This item is only available as the following downloads:
Empathy Enhancement in Text Base d Computer Mediat ed Communication BY Sivens J. Glaude A Thesis Submitted to the Division of Psychology New College of Florida In partial fulfillment of the requirements for the degree Bachelor of Arts Under the sponsorship of Dr. Michelle Barton Sarasota, Florida May, 2012
ii Acknowledgments There are many people who helped me ove r these past few years and cant thank them enough, but I will try. I would like to thank Dr. Barton, my thesis sponsor, for her wisdom and guidance during the entirety of my th esis process, and her constant efforts to help me hone my skills as a writer and development as a research student. I would like to thank Dr. Graham for his never ending support of me, from the first day I took a step into a course at NCF into these final moments just before I take my last graduating steps. I would like to think Dr. Harley for being hers elf, that is, a wonde rful professor and a magnificent person. I would like to thank the new college psychology faculty as a whole for being some of the most insp irational individuals that I k now and Im really lucky I got the chance to work with you all. I would lik e to think Wendy Bashant for her diligence in making sure I stayed focused and on track. I would also like to thank the invaluable people at the new college offices especially those who have worked with me at the counseling and wellness center, the business offi ce, the office of fina ncial aid, the provost office, and the student affairs office. Soni a Wu, Dr. J, Kathy Allen, Konnie Kruzec, you all are great people! I would also like to thank my peers and roommates who have given me much caring support throughout the years (R achel S., Carmen G., Brian W., Wesley D., Mindy T., Ben G., Tian M., Laurel C., Sa m E., Nick M., Beck F., Naushin J., Raluca C., Jyoti D., Mark W., Greeley D., and Jeremy B.)! I would like to give a very special thanks to Susan Castlen, my high school c ounselor, who kept me from going down the path of mediocrity and convinced me to appl y to New College. Last but not least I would like to thank my family who have been ther e for me since before I could remember, I love you all!
iii Table of Contents ACKNOWLEDGMENTS ii TABLE OF CONTENTS iii LIST OF TABLES v ABSTRACT vi INTRODUCTION 1 Communication and Empathy 3 Face-to-Face Communication a nd Nonverbal Behavior 8 Computer-Mediated-Communication 12 Problems with Text Based CMC 14 Text Based CMC Potential Solutions 20 Typography 26 CURRENT STUDY 29 METHOD 31 Participants 31 Materials 32 Procedure 36 Dependant variables 37 RESULTS 39 Actual Communication Accuracy 39 Reader Confidence 41 Writer Confidence 42 DISCUSSION 42 Accuracy 42 Reader Confidence 45
iv Writer Confidence 46 Limitations 47 Future Research 47 Conclusions 48 References 49 REFERENCES 52 TABLES 53
v List of Tables TABLE 1: Average Reader Accuracy for Each Emotion Across all Conditions 53 TABLE 2: ETS Effect on Reader Accuracy for Each Emotion 54 TABLE 3: Average Reader Confidence per Emotion 55 TABLE 4: Average Writer Confidence pe r Emotion 56 TABLE 5: ETS Effect on Writer Confidence for Each Emotion 57
vi Empathy Enhancement in Text Based Computer Mediated Communication Sivens Glaude New College of Florida, 2010 Abstract Empathy, defined as the ability to put oneself into the mental shoes of another person to understand her emotions and feeli ngs (Goldmann was cited in Decety & Lamm, 2006, p. 1147 ), is a fundamental part of human behaviors su ch as communication, cooperation, and altruism. Many of the signals we use to understand the emo tions of others, or to be empathetic, are based off of nonverbal behavior such as facial expr essions, tone of voice, and gestures. However, traditional text based communication lacks many of the nonverbal cues we normally use to communicate emotions during face-to-face interacti ons. Media richness theory states that because of its lack of cues, text based communication is a poorer form of communication than face-to-face communication, in short it takes longer to communicate a lot of information over text than in a face-to-face conversation. Several studies have also suggested that the lack of cues in text based communication is associated with problems such as miscommunication over text. The present study investigated the effect of typographical cues, such as bold type, font type, and emoticons on participants ability and confidence at communica ting over e-mail. Overall the results showed that the use of typographical cues in email to convey emotions significantly increased the accuracy of empathetic communication between pa rticipants. These effects were found to be particularly significant for the accurate communication of humor and sarcasm. _______________________ Dr. Michelle Barton Division of Social Sciences
Empathy Enhancement in Text Base d Computer Mediat ed Communication You know, there's a lot of talk in this country about the federal deficit. But I think we should talk more about our empathy deficit -the ability to put ourselves in someone else's shoes; to see the world through the eyes of those who are different from us -the child who's hungr y, the steelworker who's been laid-off, the family who lost the entire life they bui lt together when the storm came to town. When you think like this -when you c hoose to broaden your ambit of concern and empathize with the plight of others, whether they ar e close friends or distant strangers -it becomes harder not to act; harder not to help. Barack Obama (2006) This quote from the 44th president of the USA is im portant to society and even humanity in more ways than one. Given the not so small problems such as global wars, global hunger, and global warming it becomes clear that the need for more empathy is not only an American issue but al so a world problem. Through empathy we can be moved to make peace with our enemies, make food for the hungry, and make healthy our environment. Thus, empathy is a universal tool and motivator fo r individuals within society as well as all humanity. However befo re we can even begin a search for empathy a very important question must be asked, What is empathy ? A straight answer to this question would be difficult due to the many va rying definitions of empathy. Nevertheless Alvin Goldman provides a general yet adequa te description of empathy; according to Goldman (1993) empathy is The ability to pu t oneself into the mental shoes of another
2 person to understand her emotions and fee lings (as cited in Decety & Lamm, 2006, p. 1147 ). This definition leads to anot her important question which is How does one come to understand or share the feelings of anot her? or How does one obtain empathy and become empathetic? The simplest answer to these que stions is communication, the act of transmitting and exchanging information. It is important to note, though, that while communication is a way of achieving empa thy, empathy is also used to achieve communication. Furthermore, depending on th e medium of communication empathy can be difficult to achieve despite the efforts ma de by communicators. For instance in text based digital communication, su ch as emails, short-messagi ng-service (sms) used in phones, or instant messaging people often e xperience difficulties communicating emotion resulting in miscommunication (that may lead to other negative outcomes), and negative interpretations of otherw ise neutral messages (Bryon, 2008). These problems are all related to a lack of understanding of the emotions being felt and expressed between partners or in other words a lack of empathy. Problems communicating through text are magnified when we have some perspective of how prevalent text based di gital communication is. Recent Pew internet and American life polls show that 78% of men and women us e the internet, 50% of all adults use social networking sites, and 73% of adults send and receive text messages (Madden & Zikuhr, 2011; Smith, 2011). These ra tes are expected to grow as more and more people acquire smart phones and com puters around the world. Thus, there is a growing need to solve problems related to communication through text. Given that much text based communication problems are linke d to a lack of effective empathetic
3 communication between individuals, one may ask what are ways to enhance the accuracy of empathetic communi cation through digital text? In response to this question the lite rature related to various forms of communication will be investigated with an emphasis on understanding what qualities of communication are related to and improve empathy. Next, problems with text based digital communication will be examined with a focus on their relations to empathetic understanding and emotional expression. Afte r inspecting these pr oblems a review and critique of the potential solutions psychologi sts have researched regarding text based communication will be offered. All of the data from these analyses are used towards the construction of an experimental study that aimed to help discover new ways of incorporating empathy and emotions in our current text based forms of digital communication. Communication and Empathy Derived from the Latin word communis, meaning to share, communication is now commonly defined as the activity of conve ying information. Although there are several theories and definitions regarding communica tion, the most basic form of communication involves a sender, a message, and a receiver. In general, communication occurs whenever a sender (human, plant, animal, or even computer) transmits a message (of any type of information) to a receive r (Heil & Karban, 2010). Here we will define human communication as Any act by which one pers on gives to or receives from another person information about that person's n eeds, desires, perceptions, knowledge, or affective states. Communication may be inte ntional or uninten tional, may involve
4 conventional or unconventional signals, may ta ke linguistic or nonlinguistic forms, and may occur through spoken or other mode s (National Joint Committee for the Communicative Needs of Persons with Severe Disabilities, 1998, p. 3) As suggested in the definition a sender doesnt have to intentionally transmit the message for communication to occur nor does the receiver have to correctly interpret the message. However, there has to be an accurate interp retation of a senders intended message for good communication to occur. Depending on the medium of the message, messages can be sent instantaneously face-to-face, across the world through telephone calls, and even into the future by freezing them in writin g. It is suggested that Homo sapiens were always capable of some form of communication (like all animals are) either verbally or nonverbal, and that our ability to communicate effectively played a key role in helping us survive as a species. Two of the most common mediums/modes of communication include Face-to-Face (F2F or FtF) communica tion, our oldest and most natural form of communication, and computer-mediated-comm unication (CMC), our newest and most technological form of communication. Pe ople can also communicate through smelling, touching, seeing, tasting, feeling, timing and he aring nonverbal signs, which all fall under nonverbal communication. Communication is an important func tion for survival, reproduction, and socialization, of many species. A few ways in which people use communication include warning others of potential th reats, sending signals of intere st to potential mates, reducing ambiguity/equivocality of a situ ation or problem, learning or sharing information about the world, and forming various types of re lationships (e.g. exchange, communal, and romantic) with others (Tomasello, 2008). In fact the human brain can be considered a
5 type of communication system in which billions of neurons constantly send signals to each other in a process involving the firing of action potentials and the releasing of neural transmitters. Thus communication is a very im portant and common aspect of our internal and external world. Much of human communication intentiona lly or unintentiona lly involves the expressing and perceiving of emotional conten t. However, to be capable of successfully expressing or perceiving emotional informa tion one must not only have communication skills, but must also have empathy. Though it too has many varying definitions, empathy is closely related to communi cation. A few examples of defi nitions of empathy reviewed by Decety and Lamm (2006) include: The ability to put oneself into th e mental shoes of another person to understand her emotions and feelings (p. 2). An other-oriented emotional res ponse congruent with the others perceived welfare (p. 2). An affective response that stems from the apprehension or comprehension of anothers emotional stat e or condition, and which is similar to what the other person is feeling or woul d be expected to feel in the given situation (p. 2). No matter which definition you prefer empathy can be interpreted as a prerequisite for the successful communicati on of emotion. Given th at communication is the usual means by which people come to unde rstand each others emotions, one can say that communication and empathy are interdependent on each other. In other words, it is
