E XAMINING THE E FFECT OF VIEWING SOCIAL NETWORKING PROFILES ON INITIAL ONLINE CONVERSATIONS By Aditi Paul A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Communication Doctor of Philosophy 2015 ABSTRACT EXAMINING THE EFFECT OF VIEWING SOCIAL NETWORKING PROFILES ON INITIAL ONLINE CONVERSATIONS By Aditi Paul Much of the research exploring social networking sites as a tool for initiating relationships between people has focused on the relational outcomes of such online - initiated relationships. Little to no attention has been given to the nature of the initial interactions that ensue between unac quainted individuals who meet o n such websites. The primary purpose of thi s study is to examine the impact of viewing personal information from social networking profiles on the verbal behaviors during ini tial online conversations between strangers. The online conversati on transcripts between forty one stranger dyads from Paul and were used. Out of these, twenty two dyads viewed their conversational profiles before having an online conversation wi th them, while nineteen dyads viewed their conversational after the conversation . These conversation transcripts were coded for number and types of questions, breadth and depth of self - disclosure , and the topics of conversation to gauge if individuals discussed topics they had in common . The Facebook profiles were coded to measure the extent of actual similarity between dyads. Results showed that viewing Facebook profiles prior to c ha t t ing le d individuals to ask fewer questions but make more intima te disclosures to their conversational partners . However, there was no significant difference in the number of similar topics discussed between the two groups of participants. Intimate disclosures and discussing topics of similarity were found to be positively related to feelings of p redictability toward conversational partners. Also, the perce ived similarity was a stronger predictor of liking their conversational partners than actually being similar to them. iii ACKNOWLEDGMENTS This investigation would not have been possible without the unrelenting support of a number of key individuals. First, I would like to thank Dr. Kelly Morrison, my committee chair, for her unrelenting suppor t and for being there by my side when I needed her the most. Thank you so much Kelly for taking me up when all bets were off, you know what I mean! Second, I would like to thank my committee members, Dr. Lourdes Martinez, Dr. Steven McCornack, and Dr. Wei Peng for sticking with me through the process as I endeavored to complete this project. A special thank you to Dr. Vernon Miller who supported me through the preliminary study which helped me launch the dissertation study. Third, I would like to thank all the professors and staff members of the Department of Communication at Michigan State University. Thank you to the Spartan intelligentsia who taught me the key roles that theory and the robustness of research methodology play in advancing the field of comm unication. I am taking away an amazing wealth of information and precious experiences from my tenure at Michigan State University that will have a profound positive impact on my academic career and also on my personal life. Now for the recognition of peop le who matter the most and who had an extremely significant role to play ( p <.00000) to bring me to the place I am today. My parents, Dipak and Namita Paul, my sister and brother - in - law, Atrayee and Anirban Sett, and the most important person, my guru and s oul guide, Shrimati Kinkori Jogeshwari Devi, are THE reasons behind this pursuit and the successful fruition of the same. Without their prayers, their emotional support, and their best wishes, this would not have been possible. I offer the successful compl etion of the iv doctoral degree as an obeisance for their selfless love. Thank you Ma, Baba, Didi, Jiju, and the topmost persons, Shibanidi and Gurudev. Joy Guru. v TABLE OF CONTENTS LIST OF TABLES v i i i Introduction 1 Literature Review 5 Uncertainty Reduction Theo ry 5 Conversational Behavior Hypotheses 12 Questions and s el f - d isclosure 12 Topics of d iscussion 17 Conversational predictors of uncertainty r eduction 19 Further exploration: The relationship between similarity and liking 19 Method 23 Unitizing 23 Qualitative Measures 24 Questions 24 Self - disclosure 25 Intimacy level o f self - disclosure 27 Actual s imilarity 28 Communicated s imilarity 29 Control for covari ates 30 Amount of information shared 30 Quantitative Measures 31 Uncertainty 31 Similarity 31 Liking 32 Results 33 Hypotheses 34 Hypothesis one 35 Hypothesis two 35 Hypothesis three 35 Hypothesis four 36 Hypothesis five 36 Hypothesis six 37 Hypothesis seven 38 Discussion 3 9 Hypothesis one 40 Hypothesis two 41 vi Hypothesis three 42 Hypothesis four 44 Hypothesis five 45 Hypothesis six 46 Hypothesis seven 47 Limitations and Future Direction s 50 Conclusion 52 APP ENDICES 54 Appendix A: Tables 55 Appendix B : Unitizing Conversations 60 Appendix C : Identifying Questions 66 Appendix D : Identifying Types of Questions 68 Appendix E : Identifying Self - Disclosure 72 Appendix F : Ident ifying Depth of Self - Disclosur e 77 Appendix G : Identif ying Breadth of Self - Disclosure 81 Appendix H : Identifying Communicated Similarity 86 Appendix I : Identifying Actual Similarit y t hrough profiles 89 Appendix J : Counting Amount of Information shared in profiles 92 Appendix K : Procedures in the Paul and Morrison (2014) study 95 BIBLIOGRAPHY 97 vii LIST OF TABLES Table 1: Participant Characteristics 55 Table 2: Inter - coder Reliabilities 5 6 Table 3: Significance Tests for Questions, Proportion of Secondary Questions, Depth and Breadth of Self - Disclosures and Communicated Similarity by Experimental Condition (PTC or CTP) 5 7 Table 4: Correlation Matrix of Uncertainty Levels, Number of Quest ions, Proportion of Secondary Questions, Depth of Self - Disclosures (Peripheral, Intermediate, and Core), Breadth of Self - Disclosures, and Communicated Similarity 5 8 Table 5: Regression Analysis with Liking at Time 1 as Dependent Variable in PTC condition 5 8 Table 6: Regression Analysis with Liking at Time 1 as Dependent Variable by Experimental Condition 5 9 Table 7: Regression Analysis with Liking at Time 2 as Dependent Variable by Experimental Condition 59 1 Introduction The last decade has seen an exponential increase in the popularity of social networking sites (SNS). For example, one of the most popular SNS, Facebook, had a membership of 5.5 million in 2005 . Since that time, Facebook membership has increased by 200 ti mes over the last ten years, standing at 1.1 billion in 2015. Boyd and Ellison (2007) define such SNS as web - based services that allow individuals to (a) create a profile where they can include personal information, (b) build a list of contacts usually cal led the friends list, and (c) share different types of information with the members on this list. The widespread popularity of such websites can be ascribed to the fact that these websites help people connect with each other. For example, the mission state e connection s fostered by SNS are mostly between peop le with others who they already know from their interactions in the physical world (Ellison, Lampe, & Steinfield, 2007). However, the burgeoning membership s of these websites have made them popular platform s for initiating and building relationships between people who may share no prior social ties (Krasnova, Spiekermann, Koroleva, & Hildebrand, 2010). SNS like Facebook are becoming increasingly popular for relationship initiation . The information shared by users in their SNS profiles helps others learn personal details , and provides a sense of their tastes in topics like movies and music . Indeed, McKenna, Green, and Gleason (2008) stated that: By following online group discussions or by reading through prior to interacting, a reader can initiate a discussion with a new online acquaintance already armed backgrounds, and behavior. It is rare, indeed, to be privy to this depth of information 2 prior to making the acquaintance of another through traditional means and venues (p. 236) . Thus, these websites fulfill several relational functions, namely providing for convenience, a rich source of information regarding similarity, and relational acceleration. First, SNS have made it more convenient for people to get to know one another. According to interpersonal scholars Berger and Calabrese (1975), the primary concern that people have when initiating relationships is to get to know another person. In general, as an individual gathers more information about another, the ability to predict another increases, and uncertainty decreases. In traditional face - to - face encounters, this information is gathered via face to face conversations. However, the unique affordances of SNS have simplified this information - seeking process because a consi derable amount of information about others can be gained through viewing their SNS profiles. This is because SNS users display a lot of information about themselves across a wide array of topics. A longitudinal analysis of Facebook data by S tutzman, Gross, & Acquisti (2012 ) revealed that seventy percent of 5,016 users reported sharing their hometown, birthdate, and high school information in their profiles. Thus, the desire to gather information during the initial stages of an encounter can be satisfied sim ply by viewing SNS profiles, rather than actually engaging in a conversation. Furthermore, this information is accessible whenever the user wants to use it, at any time of day or night, thus making it quite convenient to the busy schedules of contemporary life. SNS not only facilitate conveniently gathering information about others, they also provide the opportunity to discover points of similarity with others. SNS user profiles usually display a substantial amount of attitudinal information, such as use r preferences regarding movies, music, sports, and other superficial topics. Access to such information helps people 3 gauge if they share similar interests and tastes before they initiate a face to face encounter. Thus, if an individual chooses to initiate a face to face encounter with another individual, the person can do so already armed with an arsenal of knowledge about mutual interests. In traditional face to face settings (i.e. without accessing online user profiles), discovering such similarities oc curs during the course of a conversation, or over the course of several conversations. Linked to gauging similarity is the function of relational acceleration. Specifically, the nature of information disclosed in SNS allows for the development of intim acy at an accelerated pace between people. In traditional face - to - face conversations, people learn superficial information about the other person, and as the relationship progresses, they learn about more intimate details. Typically, this process requires time, as people negotiate the boundaries of trust and vulnerability. However, in SNS, people display highly personal information, such as relationship statuses and their religious views, and this information is available to anyone who peruses their profil es (Stutzman et al, 2012). The simple act of displaying personal information, for all, with the click of a button, replaces the time previously required to negotiate intimacy levels through multiple conversations and decisions about the appropriateness of questions and personal disclosure. Thus, displaying such highly intimate information in user profiles may accelerate the perceived sense of intimacy between people who are trying to get to know one another. In summary, SNS profiles provide a unique platfo rm for initiating relationships which is distinctly different from relationships that are initiated in traditional face - to - face environments. Research examining the relational development functions of SNS has mostly focused on the outcome variable of rela tionship initiation, particularly uncertainty reduction (Antheunis, Valkenburg, & Peter, 2010; Paul & Morrison, 2014; Hancock, Toma, & Fenner, 2008). However, these studies have not focused on the conversational interactions that ensue between 4 two individu als who are trying to get to know one another through such websites. In order to gain a holistic understanding of how SNS aid the acquaintanceship process, it is important to acknowledge the interactional aspect of this process. This study aims to fill thi s gap in the research by examining the structure and content of actual conversations that ensues between strangers over SNS. In particular, this research will compare the conversations and perceptions ey converse to the conversations and reduction theory by Berger & Calabrese (1975) will be used as the theoretical framework for this study because this theory is fou ndational in understanding the conversational behaviors people engage in during initial interactions. First, a n explication of the propositions of this theory and a summary of the findings of studies that have looked at the acquaintanceship process through the lens of URT is provided. Second, predictions about the content and structure of conversations in terms of (a) the quantity and intimacy level of verbal strategies, and (b) topics of discussion will be presented. The role of perceived versus actual similarity in initial conversation also will be examined. Finally, a detailed description of the methods involved in conversation analyses will be provided , and several hypotheses will be explicated . 5 Literature Review Uncertainty Reduction Theory Uncertainty reduction theory (Berger & Calabrese, 1975) has been widely used as a theoretical background to explain the ini tiation of relationships between individuals. According to this theory, when strangers meet, they experience a high level of uncertainty toward each other. This happens due to the lack of knowledge or information about the other person. Therefore, the main motivation is to acquire more information in order to better predict and , and reduce their uncertainty. Berger and Calabrese (1975) proposed seven axioms and twenty - one theorems that explain the asso ciation of uncertainty reduction with verbal communication, non - verbal affiliative expressiveness, information - seeking, intimacy level, rate of reciprocity, similarity and liking. The underlying theme of these axioms and theorems is that as communication b etween strangers increases, uncertainty gradually decreases , and attraction and liking gradually increase. In further developing uncertainty reduction theory, Berger (1979) and Berger and Bradac (1982) proposed three types of strategies that people use to reduce uncertainty. These include interactive strategies whereby one person dire ctly interacts face to fac e with an other person, passive strategies, whereby one person observes the other person unobtrusively in public settings, and active strategies, whereby one person elicits information about an other person from third parties. A large number of studies that have examined the getting - acquainted process between stra ngers in face - to - face settings have corroborated the propositions of uncertainty reduction theory. For example, Douglas (1990) conducted an experimental study where he had thirty - nine pairs of previously unacquainted participants talk face - to - face with eac h other for a period of 2, 4 or 6 minutes and then asked them to report the level of uncertainty and liking toward their 6 conversational partners. Overall, the results provided broad support for uncertainty reduction theory. Feelings of uncertainty toward e ach other decreased with increased interaction time, and the more the participants got to know about their conversational partners, the more th ey seemed to like their partners . This confirmed the seventh axiom of uncertainty reduction theory that predi cted a positive correlation between uncertainty reduction and liking. Follow up studies by Douglas (1991, 1994) and other scholars (Harvey, Wells, & Alvarez, 1978; Prisbell & Andersen, 1980) have supported the predictions of uncertainty reduction theory. Even though uncertainty reduction theory (Berger & Calabrese, 1975) originally was designed to explain and predict initial interactions in face - to - face communication, th is theory also has successful ly describ ed and predict ed the communication processes