion time was approximately the same for both letter tracing (M = 12.85, SD = 3.57) and sentence decis ion task (M = 11.64, S D = 3.54). The effect s ize for the mean difference in complet ion t imes for Brook 's study is d = 1.54) and d = 0.34 for the mean difference in complet ion t imes for Exper iment 3. O n c e again, the difference in mean complet ion t imes found in Brook 's study and my experiment may be expla ined by the method used to record the responses . In the original study, Brooks recorded the answers by having participants touch either the Y or N printed on a piece of paper for their responses. Part icipants were likely reading the Y or N prior to touching the answer. Thus , the response record method was likely to require some visuo-spat ial resources, which resulted in greater complet ion time for performing the letter tracing task. Finally, a much larger difference in complet ion time was found in this experiment between the letter tracing and sentence decis ion tasks when participants responded by 45 saying the answers compared to what was reported by Brooks (1968). Specif ical ly, in this experiment, an eight second difference was found between the two input tasks (d = 1.66), while the original study found only a two second difference between the same tasks (d = .77). The difference in complet ion t imes and effect s i zes may be explained by the increased number of participants in the study, which may reflect a more accurate understanding of the speci f ic type of processing required for these tasks. 46 Exper iment 4 The objective of this experiment was to explore the type of attention required of two P D A task types: Navigat ion and data entry. The dual task methodology was used to approach this objective. The input tasks validated in Exper iment 3 (letter tracing task and sentence decis ion task) were used as the secondary tasks. Part icipants were required to perform either a letter tracing task or a sentence dec is ion task concurrently with either a navigation or data entry task on the P D A . Method Participants Thirty undergraduate students were recruited through the subject pool in the psychology department at the University of British Co lumb ia . They were compensated with one course credit in return for their participation. The experiment was conducted with the approval of the University of British Co lumbia behavioral ethics review board. Apparatus The s a m e apparatus was used as in Exper iment 1. Input Tasks The letter tracing task and sentence decis ion task were used as the secondary tasks. The stimuli for the tasks, as well as the setup for the tasks are descr ibed in the Methods sect ion of Exper iment 3. Output Tasks Part icipants responded in one of two methods: verbal ly saying the answers or tapping the answers . T h e s e methods of response are descr ibed in the Methods section of Exper iment 3. 47 PDA Tasks Sixteen P D A tasks were se lected, two per condit ion (a factorial combination of input task, output task and P D A task type yielded eight condit ions). Eight tasks required searching through the menu layers to find information (find font s ize , retrieve appointments, check the battery, find voice message , create new folder, find Bluetooth version, find themes, and load webpage) , and eight tasks required entering data. Two of the navigation tasks were identical to the ones used in Experiment 1 (check the battery, retrieve appointments) and one of the navigation task was the same one used in Exper iment 2 (find voice message) . Be low includes a brief description of each navigation task. Find font size. The instructions for this task directed participants to find the font s ize that d isplays the text on the P D A . Create new folder For this task, participants were required to find the program that had the option to create a new folder. Find Bluetooth version. This task required participants to find the Bluetooth application and to display the current version of the appl icat ion. Find themes. Part ic ipants were instructed to find the screen that showed the avai lable themes of the device. Load a webpage. For this task, participants were instructed to locate the saved webpage www.hotmai l .com under the Favori tes category and to load this webpage. The information for the data entry task consisted of eight famous Engl ish aphor isms, each one sentence in length. The sentences ranged from 31 to 35 presses, five to seven syl lables and five to seven words. Part icipants entered one sentence for each data entry task. E a c h of the P D A tasks can be completed in a number of different 48 ways. Tab le 4 shows the number of number of steps required for complet ing each in the most efficient or optimal manner. Tab le 4 Design The des ign of this experiment consisted of a 2 x 2 x 2 that had input task (letter tracing task, sentence decis ion task), output task (tapping, saying) and P D A task (navigation, data entry) as the within subjects factors. Procedures I tested participants individually in a sess ion that lasted 60 minutes. Upon obtaining their written consent , each participant completed the tasks descr ibed below. E a c h participant completed eight condit ions in this experiment. E a c h condition consisted of a set P D A tasks listed in Table 4 combined with an input task and an output task. The eight condit ions were presented in a counterbalanced order by means of a complete Latin Square des ign. Prior to beginning each condit ion, participants were given two practice trials to familiarize them with the input task. Specif ical ly, participants were given detailed instructions for how to complete the secondary task and the procedures for the method of response. T h e s e instructions were the same as Exper iment 3 and are descr ibed in the Procedures sect ion of Exper iment 3. The sequence of events for each trial is a lso descr ibed in the Procedures sect ion in Experiment 3. Part icipants then completed the P D A tasks listed in Tab le 4 together with an input task (i.e., letter tracing task, sentence decis ion task) and method of response for the secondary task (i.e., tapping, saying). I instructed participants to focus on complet ing the P D A tasks accurately and quickly, but to make responses for the input task whenever possib le. 49 Before participants started each P D A task, I gave a verbal explanation of the goal of each task. For navigation tasks, I showed participants a written descript ion of the task and expla ined the goal of the task. I removed the written material from view once participants indicated that they understood the task. For data entry tasks, participants were first provided with the sentence of the to-be-entered information. Next, I asked participants to recite the sentence to ensure that they were able to rehearse the material from memory. I corrected any incorrect words recited by the participant. After correcting the participants, I asked them to repeat the complete sentence again. W h e n participants