6 plausible to assume that as empathy or co mmunication increases the other increases, and as empathy or communication decreases the other decreases between individuals. In addition to this some scholars theorize that empathy is closely tied to moral reasoning and motivates pro-social behaviors such as al truism (Batson et al., 1991; Golman, 1993). Thus communication of emotions, aka empathe tic communication, is at the heart of moral reasoning and a func tioning society. Failure to communicate an intended message clearly or effectively is generally known as miscommunication. Miscommunica tion is usually caused by communication noise. Communication noise is considered to be anything (not ju st sounds) that can interfere with the decoding (or understanding) of a message, and is detrimental to all types of communication incl uding empathetic communicat ion (Narayanrao, 2011). There are several types of communication noise in cluding: environmental noise, physiologicalimpairment noise, semantic noise, syntactical noise, organizational noise, cultural noise, and physiological noise (Narayanrao, 2011). Envi ronmental noise is noise that physically disrupts communication, such as loud construction noises du ring a conversation, static during a telephone call or frame lag dur ing a video call (Narayanrao, 2011). Physiological-impairment noises are physical maladies that prevent effective message communication, such as actual deafness or bl indness (Narayanrao, 2011). Semantic noise occurs when a word or sentence has different interpretations of its meaning. Semantic noise is sometimes created intentionally for the purpose of making puns or double entendres such as in this old newspaper h eadline Presidential Candidates Promise Larger Budget for Education; More Lies Ahead. S yntactical noises are grammar mistakes that can disrupt communication, such as improper pro noun use. If a receivers interpretation is
7 impeded by poorly structured communicati on such as complex or bad directions organizational noise has occurred. Cultural noises are stereotypical assumptions which can lead to misunderstandings, such as uni ntentionally offending a non-Christian person by wishing them a Merry Christmas. Psyc hological noises o ccur when a persons attitude or psychological state leads them to lose focus on or incorrectly interpret communication. One example of a case in whic h psychological noise interferes with the interpretation of a message can be when an outraged person interprets comments such as is everything ok? as being c ondescending or offensive. In terestingly, empathy can be deterred by communication noise and lack of empathy can be a type of communication noise (i.e., psychological noise). It is important to note th at miscommunication is not negative communication in the sense of hostile communication; nonetheless miscommunication can lead to many negative outcomes. In contrast to communication, miscommunication can lead to decreased c oordination, incorrect assumptions, and in some cases even unintentional hostility between individuals. Thus noises should be kept at a minimum in order to have successful communication. Communication is also divide d into various categories based on the way in which messages are sent and received, also known as the medium or channel of communication. Not surprisingly these media also vary in th e amount of empathetic information they can easily express (Daft & Lengel, 1986). The following reviews several communication mediums beginning with text based communication.
8 Face-to-Face Communication and Nonverbal Behavior As the name describes, face-to-face (F2F or FtF) communication is communication made by people who ar e facing each other. Although F2F communication is an everyday form of communi cation, it is considered a rich medium of communication based on media richness theo ry (Daft & Lengel, 1986). According to Daft and Lengel (1986, p. 560) Information richness is defined as th e ability of information to change understanding within a time interval. Communication transa ctions that can overcome different frames of reference or clarify ambiguous issues to change understanding in a timely manner are c onsidered rich. Communications that require a long time to enable understandi ng or that cannot overcome different perspectives are lower in ri chness. In a sense, richne ss pertains to the learning capacity of a communication. Therefore media richness theory describes the varying degrees in which communication media can provide rich information and the ch aracteristics of rich and poor media. Daft and Lengel (1986) also wrote that In order of decreasing richne ss, the media classifications are 1.) face-to-face, 2.) telephone, 3.) personal documents such as letters or memos, 4.) impersonal written documents, and 5.) numeric documents. The reasons for richness differences include the mediums capacity for immediate feedback, the number of cues and channels utilized, persona lization and language variety (Daft & Wiginton 1979). Face-to-face is the ri chest medium because it provides
9 immediate feedback so that interpreta tion can be checked. Face-to-face also provides multiple cues via body language and tone of voice, and message content is expressed in natu ral language (p. 560). Being the richest medium of communicat ion, face-to-face communication offers several benefits over other mediums of communication. When compared to other mediums of communication (such as telephone, or text) face-to-face communication is more personal, facilitates more communicat ion by group members, makes it easier to communicate emotions, and decreases message ambiguity (Menchik & Tian 2008). Much of this is due to the availa bility of nonverbal cues/behavio rs naturally used to provide richness to information transmitted during F2 F conversations. Nonverbal cues/behaviors are described as any observable wordless sym bols/cues that may convey information to communicators. They include gestures posture, body language, clothing style, appearance, voice, quality and even the passage of time. Many scholars believe that about 2/3 of the information transmitted during F2F communication is nonverbal in nature. Thus nonverbal behaviors are a broad range of flexible symbols and signs that not only accompany most of F2F communication, but al so help make it the richest form of communication. Nonverbal behaviors/cues have been s hown to serve several unique purposes during F2F communication. Ekman (1965) iden tified six functions of nonverbal behaviors during communication, they includ ed: repeating, conflicting, complementing, substituting, accenting/moderating, and regulat ing information. According to Ekman nonverbal information can repeat what was sa id verbally, for example if you told a person they had to go left to find their dest ination and you pointed left while giving them
10 directions, this would be an instance of using nonverbal behavi or to repeat what was said. Sometimes nonverbal messages can conflict which may lead to ambiguity, imply sarcasm, or suggest an altern ative interpretation of signals. Nonverbal cues complement verbal messages by modifying or elaborati ng on them. For example a person who refuses to give in to a certain argument may cross his/her arms as he/s he declines or says no to send a clearer message. Another f unction of nonverbal behavior is that it can be used as a substitute for verbal messages. The behavior s used for substitution of words are usually referred to as iconic signals because they each have a clear and reliable meaning. One example of using an iconic verbal gesture as a substitute for words is when someone uses the ok gesture to indicate that nothi ng is wrong. However iconic symbols are like words in that different cultures and groups vi ew many of them differently; for example in Spain the same ok gesture is considered an insult (Knap & Hall, 2010). Nonverbal behaviors are also capable of modera ting (toning down) and accenting (amplifying) messages. This is usually done through hand movements and head tilts. One example of amplification can be observed when a teacher tells a student he/she did a great job on an assignment and delivers the message with a smile. An example of moderation is when someone agrees to a condition by saying suuuu uurre while tilting his/her head down or to the side to hint at their disagreement. N onverbal behavior can also be used to regulate communication. We can achieve this effect by coordinating our own nonverbal behavior with our words or by coordinating our nonverbal behavior with the behaviors and words of someone else during comm unication. One example of how people sometimes regulate their communication with their nonverbal beha vior is when people use faster and more intense hand motions as a c onversation heats up, and slower hand motions to try to calm
11 down the conversation. Though Ekman describe d these functions as unique categories, behaviors can always serve more than one function, which further adds depth to the amount of information capable of being tran smitted through something as simple as a smile. Thus nonverbal behaviors/cues are s ilent forms of communication that play several major roles in F2F communication. Nonverbal behaviors have also been shown to help people communicate with each other even when cultural or language ba rriers prevent them from using formalized language and this is especially true for the communication of emotions. Ekman showed that certain facial expressions unlike word s can be used to successfully communicate emotions across several unique and even pr eliterate cultures (Ekman, 1970). Ekmans research was used develop the theory of unive rsal expression of emotions which states that there are at least 6 emo tions (happiness, anger, sadness fear, surprise, and disgust) that can be found and expressed in similar ways across the world, which suggest that these behaviors are uni versal to people. Given that these universal expressions ar e used to communicate emotions there is a conceptual relationship linking empathy with nonverbal communication. Haase and Tepper (1972) conducted a study examining the actual contribution of verbal and nonverbal behaviors to judged empathy. In their study 26 professi onal counselors rated the levels of perceived empathy seen duri ng counselor and client interactions. The interactions varied using 48 unique combina tions of nonverbal and verbal cues/messages sent by the counselor. The results showed that nonverbal behavior s had a strong effect on the levels of judged empathy between th e counselor and cli ent. The authors specifically wrote Summing the relevant variance components re veals that of the
12 variability accounted for in judged empathy, the nonverbal main effects account for 45% of the known variability, the verbal message accounts for 22% of the variability, while all interactions account for the remaining 33% of the variance in judged empathy (p. 419). This study along with several others suggests that our ability to be empathetic is dependent in large part on our ability to r ead/understand the nonverbal cues that others transmit (Mehrabian & Ferris, 1967; Mehr abian, 1969). This is good because nonverbal communication is our first and most basic form of communication (e.g., a baby crying communicates that he/she is in need of so mething). However, a deficiency of nonverbal cues may prove to be a problem when we move on to consider newer forms of communication commonly referred to as digital communication. Computer-Mediated-Communication Digital or computer mediated communicat ion can be described as any form of communication that uses technology, such as computers or cell phones, as a medium for the transmission of information between communication partners. One of the most unique and important functions of computer-media ted communication (CMC ) is that it gives communicators the ability to communicate w ith each other regardless of physical distances that may separate them. CMC tec hnology can be categorized into synchronous and asynchronous communication types. Duri ng synchronous forms of communication, communication occurs in real time and the co mmunicators are engaged in conversation at the same time with each ot her like in F2F communicat ion. Synchronous CMC technology includes instant messaging, live chat r ooms, telephone calls, and video calls. Synchronous technology is usually used for time sensitive tasks that require constant feedback such as a fast paced conversati on. When variable gaps of time separate