and conversational outcomes in computer - mediated communication. For instance, Tidwell and Walther (2002) compared the uncertainty reduction behaviors of participants who met each other online with participants who met each o ther face - to - face. Participants who were assigned to the online condition had to engage in a conversation with each other over an instant messenger where they were able to communicate through text - based messages. Unlike their face - to - face counterparts, the y were not able to see each other . Participants in the face - to - face condition were provi ded 15 minutes to interact whereas the participants in the online condition were provided 30 minutes due to the delay in response time involved in computer - mediated com munication. Results showed that online participants not only used interactive uncertainty reduction strategies, the frequency and directness of interactive strategies used by these participants was significantly greater than the face - to - face p articipants. Furthermore, they reported higher levels of uncertainty reduction and gathered more intimate information about each other compared to participants in the face - to - 7 face condition, likely because they did not have access to the rich nonverbal cues available in face to face encounters. uncertain t y reduct ion strategies because of the difficulty of using such strategies in online environments. At the time this research was conducted, the only online venue in which people could observe the behavior of others in public settings, i.e. use passive strategies, was in Usenet groups and mailing lists. Moreover, the difficulty of establishing a network of common acquaintances online was a deterrent for implementing active strategies . However, the advent of SNS has expanded the possi bilities of using the three uncertainty reduction strategies in online communication with relatively equal ease. For instance, Facebook allows its users to create a profile where they provide biographical and demographic information. The interactive interf ace in their profiles, called Timeline, allows users to share their thoughts and opinions in the form of text or multimedia, such as news articles, audio and/or video files, with their contacts. Users can look up plicitly having to seek permission, thus allowing for passive strategies. Facebook generates a list of common acquaintances, called mutual friends, between any two users. This feature provides an avenue for the use of active strategies. Lastly, Facebook al so has an instant messenger program where users can directly interact with each other, thus allowing for interactive strategies. Hence, due to the technological advances in SNS, these venues now provide novel and interesting frameworks for examining how th e breadth of uncertainty reduction strategies are used between previously unacquainted individuals. Antheunis et al. (2010) conducted a study examining the types of uncertainty reduction strategies that people used in SNS. They surveyed 704 users of Hyv es, one of the most popular Dutch SNS, and asked them questions about a friendship they had recently formed in Hyves . The 8 users reported the type of strategies they used to gather information about that person and also their feelings of uncertainty, simila rity, and liking toward their newly formed friend. Results indicated that users employed all three strategies but most often used passive strategies, i.e., Hyves profile. However, in spite of the predominant use of passive stra tegies, only interactive strategies significantly predicted the reduction of uncertainty. In other words, having an actual conversation with acquaintances was more effective in getting to know them compared to simply viewing their Hyves profile information . Furthermore, the use of passive and interactive strategies was found to be highly correlated , indicating that use of passive strategies was associated with the use of interactive strategies. For instance, one of the questions used for measur profil passive strategies, two explanations f or their results are: work or education from their profile and then asked them questions about it; or (b) the users asked about their work or education and then checked their profile to see if the information was present and/ or consisten t with their conversation . Since Antheunis et al. (2010) used a survey - based methodology, they could not determine the direction of this influence, i.e., if a certain strategy was used first followed by another, and how the sequence of strategy use would a ffect the outcomes of uncertainty reduction, similarity, and liking. In a nother study involving SNS, Hancock et al. (2008) examined Facebook information during con versations in which the goal of the encounter was varied. Fifty - five dyads of previously unacquainted participants were formed, and in every dyad one person 9 was randomly assigned to the role of ingratiator or partner . Participants in the ingratiator role were asked to get their partne rs to like them whereas participants in the partner role were just asked to get to know their partners better. Thirty of the 55 dyads were assigned to the experimental condition and the remaining 25 dyads were treated as a control group. In the experiment al condition, only the ingratiators were given access to their conversational Facebook profiles and were asked to familiarize themselves with the information . Ingratiators in the control group did not get to see their conversational then engaged in short fifteen minute online conversation after which they reported their feelings of liking toward each other. Results showed that participants reported liking the ingratiators more in the experimental condition than in the control condition. Content analyses of conversations between ingratiators and participants in the experimental condition indicated that ingratiators frequently used probes, i.e., questions whose answers they already knew from viewing the Facebook profile information . They also tended to surreptitiously interject the conversation with information from their conversational pertained to common ground topics and attitudinally similar references. Based on th ese results , Hancock et al. (2008) suggeste d that having access to others facilitate feelings of homophily and aid in relationship initiation and development. However, these claims were made based on asymmetric information access, i.e., when only one person in the dyad had access to the Facebook p rofile of the other person. Hancock experimental manipulation could not account for the examination of the effect of symmetrical information access (i. uncertainty reduction and liking. 10 Paul and Morrison (2014) used an experimental design to address the limitations of Antheunis et al. (2010) and Hancock pecifically, they considered (a) which strategy was the most effective in reducing uncertainty in social networking sites, (b) which sequence of strategy use led to greater uncertainty reduction, and (c) the effect of different sequences of strategy use on feelings of similarity and social attraction. Previously unacquainted individuals were paired up in 49 dyads and randomly assigned to one of two experimentally manipulated sequences of uncertainty reduction strategy (URS) use: profile - then - chat (PTC) and chat - then - profile (CTP). In the PTC condition, participants viewed their conversational Facebook profile information before chatting with them online , and in the CTP condition, participants viewed their conversational nformation after chatting with them online . The participants were asked to self - report their levels of liking, similarity, and uncertainty reduction toward their conversational partners. Results indicated that participants in the PTC condition who viewed t heir conversational chatting with them online (a) reported greater uncertainty reduction, (b) liked their conversational partners more, and (c) felt more similar to their conversational partners compared to participants in the CTP condition. A counter - intuitive result from the study was that participants who viewed their conversational of them , resulting in reduced levels of certainty, liking, and similarity. Several possible explana tions were proposed by Paul and Morrison (2014) to account for the difference in the nature of communication outcomes as a result of viewing or not viewing profiles before the conversation. One explanation for the favorable communication outcomes of partic ipants who accessed their conversational conversation was that a higher level of intimacy was achieved in the conversation because 11 participants were able to ask more personal questions by probing information they view ed on their conversational also were able to make deeper disclosures about themselves. Moreover, viewing profiles prior to chatting online might have provided them with cues of attitudinally similar and common ground topics , thus allowing them to steer the conversation in a way where they talked about these similarities , thereby making the conversation more substantive (McKenna et al., 2002). The purpose of this research is to empirically test these speculation s by examining the conversations that participants had with each other in the Paul and Morrison (2014) study. The chat transcripts of the participants collected by the researchers will be used for the analyses of this study. Because questions and self - dis closure are important components of conversation, these two conversational devices are explicated first. Social penetration theory (Altman & T aylor, 1973) is then discussed as a theoretical foundation to posit the nature and content of questions and self - d isclosure likely to occur in SNS, and to characterize them in terms of intimacy. Predictions are made about the conversational behaviors of participants in the two experimental conditions to explain the nature of the outcome variables of uncertainty reduct ion, liking, and similarity. 12 Conversational Behavior Hypotheses Questions and s el f - d isclosure The use of questions and self - disclosure in conversation are two primary ways in which people exchange information. Questions are defined as expressions of inquiry that elicit a response from another person (Morris, 1976), whereas self - disclosure is defined as the process of making oneself known to others by voluntarily and intentionally revealing information about oneself to o thers (Cozby, 1973; Jourard & Lasakow, 1958). Kellerman and Berger (1984) asserted that question - asking is one of the predominant and most effective methods of seeking social and personal information about others. Research in face - to - face communication has repeatedly corroborated this claim by showing that the initial st ages of conversation between strangers are marked with greater use of questions than the later stages of conversation (Berger, 1973; Berger & Bradac, 1982; Calabrese, 1975; Douglas, 1990, 1994; Knapp, 1978). The analysis of online interactions between stran gers has shown similar trends for question - asking ( Andersen & Wang, 2005; Antheunis, Schouten, & Valkenburg, 2009; Mongeau, Jacobson, & Donnerstein, 2007; Pratt, Wiseman, Cody, & Wendt , 1999; Tamborini & Westerman, 2008; Tidwell & Walther, 2002 ). For insta nce, the content analysis of asynchronous email conversations over a period of six months between intergenerational dyads consisting of senior citizens and youngsters conducted by Pratt et al., (1999) showed that the highest number of questions was asked i n the first stage of the relationship with a steep decline in the number of questions during the later stages. This declining trend in question - asking as the relationship progressed with time reflects the third axiom of uncertainty reduction theory, which states that high levels of uncertainty cause increases in i nformation - seeking behavior. As uncertainty levels decline, information - 13 there is an inverse relationship between uncertainty re duction and the rate of asking questions. Based on this axiom and results from previous research, it can be assumed that individuals who already have some information about their conversational partners through their SNS profiles will ask fewer questions c ompared to individuals who have no information about each other . This is because individuals in the latter condition have a higher sense of ambiguity and uncertainty about their conversational partners compared to the individuals in the former condition due to their lack of information. Thus it is hypothesized that, fewer questions during their initial conv ersations compared to participants who do not view Prior information also may affect the nature of the questions asked during initial conversations. An experimental study conducted by Dipboye, Fontenelle, and G arner (1984) provided empirical evidence for this claim by demonstrating how exposure to prior information changed the nature of questions asked in subsequent conversations. They created two groups of participants who had to interview a potential candidate for a job position. The first group of participants reviewed the resume of the candidate before the interview while the second group of participants did not. Results indicated that compared to participants who did not have access to the resume, the partic probes, i.e., questions following up and probing answers to previous questions, because the resume provided a basis for probing information from the candidates. Jablin and Miller (198 0) label such probes secondary questions . Secondary questions are questions that request further information on topics that already have been introduced in to the conversation. Since viewing Facebook profiles can be considered akin to reading a resume (both are sources of superficial 14 information about an individual), it is predicted that participants who had access to their conversational s will use mor e secondary questions compared to participants who did not have access to their conversational s . Hypothesis 2: greater proportion of secondary questions during their initial conversations compared to Studies in face - to - face communication have accounted for the nature of information exchanged during initial interactions, and how the nature of this exch ange changes over time ( Kellerman & Berger , 1984; Berger, Gardner, Clatterbuck, & Schulman, 1976; Douglas, 1990; Maynard & Zimmerman, 1984; Rubin, 1979). For example, Berger et al. (1976) studied the temporal span of information exchang e between strangers. They recruited 204 participants who were informed that they were observing a two - hour conversation between two previously unacquainted individuals. They were given two decks of cards: one deck including 150 statements with varying leve ls of intimacy that could be exchanged during a two - hour conversation, and the second deck including time interval cards that ranged from 0 - 15 minutes all the way to 105 - 120 minutes. Berger et al. (1976) asked the participants to stack these 150 statements under the appropriate time slots in which they thought the statements could be mentioned in a two - hour conversation period. Results showed that the majority of participants sorted the exchange of biographical and demographic information in the first few m inutes of the conversation, whereas information regarding intimate and sensitive issues, such as family and empirical support for s. Rubin (1979) placed previously unacquainted participants in ambiguous situations , such as forming an impression about someone , and found 15 that the participants indeed started the conversations exchanging superficial information about each other and progr essively moving to ward more personal information. The types of information exchanged varying as a function of time can be explained by social penetration theory ( Altman & Taylor , 1973). Altman and Taylor (1973) suggested that as relationships develop, partners tend to disclose more intimate details about themselves . The central assumption of the theory stated them gradually, first revealing less intimate information and only later making more perso nal characterize d intimacy in terms of breadth and depth. Breadth pertains to the range , or number of different topics , that partners discuss, and the depth pertains to the magnitude of intimacy of the information exchanged. The depth dimension is further classified as peripheral, intermediate, and core. Peripheral information