were able to recite the sentence perfectly, I removed the materials with the written sentence. The procedures for starting and ending the letter tracing and sentence decis ion task were the s a m e as in Exper iment 3 and are descr ibed in the Procedures section of Experiment 3. The procedures for recording participant interactions with the P D A were the same as in Exper iment 1 and are descr ibed in the Procedures sect ion of Experiment 1. Fol lowing the complet ion of the last condit ion, participants were given a verbal debriefing, as well as a written debriefing form, and a course credit. Results Data Preparation The coding manual for the P D A tasks was deve loped in the same way as in Experiment 1. The method for coding P D A task performance was s a m e as in Experiment 1. Per fo rmance for the concurrent tasks was calculated in the same way as in Exper iment 3. I examined the data from the input tasks and corrected for coding errors until accuracy was greater than 99%. The results from the input tasks and P D A tasks were 50 screened for univariate outliers, defined as a score falling more than three standard deviat ions away from the sample mean. No outliers were found from the input tasks. Three outliers were found from the P D A task data. E a c h outlier was replaced with a non-outlying value, a number either 3 standard deviat ions above or below the sample mean, respectively. Twelve participants did not complete at least a single trial on the concurrent tasks while they were performing either the data entry or navigation tasks. The data from these participants were exc luded from the analysis. Data from the remaining 18 participants were used for the ana lyses. PDA Tasks The dependent measures for the P D A tasks were the percentage of correct presses, and the percentage of errors for data entry and navigation tasks, speed of data entry, and the amount of t ime to complete navigation tasks. Overal l , two-thirds of the presses in navigation tasks were correct {M = 62.04). The percentage of P D A tasks errors as a function of input task, output task, and P D A task type are shown in Figure 11. Overal l , error rates for navigation tasks were similar when combined with either the letter tracing task (M = 40.41) or sentence decis ion task (M = 44.33). However, error rates for navigation tasks varied ac ross the input task and the output task. W h e n concurrently performing the letter tracing or sentence decis ion task, error rates for navigation tasks were greater while making motor responses (M = 47.63) compared to voca l responses (M = 37.11). However, this difference was greater when combined with the letter tracing task as compared to the sentence decis ion task. Part icipants made a greater percentage of errors while concurrently performing the letter tracing task when making motor responses {M = 47.36, S D = 20.34) compared to when making vocal responses (M = 33.45, S D = 19.72). W h e n combined with the 51 sentence decis ion task, errors from the navigation tasks were only slightly when making motor responses {M = 47 .90, S D = 15.88) compared to making voca l responses (M = 40.77, S O = 16.44). Figure 11 A n in-depth examinat ion of the error rates for navigation tasks revealed that the percentage of errors ranged from 28 .23% (SD = 28.42) to 57 .16% (SD = 26.00). Part icipants made the most errors when finding the themes in the P D A , fol lowed by finding the font s ize (M = 55.42, S D = 22.18), creating a new folder (M = 47.95, S D = 20.21), finding the Bluetooth version (M = 47.84, S D = 22.30), finding the appointment date (M = 39 .31, S D = 26.22), finding the voice message (M = 37 .11, S D = 24.47), finding the battery (M = 30.30, S D = 25.88), and loading a webpage . The majority of p resses while entering data were correct (M = 87.00). Overal l , the percentage of data entry errors were similar when concurrently performing a letter tracing task (M = 11.96) or sentence decis ion task (M = 14.44). Part icipants made a slightly greater percentage of errors in data entry tasks while concurrently performing the letter tracing task and making vocal responses {M = 12.24, S D = 4.84) compared to when making motor responses (M = 11.69, S D = 6.72). Similarly, participants made a slightly greater percentage of errors for the sentence decis ion task when making vocal responses (M = 15.42, S D = 8.29) compared to when making motor responses (M = 13.47, S D = 7.39). Error rates for the data entry tasks ranged from 10 .45% ( S D = 8.69) to 17.04% (SD = 11.58). Part icipants made the most errors while entering ' chances favors the prepared mind' , fol lowed by entering 'don't bite the hand that feeds you ' (M = 15.92, S D = 18.22), 'all things come to him who waits' (M = 14.27, S D = 9.38), 'birds of a feather 52 flock together' (M = 13.16, S D = 8.60), 'a rolling stone gathers no moss ' (M = 12.93, S D = 7.58), 'look on the sunny side of life' (M = 12.67, S D = 7.46), 'every cloud has a silver lining' {M = 11.36, S D = 6.42), and 'good fences make good neighbors' . A n A N O V A conducted on the error steps data with the input task, output tasks and P D A task as the within-subject factors revealed a significant main effect of P D A task, F (1, 17) = 107.90, MSE = 283.85, p < .05, / = 2.44, such that participants made a significantly greater percentage of errors in navigation tasks compared to data entry tasks. A two-way interaction was found between output task and P D A task, F (1, 17) = 5.66, MSE = 220.23, p < .05, / = 0.51. Fol low-up tests using a Bonferroni correction were conducted to explore the differences in percentage of errors. For navigation tasks, participants made a significantly greater percentage of errors while indicating their answers for the input task by tapping compared to by say ing. No other differences were significant. Typing speed ranged from 4.28 W P M ( S D = 1.49) to 8.20 W P M (SD = 4.19). Typing rate was the quickest for 'Every cloud has a si lver l ining', fol lowed by 'Birds of a feather flock together' (M = 7.43, S D = 2.66), 'Good fences make good neighbors' {M = 7.21, S D = 2.38), 'A rolling stone gathers no moss ' (M = 6.87, S D = 1.99), 'Look on the bright s ide of things' {M = 6 .11, S D = 3.63), 'Don't bite the hand that feeds you ' {M = 5.42, S D = 1.80), 'All things come to him who waits' {M = 5.18, S D = 1.53), and 'Chance favors the prepared mind' . The summary of typing rate by the input task and output task is shown in Figure 12. Typing rate was slightly s lower while participants were concurrently performing the sentence decis ion task (M = 5.23) compared to the letter tracing task (M = 7.42). Figure 12 53 A n A N O V A conducted on the typing speed data from the data entry tasks with the input task and output task as within-subject factors revealed a significant main effect of input task, F (1, 17) = 23.28, MSE = 3.73, p < .001, / = 1.11, such that typing speed was slower while concurrently performing a sentence dec is ion task. A