13 conversation turns between individuals, th e communication is said to be of the asynchronous type. Asynchronous CMC technol ogy includes e-mail, electronic bulletin boards, and text messaging. Asynchronous co mmunication is generally used when responses from communication pa rtners take a significant amount of time and effort to compose, or when communication partners are too busy to synchronize their schedules to allocate a specific time for synchronous co mmunication. Despite the differences between asynchronous and synchronous communication, both forms of communication are very popular and are continuously evolving along with communication technology. In the past online CMC was purely text based and used only by a minority of the population. Now, however, online CMC has a vari ety of uses including (but not limited to) virtual work/business teams, online dating, online support groups, and social networking. We now live in a world where each day millions of people actively use social networking sites such as Reddit and Facebook. This wide spread use of CMC has led to a wave of new research regarding how people communicate with each other when surfing the web, and as a result several m odels/theories on CMC have been developed. The social identity of deindividuation eff ects (or SIDE model), and the hyper-personal interaction model are just a couple examples of models/theories used to specifically explain online social behavior (Yzer & Sout hwell, 2008; Postmes, Spears, & Lea, 1998). A more general theory that may also be appl ied to CMC is media richness theory, which as stated earlier, posits that the different media of communication vary in the amount of rich information they provide to partne rs during the time of communication. Based on media richness theory CMC technology woul d be categorized such that Video
14 Calling/Conferencing technology is considered the richest form of CMC, voice calls the second richest, and text based technol ogy the third richest form of CMC. Despite not being the richest form of communication and the availability of other forms of CMC, text based CMC is a very popul ar form of communication. This is due in large part to low cost, availability, an d ease of using text based communication technology, such as e-mail, forums, cellular te xt messaging (sms), and instant messenger. However there are some drawbacks to using te xt based CMC, many of which seem to be related to the mediums low richness. Problems with Text Based CMC Modern text based communication media such as email, blogs, online forums, sms, and IM all have the potential to spread ideas and thoughts to anyone around the world in a matter of seconds. In the past peop le believed these internet tools would help us build a cyber-utopia (Cowles, 2009). T oday, however, we find that flame wars (informal term for hostile online environmen ts) and trolls (colloquialism for someone who posts inflammatory messages) run ramp ant in cyberspace, constantly destroying communication bridges. The following is a re view of popular text based CMC problems and an analysis of their potential causes. Flaming is a very common problem online in which individuals display hostility towards each other through insults, sweari ng, and offensive language (Moor, Heuvelman and Verluer, 2010). Moor et. al., (2010) conduc ted an extensive analysis of flaming on YouTube. Based on their review of the litera ture the authors divi ded the most popular explanations as to why people flame (the act of flaming) on YouTube into three
15 categories. Their first category was the Changed awareness of self and others explanation, which suggested that flam ing on YouTube could be caused by the perception of a flaming norm or by reduced aw areness of other peoples feelings (or in other words reduced empathy). In their se cond category of explanations the authors suggested that flaming on YouTube could be th e result of miscommunication. Finally in their third category of explanations the au thors hypothesized that flaming on YouTube may be the result of intentional acts of individuals who seek to flame others. To test these theories Moor et al. conducted a survey among 157 YouTube users. The results showed support for all three type s of explanations, suggesting that flaming is a complex behavior with multiple possible causes including misc ommunication which may be related to the lack rich cues in the communication medium. Research has also shown that old psyc hological constructs can also lead to problems in text based CMC. For example th e correspondence bias theory states that people have the tendency to rely more on dis positional explanations th an situational ones when explaining the behaviors of others, espe cially when the situation is ambiguous or the other is a stranger (G ilbert & Malone, 1995). Vignovic and Thompson (2010) wanted to examine how the correspondence bias theory applied to behaviors that occurred during computer mediated communications. To this aim the authors employed a 2 (cultural cues: none or some) x 3(email-linguistic-deviations: control, etiquette, or technical) design using 435 participants from a university. During the experimental procedure participants were asked to imagine that they were about to work with a person from a different division in their organization who was or wasnt from anot her country. The participants were then shown an email from their co llaborator that had no linguistic-deviations,
16 deviations in etiquette (i.e., the emails were blunt/terse), or technica l errors (i.e., grammar and spelling errors). Finally, after reading the emails participants completed a survey measuring various personality traits (e.g., inte lligence, extraversion, and affective trust) of their partners. The results showed tec hnical and etiquette e rrors in the emails negatively affected participants perceptions of the sender, which further supports the correspondence bias theory. In addition to this a main effect of etiquette errors was found meaning participants attributed etiquette errors to an email senders disposition in spite of extenuating circumstances such as cultural diffe rences. In sum, the study suggested that during email communication, in which situa tional factors are usually unknown, people generally adhere to the corres pondence bias by attributing mistak es made in emails to the senders disposition; and in some instances this attribution error happens despite extenuating circumstances. Given that this type of misconcepti on of others could lead to miscommunication, negative eval uations, or even hostility, th is provides more reason to reduce ambiguity of email communication by enriching the medium. In order to gain a deeper understanding of the problem s related to text based communication and the limits of the communi cation medium, Menchik and Tian (2008) conducted a study in which they recruited a pa nel of 12 highly motivated and experienced activists as participants for the study. Particip ation in the panel cons isted mainly of online group discussions which were recorded by th e researchers for a 6 month period. There was one time during the study in which the partic ipants were given a chance to meet as a group f2f and discuss topics of their interest. The discussions from this f2f meeting were also recorded, analyzed, and compared w ith the email related findings. In addition multiple surveys and interviews were conducte d to further add to the research. Before
17 going into the results it is important to note that all of the participants were very comfortable using online comm unication; 92 percent of the participants had a high (or very high) desire to contribute to group disc ussion, and given the nature of the panel it was to each participants advantage to cont ribute to the group discussion. Going against all these factors (that made it seem likely that participants would have lively discussions during the 6 month project) the overall data showed that participants did not have much discussion (in fact many discussion threads and questions were left unanswered) and instead had many difficulties communicating ov er email. In addition to this, when compared to their face to face meeting partic ipants were much less likely to add to a discussion or respond to othe rs when communicating over em ail. The researchers even noted that in the F2F meeting they seamle ssly discussed subjects ignored online (p. 340). Furthermore, based on self-reports and inte rviews from particip ants, the researchers found that communication over email to be a lot more difficult than face to face communication for several reasons. The problems with communicating over ema il were categorized into three main types. They included problems with term inology, problems of relevance, and problems with situational and b ackground ambiguity. Terminology problems involved misinterpretations of the meaning of a sent ence or its components due to terminological ambiguity. To deal with this problem participants sent ema ils in which they asked for group clarification of certain terms, such as c ivil society but many of these emails were left unanswered. However, surprisingly duri ng offline meetings participants easily negotiated concerns related to ambiguous te rminology, which actually would be expected based on media richness theory. Another probl em participants en countered over email
18 was ambiguity around relevance. In normal conve rsation there tends to be a turn taking format between speaker(s) and listeners(s), which, combined with referential behavior (such as pointing to establish joint attent ion), makes it easy for communicators to know how statements relate to past statements a nd to whom statements are directed. However this communication flow is disrupted in CM C because participants turn taking behavior is unregulated and instant nonverbal feedback re lated to statements is nonexistent. Thus, participants remained silent during online communication, despite having relevant ideas to contribute because of the lack of fee dback and referential cues from nonverbal behaviors. Furthermore, because there was a l ack of turn taking behavior, it was even more difficult to determine the relevance of each statement, whic h further decreased participants motivation to make a post. Me nchik and Tian (2008) also found that participants inability to draw upon certain information from others appearances and inflections provided another set of frustrations (p. 347). According to the data analyses participants were reluctant to post because it was difficult for them to determine how the others might react to their message, especi ally if the message involved controversial issues. Thus, the researchers found that particip ants would rather remain silent than risk misinterpretation and the accompanying anxiety. Overall these three problems combined provide a lot of possible explanations as to why people find communication over email so difficult. As possible solutions to the problems above three semiotic tactics, i.e., empathetic tactics, referential tactics, and characterizi ng tactics were categori zed by the researchers based on analyses of the data from the 6 month study. Empathetic tactics, which enunciate or assign a general t one to particular parts of a message (p. 355), helped