is impersonal and pertains to demography, intermediate information pertains to attitudes, bel iefs, and preferences, and core beliefs, values, and needs. Thus, the theory contends that during the relationship initiation process, when two people are trying to get to know one another, they start with the exchange of info rmation that is lower in breadth and depth. As the relationship progresses, the breadth and depth of information also increases. Memory Organization Packets (MOPs) and Svennev - Presentation Sequence base d their predictions on the tenets of social penetration theory (Altman & Taylor, 1973). For example, Kellerman and Lim (1989) proposed a structure of Memory Organization Packets (MOPs) in which topics of discussio n are categorized according to three phases of conversations: initiation phase, maintenance phase, and termination phase. According to this model, 16 introduction of self, weather, demographic, and other superficial and inconsequential topics constitute initi ation, the first phase of conversation. The maintenance phase of conversation is marked by information exchange about individual preferences and attitudes, while the termination phase consists of positive evaluations of each other and discussion or plannin g of model also explains how certain communicative acts are ordered in initial conversations. The model proposed ploy in accomplishing the p.92). The first step in SPS is a partner - affiliation to a cultural community. Exam hometown, past experiences, and occupation. The responses to these questions allow individuals to associate their partners with a larger group of people and infer traits about their partners based on the information they have about this larger group of people. For example, if an individual discloses that he/she is a professor at a research - intensive university, then the conversational recipient can use his or her knowledge of others in this profession t o make infer ences. The second step in SPS is eliciting further self - presentation from another by using the previously disclosed information in order to ask questions that are more specific and focused than the preceding topic initiation questions. The final step of SPS is what Svennevig (2000) labels an acknowledgement token , when the initiator either signals the respondent to continue self - presenting or when the initiator relates the information previously disclosed by the respondent back to his/her own experience s . When individuals do not have any information about each other, their selection of conversational topics will be in accordance to the topics delineated in conversational models like 17 MOP and SPS. Thus, participants will start the conversation by exchanging demographic information and progressively mov ing beliefs. In terms of intimacy, and as e xplained by Altman and Taylor (1973), participants will start with information exchange characterized by low breadth and depth. However, conversations between participants who viewed their conversational expected to proceed in a different manner. This is because these participants already have been exposed to the information that generally is asked in the initial phases of conversation. Thus, they can skip over the initiation phase and move on to the maintenance phas e of conversation. In terms of intimacy (Altman & Taylor, 1973), they will exchange information with greater breadth and depth. higher intimacy of self - disclosure , as demonstrated by (a) greater breadth of topics, and (b) greater depth of disclosure . Topics of discussion Another dimension on which conversations can differ based on the presence or absence o f prior information about a partner, is the topics of discussion chosen during initial interactions. Individuals can alleviate feelings of uncertainty by looking for similarities between each other. Berger and Calabrese (1975) articulated this relationship between similarity and uncertainty reduction in their sixth axiom of URT, which stated that similarities between people decrease uncertainty while dissimilarities increase uncertainty. When individuals discover similarity, they then can employ self - knowl edge to make inferences about their partners and increase their predictability (Antheunis et al., 2010). These similarities can pertain to attributes such as birthdate, hometown, and also to attitudes and beliefs (Byrne, 1971). 18 One of the ways in which people find similarities is by establishing common ground between each other . According to Clark (1996), common ground consists of shared knowledge, beliefs, and assumptions. Clark (1996) stated that for a piece of information to b e considered common ground between two people, they should both have access to this information and both should be aware that the other person also can access this information. Such common ground can be experienced either through direct personal experience s or through joint membership in cultural communities. Common ground through direct experiences or co - presence (Clark, 1996) can be experienced by individuals when they share a common experience such as being invited to the same party or being in the same study - abroad program. Common ground through cultural community membership is experienced when individuals belong to certain idiosyncratic groups such as the same national ity, education, and subculture (Svennevig, 2000). Research has shown that strangers and acquaintances actively seek information on common ground during initial interactions. Once they discover topics that establish common ground, they tend to go deeper into these topics and disclose more information about them. This leads to more effectiv e and substantial communication, and helps people comprehend and predict (Planalp 1993; Planalp & Benson, 1992; Planalp & Garvin - Doxas , 1994). Social networking pro files have enabled people to easily discover such objective, example, discovering that another person is from the same hometown is as simple as looking at the past loc ations listed in his or her Facebook profile. It follows then, that individuals with access to Facebook profile gain a head start in discovering such similarities, thus increasing the odds of discussing similar and common ground topics in initia l conversations. 19 communicate more similarity to online chatting. Conversational predictors of uncertain ty r eduction The central goal of this study was to examine the ways in which having access to information about another person prior to engaging in a getting - acquainted online conversation influences conversational behavior . This knowledge, in turn, should help illuminate how online conversational behaviors impact levels of uncertainty after interaction . Thus far, it has been hypothesized that when individuals have prior access to their conversational information, they will ask more probing questions, will make deeper disclosures about themselves, and will steer the conversation toward common topics in order to increase their levels of certainty toward their conversational partners . Individuals who lack access to prior information will no t exhibit such conversational behaviors, thereby resulting in lower levels of certaint y and predictability . Thus, based on these predictions, it can be summarized that, Hypothesis 5 : The use of the three conversational elements, (a) probing questions, (b) intimate self - disclosures, and (c) references to topics of similarity and common ground, will be positively related to certainty levels . Further exploration: The relationship between similarity and liking Similarity has been shown to predict feelings of liking or social attraction , i.e., we like people who are similar to us (Berschied & Reis, 1998). The scrutiny of the similarity - attraction hypothesis led researchers to make a distinction between actual and perceived similarity to better articulate the ef fect of similarity on attraction. Actual similarity is t he degree to which two people share attributes that are exactly similar to each other (Bryne, 1971). For instance, two 20 people are said to have actual similarity if they have the same birthdate, have t he same major, or like the same movie. Perceived similarity is the degree to which an individual believes s/he holds similar views to another on an issue or topic (Montoya, Horton, & Kirchner, 2008, p. 890). Researchers have analyzed the effect of actual s imilarity and perceived similarity on relational outcome variables , including uncertainty reduction and liking. They found that perceived similarity positively predicted liking when individuals did not interact with each other as well as when they interact ed with each other for a short duration of time (i.e., between 5 - 10 minutes). On the other hand, actual similarity positively predicted liking only when there was no interaction between two individuals. (Byrne, 1971; Byrne, Ervin, & Lamberth, 1970). Most of the previous research that has studied the impact of actual similarity regarding the outcomes of initial conversations has done so using the phantom other technique ( Byrne, Ervin, & Lamberth, 1970; Capella & Palmer, 1992), or the attitude and personalit y similarity questionnaire (Byrne et al., 1970; Ti dwell, Eastwick, & Finkell, 2013 ). In the phantom - other technique , participants are asked to complete a self - report measure of their attitudes. Following this, they are paired with a target person whose att itudes are experimentally manipulated to be the extent to which they are similar with their partners, they are asked to complete a task. Following this, the participants are asked to report their degree of liking or attributional confidence toward their partners. Results show that individuals tend to like those with whom they share similar views and dislike those with whom they share dissimilar views. In the questionnaire technique, participants are asked to scale their attitude and personality traits on a continuum list of topics including friendliness, confidence, extraversion, or assertiveness before conversing face - to - face with their partners. The researchers use the 21 ems and correlate them with their partners to formulate a measure of actual similarity. The participants are then informed that they are going to be paired up with another person with whom they share similar interests and tastes and that they will engage in a thirty minute face to face conversation in order to get to know each other. They recorded their level of attraction to their partners after this conversation. Results indicated that actual similarity and attraction were highly related to each other. P erceived similarity has been measured and studied by using the responses of individuals interacting face - to - face with another ( Byrne, 1971; Condon & Crano, 1988; Hoy le, 1993; Sillars, 1985; Sunnafrank & Ramirez, 2004). A meta - analysis of research examining the effect of actual versus perceived similarity on liking by Montoya, et al. (2008) indicated that the duration of interaction between individuals moderated the ef fect of the two types of similarity on liking. Studies showed that the effect of actual similarity on liking is significant when there is either no - interaction or short interaction between individuals. Thus the effect of actual similarity on liking decreas ed with an increase in the amount of interaction time. On the other hand, perceived similarity significantly predicted liking in both non - interaction and short interaction scenarios. The experimental set - up of Paul and Morrison (2014) allows for the empi rical testing of the effects of actual and perceived similarity on liking and uncertainty reduction. Viewing objective measure of actual similarity. Consistent with t he findings of previous research on the effect of actual similarity on liking, it follows that if participants are only allowed to view their conversational partners profiles (i.e., they do not have an opportunity to chat online), then actual similarity w ill have a positive effect on liking. Thus, 22 Hypothesis 6: Actual similarity, as measured by matching Facebook profiles, will be a significant predictor of liking when there is no interaction between participants . Additionally, reporting perceptions of similarity through responses to survey - based questions serves as a measure of perceived similarity. In Paul and Morrison (2014) study , participants reported perceived similarity and liking twice: once after chatting with each other and again after they s aw each other s Facebook profiles. Previous research has shown that perceived similarity always predicts uncertainty reduction and liking more strongly than actual similarity when individuals interacted with each other or did not interact with each other. Thus, it can be proposed that Hypothesis 7: Perceived similarity will be a stronger predictor of liking compared to actual similarity in both no - interaction an d short - interaction conditions. 23 Method Th e conversation transcripts of forty - five participant pairs collected by Paul and Morrison (2014) were analyzed for this research (see a ppendix for a description of the procedures used in their study) . However, due to technical issu es, the chat transcr ipts of four pairs of participants were not correctly captured by the computer software. Therefore the final conversational analyses were conduc ted on the chat transcript of forty - one pairs of participants. Six undergraduate coders w ere recruited for coding the conversation al transcripts in exchange for course credit. The items in the chat transcripts that were coded by the undergraduate coders were : (i) the total number of questions asked (Hypothesis 1), (ii) the proportion of second ary questions asked (Hypothesis 2 ), (iii) the intimacy level of self - disclosure as indicated by breadth and depth (Hypothesis 3), and , (iv) instances of communicated similarity in the online conversations (Hypothesis 4). In addition to coding the transcripts of conversations between partners, information in the s was also coded for objective measure s of actual similarity (Hypothes es 6 and 7 ). Finally, the total amount of information revealed by participants in their profiles also was assessed . Unitizing T he chat transcripts were unitized in to single thought units. A thought unit is a non - reflective clause, that is, one that can stand alone and does not distort the meaning of the rest of the sente nce if it is taken away. It is a single complete thought with a subject and a verb ( Dollard & Mowrer, 1947 ). were included with the independent clauses to which they were subordinate . The complete set of 24 instruc tions and examples used to train the coders for identifying such thought units by the researcher and one coder is detailed in a ppendix . Inter - coder reliability was measured by dividing the number of consensual utterances (number of utterances that will be agreed upon divided by total number of utterances) with the total number of utterances. Qualitative Measures Questions . Two coders were responsible for identifying the total number of questions a nd secondary questions asked in the conversational transcripts. explanation of questi ons was used as the operational definition to guide the coding of questions. According to es or calls for a reply; an ended with a question mark were coded as questions. Additionally , statements that did not end with a question mark but were still invitations for elicitin g information from conversational partners also were coded as questions. For instance, statements like h these statements missed appropriate linguistic symbols, they were still used as means to request further information from participants , or information - seeking from uncertainty reduction theory (Berger & Calabrese, 1975) . The complete set of instructions and examples used to train the coders for identifying questions is detailed in a ppendix . Inter - coder reliability was measured Kappa with .74 set as the critical acceptable value . Categorization of questions as secondary questions, or probes, was done using Jablin and As defined above, s econdary questions are questions that request further information on topics that already have been introduced in the conversation. These questions do not make sense when taken out of context (Tengler & Jablin, 1983). For instance, 25 coded as secondary questions. When coding for secondary questions in the cond ition in which participants viewed their conve rsational partners profiles, certain additional instructions were provided for coding . Because some of the participants viewed Facebook profile screenshot s before online chatting, it was expected that they would use the information from their profiles in the conversatio n. Thus, when a participant asked a question after alluding to a piece of information fro m the profile, even though it was a new topic in the conversation, it was coded as a secondary question beca use it was based on a topic that had already been disclosed in the profile. For example , if a participant saw in his/her conversational ofile that he/she worked in an MSU Cafeteria, and in t he subsequent conversation asked questions such as, saw you worked in MS So you T hese instances were coded and categorized under Facebook profile - based secondary questions . The complete set of instructions and examples provided to coders for identifyin g probing questions is detailed in a ppendix . The two coders were trained on identifying this type of question by using the conversational transcripts not used for the final analyses of this study. - coder reliability, with a critical acceptable value of . 