significant interaction was a lso found, F (1, 17) = 11.46, MSE = 5.16, p < .05, / = 0.48. Fol low-up tests using a Bonferroni correction were conducted to explore the dif ferences in the typing speed ac ross condit ions. Overal l , typing speed while performing the sentence decis ion task was significantly s lower than concurrently performing the letter tracing task. In addit ion, typing speed was significantly s lower for the sentence decis ion task when responding vocal ly compared to tapping the answers . Complet ion t imes ranged from 40.22 s (SD = 32.45) to 94.67 s (SD = 57.70). Part icipants took the longest while trying to find the option to change the font s ize, followed by creating a new folder (M = 88.56, S D = 24.86), f inding the avai lable interface themes (M = 87.78, S D = 52.03), finding the vers ion of the Bluetooth (M = 77.39, S D = 38.09), finding a voice recording (M = 69.94, S D = 50.00), finding the battery status (M = 61 .71 , S D = 44.02), finding the appointment date (M = 58.56, S D = 34.32), and loading a webpage . Overal l , complet ion time for navigation tasks was greater when participants concurrently performed the sentence decis ion task (M = 73.49) compared to the letter tracing task (M= 66.61). A s shown in Figure 13, complet ion time was greatest when participants performed the Sen tence Decis ion task by tapping the responses (M = 82.97, S D = 21.17). Complet ion t imes were similar in the three other condit ions. Figure 13 54 A n A N O V A was conducted on the complet ion time data from navigation tasks with the input task and output task as the within-subject factors. The results revealed no significant effects. Secondary Tasks Complet ion time for the letter tracing and sentence decis ion task as a function of input task, output task, and P D A task type is shown in Figure 14. Overal l , participants took slightly longer to complete the sentence decis ion task (M = 28.20) compared to the letter tracing task (M = 26.03). Whi le performing a navigation task, complet ion time for the letter tracing task was slightly greater (M = 28.61) compared to the sentence decis ion (M = 27.64). Whi le performing a data entry task, complet ion time for the sentence decis ion task was greater (M = 28.77) compared to the letter tracing task (M = 23.45). Figure 14 A n A N O V A w a s conducted on the complet ion time data with the input task, output task and P D A task type as the within-subject factors. The results revealed a significant main effect of output task, F (1, 17) = 4 .93, MSE = 98.43, p < .05, / = 0.47. A significant two-way interaction was found between input task by output task, F (1, 17) = 11.54, MSE = 60.52, p < .05, / = 0.77, and input task by P D A task type, F (1, 17) = 4.52, MSE = 78.82, p < .05, / = 0.44. No other main effects or interactions were significant. Fol low-up tests using a Bonferroni correction were conducted to explore the differences in the complet ion t imes across condit ions. The sentence decis ion task and tapping was significantly quicker than by tapping. Complet ion time was significantly greater when participants performed the letter tracing task and navigation tasks. 55 Complet ion time w a s greater when participants combined the sentence decis ion task with data entry tasks compared to navigation tasks, but the difference was not significant. Discussion The primary objective of Exper iment 4 was to explore the types of attention, required by navigation and data entry tasks. W e required participants to perform a second task that either required visuo-spatial (i.e., letter tracing task) or auditory/articulatory (i.e., sentence decis ion task) resources in combinat ion with a data entry or navigation task. A s was found in Exper iment 1 and 2, the overall error rates for navigation tasks were greater than data entry tasks. However, performance for the P D A tasks from this experiment was worse than what was found in Exper iment 1 and 2. Part icipants made a greater percentage of errors for both navigation and data entry tasks, entered data at a slower rate, and took longer to find the target screen for navigation tasks. Complet ion time for the input tasks (i.e., letter tracing task, sentence decis ion task) by output tasks (i.e., say ing, tapping) revealed the s a m e patterns as in Experiment 3. Complet ion time for the letter tracing task did not differ whether participants indicated their answers by tapping or call ing out the answers. The complet ion time for the sentence dec is ion task was greater when participants indicated their responses by saying the answers than by tapping out the answers . T h e s e f indings are consistent with previous notions that tapping the answers is primarily a motor activity, and thus should not interfere with process ing with either input task, while say ing the answers requires mainly articulatory/auditory resources. For navigation or data entry tasks, the error rates did not differ while performing the letter tracing task or the sentence decis ion task. The di f ferences in P D A task error 56 rates were revealed by the combinat ion of input and output task. Navigation error rates were significantly higher when participants were asked to indicate their responses by tapping the answers compared to saying the answers , but the difference was quite smal l . The opposi te was found for data entry tasks. For both the letter tracing and sentence decis ion task, participants made a greater percentage of data entry errors while responding vocal ly compared to tapping the answers . However , none of the data entry error rates ac ross condit ions were significantly different. Overal l , typing speed was slower while performing the sentence decis ion task compared to the letter tracing task. Moreover, typing speed was slightly s lower while performing the sentence decis ion task and responding vocal ly compared to responding by tapping. However , none of these differences were statistically different. These findings suggest that the s lower typing rate is a result of the demands placed on the auditory/articulatory resource when performing data entry and concurrently making verbal responses . One of the poss ib le reasons for the elevated error rates, greater complet ion t imes and s lower typing speeds for the P D A tasks is the greater resource demands for the letter tracing and sentence decis ion task, compared to a tone discrimination task. This is understandable s ince the letter tracing and sentence dec is ion task required much more resources at any moment compared to a tone discrimination task. Performing tone discrimination simply requires the participant to attend to and to compare whether the previously presented tone and the present tone sound different. The response is a binary dec is ion, either the tones are the s a m e or they are not the same. W h e r e a s for the input tasks in this experiment, participants were required to first hold in mind the stimuli (e.g., a picture of the letter or the words in the sentence), then retrieve the categorizat ion instructions, and finally make categorizat ion judgements. 