19 participants make messages seem less aggressi ve, call attention to keywords or phrases that they considered important for fully understanding a message, and mitigate disagreement. These tactics included emo ticons (which convey a lighter tone in messages), strategic text formatting such as the use of capitaliza tion and quotation, and qualifiers like I may be wrong but T hough the researchers did find that these empathetic tactics helped participants communicate better over email, there is no consensus on exactly how they should be impl emented and some tactics such as use of emoticons are upon. Referential tactics were defined as tactics which specify others contributions to a message or its intended re cipient (p. 355). Due to the asynchronicity of email it is often difficult to determine the object or person of reference unless they are explicitly stated. Thus it is not surprising th at asynchonicity tactics such as assigning a moderator to a discussion group, or copying and pasting of past posts helped participants in this study communicate better. Finall y, characterizing tactic s which communicate demographic information that the sender thinks will be relevant to others interpretations of his message (p. 355) were also discussed by the researchers. Participants engaged in these characterizing tactics by adding signatures to their me ssages (which give readers a general idea of a senders background), incl uding explicit information on their physical states (such as I am writing this with a headache or On the way out the door), and using punctuation such as exclamation points which Barnes and Leudar argue (as cited in Menchik & Tian, 2008, p. 359) c onvey friendliness or inte rest in interaction. Taken together this study provided a lot of in depth information related to the nature of e-mail communication. The 3 categories of problem s (problems with terminology, problems of relevance and pr oblems with situational and background
20 ambiguity) and 3 semiotic tactics (empathe tic tactics, referential tactics, and characterizing tactics) seem to provide a good theoretical framewor k for future studies related to CMC. However due to the obser vational nature of this study the theories provided lack true experimental support, and as the authors noted, they do not convey all the possible problems and solutions related to online communication. Text Based CMC Potential Solutions To address the problems regarding text based CMC mentioned above along with other problems several studies have been conducted providing various solutions. For example Derks, Bos, and van Grumbkow (2007) examined the relationship between social context and emoticon use in Inte rnet communication. Participants were 158 students at a secondary school with a m ean age of 16.5 years. The experimenters conducted a 2 (Valence of social context: pos itive vs. negative) x 2 (Kind of social context: socio-emotional vs. task-oriente d) between subjects design measuring the frequency of emoticons used and the valence of the used emoticons. Participants were given no restrictions on when and how to us e emoticons. A questionnaire was also given out to participants. The questionnaire asked participants to respond to various internet chats that were presented on the survey. The ch ats varied in the kind of social context and in the valence of the social c ontext. Participants were aske d to respond to these chats in any way they wanted with the option of usi ng any of the provided six emoticon symbols. The results showed that stude nts used more emoticons in the socio-emotional context (where communicators state their opinions about a social or emotional issue, e.g., how they were afraid of not passing their exams) than in the task-oriented context (where communicators use CMC to stat e their opinions about a ce rtain task, e.g., planning a
21 graduation ceremony). Furthermore, students we re more likely to use positive emoticons in the positive context, and negative emoticons in the nega tive context. These results suggest that when placed in a socio-emoti onal context students communicating over text believe it is necessary to go through the effo rt of adding emoticons to their messages. This is interesting because it makes it seem as if people recognize text as being a poor or not rich medium of communication, and on their own decide to enrich it through the use of emoticons. However, the question remain s on whether or not using emoticons is an effective way to communicate emotions over text. Derks, Bos and Grumbkow (2008 ) also inve stigated past literature pertaining to emoticons and text based message interpre tation and presented their own study on the subject matter. The authors conducted a 2 (p artner: stranger, good friend) x 3 (valence: positive, negative, neutral) x 4 (emoticon: smile, frown, wink, blank) within-subjects design to measure how emoticons affect the interpretation of various messages. The 105 participants were students at secondar y schools who individually filled out a questionnaire asking them vari ous questions regarding their experiences using text based computer mediated communication (e.g., ema il and instant-messenger). During the first part of the study participants were presente d various e-mail messages which they had to read and refer to when answering specific questions from the questionnaire about the messages that were provided. Examples of m easures used during the first part of the study included Likert scales such as How do you feel after r eading this message? (1 very negative 7 very positive ) How positively do you rate the message? (1 very negative 7 very positive ) How familiar was the sender of the message? (1 very unfamiliar 7 very familiar )
22 How does the sender evaluate your performance as presenter? (1 very negative 7 very positive ) How ambiguous was the message? (1 very unambiguous 7 very ambiguous ) How serious was the message? (1 not serious at all 7 very serious ) How easy was it to understand the message? (1 very easy 7 very difficult ) How happy was the writer of the message? (1 very sad 7 very happy ) How sincere was the writer of the message? (1 very insincere 7 very sincere ) On a scale from 1 to 100, with 1 being the least and 100 being the most, how much happiness did the write r of the message portray? On a scale from 1 to 100, with 1 being the least and 100 being the most, how much sarcasm did the write r of the message portray? On a scale from 1 to 100, with 1 being the least and 100 being the most, how much humor did the writer of the message portray? (p. 383) The partner who sent the messages, the vale nce of the messages, and the use of an emoticon to end the messages varied for each message. The next part of the scale asked participants to use 1 (totally disagree) to 7 (totally agree) scales and describe how much they agreed with statements about the se nders possible motive for sending the message. An example of one of these statements includ ed When the writer of a message uses the emoticon smile, he/she wants to express his/her emotions. The writer motives included to express emotion, to strengthen the message, to manipul ate the interaction partner, to express humor, to put a remark in to perspective, to re gulate the interaction, and to express irony. Each motive was measured for each emotion. The results of the study showed that emo ticons can be and are used to express or strengthen positive and negative messages though not in all cases. Emoticons can be used to decrease the emotions being expressed by th e message content such that messages that were originally positive or negative becam e more negative or positive (respectively) when a or was added to them (respectively). Emoticons were also shown to be capable of conveying sarcasm and creating am biguity within sentences. However, no evidence was found for the notion that emo ticons can turn aro und the valence of a
23 message; that is change the tone of a st atement from positive (e.g. happy) to negative (e.g. angry). Thus this study provides support for a positive answer to the question of whether or not emoticon use is an effective way to communicate emotions over text. Overall the authors concluded that to a large extent, emotic ons serve the same functions as actual nonverbal behavior (p. 379). This suggests that text based communication could and should be enriched through the use of emoticons. Lour, Wu, Lu, and Tao (2010) investigat ed how different emoticons used in various types of IM (instant messaging) based communication at the workplace could affect how messages are interpreted. To answ er questions pertaini ng to task oriented CMC in the work place the researchers conducte d a three stage experimental study. In the first state 19885 message logs were colle cted from 199 employees working at a Taiwanese financial institute. The logs were then categorized into 5 groups according to function: (1) discussing or coordinati ng tasks (complex comm unication) 73%, (2) using emoticons to express emotion 12%, (3) arranging or scheduling meetings or appointments (simplex communication) 6%, (4) greeting (simplex communication, example: saying hi) 5%, and (5) linking Web site addresses and miscellaneous 4%. As shown in the parentheses, groups 1, 3, and 4 are associated with either the complex or simplex types of communication in the workplace (p. 891). During the second stage (which examined how emoticons are used to express emotions in the workplace) 32 employees (from the previous participant group) participated in a survey and were asked to c hoose an emoticon (from a list of 48) that best expressed their positive, nega tive, and natural emotions respectively. The third stage of
24 the experiment involved a 2 (message type : simple, complex) x 4 (emoticons: positive, negative, neutral, or no em otion) x 2 (gender: male, female) design. Each of the participants received instant messages that included both (1) scheduling for a meeting which belonged to the simplex message-type scenario, and (2) di scussing and requesting discussions which belonged to the complex message-type scenario regarding the topic addressed. Each text message was combined with one of the following four emoticon conditions as the baseline condition: positive, negative, neutral, or none. During the sampling plan experiment, all the participants were required to answer a questionnaire which assessed their emotional reactions to the text messages. They were asked to express their feelings, which were measured on a ve-point Likert type scale ranging from 1 (very unhappy) to 5 (very happy). In sum the results showed that: (a) per ceptions of negative emotions increased when negative emoticons were combined with text messages in both simplex and complex communication; (b) perceptions of positive emoticons increased positive emotions only for complex messages in both genders; (c) perceptions of positive emoticons increased positive emotions for the simplex message type in females only; (d) the use of neutral emoticons in IM had no effect on perceptions of emotions. Overall these findings further support the use of em oticons to convey emotions through text based CMC. Kato, Kato, and Akahori (2007) inves tigated the influence the degree of emotional cues transmitted has on the emotions experienced by people (senders and receivers) communicating over email. To do th is they conducted an experiment in which participants were randomly paired with anonymous partners (who were given an
25 anonymous e-mail user name), and asked to discuss over email exchanges the topic of juvenile crime. In their email discussion par ticipants sent emails one at a time until each had sent and received 3 emails. Whenever participants sent or received an email they had to respond to questionnaire items The senders responded to an expected emotion questionnaire (which analyzed the emotion th ey expected the receiv er to interpret the message as) and emotional state at sending questionnaire (which analyzed the emotion they were feeling at the time of sending each message). On the other hand the recievers answered questions from the perceived emotion questionnaire (which analyzed the emotion they believed the message was conveying), and emotional states at receiving questionnaire (which analyzed their own emo tional states when receiving each message). Kato et al. divided participants into high a nd low groups based on the levels of emotions transmitted in the messages sent between partic ipants. They then compared the emotions actually experienced by particip ants in the high group with those in the low group. Their results showed a tendency for unpleasant emo tions (e.g., anger and anxiety) to increase when emotional cues transmitted are low. The authors believe that the findings suggest that low degrees of emotional cues transmitted between senders and receivers in e-mail communication tend to cause some misunders tanding (p. 1894), which confirms the general findings in the literatu re, and implies that more emotional cues are needed in email communication in order to decrease misunderstandi ngs between communicators. Again media richness theory applies to this study because the results suggest that low emotional cues, in other words lack of emo tional information, lead to misunderstanding.