74 . After reaching reliability, the coders were asked to code the rest of the tra nscripts independently . Any disagreements that arose were resolved through discu ssion with the primary researcher . The proportion of secondary questions was calculated by dividing the total number of secondary questions (secondary questions and Facebook profile based secondary questions) by the total number of questions. Self - disclos ure . A s eparate pair of coders coded the transcripts fo r the intimacy of self - disclosure. Self - disclosures were operationalized as utterances that reveal ed personal information 26 about the sender ( Dindia, 1983 ). Thus utterances that described the participant in some way or referred to certain affective res ponse s , such as preferences and interests of the participant , were coded as self - disclosure all were coded as self - disclosure . Similar instructions for identifying self - disclosure were used by Sermat & Smyth (1973), Tidwell & Walther (2002), and Won - Doornik (1979) . In certain instances, participants tend ed to discuss things that d id not communicate anything personal . For example, if a participant was participant said 20 object ive factual information and d id not reveal anything personal about the participant. Such utterances that pertain ed to objective factual information about objects like third parties, places, movies, music, books, t elevision shows, and sports were not coded as self - disclosure , and instead were coded as i nformation . Additionally, f eedback words, greetings, and back - were not c oded as self - disclosure , and instead were coded as o ther . The complete set of instructions and examples provided to the coders for identifying self - disclosure , information, and other are provided in a ppendix . The two coders were trained on identifying self - disclosure , information and other by using the conversational transcripts not used for the final analyses of this study. After the inter - coder reliability m reached . 74 , they were asked to code the rest of the transcripts independent ly. Following this, if there were any disagreements, they were resolved through discussion with the primary researcher . 27 Intimacy level of self - disclosure . Both breadth and depth of self - disclosure were used to gauge the intimacy level of the online chat conversations. I dentify ing breadth and depth (i.e., intimacy) of self - disclosure was . Taylor and Altman (1966) created an inventory of 671 statements that individuals can potentially disclose about themselves during a conversation. They had judges scale these statements in terms of intimacy of content and topical category. Judges sorted these statements in to thirteen categories . Tidwell (1997) reviewed this schema and found potential drawbacks and modi f ied this schema to establish eleven ca tegories . These were: demographic, biographic, sex, hobbies / tastes and interests, school and work, love and relationships with others, physical appearance /body , money, task - related, non - experiential current and historic events, and other . The other category involved greetings, parting remarks, and meaningless feedback. The modified codin g scheme d eveloped by Tidwell (1997) was used to code for the breadth dimension of self - disclosure with one minor change the eleventh category was not used because it overlapped with the other category cod e that distinguished between self - disclosure from information in this study. The depth of self - disclosure was coded of peripheral, intermediate, and core levels of informational intimacy. The complete set of instructions and examples provided to coders for identifying breadth and depth of self - disclosure are provided in the appendix . The coders were trained on identifying the ten diffe rent categories of breadth and three categories of depth of self - disclosure by using the conversational transcripts not used for the final analyses of this study. Aft er the inter - coder reliability assessed , they were asked to code the rest of the trans cripts independently . Following this, if there wer e any disagreements, they were resolved through discussion with the primary researcher . 28 Act ual s imilarity . In the previous study, prior to the experiment, the participants were asked to provide the link to their Facebook profiles with information in the following fields: first and last name, education (past and present), location (current and hometown), work in formation (past and present), and at least one choice in sports, music, movies, television shows, and books. Once the researcher s received their profile information in the proper format, a screenshot of the profile was take n. The primary researcher create d a picture file to be used by the coders that combined the profile screen shots of two parti cipants paired in the same dyad. These screen shots were placed side by side so that exact matches shown in the profiles could be assessed by the coders. The coder s used a binary coding technique to assess the presence or absence of exact information For example , if pa current location wa s Wixom, Michigan and par East Lansing, Michigan, then the similarity score of both participant 1 and 2 was 1 to denote an exact match for the state of residence. In addition to matching biographic information, the choices in movies, music, televisi on shows, books, and sports also were coded. The exact matches were added up to form a score for ac tual similarity . Such a matching system was used by Tidw ell, Eastwick, and Finkell (2013 ) as well as Fiore and Donath (2005). Tidwell et al. (2012) gave participants a score of 1 when both participants had an exact match on their academic major, religion, and state and were given 0 if there was no exact match. Fiore and Donath (2005) also used a similar actual similarity coding s ystem in online dating profiles where they examined exact matches of attitudes and characteristics on dating profiles (such as similarity in smoking habits, drinking habits, and physical buil d ) between dating partners. 29 Communicated s imilarity . Similarities are discovered and common ground is established during conversations as participants exchange information with each other. For the purposes of this study, this type of similarity was labeled as communicated similarity because the participants discover ed such similarities through verbal exchange of information. A coding system was developed to recognize instances in the chat conversation s when points of similarity were discovered . Baseline keywords and phrases that indicate d an occ urrence of similarity and common ground were phrases such as I thought so agreement or similari ty on opinions or mutuality of feelings on topics ranging from movies, music, sports, books , travel , politics , and other such topics were coded as instances of attitudinal similarity. I nstances of common ground included (a ) simila rity in past experiences ( e.g., working at the same place or visiting similar vacation spots), present experiences ( e.g., affiliation to the same university or taking the same classes), and future experiences ( e.g., planning to take the same classes or goin g to the sam e vacation spot); (b ) similarity in social circle ( e.g., having someone in a social circle who has the same educational background as p artner, or having siblings); (c ) affiliation to the same community or group in the past, present, or future ( e.g., havin g the same major, same seniority in college, same professional goals , taking the same number of classes, sharing an association with a place or company); and (d ) shared knowledge (e.g., sharing an awareness or understanding of the same things). The instances of agreement, s imilarity and common ground were tallied to form a composite score for communicated similarity. 30 Control for covariates . Amount of information shared . Because the amount of information shared on the profiles varied across participants , the amount of information also was coded and treated as a covariate in the statistical ana lyses to control for confounds . In order to measure the amount of information in each profile, each uni que piece of information was counted. Facebook mandates that users provide their first and last name s , thus, thes e two pieces of information were not included in the tally since they were constant across all participants. The u nique pieces of information in the following fields that were counted were : (i) work information (past and present) , (ii) education (past and present) , (iii) location (past and present) , (iv) movies, (v) music , (vi) television shows , (vii) books , and (viii) sports. For each unique piece of info rmation, the participant was given a score of 1. For instance, if a participant provided the following information in his/her profile: Works as Phone Operator the Department of Communication at Michigan State University. Past: Front des k operator at The Marriott East Lansing, the sco re that the participant received was 6. This is because the following pieces of information were counted: Phone operator, Department of Communication, Michigan State University, Front Desk Operator , Marriott , East Lansing. Sim ilarly, if the participant listed the following information in his/her education: Studying Bachelors in Political Science at Michigan State University, Past: Okemos High School, then the participant receive d a score of 4 for the followin g pieces of information: Bachelors, Political Science, Michigan State University, Okemos High School. On the other hand, if a participant provided the following information in his/her sports section: Michigan State Football, MSU Football, Green Bay 31 Packers , Detroit Piston s, then the participant received a score of 3 . This is because Michigan State Football and MSU Football refer to the same thing and are thus not unique. The combined score of each participant o n the information shared constituted the variable tot al amount of information. This variable was used a co variate in the regression analyses to check if the variability in the amount of information affected the outcome variables of perceived similarity and liking. Quantitative Measures 1 Uncertainty . uncertainty was used. For this research, f our of the five items from the CLUES 7 scale from Clatterbuck (1979) , e.g., can predict very well how this person wil and one item from Kellerman and Reynolds (1990) , e.g., , . Participants responded on a 7 - point Likert type scale ranging from 1( entirely disagree ) to 7 ( ent irely agree ) , such that higher scores indicated greater certainty . An average of the responses to the items on this scale provided the aggregate uncertainty score for each participant in the study . Similarity . Similarity was measured by using the attitudinal homophily dimension from the revised and updated perceived homophily measure proposed by Mccroskey, Mccroskey, and Richmond (2006). The original scale (Mccroskey, Richmond, & Daly, 1975) was revised to incr ease the reliability estimates of the attitudinal and background homophily dimensions of the scale. Instead of just four items, the new scale has nine items and five reverse coded items that 1 These measures were previously used in the Paul and Morrison (2014) study 32 measure attitudinal homophily. Some of the new items include: Th is person shares my values", were measured on a 7 - point Likert ty pe scale ranging from 1( entirely disagree ) to 7 ( entirely agree ) , such that higher scores indicated greater similarity. An average of the responses to the items on this scale provided the aggregate similarity score for each participant in the study. Liking . Interpersonal liking was measured by using six items from the social a ttraction , including items such as, of the responses were measured on a 7 - point Likert type scale ranging from 1( entirely disagree ) to 7 ( entirely agree ) , such that higher scores indicated greater liking toward the conversational partner. An average of the responses to the items on this scale provided the aggregate perc eived liking score for each participant in the study. 33 Results Forty - one chat transcripts were present analysis. Thus, there were 82 participants in total. Out of these, 46 were female and 36 were male participants. These participants were assigned to either the PTC or the CTP condition. There were 44 dyads in the PTC condition and 38 dyads in the CTP condition. There were 8 same - sex dyads and 30 opposite - sex dyads in the CTP condition. In the PTC condition, there were 26 same - sex dyads and 18 opposite - sex dyads. Further information about the pa age, race, and seniority in school is provided in Table 1. The 41 transcripts were unitized by the researcher and an undergraduate coder to generate a total of 2,918 conversation al units. The inter - coder reliability for unitizing chat transcrip ts was calculated as a proportion of agreements (total number of agreements divided by the total number of utterances). The in ter - coder reliability was .91. After unitizing, the chat transcripts were given to the first team of undergraduate coders who ide ntified utterances that were questions. On the first round of coding, the coders exceeded independently. The final agreement on question units between the coder s was acceptable ( .97 ) . After identifying questions, their next task was to identify probing or secondary questions. were asked to code the rest of the transcripts independently. The final agreement between the coders on this task was also acceptable ( ) . After compl eting the coding of questions, the unitized chat transcripts were given to the second team of coders who were responsible for identifying utterances as self - disclosure , information, or other. On the first round of coding, the coders exceeded the acceptabl 34 = .74. Therefore, they were asked to code the rest of the transcripts independently. The final agree ment between the coders was acceptable ( ) . Next, they had to code the utterances identified as self - disclosure in terms of breadth and depth. As mentioned before, there were three categorie s in the depth dimension and nine categories in the breadth dimension of self - disclosure . both breadth and depth of self - disclosure . Therefore, they were asked to code the rest of the transcripts independently. The inter - coder agreement on th e depth dimension was acceptable ( ), as was the reliability for the breadth dimension ( ) . Communicated similarity in conversations was coded by two undergraduate coders and inter - coders. However, the inter - coder reliability was poor with the ratings of th e third coder included. Therefore, the ratings of the third coder were dropped. The final inter - coder reliability between counted separately by two coders. Their re sponses were highly correlated to each other thus A composite list of the inter - coder reliabilities is presented in Table 2. Hypotheses The first four hypotheses were assessed using mu l tivariate analysis of variance, with (a) the total number of questions asked , (b) the proportion of secondary questions asked , (c) breadth of disclosure , (d) depth of disclosure (peripheral, intermediate, a nd core), and (e) communicated similarity as the dependen t variables , and the experimental