57 Thus the elevated P D A error rates can be partly attributable to the fact that performing the concurrent tasks by itself is fairly difficult. Complet ion t imes for the secondary tasks varied as a function of the output task. The most prominent effect can be seen when participants were performing the sentence decis ion task. Complet ion time was greater when participants indicated their responses by saying the answers compared to tapping their answers on a keyboard. This finding is consistent with the notion that categorizing words in a sentence and indicating responses verbal ly both require auditory/articulatory resources. The critical results that are pertinent to the objective of this experiment are the complet ion time for the secondary tasks when combined with the P D A tasks. The results revealed that data entry requires more auditory/articulatory resources, while navigation requires both visuo-spat ial and auditory/articulatory resources, but more of the former. 58 Genera l D iscuss ion The overall goal of this thesis was to investigate the attentional demands of various P D A tasks. To ach ieve this goal , I used several approaches . The first approach was to validate a modified version of the dual-task methodology to a s s e s s the amount of attentional demands of the P D A tasks. The second approach was to explore the amount of attention required by the P D A tasks within the dual- task methodology framework. The third approach was to validate two tasks that would reveal the different types of attention. The fourth approach was to explore the types of attention required by these P D A tasks using the dual-task methodology with the tasks val idated in the third approach. The main f indings of these four approaches were d i scussed in Exper iments 1 through 4, and will be briefly summar ized here. The implications for design will be d iscussed , fol lowed by limitations and directions for future research. Summary of Research Findings The first approach towards achieving the goal of this thesis was to explore the attentional demands of navigation and data entry tasks using a dual-task methodology developed for this purpose. Th is approach forms the investigation reported in Experiment 1. Us ing the tone discrimination task as the secondary task, the results revealed that performance for the tone discrimination task was lower when concurrently completed with the navigation tasks compared to data entry tasks. The results reported in Exper iment 1 were fol lowed up in Exper iment 2 in order to clarify whether the deve loped methodology a s s e s s e d the attentional demands or the difficulty of task switching. A new level of difficulty was added to the tone discrimination task (i.e., easy and hard) that would manipulate the required attention for the tone discrimination task, but not the difficulty of switching between tone discrimination and a P D A task. The pattern of results from the tone discrimination task revealed that the 59 methodology a s s e s s e d attention. B a s e d on these f indings, it w a s suggested that navigation requires more attention than data entry. In an investigation to find out whether navigation and data entry tasks requires different types of attention, which forms the basis of Exper iment 3 and 4, I first validated a method descr ibed by Brooks (1968) which may be used to reveal the types of attention. The results suggested that the method revealed two tasks that drew on visuo-spatial or auditory/articulatory resources. Using the dual-task methodology in Exper iment 4, I had participants complete a P D A task concurrently with one of the tasks val idated in Exper iment 3 in order to find out whether different types of attention were required for the P D A tasks. The findings revealed that data entry drew heavily articulatory/auditory resources, whereas navigation required both articulatory/auditory and visuo-spat ial resources, but more of the latter. General Limitations There are severa l limitations to the experiments reported in my thesis. First, all of the participants were undergraduate students and may not be representative of the population, especia l ly of older adults. There are several reasons why this is a concern. First of al l , attentional capaci ty, in addition with other cognit ive abilit ies, has been proposed to reduce with aging. S e c o n d , older adults are less famil iar with technology, a factor known to inf luence the attention required to use technology. Due to both of these reasons, the secondary task decrements are likely to reveal a larger effect with older adults than with undergraduates. A s a result, caution must be taken when general iz ing the results of this study to an older population. A second limitation concerns the research des ign of Exper iment 2 and 4. Ideally, the design should a lso have included complete counter-balancing of the combinat ion of 60 P D A task sets with the condit ions of the secondary task, such that each P D A task appeared equal ly likely with each combinat ion of the condit ions of the secondary task. Thus , the f indings are confounded with both the intended manipulat ions from the exper iments, as well as possibly the nature of the speci f ic P D A tasks that were combined with each of the condit ions. In order to perform complete counter-balancing with the P D A task sets, I would require a total of 64 participants for Exper iment 2, and 512 participants for Exper iment 4. However, obtaining a sample of this s ize was not feasible during the course of the academic year. Fol low-up studies in the near future can be conducted to increase the samples to appropriate s izes for further investigation. For the moment, caution should be taken when interpreting these results. The third limitation involves the input tasks used in Exper iment 4. Whi le the tasks were valid for a s s e s s i n g the types of attention required by the P D A tasks, I bel ieve that the input tasks were too difficult to be performed s imul taneously with the P D A tasks. The elevated percentage of errors, greater complet ion t imes for navigation tasks, and slower typing speed compared to Experiment 1 and 2 indicate that participants were having difficulty performing the P D A tasks well . This finding may suggest that performing a input task and a P D A task were simply too attention demand ing , and found it difficult to maintain P D A task performance while trying to perform a input task. To fully a s s e s s the type of attention required, another set of concurrent tasks may be needed. Finally, technology familiarity of the participants was not a s s e s s e d in these experiments. Familiarity of the task is known to influence the amount of attention required to perform the task. However, while technology familiarity within the undergraduate population is likely to differ among individuals, the variability