26 Typography One characteristic of text seldom st udied by psychologists is typography, the art and technique of arranging type in order to make language visible to the reader. Though typography is usually considered an art form, the modern usage of typography in advertising has caused the field to become more interdisciplinary and scientific. The following study on typography has a direct rela tionship with the current psychological research at hand because it proposes another way in which text based communication can be enriched besides using emoticons. VanLeeuwen (2006) created a proposal fo r the enhancement of text which borrowed concepts from linguistics and typogr aphy. He based his proposal on semiotics, the study of signs, symbols, and signifi cation, while taking a multimodal approach towards typography. The distin ctive features of typogra phy which he believed had semiotic potential included weight, expansi on, slope, curvature, connectivity, orientation, and regularity. VanLeeuwen believed that each of these features can potentially provide added meaning to text, and chose them on the basis of shared experience. The theory behind the shared experience is that every sign has a meaning potential that derives from our physical experience of it, from what it is we do when we articulate it, and from our ability to extend our practical, physical experience metaphorically (VanLeeuwen, 2006, p. 146), which VanLeeuwen believed would in crease their potential for effective communication. There were several distinctive featur es which VanLeeuwen believed could be used to bring added meaning to text. Weight (e.g., Low weigh vs. High weight ) is the
27 boldness of a typeface, and is measured on a continuum of boldness. According to VanLeeuwen bold can mean daring, asser tive, or solid and substantial for instance, and its opposite can be made to m ean timid or insubstantial (p. 148), however he also states th at boldness may be made to mean domineering, or overbearing (p. 148). Expansion (e.g., Ariel vs. Ariel Narrow), which is based on how narrow or wide a typeface is, can be used to imply room to breathe (p. 148) (through the use of wide typefaces) and restrictive movement (p. 148) (using narrow typefaces). Slope (e.g., a cursive sloping typeface vs. an upright typface ) is the degree of curvature in a typeface. In regards to slope VanLeuween states depending on the context, it might signify a contrast between the organic and th e mechanical, the personal and the impersonal, the formal and the informal the mass-produced and the handcrafted, the new and the old and so on (p. 148). Curvature (e.g. an angular letter form vs. a curved letter form ) is based on the angularity vs curvature of a letterform and/or its descenders (portion of a letter that extends below the baseline of the font). VanLeeuwen belie ves that roundness can come to signify smooth, soft, natural, and so on, and angularity abrasive, harsh, technical masculine and so on. Both may either be positively or negatively valued (p. 148). Connectivity (e.g. a letterform that is connected to each other a letterform with hooked feet vs. a letterform that is separate and s elf-contained ) according to VanLeeuwen is, again, associated with handwriting, and therefore sh ares much of its meaning potential with slope (see above). But it al so has its own metaphoric pote ntial. External disconnection
28 can suggest atomisation, or fragmentation and external connection wholeness, or integration. But the values may also be reversed, with disconn ection signifying the distinctive individuality of the elements of the whole, and connection its opposite. Internally disconnected letter forms, finally, have a sense of not being buttoned up, which may be negatively valued, as unfinishe d, or sloppy, or positively, as, say, easygoing. (p.149) Orientation (e.g. a flattened typeface, vs. an elongated typeface ) is based on how much a typeface is oriented towards the horizontal dimension or stretched in the vertical direction. According to Leeuwen horizonta l orientation, for instance, could suggest heaviness, solidity but also inertia, selfsatisfaction while vert ical orientation could suggest lightness, upwards aspiration but also instability (p.149). Regularity (e.g., a regular typeface vs. an irregular typeface ) of a type face is based on whether or not a typeface has deliberate irregu larities in its features. Irregularities can be created using shading, letter si zes, letter shapes, and by not staying within the lines. Irregularity can be added to typefaces to imply physical attributes such as coldness VanLeeuwens (2006) proposal above seems plausible because it adheres to the principles of media richness theory in that it pr edicts that variations in the presentation of text can provide added meaning to the text making it a richer medium of communication. Unfortunately there are no st udies that tested his proposal or something similar to it.
29 The Current Study Despite the possible solutions for text based communication pr eviously discussed, there still remain several problems with text based communication such as flaming, miscommunication, and difficultie s communicating. Researchers believe at the root of these problems is the fact that text is a poor medium for communication (Menchik & Tian, 2008), meaning text doesn t have a richness of cues found in other mediums of communication such as speech. The issues with text are also linked to the lack of clear rules on how to communicate effectively, esp ecially when it comes to communicating emotions through text (Menchik & Tian, 2008). VanLeeuwen (2006) addressed both of these core problems with text based comm unication. In his proposal the text based medium of communication can be made rich er through the addition of typographical cues, and the ways in which we could comm unicate emotions through text was explicitly outlined. However his proposal has yet to be tested and the problems with text based communication, (introduced earlier in this pape r) remain. Thus several questions remain unanswered: Do typographical features such as weight, expansion, slope, curvature, connectivity, orientation and regularity, wh en used in a specified emotion based framework provide an alternative mode for understanding text? or in laymens terms, can typography be used to provide added em otional understanding or empathy to text based communication? If the answer to these general questions is yes the next question is, can the use of a typographical guidel ine for the construction of text based communication of happiness, humor, sar casm, sadness and anger enhance the communication of these specific emotions?
30 The current study addresses these questions by analyzing the effects of the use or non use of an empathy technique sheet (ETS; instructions on how to type emotionally evocative statements based on VanLeeuwen s earlier proposal) on email communication accuracy and communication confidence of partic ipants. Participants in the experiment were placed into one of f our conditions based on a 2 (Co mmunicator: Writer, Reader) x 2 (ETS use: yes, no) between group design. Write rs were given a list of topics matched with emotions and asked to compose an ema il about their given topics that conveyed the paired emotions. The writers were also inst ructed to use or not use typographical cues depending on which condition they were i n. The ETS was used by half the writers (yETS) and not used by the other half (nETS ). The empathy technique sheet (ETS) gave writers specific instructions on how to c onvey happiness, sadness, sarcasm, anger and humor using typography. Both groups of writers sent their emails to be reviewed by participants in the reader condition. The reader s were presented with the emails and asked to guess from a list which emotion the writers were trying to convey in each statement. Both the writers and the readers responded to questions gauging their confidence in their ability to write or interpret emotions within the emails. Their answers to these questions were used to determine each participan ts communication confidence, a dependent measure that ranged from 0-10. Communicatio n accuracy also ranged from 0-10 for each reader and was determined based on how fre quently readers successfully interpreted the correct emotional tone of their writers statements. Based on the scales and variables above the followi ng predictions were made. I. Statements in which the ETS was used would be more accurately interpreted for their intended emotional tone than
31 statements in which the ETS was not used; this positive effect of the ETS would be seen for each of the five emotions used during the experiment (sarcasm, humor, anger, sadness, happiness). II. Participants confidence in their ability to communicate effectively would be closer to th eir actual abilities in the yETS condition than in the nETS. III. Participants in the nETS condi tion would be over confident in their ability to communicate eff ectively. That is both writers and readers would believe they were more successful at conveying and interpreting emotions than they actually were. Method Participants Participants consisted of 60 students at a small liberal arts college in southwest Florida (41 women, 19 men, age range: 18 to 23, Mage = 20.46 years). Participants were recruited via flyers and forum group emails. A ll participants were 18 years of age or over and provided their informed consent before participating in the study. Furthermore 88% of the participants reported that they frequently used te xt based digital communication such as email or text messages in their daily lives. Prior to arriving at the research lab participants were randomly placed into one of 4 experimental conditions reflecting the 2 (Communicator: Writer, Reader) x 2 (ETS : nETS, yETS) between group design.