condition (PTC or CTP) as the independent variable. Since the allocation of same - sex and opposite - sex dyads was not perfectly 35 randomized, the dyad composition variable (0 = same sex, 1 = opposite - sex) was entered as a covariate in the analyses. Results of the significance tests are summarized in Table 3 . Hypothesis one. The first hypothesis predicted that CTP participants would ask a greater number of questions compared to PTC participants. The data wer e consistent with this hypothesis . Results showed that the average number of questions asked my CTP participants ( n = 38; M = 9.82, SD = 3.6) was significantly higher than the average number of questions asked by PTC participants ( n = 44 ; M = 7.86 , SD = 3.49 ), F (1, 79) = 5.46 , p <.05 , p 2 = .07 . Hypothesis two . The second hypothesis predicted that PTC participants would ask a greater pr oportion of secondary questions compared to their CTP counterparts. The data were not consistent with this hypothesis. The pr oportion of secondary questions asked by PTC participants ( M = 0 .78 , SD = 0.18 ) did not significantly differ from the CTP participants ( M = 0 .73 , SD = 0.16 ), F (1, 79 ) = 1.30 , p = 0.23 , p 2 = .02 . Post - hoc analyses revealed that, in the PTC condition, only 9% of the secondary questions that were asked were based on Facebook profile information, whereas 90% of the secondary questions asked were based on information previously disclosed in the convers ation. Hypothesis three . The third hypothesis predicted that the intimacy of conversation for PTC participants would be higher than CTP participants. Both breadth and depth of self - disclosure were used to assess the intimacy of conversations between participants. In the breadth category, PTC participants had greater breadth ( M = 4.91, SD = 1.18 ) than CTP participants ( M = 4.34, SD = 1.21 ), F (1, 79) = 4.30 , p <.05 , p 2 =.05. In the depth category, PTC and CTP participants were significantly different from each other in the peripheral and intermediate levels but not in the core level. CTP participants reported greater number of peripheral self - disclosures ( M = 9.63, SD = 4.99) compared to PTC participants ( M = 7.48, SD = 3.25 ), F (1, 79) = 5.28 , 36 p <.05 , p 2 = .06 . However, PTC participants had a greater number of intermediate self - disclosures ( M = 16.71, SD = 8.19) compared to the CTP participants ( M = 13.45, SD = 5.70 ), F (1, 79) = 4.69 , p <.05 , p 2 =.06. The difference in the number of core disclosures made by PTC participants ( M = 0 .25 , SD = 0 .78 ) and CTP participants ( M = 0 .11 , SD = 0 .39 ) was not statistical ly significant, F (1, 79 ) = 3.19 , p = .08 , p 2 =.0 4 . These re sults indicate that PTC participants reported greater breadth of self - disclosure and greater interm ediate depth of self - disclosure than CTP participants . Thus, the data were partially consistent with the third hypothesis . Hypothesis four . The fourth hypothesis predicted that PTC participants would have greater communicated similarity, i.e., use a greater number of topics that established conversational similarity , than CTP participants. Even th ough PTC participants tended to use more topics that established conversational similarity ( M = 6.41, SD = 2.81) compared to CTP participants ( M = 5.53, SD = 3.52), there was no significant difference between the two groups , F (1, 79) = 1.54 , p =.22 , p 2 =.0 2 , thus the data were not consistent with the fourth hypothesis . Hypothesis five . The fifth hypothesis pertained to the relationship between the use of secondary questions, intimate disclosure , and communicated similarity, and the resulting levels of certainty . Specifically, this hypothesis predicted that the use of these conversational elements w ould be pos itively related to certainty levels. Thus, according to the hypothesis, the greater the use of secondary questions, intimate disclosure , and communicated similarity , the higher the certainty levels . The measurement of certainty levels were self - reported by participants in the Paul and Morrison (2014) study. As described earlier, according to the scaling used, higher scores indicated higher certainty. Participants had recorded their levels of certainty at two times: in the PTC condition, the first time they reported was after viewing 37 the second time was after having an online conversation with them. In the CTP co ndition, the first time participants reported their levels of certainty was after having an online conversation with their partners, and the second time was after viewing Because this hypothesis was meant for measu ring the resulting certainty levels after having a conversation with partners, the self - reported levels of certainty from the second time stamp w ere used for PTC participants and the self - reported levels of certainty from the first time stamp w ere used for the CTP participants. A one - tailed b ivariate correlation analysis was conducted between the proportion of secondary questions, breadth and depth of self - disclosure , communicated similarity , and certainty levels. Results indicated signif icant correlations between only two out of the six conversational elements and certainty levels. C ertainty levels and intermediate self - disclosure were moderately correlated, r (82) = .2, p <.05. A similar trend was observed in the correlation between certai nty levels and communicated similarity , r (82) = .2, p< .05 . Post hoc analyses revealed that the correlation betwe en intermediate self - disclosure , r (44) = .28, p <.05 , and communicated similarity topics , r (44) = .27 , p <. 0 5 , was stronger in the PTC condition . In the CTP condition, none of the conversational elements were significant. Thus, the data were partially consistent with this hypothesis. The correlation matrix is summarized in Table 4. Hypothesis six . The sixth hypothesis predicted that actual similarity would be a significant predictor of liking when there was no interaction between participants. This was tested using a multiple regression analysis with the independent variable actual similarity in time 1 and the dependent variable liking in time 1 in the PTC condition. The a mount of information shared on Facebook profile was entered as a covariate . Co ntrary to th is hypothesis, the results indicated that actual similarity was not a significant predictor of liking in the no interaction 38 condition , = - 0.22, t (41) = 1.46, p = .15. T hus the data were inconsistent with this hypothesis. Detailed result s of the regression analysis are provided in Table 5 . Hypothesis seven . Lastly, the seventh hypothesis predicted that perceived similarity would be a stron ger predictor of liking compared to actual similarity in no interaction as well as interaction conditions. Four separate regression analyses were conducted , twice in each condition of PTC and CTP with amount of information shared as a covariate . This is because perceived similarity and liking were measured in two time stamps . Results indicated that, perceived similarity was a stronger predictor of liking than actual similarity in both CTP and PTC conditions and for both time stamps . I n the CTP condition , at time 1 (only chat), there was no effect of actual similarity on liking, = - 0 . 23 , t ( 35 ) = - 1.0 , p = . 10 , and perceived similarity emerged as the only stro ng predictor of liking, = 0.67 , t ( 35 ) = 5.03 , p < . 001. At time 2 (after chat and profile), the effect of actual similarity on liking was still non - significant, = - 0.15 , t (35) = - 1.0 , p = .32, and perceived similarity was still a strong predictor of liking, = 0.62 , t (35) = 4.21, p <.001 . Similar trends were seen in the PTC condition. At time 1 (only profile), actual similarity did not predict liking, = - 0.01 , t (40) = - 0.06 , p = .95 , but perceived similarity did, = 0.63 , t (40) = 4.75 , p <.001 . At time 2 ( after profile and chat), the effect of perceived similarity on liking was sign i ficant , = .77 , t (40 ) = 7.43 , p <.001, compared to actual similarity, = 0.05 , t (40) = 0.45 , p = .66 (see Table 6 ) . These results indicated that perceived similarity was a consistently stronger predictor of liking than actual similarity in both the conditions, and when . 39 Discussion Previous studies on SNS have argued about the positive effects that SNS profiles have on relationship initiation and friendship forma tion between previously unacquainted individuals (Krasnova, Spiekermann, Koroleva, & Hildebrand, 2010; McKenna, Green, & Gleason, 2002; Paul & Morrison, 2014). The rationale proposed to explain these positive effects is that, knowing surface level informat ion from SNS profiles such as hometown, educational affiliation, and preferences in superficial topics such as movies and music helps to form an impression about unknown others and facilitates initial conversations. Even though these benefits of knowledge about SNS profile information have been hypothesized, they have not been empirically tested. The present study took the first step in this direction by examining the impact of viewing SNS profiles prior to interaction, on the conversational behaviors between previously unacq uainted individuals. These conversational behaviors were then compared with those that ensued between individuals who did not view behaviors were explored in terms of three conve rsational elements - questions, self - disclosure , and topics of conversation. It was hypothesized that exposure to SNS profiles prior to interaction would affect the frequency and nature of questions, breadth and depth of self - disclosure , and topics of conversations that established similarity between individuals. These conversational behaviors would in turn explain the outcome s of initial conversations in terms of feelings of predictability or certainty that individuals experienced toward their partners . Some of these predictions were confirmed by the data , while others were not. Possible explanations from theory and extant research are now discussed that can explain the confirmation or disconfirmation of these predictions. 40 Hypothesis o ne It was hypothesized that being exposed to SNS profile information would reduce the frequency of interrogation or question - asking between participants. The data w ere consistent with the prediction. When participants in the PTC condition viewed their conversational s , they learned some information about them. Even though this information was superficial in nature, it reduced the absolute sense of uncertainty about their conversational partners, something that the participants in the CTP condition experienced when they chatted with their conversational partners without any prior information. The initial alleviation of uncertainty before chatting , as a result of viewing their Facebook profiles, eventually led participants to ask fewer questions during their initial online conversations. The confirmation of this hypothesis support s the third axiom of uncertain ty reduction th eory , which state d that dec reases in uncertainty levels leads to decreases in information - seeking (Berger & Calabrese, 1975). When Berger and Calabrese (1975) postulated that a decline in uncertainty levels would lead to a decline in information - seeking behavior s , they only considered the role of interactive strategies in decreasing the levels of uncertainty. In other words, Berger and Calabrese (1975) argued that unacquainted individuals would ask more questions in the initial phases of conversation, and the answers to these questions would reduce the levels of uncertainty, which in turn, would reduce the question - asking behavior. T his research has demonstrated that passive uncertainty reduction strategies can facilitate uncertainty reduction in the initial phases of online interaction s. Speci fically, i n this study, the information gained by the PTC participants from passively vi allowed them to ask fewer questions in their initial online interactions. Thus, this research 41 extend ed Berger and Calabrese interactions between strangers in online chat interactions, an d additionally allowed for the examination of a broader variety of uncertainty reduction strategies. Hypothesis t wo It was hypothesized that being exposed to SNS profile information would increase the proportion of secondary or probing questions between participants. The data did not support this hypothesis because the proportion of secondary questions asked by participants in the PTC condition did not significantly differ from the participants in the CTP condition. The lack of significant difference in the proportion of secondary questions asked by PTC and CTP participants could be explained by the fact that most of the secondary questions asked by PTC participants were based on information revealed during the co nversation and were not based on Facebook profile information as predicted. Post - hoc analysis revealed that only nine percent of the secondary questions asked by PTC participants were based on Facebook information whereas ninety - one percent of the secondar y questions were based on information that participants discussed in the conversation. This finding reveals that even though Facebook profile information provided grounds for probing, most of the probes that were used to elicit further information were based on disclosures that happened during the course of the conversation itself. Two reasons can explain the l imited use of Facebook profile based probes during initial online conversations. The first is the lack of retention of information from the profiles. In the experimental study conducted by Paul and Morrison (2014), the researchers did not provide clear ins chatting with them. Thus, there could be a possibility that some of the PTC participants had did not remember the 42 information from the profiles well enough to use them as probes. Another reason why PTC participants used Facebook profile information less frequently in conversations could have been social information presented in the profiles by others (Dubow, 2007). Excessive probing o f Facebook profile information may have led people to think that they were being stalked online, and this beha vior can be construed as socially inappropriate. Therefore, in an attempt to balance the social appropriateness and the effectiveness of using an uncertainty reduction strategy (Berger & Kellerman, 1983), the participants may have limited the use of Facebo ok profile based information in their online conversations . Hypothesis t hree It was hypothesized that exposure to SNS profiles would lead participants to reveal more intimate information about themselves. The intimacy of information disclosed was measured in terms of breadth and depth of self - disclosure . The data were found to be partially consistent with the hypothesis. PTC participants had greater breadth of disclosures than CTP participants meaning PTC participants disclosed personal informatio n on a wider range of topics than CTP participants. Thus, the presence of Facebook profiles led participants to disclose about a greater number of topics. The presence of Facebook profiles increasing the breadth of conversations can be understood from the standpoint of impulsive behavior. Goldenson (1984) defines impulse as a This means that impulse s are not planned, but they arise immediately when exposed to a stimulus. This psychological concept of impulse has been used to explain various human behaviors such as impulsive buying (Rook, 1987), the unplanned decision of buying a product upon being exposed to the product. Impulsive behavior can also be extrapolate d to understand 43 why participants chatted about a wider range of topics after being exposed to SNS profiles. Categories of information from Facebook profiles such as movies, music, television shows, and so on might have acted as stimuli that may have given rise to an impulse in the participants to use these categories of information as conversational material. Participants might not have had any intention s of using certain topic s as conversational material, but the mere exposure to the stimuli (i.e. information from Facebook profiles), might have led them to impulsively use it in conversation. Thus, it can be contend ed that Facebook profiles leads to impulsive conversatio ns , the u nplanned decision of talking about something upon being exposed it. PTC participants also reported making disclosures that were greater in depth than CTP participants. Since CTP participants did not have access to any info rmation about their conversational partners prior to chatting with them, they had to disclose superficial information about themselves before moving on to more personal information. On the other hand, PTC participants, already having access to such surface - level information about their conversational partners from their Facebook profiles, could move on to the disclosure of information higher in intimacy such as their attitudes, preferences, and opinions about topics such as personal experiences, entertainme nt (music, movies, television shows, books, and sports), work, and family . Therefore, CTP participants reported greater number of peripheral (or superficial) self - disclosures and PTC participants reported greater number of intermediate sel f - disclosures. As far as core self - disclosures were concerned, there was no significant difference between the two groups of participants. This stands to reason because it is highly unlikely that individuals will disclos e extremely personal information about themselves to strangers within fifteen minutes of initial conversations. Thus, overall, PTC participants were able to achieve higher conversational intimacy by disclosing more intimate details about themselves compared to CTP participants. 