should be quite smal l . In addit ion, at the time that these exper iments were conducted the device was new and it was unlikely that participants had any exper ience in using this specif ic 61 device. This reasoning is supported by the smal l variability seen in the performance measures of the P D A tasks. However, a follow-up study that a lso a s s e s s e s the technology familiarity would strengthen the validity of the f indings. Implications for Design A s ment ioned in the introduction, P D A s are frequently used in various environments. Thus , a number of guidel ines and principles can be implemented to ease the attentional demands in order to facilitate efficiency and eff icacy. First, to reduce the amount of attention p laced on navigation, one of the suggest ions would be to reduce the number of menu layers required to find the target sc reen . A c r o s s the experiments, trying to locate the Exce l worksheet was one of the tasks performed most efficiently and easi ly (i.e., least amount of time and errors). O n e of the possib le reasons is that the Exce l shortcut shows up on the Start menu after the first a c c e s s . Thus , after pressing the Start menu, participants could easi ly recognize the option. For younger adults where technology use is frequent and var ied, this design suggest ion may be even more effective for older users where technology familiarity is low. However, reducing the levels in the hierarchy may lead to interfaces with many menu options, creating an environment with too many cho ices that may overload attention. Th is is especia l ly true for P D A s , where screen space is limited and there is a trade-off with displaying the amount of information and the visual clarity of which the information can be v iewed. Therefore, it is important that the navigation structure is balanced between depth and breadth, so that the items in each menu and number of steps that have to be taken to complete a task are ba lanced (Westerman, 1995). There is previous research that demonstrates that the organizat ion of menu structures can inf luence attentional demands . Wes te rman , Dav ies , G lendon , S tammers and Matthews (1995) found that a l inear information structure was beneficial for all age 62 groups, in terms of search t imes. Stanney & Sa lvendy (1995) found that 2D visual hierarchies (all levels visible) and linear structures (open folders with their fi les presented) were more efficient in supporting individuals with low spatial ability than interfaces where s o m e parts of the information structure were hidden (buttons presenting only main categories). Based on these f indings, interfaces that are rich in presenting the structure of the information space may remove the need to mentally construct the environment, which in turn could be beneficial for individuals with low spatial ability (Stanney & Sa lvendy, 1995; Vincente & Wi l l iges, 1988). The second recommendat ion concerns the visual presentat ion of information. From informal observat ions, I noticed that participants spent longer looking at screens that d isplayed more icons and menu options than when fewer options were avai lable. It is important that the user is able to focus the attention on the task at hand. Whi le engaged in multiple activities simultaneously, irrelevant information or cluttered backgrounds on a computer screen can be distracting (Connel ly & Hasher , 1993). Interfaces with reduced information content make it eas ier to focus attention on relevant information and reduce the time spent on information searchers . A id in focusing attention can be provided by structuring the information, providing spatial and temporal cues , and manipulat ing the screen layout (Singh, 2000; see P reece , Rogers , Benyon Hol land & Carey , 1994). Present ing information to various sensory channe ls can reduce the attentional load on one single sys tem. The capacity of the sensory sys tem limits the amount of information that can be learnt or absorbed. Taking advantage of all the avai lable sensory sys tems by presenting information via different sys tems can facilitate acquisit ion of knowledge. For example, Brunken, Ste inbacher , P l a s s and Leutner (2002) found that audiovisual presentation of text-based and picture-based learning materials 63 induced less cognit ive load, and facilitated knowledge acquisi t ion, compared to the visual-only presentat ion of the s a m e material. Tardieu and Gyse l inck (2003) investigated whether the use of multimodal information presentation would reduce cognitive over load on working memory by using the subsys tems in working memory. Finally, providing environmental support might enhance learning and subsequent search performance. Environmental support consists of information in the environment that facilitates encoding or retrieval of information and can reduce the amount of cognitive process ing that is needed (Jones & Bayen , 1998). Navigat ion is effortful and attention demand ing . O n e suggest ion is to provide certain pop-up dialog boxes that may guide finding the task based on the statistical probability of the most commonly used appl icat ions. However , it is a lso important to consider whether or not the environmental support increases the cognit ive demands to the point that it becomes too demanding. It is therefore important to investigate which type of support cons idered to be useful for different tasks, and to what extent individuals can make use of the environmental support that is provided (McDowd & Shaw, 2000). Future Directions A s descr ibed in the introduction, one of the primary motivations for conducting this line of research is to understand the factors that affect usability of P D A s . The ser ies of exper iments descr ibed in this thesis formed as pilot studies to which the methods could be appl ied to a s s e s s the attentional demands of P D A use across the l i fespan. One line of future research can explore how attentional demands of different P D A tasks change across the adult l i fespan. The findings from the exper iments in this thesis have suggested that navigation is more demanding than data entry. From what is known in the area of cognit ive aging, aging is assoc ia ted with loss of avai lable resources (Craik, 1983; Cra ik & Byrd, 1982; Rabinowitz, Cra ik & Acke rman , 1982). 