32 Materials Gmail accounts. Anonymity of all emails was ensured by having writers (i.e., participants in the writer condi tion) and readers (i.e., particip ants in the reader condition) use anonymous email accounts created specifica lly for the experiment. Writers were instructed to compose and send their emails from 1AnonTextWriter@gmail.com to 1AnonTextReader@gmail.com. Writers were also asked to not include any information that might identify themselves or others in their emails. When readers read their emails they used the 1AnonTextReader@gmail.com account. Only the experimenter knew the passwords for the accounts. Furthermore all the emails (written and read by the participants) had preset subject lines in order to keep trac k of the participants numbers and conditions. Writers prompt. Participants in the writer condition ( n = 20) were given a prompt with their task instru ctions and materials. Each wr iter was asked to compose and send an email that would later be read by anot her participant. Writers were given one of two lists of topics partially randomly paired with emotions, and were instructed to write statements on the topics conveying the paired emotion, but without explicitly stating the emotion. The first 10 writers received list A th e second 10 writers used list B. The list of topics included parties, dating, TV, food, literature, campus life, movies, anything, Florida, and music whereas the list of emotional tone s (that were partially randomized with the each topic) included sarcasm, sadness, anger, happiness, and humor The topics and emotions were selected using previous studies on text based communication as a template (Kruger et al., 2005; van Leeuwe n, 2006). List A and B only differed in the ways in which topics were paired with emo tions in order to cont rol for the effects of
33 specific topic and emotion pairs. Given there were only 5 emotions and 10 topics, writers had to convey each emotion twice within th eir emails. To help them better understand their task writers were also given the follo wing example of a statement using the topic Cars and emotion Happiness: Having a fuel effi cient car is really great since gas prices are so high. However writers in the ETS-u se condition were instructed to use the empathetic techniques sheet (ETS) when typing each of their statements, therefore these writers were given the following version of the sentence as an example Having a fuel efficient car is really GREAT since gas prices are so high! :). The writers in the ETSno-use condition did not use the ETS when writing their statements. Instead they were instructed to refrain from us ing all caps, italics, bolded t ype, or emoticons, and to only write in sans serif (the default text in gmail). See below for more information regarding the ETS sheet. ETS sheets. The empathetic techniques sheet (ETS) is a treatment tool designed using VanLeeuwens theory regarding t ypography and semiotics (Van Leeuwen, 2005). VanLeeuwen believed that certain nonverbal behaviors that are a ssociated with our emotions and also help us communicate mo re effectively can be translated to typographical cues. Furthermore he suggested that the translated typographical cues would be easily recognized and interpreted by us because similar cues are found in our natural face-to-face communica tion. In his proposal VanLeeuwen described how several emotions could be conveyed in text us ing typographical cues. Using VanLeeuwens descriptions and other typographical cues re lated to emotions (i.e. emoticons) the empathetic technique sheet was created. The sh eet gives writers spec ific instructions on how to type statements when trying to c onvey sarcasm, sadness, anger, happiness, or
34 humor. Using the ETS writers can convert pl ain text, which is l acking in typographical cues, into rich text full of extra cues such as bolded type, contrasting fonts, and emotional-icons. An example of one of the ETS instructions is shown below: Convey happiness using capit alization, exclamation marks, and happy emoticons such as Ex: I JUST GOT MY TEST BACK! :-) Only writers in the yETS condition used th e ETS when writing; the nETS writers and their readers were not made aware of th e ETS until the debriefing period (see the Appendix for the complete ETS). Readers prompt. Participants in the reader cond ition were given directions to read the emails that were pr eviously composed by particip ants in the writer condition. The readers were given the list of five emoti ons previously used by the writers (sarcasm, sadness, anger, happiness, humor) and were in structed to assign one emotional tone to each statement based on their interpretations of what emotion they believed the writer was trying to convey. The readers were deceiv ed into believing that each emotion could show up in the statements any number of times (in reality each emotion showed up twice), so they did not choose their answer s through process of e limination. The readers did not have the empathy techniques sheet to help them, so they had to make their own assumptions as to what the typographical cues in some of the emails meant (that is if they thought the cues mean t anything at all). Readers stimuli. Paired controls were created by enriching (adding ETS cues to) or stripping (removing ETS cues from) emails originally written by participants in the
35 writer conditions. If a writer was in the yETS condition his/her original email was rich with typographical cues as a re sult of using the ETS. The experimenter created a stripped copy of emails (originally written using the ETS), by converting them to plain sans-serif text, and removing the emoticons from them. Th is meant that there was an original copy and a stripped copy of emails written by write rs in the yETS condition. Each pair of copied emails had the same content, that is all the words in the emails were the same, but in one email the ETS was used, whereas in the other it was not used. These two copies were paired (using identifiers) and separated during the experiment so that each copy was evaluated by only one reader. In contrast, if a writer was in the nETS condition his/her original email was lacking typographical cues such as bolded type, all caps, or emoticons. Therefore the experimenter created an enri ched copy of the plain emails (originally written by participants in the nETS conditi on) by adding these ETS cues, but without altering the wording of the message. As a re sult each of these copies (original and enriched) had the same content and only di ffered based on whether or not the ETS was used. Again the copies were paired using identifiers and evaluated individually by participants. This process meant that conten t variation between paired emails in which the ETS was used and not used was experimentally controlled. Survey items. All participants were given response sheets based on which condition they were in. Participants in th e writer condition were given a response sheet which instructed them of their task and asked them to answer (yes or no) to the question Are you confident that the reader of your message (above) would correctly guess the emotion you tried to convey? After completing each of thei r 10 statements. Participants in the reader condition were given a different response sheet which instructed them of
36 their task and asked them to match each st atement from the email they were viewing (previously composed by a wr iter) to one of the five emo tions listed (anger, happiness sadness, humor, sarcasm). The readers also ha d to answer (yes or no) to the question Are you confident in your answer? for each of their answers rega rding the emotional tones of the statements. At the end of all response sheets were 4 general survey questions which asked participants for their demographic informa tion, and information related to their text message use. The demographic questions aske d participants to describe their age (in years) and gender (male or female). The othe r survey items asked participants how much effort did they put into their task and How often do you use text based communication such as text messages or emails? Procedure Upon arriving to the resear ch lab (between the hour s of 12:00pm and 10:00pm) participants were seated at a computer with an opened g-mail acc ount (made specifically for the experiment). Participants in the writer condition were seated in front of a blank draft of an email with a prompt that instruct ed them to compose 10 statements to be read by a reader at a later time. The writers (aka participants in the writer condition) constructed their statements such that each statement was on a predetermined topic and conveyed a predetermined emotion. While co mposing their email the writers also responded to a survey which assessed their confidence in their ab ility to convey the specified emotions in each of their statemen ts. When they were finished with their
37 statements the writers sent their emails, and completed the rest of the survey. The sent emails were presented (individually) to pa rticipants in the reader condition. Readers (i.e., participants in the reader condition) rece ived a set of 10 statements all previously written by the writers. All read ers were given the task of reading each of the statements and trying to guess from a list which emotion they believed each statement best conveyed. While determining the emotiona l tone of each statement, readers also completed a survey. Readers, unlike the writ ers, responded to questions asking them how confident they were about their interpre tation of the emotions conveyed in each statement. After completing all of their tasks and su rveys participants were thanked for their help and debriefed. They were then gi ven handouts of tips for effective online communication and a gel pen as gi fts for their participation. Dependent variables Actual Communication Accuracy (ACA) was the main dependent variable of focus for this study and was determined using pairs of readers. Every writer was paired with two readers (using the writers original email and the enriched or stripped control copy). Each reader recorded his/her interpreta tion of the emotional tone conveyed in each of the 10 statements. The emotional tone that writers intended to convey was based on the list (A or B) of 10 topic-&-emotion pa irs they were originally given. Thus, communication accuracy was measured based on how accurately reader s interpretations of emotional tone matched the intended tone for each statement. For example, if a writers first statement was based on the topic-&-emotion pair art-&-sad and the reader
38 correctly interpreted the writers statement (e.g. Funding for art is always first to go) as being sad, then the statement would be give n 1 point for accuracy. Therefore each writer&-reader pair could score from 0 to 10 poi nts for accuracy. Given that each writer was paired with 2 readers, one read ers (in the ETS: no-use condition) ACA was used as a control score while the other readers (in the ETS: use condition) ACA was used as a treatment score during analyses. Communication Confidence (CC) was measured for each participant using their responses to the questions on confidence. As writers wrote each of their messages they answered yes or no to Are you confident that a future reader of this statement will be able to correctly guess the emotion you conveyed? Similarly, readers answered yes or no to Are you confident that you corre ctly interpreted the emotional tone that the writer of this message tried to convey? Participants who answered yes for each statement (read or written) received the maximum score of 10 points for their CC rating which also equated to being 100% confident. Therefore, for each statement there was an assessment of writers and readers confid ence pertaining to their antici pated accurate interpretation of the emotional tone conveyed. Relative Confidence (OC) for each participant was determined by finding the difference between their ACA (actual co mmunication accuracy) score and their CC (communication confidence) score. A higher CC score than ACA score indicated that the participant was over confident in his/her ab ility to communicate/interpret emotion over email. The following are the possible relati onships between a participants confidence and accuracy:
39 a) if CCACA > 0 then the pa rticipant was overconfident b) if CCACA < 0 then the pa rticipant was under confident c) if CC ACA = 0 then the partic ipant was accurately confident Though there are three possible ou tcomes above, it was expected based on Kruger et als. (2005) study that in genera l participants would be ove rconfident in their abilities regardless of th eir condition. Results Actual Communication Accuracy Participants in the reader section ( n = 40) engaged in a study in which they were required to intepret the emotional tone conveyed in the sentence. Any reader interpretation that matched the assigned wr iters intent was counted as correct. Actual communication accuracy scores were determined by summing the readers correct responses. The possible range in these scor es was 0-10 with an average score of 2 predicted by chance alone. Readers actual sc ores ranged from 3-10 and their mean score was 7.55. Readers were assigned into several groups before the analyses were conducted. Readers were evenly divided into two groups ( n = 20) based on whether they did (yETS) or did not (nETS) receive statements that included cues based on the empathy technique sheet (ETS). Readers were also divided into groups ( n =20) based on whether the statements they were reading were original unmodified statements written by the writer (Original) or non-original statements in which cues were added or removed by the experimenter (Modified). Finally Reader data were analyzed based on which form
40 Writers were assigned to use when writing thei r statements (A or B) into two groups of 20 participants. The results related to these scores and variables are as follows. Several statistical tests were conducted us ing readers accuracy scores as the dependant variable. Readers who read st atements that included ETS cues had significantly higher accuracy scores ( M = 8.2, SD = 1.281) than those who read statements that did not include ETS cues ( M = 6.9, SD =1.774), t (38) = -2.66, p = .011. Results indicated a near significant difference between ETS cue use and no use within the modified statement condition (Modified), such that readers who read statements with the ETS cues included scored higher ( M = 8.2, SD = 1.317) than readers who did not receive cues in their statements ( M = 7.1, SD = 1.101), t (18) = -2.03, p = .058. Similarly within the original statement conditi on (Original) readers (yETS) w ho read statements using the ETS had higher scores ( M = 8.2, SD =1.317) than readers (nETS) who did not receive ETS cues in their statements ( M = 6.7, SD = 2.311); however these scores were not significant t (18) = -1.78, p = .091. For each emotion (anger, happiness, sarcasm, humor, sadness) a t -test was run comparing the scores of readers (yETS) w ho received ETS statements and those (nETS) who received statements without ETS cues. Results showed that readers in the ETS condition were significantly better at detecting humor ( M = 1.55, SD = .510) than readers who did not receive ETS cues in their statements ( M = 1.15, SD =.489), t (38) = -2.53, p = .0157. A similar outcome was found for sarcasm in that results indicated readers (yETS) who reviewed statements containing ETS cues were significantly more accurate at determining the right emotion ( M = 1.70, SD = .470) than readers in the nETS condition ( M = 1.25, SD = .786), t (38) = -2.20, p = .034. No significant effect of ETS use was
41 found on participant accuracy fo r the other three emotions anger, happiness, sadness (see Tables 1 & 2). Furthermore at the p < .05 level no significant eff ects of form (A vs. B) or statements (Original vs. Modified) we re found on actual comm unication accuracy. Reader Confidence Reader confidence was measured by taki ng the sum of Readers (Yes vs. No) responses to the question Are you confident in your answer after having evaluated the emotional tone conveyed in the statements they received. Reader confidence scores ranged from 1-10 and had a mean of 7.3. As expected reader confidence ( M = 7.30, SD = 2.139) had a significant correlati on with reader accuracy ( M = 7.55, SD = 1.633), r (38) = 0.369, p = .019. Reader relative-confidence was m easured by subtracting each Readers confidence score from their previous accura cy score. Reader re lative-confidence is measured on a scale of -10 to +10 however the range reflected in this study was -6 to +4, with a mean of -0.25. Several tests were run using reader confidence and relativeconfidence as the dependant variables, however no significant results were found. Notably no effect of ETS [nETS ( M = 7.1, SD = 2.382) or yETS ( M = 7.5, SD = 1.906)] was found on confidence t (38) = -.59, p = .561. No significant difference was found in reader confidence scores between readers in the original statement condition ( M= 7.3, SD= 2.43) and the readers in the modified statement condition ( M = 7.3, SD = 1.87). Furthermore relative confidence was not signi ficantly different from 0 for the nETS condition M = 0.2, t (18) = .456, p = .654 nor for the yETS condition M = -0.7, t (18)= .152, p =.144. A t -test also showed that relative c onfidence scores in the nETS condition ( M = 0.2, SD = 1.963) were not significantly different from yETS ( M = -0.7 SD =2.05) relative confidence scores, t (38) = 1.42, p = 0.165.