44 This trend of self - disclosures as a result of viewing SNS profiles has an important implication on the process of letting ourselves be known to others as postulated by social penetration theory (Altman & Taylor, 1975). In this theory, Altman and Taylor (1975) use th e onion metaphor to explain the trajectory of disclosing information about oneself in order to be known to others in the course of relationship development . They argue d that the outer layer of an onion is representative of the surface level information, and the core of the onion is representative of more personal details of an individual. Hence the process of disclosing oneself to others starts by peeling the outer layer first and then proceed ing to the inner core. In other words, individuals disclose superficial information about themselves before divulging intimate details. The results of this study show that SNS profiles make the peeling of the outer layer of the onion redundant - it is already peeled. This means that the surface level information is already available to others through the profiles. This makes the core of the onion more readily available to the outer world, i.e., the more intimate details of an individua l become easily accessible to others . Thus, i ndividuals can begin the acquaintanceship process with an unknown other with a head start, and divulg e more intimate information about themselves instead of surface - level information. Hypothesis f our Given that SNS profiles provide a means of finding out similarities with one another in terms of work place, educational affiliation, and preferences in things like movies and music, it was hypothesized that participants with access to eac conversation in a way to probe and discuss such similarities. Results showed that even though participants in the PTC condition exhibited such a conversational pattern, the average number of topics of similarities dis cussed by them was not significantly different from CTP participants. 45 Th ree reasons can account for wh y viewing the SNS profiles did not bolster the chances of PTC participants pursuing similar topics with their conversational partners. First, post - hoc analyses revealed that the total number of topics on which PTC participants were similar to their conversati onal partners was very few. Only 13% of the total information shared in the PTC have majorly curtailed the possibility of talking about topics that were found to be similar in their profiles during their conversations. Second, the lack of retention of information from profiles by the PTC participants might have affected the use of similarity probes on conversations , as there were no measures accounting for whic h aspects of a profile page were viewed and/ or retained . Third, the reason CTP participants were just as likely to talk about topics of similarity as PTC participants is because both these participants had a need to stick to the social norm s for initial conversations , which are typically marked by high levels of agreeableness and low levels of disagreements and dissimilarities (McLaughlin, Cody, & Rosenstein, 1983). , Hence, it may have been easier to probe for topics of similarity that could be verified in conversation rather than assume similarity based on profile viewing and r isk disagreement in the initial conversation. Hypothesis f ive The main motivation for comparing the nature of initial conversations in terms of types of questions, intimacy of self - disclosure , and communicated similarity, was to see if the difference in t he nature of these conversational elements c ould explain the uncertainty levels experienced by participants at the end of the conversations. Thus, it was hypothesized that the proportion of probing questions, intimacy level of self - disclosure , and topics of conversational similarity would have a significant correlation with uncertainty, such that the more these elements were used in conversations, the greater the resulting certainty about the conversational partner. 46 Results indicated that, out of all the conversational elements, only intermediate - level disclosures and topics of conversational similari ty were statistically significant. In other words, only these two elements were seen to have a positive relationship with feeling s of predictability between previously unacquainted partners after initial interactions. These finding s were consistent with th e third and the sixth axiom s of uncertainty reduction theory. These axioms propose d a positive relationship between levels of certainty , intimacy and simila rity (Berger & Calabrese, 1975), i.e., the higher the similarity and intimacy established between individuals, the higher their feelings of certainty . The non - significant correlations between proportion of secondary questions, breadth of self - disclosure , and peripheral and core depth of self - disclosure , and levels of certainty can be attributed to the lack of power. Post - hoc power analysis using the G*Power the power was .6 which according to Cohen (1988) is not suffici ent to yield significant effect sizes. Thus, one way of detecting significant correlations between the conversational elements and levels of certainty is to increase the sample size and replicate this study in the future . Hypothesis s ix Social attraction, or feelings of liking toward a partner , was assessed as a function of perceived versus actual similarity. Previous r esearch has demonstrated that actual similarity, or sharing attributes that are objectively similar, is a significant predictor only when there is no interaction between partners (cf. Montoy, Horton, & Kirchner, 2008). Therefore, it was hypothesized that actual similarity, as measured by matching Facebook profiles, would be a significant predictor of liking when participants did not interact with each other. This research assessed the relationship between actual similarity and liking for PTC participants for time stamp 47 1 , where participants Actual similarity was not found to be a significant predictor of liking. The lack of support for a ctual similarity as a significant predictor of liking in this study can be attributed to a lack of measurement validity. The technique used in this research to see the effect of actual similarity on liking is akin to the phantom - other technique. However, u nlike the phantom - other technique, where previously unacquainted individuals were provided with attitudinal information on salient topics about each other, this study provided no such attitudinal information. In this study, participants were provided with information such as location, education, work place, and interests in entertainment, which may not have had as much salience as the attitudinal information in previous studies. Additionally, this study did not assess the degree to which the profile pages were attended to and retained. Simply put, just because matches were evident on the profile pages does not mean that the participants recognized and retained the information. Thus, th is flaw in measuring actual similarity could have impeded the effect o f actual similarity on liking in the no - interaction condition. Future research could employ a technique in which participants assess the degree to which their profile pages match in order to make actual similarity salient. Hypothesis s even T he effect of perceived similarity, or the belief of being similar to someone, on liking was also assessed. It was hypothesized that perceived similarity would be a stronger predictor of liking than actual similarity when there was interaction as well as when there was no interaction. The data were consistent with the hypothesis as the results indicated that perceived similarity emerged as the only significant predictor of liking in both the PTC and CTP conditions . The 48 results of this study corroborated the existing support that the perceived similarity and liking relationship has received from previous studies ( cf. Montoya, Horton, & Kirchner, 2008). Before understanding the reason be hind the higher potency of perceived similarity for predicting liking than actual similarity, it should be understood why similarity, in general, has a positive effect on liking. Byrne (1971) explained the reason why individuals like others who are similar to them. According to Byrne (1971), individuals have a basic need for a consistent view of the world, especially in times of uncertainty. Thus when individuals meet others who are similar to them, their ideas and attitudes get reinforced which in turn inc reases the consistency of their world . This need for congruence may lead individuals to delude themselves into believing that they are similar to each other, even when in reality , they are not. Ross, Green, and House (1977) labeled this the false - consensus bias. Such a bias may seem to be working for the participants in this study. In both the conditions, the level of uncertainty was very high. In the their level of uncertainty on a 7 point scale , where lower scores indicated high uncertainty and higher scores indicated low uncertainty. On an average, the PTC participants scores 2.9 out of 7 (Paul and & Morrison, 2014). Thus, given that their scores we re lower than the mid - point of the scale, it is safe to state that PTC participants were highly uncertain about their partners. On the other hand, CTP participants had no information about their partners before the online conversation with them. Given this absolute lack of knowledge, it can be assumed that their initial uncertainty was also very high. Thus, given the high levels of uncertainty experienced by participants in both PTC and CTP conditions, it can be assumed that they were highly susceptible to the false - consensus bias. This bias seemed to be working for all the participants because even when they were not actually similar, their belief that they were similar led them to like each other. 49 In summary, a unique contribution of the present study is that it provided empirical evidence for the claim that exposure to SNS profiles affect conversational behaviors during initial interactions between strangers. Exposure to profiles led individuals to ask fewer questions to their conversational partners, disclose information about themselves on a wider range of topics, and also volunteer more intimate information about themselves to others, compared to individuals without access to their partners SNS profiles prior to interaction. In terms of these conversational behaviors predicting uncertainty, , only two conversational elements were found to be significant. Intermediate levels of self - disclosure (i.e., disclosures regarding attitudes, preferences, opinions, desires and aspirations) and topics that established similarity between conversational partners were both positively related to certainty. Moreover, the nature of liking also was assessed as a function of actual versus perceived similarity. Results indicat ed that even when individuals did not share objective similarity, their false belief s of being similar to their partners could lead them to like their partners. 50 Limitations a nd Future Direction s Since this study was the first of its kind in analyzing initial conversations after exposure to SNS profiles, it can be extended and improved in a number of ways to answer the same or different questions about the use of such SNS profiles on the interaction ensuing between pr eviously unacquainted individuals. I have listed three potential areas for future work to focus. First, this study found significant differences between PTC and CTP participants on their use of conversational elements such as depth an d breadth of self - disclosure , and frequency of question - asking, but the effect sizes were moderate to low (see Table 3) . Significant correlations between the use of intimate self - disclosure and communicated similarity and levels of uncertainty also were found , but the correlations were not strong (see Table 4) . A major reason behind the low effect size and weak correlations was the low power resulting from the low sample size. Future studies should replicate this study with a larg er sample size to find out i f these non - significant results reach the appropriate level of statistical significance to attest to the propositions made in this study. Second, t his study examined four dimensions on which the conversations between participants in the PTC or CTP condition s could differ: frequency and type of questions, intimacy of self - disclosure , and topics of conversation. This study can be extended by incorporating other factors of c onversation that could be different in the conversations ensuing from viewing versus not viewing SNS profiles. For instance, the valence of self - disclosures (positive, negative, and neutral), and the types of disclosure (evaluative and descriptive) are tw o examples of the ways in which these conversations might vary ( Dindia, Fitzpatrick, & Kenny, 1997; Wheeless, 1978; Wheeles & Grotz, 1976). 51 Third, this study considered four conversational elements that c ould be related to levels of uncertainty. It is worthwhile to also consider other factor s that may impact uncertainty . As Berger and Kellerman (1983) suggested, the use of uncertainty reduction str ategies is determined by two factors: the effectiveness of the uncertainty reduction strategies and the social appropriateness of using these strategies. Politeness strategies are used by individuals to negotiate the social appropriateness of the use of th eir uncertainty reduction strategies. It will be interesting to see if the CTP and PTC participants differ ed in the use of these politeness strategies . Since politeness has also been shown to increase liking and attraction (Bell & Daly, 1984), analyzing the use of these strategies can further inform our understanding of uncertainty, liking, and attraction and how they are related to viewing SNS profiles prior to chatting with others online . 