64 Given these two premises, it would be reasonable to predict that P D A task performance would be worse among older adults compared to younger adults, which could be partly explained by the reduced attentional capacity from aging. Al though previous research has led to the formulation of extensive design guidel ines for cell phone interfaces, web pages, and var ious software appl icat ions for the P C , there is no research that specif ically focuses on reducing attentional load on P D A tasks. S ince P D A s relies on a different interaction technique than a desktop or laptop computer (i.e., stylus and touchscreen), and has a different screen s ize than other technological dev ices (such as cell phones or personal computers) , it is reasonable to bel ieve that the findings will yield something novel. Another line of this type of research is to investigate whether aging has the same effect on the different types of attentional resources. At present, there is a lot of research that indicates that aging affects spatial ability, but not verbal ability. In addit ion, there has been no study that compares the two types of attention. One of the implications of this theoretical research is to determine whether to focus innovative research and des ign guidel ines to change P D A tasks that take advantage of the ability that is less suscept ib le to aging. The other implication would be to complement interface and device speci f icat ions with training. The third line of research can explore whether older versus younger adults are more chal lenged by navigation than data entry tasks. In genera l , most studies that have investigated the effects of age on technology use have found that older adults typically make more errors and take longer to complete the task compared to younger adults on both data entry tasks (Brewster & Cryer, 1999; C z a j a , H a m m o n d , B lascov ich & Swede , 1989; Cza ja & Sharit , 1998) and information search tasks. In a pilot study with individuals ranging from 18 to 85 years of age, we found that older adults made more 65 errors on navigation and data entry tasks, and took longer for navigation tasks compared to younger adults (Graf & Li , 2007). However, it is unclear the specif ic nature of each task that makes them more difficult to perform by older adults. Future studies can explore the speci f ic factors that may contribute to the difficulty of the task. 66 Conc lus ion The results from the exper iments of my thesis demonstrated that navigation and data entry performed on a P D A differ in the amount of attention, and type of attention. These f indings are consistent with the intuition that data entry is eas ier s ince it involves mostly a motor activity, whereas navigation requires a greater amount of cognitive process ing, such as remember ing what you need to f ind, where you are in the system, and the correct opt ions that will lead to the desired appl icat ion. T h e findings are a lso consistent with the notion that data entry involves primarily rehearsing the to-be-entered information while navigation requires creating a mental model of the menu system. A s d iscussed in the introduction, performing these tasks under highly attention demanding environments may compromise usability of certain tasks. The results from this thesis contribute to the theoretical understanding of the interaction between the attentional demands of P D A tasks and the environment. This understanding opens up the possibil i t ies of further research that could be conducted to pinpoint the specif ic characterist ics of these tasks that make them attention demand ing . T h e s e findings will have implications for the des ign guidel ines of P D A s and , and perhaps, other technologies. 67 Table 1 The number of Optimal Steps/Presses for each PDA Task as an Indicator of Task Difficulty P D A T a s k s Number of required steps Order of T a s k s Navigation Data entry Total Navigation tasks Check the battery 4 / 4 2 Retr ieve appointments 6 / 6 4 Find a picture 4 / 4 5 Data entry tasks Enter contact information 3 116 119 1 Enter expense information 4 74 78 3 Make an appointment 6 23 29 6 Total steps 25 213 238 68 Table 2 Percentage of Correct, Incorrect, and Adjusted Identifications in the Tone Discrimination Task across Conditions Identification Condit ion Test Score Base l ine Data Entry Navigation M CI.95 M CI.95 M CI.95 Correct 75.00 70.21, 48.10 39.73, 32.78 24.95, 79.79 56.46 40.61 Incorrect 9.50 5.43, 6.26 4 .39, 9.31 5.85, 13.56 8.12 12.77 Adjusted 65.50 58.39, 41.84 32.78, 23.47 14.05, 72.61 50.91 32.90 69 Table 3 The Number of Steps/Presses Required by each PDA Task across Tone Discrimination Task Difficulty P D A T a s k s Number of required steps Order within Sets Navigation Data entry Total E a s y Nav igat ion 3 Find Picture 4 / 4 1 Retr ieve Appt Information 6 / 6 2 C h e c k Battery 4 / 4 3 E a s y Entry b Enter Sen tences 4 88 92 1 N a m e New Folder 5 19 24 2 Enter Contact Information 3 116 119 3 Hard Entry c Enter Emai l M e s s a g e 4 36 40 1 Enter Appointment Information 6 23 29 2 Enter E x p e n s e Information 4 74 78 3 Hard Nav igat ion d Find Vo ice Record ing 5 / 5 1 Find Emai l Add ress 3 / 3 2 Find Owner 3 / 3 3 Total steps 48 356 404 a Set A b S e t B c S e t C d S e t D 70 Table 4 P e r c e n t a g e of C o r r e c t , Incor rec t , a n d A d j u s t e d Ident i f icat ions in the T o n e D i s c r i m i n a t i o n T a s k a c r o s s T a s k C o n d i t i o n s a n d T o n e D i s c r i m i n a t i o n Di f f icu l ty denti f ication | Condi t ion Tes t S c o r e j j Base l i ne | Data Entry j Nav igat ion E a s y : Hard I E a s y , Hard j E a s y [ Hard M \ S E M ; SE M | S E : m | S E | M \ S E M \ SE ! Correc t 90 .23 ! 2.06 . J 3 0 . 6 3 | 5.07 53.09 | 5.49 ] 43.94 | 5.36 I 45 .83 5.62 28 .86 | 5.07 ! Incorrect 1.82 0.34 3.04 I .74 1.17 I 0.23 : 1.53 | 0.27 j 1.73 0.36 1.91 ! 0.41 | Ad jus ted 88 .42 2.13 i 57.59 ! 5.62 51.92 | 5.51 42.40 | 5.42 | 44 .10 5.62 26.96 j 5.14 | ->J Table 5 The number of Steps/Presses Required by each PDA Task Across Conditions Number of Order within Sets required steps Letter Tracing/Tapping/Navigat ion Find font s ize 6 1 Retr ieve appointments 5 2 Letter Trac ing/Saying/Navigat ion Check the battery 5 1 Find voice m e s s a g e 4 2 Sentence Decis ion/Tapping/Navigat ion Create new folder 5 1 Find Bluetooth vers ion 5 2 Sentence Decis ion/Saying/Navigat ion Find themes avai lable in device 3 1 Load webpage 4 2 Letter Trac ing/Tapping/Data entry A rolling stone gathers no moss . 32 1 G o o d fences make good neighbors. 32 2 Letter Trac ing/Say ing/Data Entry Birds of a feather flock together. 34 1 Every cloud has a si lver lining. 32 2 Sentence Dec is ion/Tapping/Data Entry Look on the sunny s ide of life. 31 1 All things come to him who waits. 33 2 Sentence Dec is ion /Say ing/Data Entry C h a n c e favors the prepared mind. 32 1 Don't bite the hand that feeds you. 35 2 72 Image removed due to copyright. V iew original image at www.hp.com Figure 1. Picture of the dev ice used for the experiment. 73 1400 w 1200 E (/> c o a a: 1000 800 isilPtSf 600 Basel ine Data entry Condition Navigation Figure 2. R e s p o n s e t ime (mean of median) for correct identification of target tones across condit ions. f vertical bars represents 9 5 % conf idence intervals 74 6 ^ S2 o k_ k. UJ \u00C2\u00AB*-o 0) U) c o o L . (0 c o a. a: 900 800 700 600 tulip 's Basel ine Data entry Condition Navigation Figure 4. R e s p o n s e t ime (mean of medians) of correct identifications in the tone discrimination task as a function of tone discrimination difficulty and task condit ions, f vertical bars represents standard errors 76 Figure 5. Stimuli for the practice trials of the letter tracing task. 77 5 Figure 6. Stimuli for the test trials of the letter tracing task. 78 t ime respond top/bottom re 7. Illustration of a trial of the letter tracing task with top/bottom instructions. Figure 8. Illustration of a trial of the letter tracing task with outside instructions. 