42 Writer Confidence Participants in the Writer group ( n = 20) were evenly divided into two groups ( n = 10) based on whether they did (yETS) or did not (nETS) use ETS when writing their statements. Writer confidence was measured by taking the sum of Writers (Yes vs. No) responses to the question A re you confident that the re ader of your message (above) would correctly guess the emotion you tried to convey? for each statement they wrote. Writer confidence ranged from 4-10 with a mean of 8.1. No significant results were found related to writer confidence, notably t -tests showed that neither form [ t (18) = 1.39, p =.182], nor ETS [ t (18) = -1.70, p = .106] had an effect on writer confidence. Discussion As illustrated by the results and tables the original hypotheses were not fully supported. However they were supported to varying degrees depending on which condition is reviewed. Accuracy Overall participants were relatively a ccurate at reading the statements and determining the emotion that was intended to be conveyed. Readers had an average accuracy score of 7.55 (from a maximum of 10), meaning they accurately interpreted the emotional tone of 75.5% of the statements written by writers. This percentage is significantly higher than the accuracy percen tages Kruger et al. (2005) found which were 56% and 62%. Nonetheless r eaders who read statements with ETS cues correctly identified the emotional tone in 82% (based on a mean score of 8.2) of their statements, a number which was significantly higher than the 69% scored by readers who read
43 statements with no ETS cues. These data provide support for part one of the first hypothesis: that statements in which the ETS was used would be more accurately interpreted for their intended emotional tone than statements in which the ETS was not used. This leads us to the second part of the hypothesis. It was hypothesized that the positive effect of the ETS would be seen for each of the five emotions (sarcasm, humor, anger, sadness, happiness) such that for each emotion yETS readers would be more accurate than nETS readers. A glance at the results depicted in Table 2 would support this hypothesis in that the ETS effect was positive for every emotion with a range of 0.05 to 0.45, which equates to a 2.5% to 22.5% increase in accuracy associated with the use of the ETS. However t -tests only found statistical significance (at the .05 level) fo r only two of the five emotions; these were humor and sarcasm. Thus the second part of the hypothe sis was only partially supported as anger, happiness, and sadness were also expected to show differences. It is important to note, however, that the general literature on nonverbal communication states that judgments made us ing nonverbal behaviors of others almost always depend on a cluster of cues such as nonverbal gestures, facial expressions, the physical environment during communication, the communicators idiosyncratic habits, information about the situation or topic of conversation and much more (Knapp & Hall, 2010). Furthermore the literature argues th at iconic cues which immediately convey specific social information, such as a speakers attitude towards an object, are rare and generally culturally dependent (Knapp & Hall, 2010). In regards to the current study this would suggest that even if nonverbal cues were successfully translated into textual cues to convey specific emotions (in the form of the ETS), more social information (e.g., the
44 conversation mood, the writers personality, or the conversation topic) would be needed for readers to be able to successfully dete rmine which emotions were being conveyed. However in my experiment participants were only exposed to 10 one-sentence statements that were not related to each other. In other words participan ts in my experiment did not have access to the plethora of social inform ation that one would normally have available when communicating over text with a friend or coworker. Th erefore it is possible that even though the use or non-use of the ETS did not have a significant effect on the participants ability to accura tely interpret all 10 statemen ts by emotion, this may not necessarily invalidate the possi bility of it being effective in real world context. The simple fact that the ETS effect was found to be positive (though not always significantly positive) for every emotion may be reason enough to argue for its use, however to be sure that the positive results were not just products of chance a larger sample size is needed in order to increase the statistical significance ( t -value) of the ETS effects on accuracy. The most plausible reason why a near sign ificant effect of ETS was found in the Modified statement condition but not in the Or iginal statement condition can be the fact that the experimenter, who knew how to co mpose statements using the ETS perfectly, applied the ETS cues to the statements in the modified condition. Furthermore the writers who originally wrote their statements using the ETS sheet seldom followed the directions given in the ETS perfectly. In fact some writers had to be omitted because less than half of their statements followed the directions on how to convey emotions based on the ETS. This resulted in some variation in how well the ETS sheet was followed between the original and modified group, wh ich probably caused the differences in the ranges and the p -values.
45 Reader Confidence The average reader confidence score of 7.3 meant that on aver age readers were confident in 73% of their answers. Howeve r their average confidence score was lower than their average accuracy score of 7.55 (or 75.5%). Not on ly were these results not significant, they go against Kr uger et al.s (2005) study whic h would have predicted the participants to be overconfi dent, that is to have a hi gher score on confidence than accuracy. One possible way of explaining this un expected direction in the results is by attributing it to the relative confidence scores of the re aders in the yETS condition. Readers who read statements containing ETS cues scored -0.70 in their relative confidence scores whereas readers in the nETS condition had an average of +0.20. The negative relative confidence score from the yETS readers suggest that they were less confident in their ability to interpret the emotional tone of their messages correctly than they should have been. It is possible that th ese readers were somewh at thrown off by the ETS cues they read in their messages give n that people normally dont use the ETS sheet when writing their statements. If this is the case then it is also plausible that their confusion regarding what the cues meant or w hy they were even in the statements could have led to a decrease in their levels. Despite the fact that th ese results did not support the hypothesis they do have positive implications. In general, overconfid ence in ones ability to communicate may lead to more instances of attempts at communication that fail to be accurately interpreted yet go unnoticed. The less confident people are in their abilities to communicate effectively, the more reason/motivation they have to engage in behaviors that would make sure the message was interpreted correctly such as asking the writer if their (the
46 readers) interpretation of the message was correct. In other words, the under-confidence observed in the results is good because it can lead to behaviors that increase accurate message interpretations in real world scenarios. Writer Confidence Participants in the Writer condition had an average confidence score of 8.1 which was higher than the readers average confid ence score of 7.3, and average accuracy score of 7.55. This means writers were indeed overc onfident in their abil ity to communicate. Writers in the yETS conditi on were more confident ( M = 8.7, SD = 0.823) than writers in the nETS condition ( M = 7.5, SD = 2.07) in their ability to successfully convey their intended emotion in their statements to readers. However this difference was not significant at the .05 level ( t (18) = -1.70, p = 0.106). Nonetheless given the size of the differences between the means and the small size of the samples ( n = 10), one can expect that a larger sample would ha ve yielded significant results. R eaders in the original yETS statement condition who received unaltered statements from the writers had an accuracy score of 8.2. This would mean that the relativ e confidence score for the yETS writers is 0.5 (8.78.2) which means the yETS writers were 5% overconfident in their abilities. On the other hand the readers in the original nETS conditi on who read the unaltered statements written by the nETS writers had an accuracy score of 6.7. This would mean that the relative confidence score for the nETS writers is 0.8 (7.5-6.7) which means the nETS writers were 8% over confident in thei r abilities. Though the difference in relative confidence scores between the writers is not si gnificant, the trend within the results does support the second hypothesis. Writers confid ence in their ability to communicate effectively was closer to their actual abi lity to communicate effectively in the yETS
47 (deviation from actual score = 5%) condition than in the nE TS (deviation from actual score = 8%). Limitations There were several limitations of th is study. For one, the sample size was relatively small ( n = 40), and participants we re all students at a sm all liberal arts college between the ages of 18 and 23. Not only did this limit the ability to find significant values for some of the variable effects, but it al so limited how much these results generalize to an outside population. Another limitation of th e study is that it only looked at the effects of the ETS and not how effectively specific cu es helped convey speci fic emotions. It is possible that any cues could have been used to convey the varying emotions and lead to similar results. Also the participants were mostly female, and the literature is mixed on how gender is correlated with diffe rent variables measured in text. Future Research Future research can overcome many of th e limitations of the current study and branch out to new areas of knowledge. Future research should use a larger sample size that better represents the ge neral population. Furthermore fu ture research should compare the use of varying types of cues to each other in order to fi nd out if there are particular cues that convey specific emotions. For instance research can compare if BOLDED CAPITALIZED TYPE can convey anger more effectively than italicized type or type that is just CAPITALIZED or just bolded.