52 Conclusion This study examined the effect of exposure to SNS profiles on the conversational behaviors between previously unacquainted individuals, and compared these conversational behaviors with th ose profiles. This analysis and comparison w as conducted in an attempt to understand if uncertainty could be explaine d by the profile viewing and the conversational behaviors. In addition to uncertainty levels, the nature of liking also was analyzed as a function of perceived and actual similarity. The chat transcripts from the Paul and Morrison study (2014) were content analyzed to investigate these questions. Out of the 41 transcripts examined , 22 were those of PTC participants who had viewed prior to having an online conversation with them, and 19 were those of CTP participants who had not viewed their prior to the online conversation. Multivariate ANOVA results showed that PTC and CTP participants significa ntly differed on certain conversational behaviors. PTC participants asked fewer questions during initial interactions than CTP participants (Hypothesis 1 supported), but did not ask a greater proportion of probing questions than CTP participants (Hypothesis 2 not supported) . PTC participants had a greater breadth of self - disclosure , and also made more intermediate level self - disclosures than the CTP participants (Hypothesis 3 supported). However , the PTC participants were not more likely to use conversational topics that established similarity compared to their CTP counterparts (Hypothesis 4 not supported). Correlation analyses indicated that talking about topics that established similarity and engaging in intermediate level s of self - disclosure had a positive effect on certainty (Hypothesis 5 partially supported). Lastly, results of the regression analysis showed that having actual similarity with partners d id not affect feelings of liking 53 toward them (Hypothesis 6 not supported); rather, it was the belief that they were similar to their partners, ( i.e., the perception s of similarity ) , that positively predicted feelings of liking toward partners (Hypothesis 7 supported). In summary, this study informed us how SNS profiles affect initial conversations between previously unacquainted individuals . As is evident from the findings, SNS profiles (a) help in getting to know a person such that fewer questions are needed in an initial encounter, (b) make people susceptible to having impulsive conversations about a variety of topics, and (c) prime people to volunteer more intimate information about themselves to others during initial conversations given that the surface - level information is already provide d in the profiles. In addition to the how, this study also showed why SNS profi les are beneficial for the acquaintanceship process between previously unacquainted individuals by looking at the effect of certain conversational behaviors on uncertainty. Results indicated the medium levels of intimate self - disclosure and similarity established through conversations increased the sense of predictability and certainty toward conversational partners. Thus SNS profiles act as a facilitator of two of the most important factors of successful relationship development: intimacy and similarity . This unique quality of SNS profiles, namely fostering a sense of intimacy and similar ity between previou sly unacquainted individuals, is what make s them a very effective tool for friendship formation. 54 APPENDICES 55 Appendix A: Tables Table 1: Participant Characteristics Measures Chat - then - Profile(CTP) Profile - then - Chat(PTC) N = 82 38 44 Gender Female (F) 19 27 Male (M) 19 17 Dyad Composition Same - Sex Dyads 8 26 Opposite - Sex Dyads 30 18 Age of Participants 18 - 21 25 24 22 - 25 9 14 26 - 30 2 4 31 and above 2 2 Race of Participants Caucasian 26 23 Asian 6 11 African - American 3 6 Latino/Hispanic 1 1 Middle Eastern 2 0 Native American 0 2 Asian American 0 1 Year in School Senior 21 26 Junior 12 12 Sophomore 2 1 Freshman 0 1 Graduate 3 4 56 Table 2: Inter - coder Reliabilities Coding Categories Identifying Questions .97 Identifying Secondary Questions .93 Identifying Self - Disclosures/Information/Other .88 Identifying Peripheral/Intermediate/Core Disclosures .88 Identifying Breadth of Disclosures .78 Counting instances of Communicated Similarity .96 Counting Amount of Information in profiles .96 Counting Actual Similarity through Facebook profiles .99 57 Table 3: Significance Tests for Questions, Proportion of Secondary Questions, Depth and Breadth of Self - Disclosures and Communicated Similarity by Experimental Condition (PTC or CTP) + Dependent Variables df df error F Condition CTP = 0 PTC = 1 Means 95% CI L.Bound U. Bound Observed Power Number of Questions (#Q) 1 79 5.46 * 0 1 9.84 7.84 8.64 6.73 11.04 8.95 .64 Proportion of Secondary 1 79 1.30 0 1 0.73 0.78 0.03 0.03 0.67 0.72 .20 Self Disclosure - Peripheral 1 79 5.28 * 0 1 9.71 7.41 8.30 6.11 11.12 8.71 .62 Self - Disclosure - Intermediate 1 79 4.69 * 0 1 13.20 16.92 10.78 14.68 15.62 19.16 .57 Self - Disclosure - Core 1 79 3.19 * 0 1 0.04 0.31 - 0.16 0.11 0.25 0.50 .42 Self - Disclosure - Breadth 1 79 4.30 * 0 1 4.33 4.92 3.92 4.55 4.73 5.30 .54 Communicated Similarity 1 79 1.54 0 1 5.49 6.44 4.42 5.45 6.57 7.43 .23 *Significant univariate effects for dependent variables at the level p< .05 for 2 - tailed F test + Results are adjusted for dyad composition covariate 58 Table 4: Correlation Matrix of Uncertainty Levels, Number of Questions, Proportion of Secondary Questions, Depth of Self - Disclosures (Peripheral, Intermediate, and Core), Breadth of Self - Disclosures, and Communicated Similarity Variables URC Prop_SQ SD1 SD2 SD3 B_SD CS Uncertainty Levels (URC) --- Proportion of Secondary Questions (Prop_SQ) - .13 --- Peripheral Self - Disclosures (SD1) .02 - .06 --- Intermediate Self - Disclosures (SD2) .20 - .13 .01 --- Core Self - Disclosures (SD3) .06 .08 .06 .28* --- Breadth of Self - Disclosures (B_SD) - .03 - .02 .28* .36** .29** --- Communicated Similarity (CS) .20 - .19 .12 .45** - .02 .09 --- *Correlation is significant at the 0.05 level (2 - tailed). **Correlation is significant at the 0.01 level (2 - tailed). Table 5: Regression Analysis with Liking at Time 1 as Dependent Variable in PTC Condition + Variable B SE(B) t Sig. ( p ) Actual Similarity .11 0.08 0.22 1.46 .15 Note. Adj. R 2 = .06 + 2 - tailed t - test results are displayed. Effects are adjusted for amount of information covariate 59 Table 6: Regression Analysis with Liking at Time 1 as Dependent Variable by Experimental Condition + Variable B SE(B) t Sig. ( p ) In PTC Actual Similarity - 0.004 0.07 - 0.01 - 0.06 .95 Perceived Similarity 1 0.60 0.13 0.63 4.75* .00 Note. Adj. R 2 = .35 In CTP Actual Similarity - 0.15 0.09 - 0.23 - 1.69 .10 Perceived Similarity 1 0.65 0.13 0.67 5.03* .00 Note. Adj. R 2 = .41 Table 7: Regression Analysis with Liking at Time 2 as Dependent Variable by Experimental Condition + Variable B SE(B) t Sig. ( p ) In PTC Actual Similarity 0.03 0.06 0.05 0.45 .66 Perceived Similarity 1 0.77 0.10 0.76 7.43* .00 Note. Adj. R 2 = .56 In CTP Actual Similarity - 0.10 0.10 - 0.15 - 1.00 .32 Perceived Similarity 1 0.67 0.16 0.62 4.21* .00 Note. Adj. R 2 = .35 + t - value is significant at the 0.05 level (2 - tailed) Effects are adjusted for amount of information covariate 60 Appendix B : Unitizing Conversations This manual provides directions for coding the complete interaction transcripts between participants. You will be cod ing relevant portions of the all conversations. The first step of the coding process is for you to organize the conversation between participants in terms of thought units. A thought unit is a non - reflective clause, that is, one that can stand alone and does not distort the meaning of the rest of the sentence if it is taken away. It is a single complete thought with a subject and a verb (Dollard & Mowrer, 1947). Since this coding category provides the basis for several other codes it is necessary that it be scored as accurately as possible. A. Types of sentences with combinations of independent and dependent clauses **Understanding Independent and Dependent Clauses** An independent clause expresses a complete thought and could be considered a sentence by i tself and is, therefore, scored as 1 thought unit. A subordinate or a dependent clause does not express a complete thought and must be attached to an independent clause and is not scored. Dependent clauses begin with subordinating conjunctions such as: after although as as if along as also as though because before even if even though especially if if only in order that now that once rather than since so that than that though till unless until when whenever where whereas wherever while. I. Simple sentences (Just one independent clause) are always scored as 1 unit. Examples a. He went to the store. (1 unit) b. I am afraid of heights. (1 unit) II. Compound sentences are scored according to the number of independent clauses they contain. A compound sentence consists of two or more main clauses that are connected with a coordinating conjunction, a comma, a semicolon along, or by a conjunctive adverb and a semicolon should be classified as separate utterances since they are different - but complete - ideas. Coordinating conjunctions are FANBOYS for, and, nor, but, or, yet, so. Examples a. She drank water/ for she was thirsty (2 units). b. The visiting dignitaries were met by the President / and they were entertained at the White House. (2 units) c. Sally learned to knit during the summer / and now she spends most of her time knitting socks for Bill. ( 2 units) d. I cannot sing/ nor can I dance (2 units). e. I was face to face with my partner/ but I could not hear him. (2 units) f. You can buy the book/ or you can borrow it from the library. (2 units) 61 g. Tom studied a lot/ yet he did not pass the test (2 units) h. It w as cold outside/ so I put on a sweater (2 units) **** Positioning of dependent clauses in the sentence **** Sometimes dependent clauses are placed before the independent clause. Examples: i. If only Johnny was a little taller, he could be on the roller coas ter ride (1 unit). j. Where do you work? If you have a job that is (1 unit). III. Complex sentences contain one or more dependent clauses. Complex sentences are scored as only 1 thought unit. Examples: a. The look that she gave me was discouraging. (1 unit) b. I was not comfortable with him because he had a weird vibe (1 unit). c. What the announcer said was not clear. (1 unit) d. There is a man everybody admires. (1 unit) e. He runs as if he is afraid. (1 unit) IV. Compound - complex sentences contain two or mor e independent clauses and one or more dependent clauses. Compound - complex sentences are scored only for the number of independent clauses they contain. Examples: a. The visiting dignitaries who landed at National Airport were met by the President / and he escorted them to the White House. (2 units) V. Further understanding of AND, OR, BUT Sometimes participants will use coordinating conjunctions and, or, and but to make lists, give options of similar/same things (for example, types of movie, food, music, p lace, etc). In those cases (i.e. when unrelated things are being listed together), count the sentence as one unit. Examples: a. I like 2Chainz, Snoop Dogg, and Eminem. (1 unit). b. I would watch Stepbrothers, Road Trip, but I will not watch Pineapple Express ( 1 unit). c. I can use Gucci or Prada. (1 unit). d. She was confident but way too cocky (1 unit). e. You can have a small family or a big family (1 unit). f. He bought her fiancé a diamond ring and he is taking her on a cruise (2 units). g. The Taj Mahal was beautiful and clean and looked really nice (1 unit). h. She went to Barbados in the summer and then again in the winter (1 unit). VII. Watch out for IMPLIED WORDS thought unit is scored when the subject of an independent clause is implied and conveys a thought different from the preceding clause. Examples: 62 units) le very often. (2 units) (1 unit) i. Class was good, not boring. (2 units). B. Emphatic Clauses Emphatic statements are statements which are set off with exclamation marks. Mild interjections are statements which are set off by commas (or implied commas). Treat them as modifying subordinate words or clauses. For example: "No! I disagree" "Really, do you think so?" 63 These statements should be coded as a single utterance if the respo ndent does not follow it immediately with an explanation or a supporting statement. For example, Denzel stepping up. (1 unit). Drew: Yup, totally. (1 unit). Drew: So do you follow any other sports outside basketball? OR Denzel stepping up. (1 unit). Drew: Yup, totally. (1 unit). Drew: So do you follow any other sports outside basketball? Drew: Although, Matt Costello really needs to step up his game if we have to beat Wisconsin. (1 unit). 1. These statements should also be coded as a single utterance if the respondent follows it up with a statement that is unrelated. For example, doing very well, especially with Travis and Denzel stepping up. (1 unit). Drew: Yup, totally. You watched any movies recently? (2 units). 2. These statements should *not* be coded as a single utterance if the respondent follows it up with an explanation or a supporting statement. These statements should then be combined with the following or preceding explanatory clause. For example, up. (1 unit). Drew: Yup, totally. Those guys a re in good shape (1 unit). Drew: So do you follow any other sports outside basketball? OR Denzel stepping up. (1 unit). Drew: Yup. Although, Matt Costello really needs to step up h is game if we have to beat Wisconsin (1 unit). C. Unintelligible and Repetitive Clauses i. counted as units. ii. Redundant phrases such as "Yes, yes, yes" or "yeah that's probably... yeah that's probably single utterance. D. (i) Typos and corrections 64 Since this is a text - based conversation, you will find that sometimes participants rectify their spellings with the asterisk (*) sign. The words and phrases that follow the asterisk sign or rectifications of spelling mistakes will not be coded as units. For example, Danny: The weather in Mchjan is so unoedictable (1 unit). Danny: *Michigan (0 unit). Danny: lol I meant unpredictable (0 unit). However, if the participant provides an explanation for their typos, that will be coded as a separate unit. For example, Danny: The weather in Mchjan is so unoedictable (1 unit). Danny: *Michigan (0 unit). Danny: lol I meant unpredictable I am totally out of my game right now (1 unit). (ii)Misplaced commas, periods, question marks and other punctuations Together with spelling mistakes, sometimes commas, periods, and question marks might also be misplaced because of typing errors. Please be extra careful when figuring out types of clauses and sentences because misplaced commas and periods can confuse the p roper identification of the coding units. For instance, Danny: Hey Kim, how are you (1 unit). Danny: ? (0 unit) E. Greetings/ Conversation Starters i. unit. ii. iii. should be coded as one unit. F. Parenthetical Statements A parenthetical elements is a word or group of words that interrupts a sentence and is not part of its main structure. Parenthetical expressions include explanations and information that are typically viewed as helpful, but nonessential. Thus, they would typically be handled as descriptive dependent elements. However, in some cases in online chat, participants may put comp lete phrases/sentences within parentheses in order to provide an explanation . If a parenthetical expression contains a subject and a predicate and contains a complete, meaningful thought it should be coded as an independent utterance. If it lacks elements necessary to permit independent meaning it should be combined with the independent clause it describes or modified. For example, Demi: So what kind of music do you like? 65 Kevin: I like soft rock (3 Doors Down, Nickelback), pop (Ed Sheeren, Sam Smith), an d sometimes hip - hop too (although it is not really my scene). (1 unit). BUT, Demi: Where do you live? Kevin: I live in Schaumberg, Illinois. (It is a small suburb near Chicago, IL). (2 units) 66 Appendix C : Identifying Questions This manual provides guidelines for identifying questions and self - disclosures in the unitized chat transcripts. The coder must keep a count of the questions and self - disclosures. What is a question? The operational definition of a question that will guide the overall coding is this - "An expression of inquiry that invites or calls for a reply: an interrogative sentence, phrase, or Guidelines for identifying questions: 1. Primarily, should be coded as a questi on EXCEPTIONS: not be questions. They are backchanneling responses. Examples of such responses are . They are categorized as feedback responses that prompt the partner to continue. They serve the same function as sentences like "Tell me more." Make sure that you are not coding them as questions. These type of feedback will be coded under the category Feedback (Question). 2. Watch out for implied questions Identify statements that might not be posed as a question but these statements are expressions of inquiry nonetheless. For example, Danny: Tell me about yourself. Desiree: Well I am a sophomore in a ccounting, and I am from Okemos. statements as questions. 