80 N N N N N N N N N N N N Figure 9. Image of s taggered Y and N of the output task (pointing). 81 30 * 20 10 0 Pointing El Letter Tracing task \u00E2\u0080\u00A2 Sentence Dec is ion task Tapping Output Tasks Saying Figure 10. M e a n complet ion time as a function of input task and output task. + vertical bars represent standard error. 82 Navigation \u00E2\u0080\u00A2 Saying \u00E2\u0080\u00A2 Tapping Navigation Data Entry Sentence Dec is ion Task Data Entry Letter Tracing task Figure 11. Percen tage of P D A task errors as a function of input task, output task, and P D A task type. * vertical bars represent standard error. 83 10 \u00E2\u0080\u0094 8 Q . B 6 o Q . V> T J k. O 5 \u00E2\u0080\u00A2 Tapping \u00E2\u0080\u00A2 Saying 1 Letter Tracing task Sentence Dec is ion task Task Type Figure 12. Typing speed (words per minute) in data entry tasks as a function of input task, and output task. t vertical bars represent standard error. 84 100 _ 80 w o O (A 0) E 60 c o 1 40 Q . E o \u00C2\u00B0 20 n Tapping \u00E2\u0080\u00A2 Saying Letter Tracing task Sentence Dec is ion task Task Type Figure 13. Complet ion time in navigation tasks as a function of input task, and output task. f vertical bars represent standard error. 85 El Saying \u00E2\u0080\u00A2 Tapping Navigation | Data Entry Letter Tracing task Navigat ion | Data Entry Sentence Dec is ion Task Figure 14. Complet ion t ime of the concurrent tasks as a function of input task, output task, and P D A task type. t vertical bars represent standard error. 86 References Anderson , N. D., l idaka, T., C a b e z a , R., Kapur, S . , Mc in tosh , A . R., & Craik, F. I. M. (2000). The effects of divided attention on encod ing- and retrieval-related brain activity: A P E T study of younger and older adults. Journal of Cognitive Neuroscience, 72(5), 775-792. Baddeley, A . D., & Hitch, G . J . (1974). Work ing memory. In G . H. Bower (Ed.), The psychology of learning and motivation: Advances in research and theory (pp. 47-89). N e w York: A c a d e m i c P ress . Blair, J . R., & Sp reen , O. (1989). Predict ing premorbid IQ: A revision of the national adult reading test. The Clinical Neuropsychologist, 3, 129-136. Brewster, S . A . & Cryer, P. G . (1999). Maximiz ing screen s p a c e on mobile computing dev ices. Summary Proceedings CHI'99 (pp. 224-225). New York: M c G r a w Hill. Broadbent, D. E (1957). A mechanica l model for human attention and immediate memory. Psychological Review, 64, 205-215. Broadbent, D. E. (1958). Perception and communication. London: Pergamon P ress Ltd. Brooks, L. R. (1968). Spat ia l and verbal components of the act of recall . Canadian Journal of Psychology, 22(5), 349-368. Brunken, R., Ste inbacher , S . , P l a s s , J . L., & Leutner, D. (2002). A s s e s s m e n t of cognit ive load in mult imedia learning using dual-task methodology. Journal of Experimental Psychology, 49(2), 109-119. C o h e n , J . (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hi l lsdale, N J : Er lbaum. Connel ly , S . L., & Hasher , L. (1993). Aging and the inhibition of spatial location. Journal of Experimental Psychology: Human Perception and Performance, 19, 1238-1250. Cors i , P. M. (1972). Human memory and the medial temporal region of the brain. Dissertation Abstracts International, 34, 819B. Craik, F. I. M. (1983). On the transfer of information from temporary to permanent memory. Philosophical Transactions of the Royal Society of London, Series B, 302, 341-359. Craik, F. I. M. , & Byrd, M. (1982). Ag ing and cognitive deficits: The role of attentional resources. In F. I. M. Craik & S . Trehub (Eds.) , Aging and cognitive processes (pp. 191-211). N e w York: P lenum. Craik, F. I. M. , Govon i , R., Naveh-Ben jamin , M. , & Anderson , N. D. (1996). The effects of divided attention on encoding and retrieval p rocesses in human memory. Journal of Experimental Psychology: General, 125(2), 159-180. Cza ja , S . , H a m m o n d , K., B lascov ich , J . & Swede , H. (1989). Age-re lated differences in learning to use a text-editing system. Behavior and Information Technology, 8, 309-319. Cza ja S . J . , & Sharit J . (1998). Abi l i ty-performance relat ionships as a function of age and task exper ience for a data entry task. Journal of Experimental Psychology. Applied, 4, 332-351. 87 Dahlback, N., Hook, K., & Sjol inder, M. (1996). Spat ia l cognit ion in the mind and in the world - the case of hypermedia navigation. Proceedings of the Eighteenth Annual Conference of the Cognitive Sciences Society (pp. 195-200). S a n Diego: Lawrence Er lbaum Assoc ia tes . Emb i , P. J . (2001). Information at hand: Using handheld computers in medicine. Cleveland Clinic Journal of Medicine, 68(10), 840-853. Fernandes, M. A . , & Moscov i tch , M. (2000). Divided attention and memory: Ev idence of substantial interference effects at retrieval and encod ing. Journal of Experimental Psychology: General, 729(2), 155-176. F leetwood, M. D., Byrne, M. D., Centgraf, P. , Dudziak, K. Q. , L in , B., & Mogi lev, D. (2002). A n evaluat ion of text-entry in palm O S - Graffiti and the virtual keyboard. Proceedings of the Human Factors and Ergonomics Society 46th Annual Meeting, 597-601. France, S . J . , Rosner , B. S . , Hansen , P. C , Ca lv in , C , Talcott, J . B., R ichardson, A . J . et a l . (2002). Auditory f requency discrimination in adult developmental dyslexics. Perception & Psychophysics, 64(2), 169-179. Freudenthal , D. (2001). A g e differences in the performance of information retrieval tasks. Behavior & Information Technology, 20(1), 9-22. G o o d m a n , J . , Gray , P. , K h a m m a m p a d , K., & Brewster, S . (2004). Us ing landmarks to support older people in navigation. Proceedings of Mobile HCI, 3160, 38-48. Graf, P. & L i , H. (2007). Cognit ive ability factors which affect P D A usability across the l i fespan. Proceedings of the IADIS International Conference on Interfaces and Human Computer Interaction, 99-108. Hawthorn, D. (2000). Poss ib le implications of aging for interface designers. Interacting with Computers, 12, 507-528. l idaka, T., Ande rson , N. D., Kapur, S . , C a b e z a , R., & Craik, F. I. M. (2000). The effect of divided attention on encoding and retrieval in ep isod ic memory revealed by positron emiss ion tomography. Journal of Cognitive Neuroscience, 72(20), 267-280. Jones , B. D., & B a y e n , U. J . (1998). Teach ing older adults to use computers: Recommendat ion based on cognitive aging research. Educational Gerontology, 24, 675-689. Kahneman , D. (1973). Attention and effort. Eng lewood Cliffs, N J : Prent ice-Hal l . Kesse l s , R. P. C , van Zandvoort , M. J . E. , Pos tma , A . , Kappe l le , L. J . , & Haan , E. H. F. (2000). The corsi block-tapping task: Standardizat ion and normative data. Applied Neuropsychology, 7(4), 252-258. Klauer, K. C , & Stegmaier , R. (1997). Interference in immediate spatial memory: Shifts of spatial attention or central-executive involvement? The Quarterly Journal of Experimental Psychology, 50/4(1), 79-99. Kl ingberg, T., & Ro land , P. E. (1997). Interference between two concurrent tasks is assoc ia ted with activation of overlapping fields in the cortex. Cognitive Brain Research, 6(1), 1-8. 