48 Conclusion While the present study was not completely successful in sup porting the original hypotheses, it did find several significant resu lts. Overall the ETS was effective in helping writers better convey 2 emotions to readers (humor, sarcasm), which would suggest another solution for text based comm unication problems, and the possibility that nonverbal communication cues could be successfu lly transferred into text based computer mediated communication.
49 References Alonzo, M., & Aiken, M. (2004). Flam ing in electronic communication. Decision Support Systems 36(3), 205-213. doi: 10.1016/S0167-9236(02)00190-2 Batson, C., Batson, J. G., Slingsby, J. K., Harrell, K. L., Peekna, H. M., & Todd, R. (1991). Empathic joy and the empathy-altruism hypothesis. Journal Of Personality And Social Psychology 61 (3), 413-426. doi:10.1037/00223522.214.171.1243 Bryon, K., (2008) Carrying too heav y a load? The communication and miscommunication of emotion by email. Academy of Management Review 33 (2), 309-327. Daft, R. L., & Lengel, R. H. (1986). Orga nizational information requirements, media richness and structural design. Management Science, 32 (5), 554-557. Derks, D., Bos, A. R., & von Grumbkow, J. (2007). Emoticons and social interaction on the Internet: The importan ce of social context. Computers In Human Behavior 23 (1), 842-849. doi:10.1016/j.chb.2004.11.013 Derks, D., Bos, A. R., & von Grumbkow, J. (2008). Emoticons and online message interpretation. Social Science Computer Review 26(3), 379-388. Ekman, P. (1970). Universal faci al expressions of emotion. California Mental Health Research Digest 8 (4), 151-158.
50 Fussell, S. R., & Krauss, R. M. (1991). Accu racy and bias in estimates of others' knowledge. European Journal of Social Psychology 21 (5), 445-454. doi:10.1002/ejsp.2420210507 Goldman, A. (1993). Ethics and cognitive science. Ethics, 103, 337360. Decety J., & Lamm C. (2006). Human empathy th rough the lens of social neuroscience TheScientificWorld 6 1146-1163. Haase, R. F., & Tepper, D. T. (1972) Nonverbal components of empathic communication. Journal Of Counseling Psychology 19 (5), 417-424. doi:10.1037/h0033188 Hwil, M., Karban, R. (2010) Explaining evol ution of plant communication by airborne signals. Trends in Ecology & Evolotion, 25 (3), 137-144 Gilbert, D. T., & Malone, P. S. (1995). The correspondence bias. Psychological Bulletin, 117 (1), 21-38. doi:10.1037/0033-2909.117.1.21 Hancock, J. T. (2004). Verbal irony use in face-to-face and computer-mediated conversations. Journal of Language and Social Psychology 23 (4), 447-463. doi:10.1177/0261927X04269587 Kato, Y., Kato, S., & Akahori, K. (2007). Eff ects of emotional cues transmitted in e-mail communication on the emotions experi enced by senders and receivers. Computers In Human Behavior 23 (4), 1894-1905. doi:10.1016/j.chb.2005.11.005
51 Kato, S., Kato, Y., & Scott, D. (2009). Re lationships between emotional states and emoticons in mobile phone email communication in Japan. International Journal on E-Learning 8 (3), 385-401. Knapp, M., Hall, J., (2010). Nonverbal communication in human interaction. Boston, MA: Wadsworth, Cengage Learning Krauss, R.M., (2001) The psychology of ve rbal communication. (Unpublished article) Columbia University, NY. Lo, S. (2008). The nonverbal communication functions of emoti ons in computermediated communication. CyberPsychology & Behavior 11 (5), 595-597. Luor, T., Wu, L., Lu, H., & Tao, Y. (2010). The effect of emoticons in simplex and complex task-oriented communication: An empirical study of instant messaging. Computers in Human Behavior 26(5), 889-895. Menchik, D. A., & Tian, X. (2008). Putting soci al context into text : The semiotics of email interaction. American Journal of Sociology 114 (2), 332-370. doi:10.1086/590650 Mehrabian, A. (1969). Signifi cance of posture and positio n in the communication of attitude and status relationships. Psychological Bulletin 71 (5), 359-372. doi:10.1037/h0027349 Mehrabian, A., & Ferris, S. R. (1967). Inference of attitudes from nonverbal communication in two channels. Journal Of Consulting Psychology 31 (3), 248252. doi:10.1037/h0024648
52 Moor, P. J., Heuvelman, A., & Verl eur, R. (2010). Flaming on YouTube. Computers in Human Behavior 26 (6), 1536-1546. doi:10.1016/j.chb.2010.05.023 Narayanrao, H. L., (2011) A study on gl obal communication in English language. Research on Humanities and Social Sciences 1 (1), 1-3 National Joint Committee for the Communica tion Needs of Persons With Severe Disabilities. (1992). Guidelines for meeting the co mmunication needs of persons with severe disabilities [Guidelines]. Available from www.asha.org/policy or www.asha.org/njc Obama, B., (2006, August). Commencement and Katrina & Gulf Recover. Speech presented at Xavier Univ ersity, New Orleans LA. Postmes, T., Spears, R., & Lea, M. (1998) Breaching or buildi ng social boundaries? SIDE-effects of computer-mediated communication. Communication Research, 25 (6), 689715 Vignovic, J. A., & Thompson, L. (2010) Computer-mediated cross-cultural collaboration: Attributing co mmunication errors to the pe rson versus the situation. Journal of Applied Psychology 95 (2), 265-276. doi:10.1037/a0018628
53 Table 1 Average Reader Accuracy for Each Emotion Across all Conditions Emotion n M SD Min Max Anger 40 1.58 .55 0 2 Happiness 40 1.63 .54 0 2 Sadness 40 1.53 .64 0 2 Humor 40 1.35 .53 0 2 Sarcasm 40 1.48 .68 0 2 Note. Average for all emotions = 1.514
54 Table 2 ETS Effect on Reader Accuracy for Each Emotion ETS Condition (percent accuracy) Emotions nETS yETS ETS effect t Value p Anger 1.50 (75%) 1.65 (82.5%) 0.150 t (38) = -0.86 p= 0.395 Happiness 1.60 (80%) 1.65 (82.5%) 0.050 t (38) = -0.29 p= 0.773 Sadness 1.40 (70%) 1.65 (82.5%) 0.250 t (38) = -1.24 p= 0.221 Humor 1.15 (57.5%) 1.55 (77.5%) 0.400 t (38) = -2.53 p= 0.012* Sarcasm 1.25 (62.5%) 1.70 (85%) 0.450 t (38) = -2.20 p= 0.036* Note. Average nETS accuracy = 1.38, Average y ETS accuracy = 1.64, ETS effect = yETS nETS = significant at the p < .05 level
55 Note. Average Reader confidence = 1.46 Table 3 Average Reader Confidence per Emotion Emotion n M SD Min Max Anger 40 1.45 0.639 0 2 Happiness 40 1.60 0.591 0 2 Sadness 40 1.40 0.744 0 2 Humor 40 1.35 0.700 0 2 Sarcasm 40 1.50 0.716 0 2
56 Table 4 Average Writer Con fidence per Emotion Emotion n M SD Min Max Anger 20 1.6 0.598 0 2 Happiness 20 1.85 0.366 1 2 Sadness 20 1.75 0.55 0 2 Humor 20 1.3 0.801 0 2 Sarcasm 20 1.6 0.598 0 2 Note. Average confidence = 1.62
57 Table 5 ETS Effect on Writer Confidence for Each Emotion ETS Condition (percent confidence) Emotions nETS yETS ETS effect t Value p Anger 1.40 (70%) 1.80 (90%) 0.40 t (18) = -1.55 p = 0.139 Happiness 1.80 (90%) 1.90 (95%) 0.10 t (18) = -0.60 p = 0.556 Sadness 1.60 (80%) 1.90 (95%) 0.30 t (18) = -1.24 p = 0.232 Humor 1.10 (55%) 1.50 (75%) 0.40 t (18) = -1.12 p = 0.276 Sarcasm 1.60 (80%) 1.60 (80%) 0.0 t (18) = 0.00 p = 1.00 Note. Average nETS confidence = 1.5, Av erage yETS confidence = 1.74