88 Kotary, L , & Hoyer, W . J . (1995). A g e and the ability to inhibit distractor information in visual select ive attention. Experimental Aging Research, 21(2), 159-171. Li , H. & Graf, P. (2007). Cogni t ive, perceptual, sensory and verbal abilities as predictors of P D A text entry error and instructions across the l i fespan. In D. Harris (Ed.), Human-Computer Interaction International: Vol. 4562. Engineering Psychology, and Cognitive Ergonomics (pp. 349-358) . Heidelberg: Spr inger-Ver lag. M a c K e n z i e , I. S . , & Soukoreff, R. W . (2002). A character- level error analysis technique for evaluat ing text entry methods. Proceedings of the 2nd Nordic Conference on Human-Computer Interaction (NordiCHI), 31, 243-246. Mazzon i , D., & Dannenberg , R. (2000). Audaci ty (Version 1.2.6) [Computer Software]. Carneg ie Mel lon , U S . M c D o w d , J . M. , & Shaw, R. J . (2000). Attention and Ag ing : A Funct ional Perspect ive. In F. I. M . Craik & T. A . Sa l thouse (Eds.), The handbook of aging and cognition (pp. 221-292). N e w Jersey : Lawrence Er lbaum Assoc ia tes . Moffat, S . D., Zonde rman , A . B., & Resn ick , S . M. (2001). A g e dif ferences in spatial memory in a virtual environment navigation task. Neurobiology of Aging, 22, 787-796. Moray, N. (1967). W h e r e is attention limited: A survey and a model . Acta Psychologica, 27, 84-92. Morris, M. G . , & Venka tesh , V . (2000). A g e dif ferences in technology adoption dec is ions: Implications for a changing work force. Personnel Psychology, 53, 375-403. Murray, L. L., Hol land, A . L., & B e e s o n , P. M. (1997). Grammatical i ty judgments of mildly aphas ic individuals under dual-task condit ions. Aphasiology, 11, 993-1016. Naveh-Ben jamin , M. , Craik, F. I. M. , G u e z , J . & Dori, H. (1998). Effect of divided attention on encoding and retrieval p rocesses in human memory: Further support for an asymmetry. Journal of Experimental Psychology: Learning, Memory and Cognition, 24(5), 1091-1144. Parasu raman , R., & Dav ies , D. R. (Eds.). (1984). Varieties of attention. New York: A c a d e m i c P ress . P reece , J . , Rogers , Y , Benyon , D., Hol land, S . , & Carey , T. (1994). Human-computer interaction. S ingapore : Add ison-Wes ley . Rabinowitz, J . C , Craik, F. I. M. , & Acke rman , B. P. (1982). A processing resource account of age dif ferences in recall. Canadian Journal of Psychology, 36, 325-344. Rei tan, R. M. (1992). Trail making test: Manual of administration and scoring. Tuscon , Ar izona: Rei tan Neuropsychology Laboratory. Schneider , W . , & Shriff in, R. M. (1977). Control led and automatic human information process ing: I. Detect ion, search , and attention. Psychological Review, 84(1), 1-66. S ingh, S . (2000). Designing intelligent interfaces for users with memory and language limitations. Aphasiology, 14(2), 157-177. 89 Soukoreff, R. W. , & M a c K e n z i e , I. S . (2003). Metr ics for text entry research: A n evaluation of M S D and K S P C , and a new unified error metric. Proceedings of the ACM Conference on Human Factors in Computing Systems (CHI), 113-120. Stanney, K. M. , & Sa lvendy , G . (1995). Information visual izat ion; assist ing low spatial individuals with information a c c e s s tasks through the use of v isual mediators. Ergonomics, 38(6), 1184-1198. Tardieu, H., & Gyse l inck , V . (2003). Work ing memory constraints in the integration and comprehens ion of information in a mult imedia context. In H. van Oostendorp (Ed.), Cognition in a digital world (pp. 3-24). M a h w a h , New Jersey : Lawrence Er lbaum Assoc ia tes . Troyer, A . K., Winocur , G . , Craik, F. I. M. , & Moscov i tch , M. (1999). Source memory and divided attention: Rec iproca l costs to primary and secondary tasks. Neuropsychology, 13(4), 467-474. Vincente, K. J . , Hayes , B. C , & Wil l iges, R. C . (1987). Assay ing and isolating individual di f ferences in search ing a hierarchical file sys tem. Human Factors, 29(3), 349-359. V incente, K. J . , & Wi l l iges, R. C . (1988). Accommodat ing individual dif ferences in search ing a hierarchical file sys tem. International Journal of Man-Machine Studies, 29, 647-668. Wechs ler , D. (1981). Wechsler Adult Intelligence Scale-Revised Manual. S a n Antonio: Psycho log ica l Corporat ion. Weste rman, S . J . , Dav ies , D. R., G lendon , A . I., S tammers , R. B., & Matthews, G . (1995). A g e and cognit ive ability as predictors of computer ized information retrieval. Behavior and Information Technology, 14, 313-326. Wes te rman, S . J . (1995). Computer ized information retrieval: Individual dif ferences in the use of spatial vs . nonspatial navigational information. Perceptual and Motor Skills, 81, 771-786. Wickens , C . D. (1992). Engineer ing psychology and human performance. N Y , NY: Harper Col l ins. Wobbrock, J . O. , & Myers , B. A . (2006). Analyz ing the input stream for character- level errors in unconstrained text entry evaluat ions. ACM Transactions on Computer-Human Interaction, 73(4), 458-489. Wobbrock, J . O. , Myers , B. A . , & Aung , H. H. (2004). Writ ing with a joystick: A compar ison of date stamp, select ion keyboard, and EdgeWri te . Proceedings of the Graphics Interface Conference, 62, 1-8. Wobbrock, J . O. , Myers , B. A . , & Kembe l , J . A . (2003). EdgeWr i te : A sty lus-based text entry method des igned for high accuracy and stability of motion. Proceedings of the ACM Symposium on User Interface Software and Technology (UIST), 61-70. 90 Z h a , Y . , & S e a r s , A . (2001). Data entry for mobile dev ices using soft keyboards: Understanding the effect of keyboard s ize. In M. J . Smi th , G . Sa lvendy, D. Harris, & R. J . Koubek (Eds.) , Usability Evaluation and Interface Design: Cognitive Engineering, Intelligent Agents and Virtual Reality, (pp. 16-20). Mahwah , N J : Lawrence Er lbaum Assoc ia tes Inc. Ziefle, M. , & Bay , S . (2004). Mental models of a cellular phone menu. Compar ing older and younger novice users. In S . Brewster & M. Dunlop (Eds.) , MobileHCI, LNCS 3160, 25-37. 91 "@en . "Thesis/Dissertation"@en . "10.14288/1.0101006"@en . "eng"@en . "Psychology"@en . "Vancouver : University of British Columbia Library"@en . "University of British Columbia"@en . "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en . "Graduate"@en . "Attentional demands of different types of PDA tasks"@en . "Text"@en . "http://hdl.handle.net/2429/31888"@en .