3. Watch out for abbreviations and one word questions 4. Since the transcripts are those of online tex t - based chat, i.e., participants had to type and communicate, these transcripts contain textual errors made by the participants. One of the most Even if a statem that is making an inquiry, you should code it as a Question. For instance, 67 Danny: You are right. So what is your major Desiree: My major is accounting. In this example, even sign at the end, even if he was asking for a response. Code this as a question 5. One of the other errors that participants make while typing is the misplacemen For instance Danny: You are right. So what is your major Danny:*? OR Danny: You are right. So what is your major Danny:? invariably separated 6. Watch out for combination of feedback responses and questions In some cases, participants will ask multiple questions in one single line. One of these questions might be feedback response and the other might be a question which is not a feedback response. You should code them separately. For instance, Danny: Really? You like sports? Really? (Question - other) Sports?(Question) 68 Appendix D : Identifying Types of Questions IMPORTANT: 1. Whenever you are making a judgment of what type a particular question is, it is ALWAYS advised to see the question in context of the conversation. Since conversational utterances are linked with each other, i.e., there is always a flow (or lack of in some instances), looking at the bigger picture will always help you in making an informed decision. 2. Also, remember that you are looking at text - based chat conversations. As you know through your experience in texting and online communication, ther e is a delay in you texting and the other person responding or the other way around. Due to these small delays the sequence of these conversations can go out of order. Therefore, in situations like this, it is all the more important to look at question utt erances in context. How to place in context? You might need to look at utterances immediately above and below it OR you might need to look at utterances a coup le of lines above and below it. There are two types of questions you will be coding for: Primary questions and Secondary questions. Primary questions a. These are questions which introduce new topics of discussion. b. Even when these questions are taken out of context, the questions still make sense. c. These questions can stand alone which means that the m eaning of the question can be deciphered. For example, These questions introduce new topics of discussion and make sense even when taken out of the context of the conversation. Secondary questions a. These questions need to satisfy one of the following conditions i. Do not make any sense when taken out of context OR ii. Make sense only when looked at within the context of the conversation because they gather more inform ation about an area or topic previously introduced by a primary question or a secondary question or a self - disclosure. 69 There are two variations of these questions: a. Questions that do not make sense when taken out of context For example, rowing up there that One of the indicators of these questions are use of pronouns words that are used to replace an object that has already been mentioned or can easily be known, such as: He she him her it t hey them theirs here there. WORD OF CAUTION be correct, but it does not make sense if you randomly ask someone this question. You need to provide some precedence to this question for this question to make sense. Further examples: Jake: Yeah I am a junior at MSU. Frank: Did you transfer here? OR Jake: I took COM 225. Frank: It is hard, right? here in the question. If you were to ask this question to another person without giving them a background of what you are referring to when you say here , then the question does not make any sense. it. I f you were to ask this question to someone randomly it would not make sense because the other person does not know what is the it you are referring to. b. Questions that gather further information about a topic previously discussed These questions will requ ire you to look at the sequence of the topics discussed to that point to understand if this is a brand new topic or if this is a topic that is bringing more clarity (resolve doubts) or more depth to a topic that has been discussed thus far. This requires m eticulous monitoring of the context of conversation. These questions might look like primary questions, i.e., they make sense as a stand - alone question and they introduce a topic of conversation that has not been discussed thus far. However, these questio ns might qualify as sub - topics 70 questions getting me more refined and more particular information about This is an example of seconda ry questions getting into depth . 1. Justina: Do you go to State? 2. James: Yes, I am a Kin Major here. 3. Justina: What year are you? 4. James: I am a junior going to my senior year. 5. Justina: Oh, awesome! When do you think you will graduate from MSU? 6. James: I am hopin g 2015 summer! In this scenario, utterance 1, 3, 5 are related to each other in the following sequence academic academic background, questions 3 and 5 will be coded as secondary questions. This is an example of secondary questions getting clarity or resolving dou bt . 1. Elsa: I really like One Direction and Maroon 5. alone. However, it derives information fr Introducing topics of discussion from Facebook profiles them. Some of these participants used the informat to it ask questions to their partners. For instance, Justina: I noticed that you are from Korea. Which city in Korea did you grow up in? Seong Jun: I grew up in Seoul South Korea. It was great! In this exa discussion thus fulfilling the criterion of primary questions. However, by reading the question you understand that she is deriving the information from another source (th e Facebook profile of their partner). In such cases, even though the question makes sense by itself and introduces a new topic of discussion, it should be coded as secondary because it probes on information from the profile. 71 Normally, they preface the que stion they ask based on Facebook profile information with a statement which states that they looked the information up. Some of the phrases participants use to probe information from Facebook profiles are However, sometimes they will use particular information from the profile and use it directly in the question. For example, The way to identify that these questions are secondary questions and not primary (even though they are introducing new topics of conversation) is by looking at utterances before this question. If there were no questions or self - disclosures leading up to th is question, i.e., if there was no discussion on this topic thus far, then code it as a secondary question because, chances are, the participants have looked at a certain piece of information from the profile and basing their questions on that information. 72 Appendix E : Identifying Self - Disclosures Self - disclosure (SD) This definition will help you in recognizing the utterances that qualify as self - disclosures. Self - disclosures are defined as messages that reveal personal information about the sender or the receiver. Chelune (1975) defines a self - reference as "a verbal response (thought unit) which describes the subject in some way, tells something about the subject, or refers to some affect the subject experiences" (Chelune, 1975, p. 133 cited in Tardy, 1988). When an utterance is identified as a self - disclosure, mark it as SD. Example 1 Kelly: What do you like to do for fun? James: I like snorkeling, scuba diving, and bowling, and during weekends I hang out with my friends. Example 2 Kelly: What year are you? James: I am a sophomore going into my junior year in Communication The utterances by James in both the examples qualify as SD because it reveals something personal about him his hobbies (snorkeling, scuba diving, etc.) and his senior ity and major (sophomore going to junior in Communication). Information (I) Certain utterances by participants do not qualify as self - disclosures even though participants participate in the act of disclosing information. This is because the information th ey disclose/share does not reveal anything personal about themselves. These utterances will be coded as Information (I) . These include information about objects like third parties, places, movies, music, books, tv shows, sports, and so on. EXCEPTION: The utterance will be coded as Information if and only if they share information about an object without any personal judgments on the object . If the participants pass personal judgments on the object while describing the object or sharing information about t he object, then the utterance needs to be coded as SD. Example 1 1 Jake: Where do you live? 2 Jason: I live in Mason 3 Jason: It is a small town around 20 mins away from East Lansing wever, the third I. This is because, even though Jason is disclosing information, it is not personal to him, it is about the particulars of the geographic locati on. 73 Example 2 1 Jake: I really like to listen to Nickelback and Papa Roach 2 Jaaon: Papa Roach? In this example, the third utterance by Jake will be coded as I because he is sharing information about the particulars of the band Papa Roach. Example 3: 1 Jake: I will be graduating in June of 2016 if I stick with Mech Engineering 2 Jake: It might take me two more years if I switch it to Com p Engg. In this example, the second utterance by Jake looks like he is giving the time it takes to graduate and it is specific to him. Therefore this needs to be coded as a SD. Example 4 1 Jake: I have been to San Diego but not Santa Monica 2 Jason: I have been to Santa Monica 3 Jason: Th e weather there is amazing! 4 Jason: And the people are so nice! In this example, the third and fourth utterance by Jason will be coded as SD together with utterance 2. This is because, even though Jason is talking about a place: Santa Monica, he i s Example 5 1 Riley: What is your major? 2 John: I am a Com arts major with a concentration in business 3 John: Comm can be pretty intense at times especially COM 300 and COM 225 S imilar to the previous example, in this example the third utterance by John refers to information these classes. Therefore, 3 John must be coded as SD. Example 6 1 Jake: I went to Orlando in the summer 2 Jake: It was 95 degrees! The second utterance is going to be coded as Information because Jake is just stating the temperature of the place, he is not stating any personal judgment about the place by saying that it was too hot or too cold. Other utterances (O) An utterance that is neither a self - disclosure nor a question will be coded as Other (O). This category will include a wide range of utterances. These include: 74 (i) Stand - alone feedback words/phrases or exclamati ons/emphatic phrases a. Examples include: (ii) Greetings: These are replies to questions asked during the initial few lines of conversations or during the last few lines of conversations a. Examples include (iii) Backchanneling elements a. Examples include: In some cases the single word response of "Yes" or "No" may be coded as SD. It may mean "Yes, I have that experience or knowledge" or "No, I do not ha ve that experience or knowledge." However, the circumstances where solitary "Yes" and "No" responses may be taken to be self disclosive are not as common as those situations where the single word response is a backchanneling element. Please be conservative with your coding of single word responses. Examine the context carefully to ascertain the correct meaning in context for all utterances. Combinations of Other Utterances, Information, and Self - disclosures Certain utterances in the chat transcript might i nclude a combination of utterances that fall under - combination, follow the following rules Information + Other = Information Self - disclosure + Other = Self - disclosure S elf - disclosure + Information = Self - disclosure For example, 1 Jeremy: Oh haha no (O) it is not the Miami of Florida, it is the Miami of Ohio (I) - >I - >SD 75 3 Jeremy: There are Seminoles and Gators down in Florida (I). But I am more of a Gators fan (SD) - > SD. Some overall rules of coding: 1. Do not code utterances that a. have not been numbered , or indicate that there have been corrections to the utterances (for instance, the utter b. Render clarification to an utterance stated earlier . For example, John: I love LOTR, and GOT Jake: LOTR? John: Lord of the Rings self - disclosure be cause he is just providing clarification of a self - disclosure he has already made in an earlier utterance. c. Repetitions of the same utterance by the other person Example, Jake: Yeah I guess I am lucky! coded as self - disclosure because he is just echoing or repeating what his partner had said earlier. 2. Utterances might look like information but are self - disclosures. This becomes evident only evident when you r ead the utterances above, i.e., when you put the utterance in context of the conversation. Placing the utterance in context of the conversation is very important in text - based chat. Since text - based chat is asynchronous (i.e. there is a delay in responses) the sequence of conversation can be off. So looking at the context and then deciding if the utterance is a self - disclosure or just information is important. For example, Jake: What are the Harry Potter books you have read? Jessica: What major are you in ? Jake: I am a pre - med major. Jake:.. Jessica: The Half Blood Prince, Order of the Phoenix. 76 In this example, the last utterance by Jessica might seem to be information because it is referring to the names of the Harry Potter b question of what books of Harry Potter series Jessica has read. 7 7 Appendix F : Identifying Depth of Self - Disclosures Review of definition of self - disclosure (SD) This definition will help you in recognizing the utterances that qualify as self - disclosures. Self - disclosures are defined as messages that reveal personal information about the sender or the receiver. Chelune (1975) defines a self - reference as "a verbal response (thought unit) whi ch describes the subject in some way, tells something about the subject, or refers to some affect the subject experiences" (Chelune, 1975, p. 133 cited in Tardy, 1988). Defining depth of self - disclosure The depth of self - dislosures is defined as the leve l of intimacy of the self - disclosures (Altman & Taylor, 1973). The more intimate the information, the greater the depth of the disclosure. There are three layers of the depth dimension of self - disclosures. They are: 1. Peripheral Layer (Category 1) This laye r involves information that is superficial in nature. For instance, demographic information such as Name Age Gender Height Weight Marital Status Zodiac Education Major Minor College Class/Course title Classification Major Job/ Internship Title Job/ Internship Description Religion Family Composition Hometown Birth Place Current Location Future Location Personal History/ Past Experiences Past Experience involving third parties Concrete Future Plans Examples: 78 2. Intermediate Layer (Category 2) This layer involves information that is higher in intimacy than peripheral layer. This includes information such as Attitudes Preferences Opinions Interests Hobbies Aspirations Dreams Desires These attitudes/preferences/opinions can pertain to music, movies, tv shows, books, sports, places of tra vel, school, work, and other related things. Desires and dreams are future plans that are not concrete, i.e., one hopes or wishes for them to occur but there are no concrete plans of them materializing. Examples, Preference) NOTE: Distinction between periphera l and intermediate layer of information When you come across self - disclosures that could be either peripheral or intermediate ask yourself the question Is the person disclosing OBJECTIVE FACTUAL Information or SUBJECTIVE Information? Factual or objective information is when a participant discloses facts about him/herself. These disclosures normally do not have adjectives involved. Subjective information about a particular topic of disclosure is when participants share their point of v iew or opinion about the topic. These topics can include anything from movies, music, to family. These self - disclosures normally include adjectives. 79 For instance, look at the sequence of self - disclosures Category 1 (This is fact - based) Category 2 (This is opinion - based) Some other examples are, Category 2 (Preference/interests) Category 1 (Personal History/Past Experiences) Category 2 (Preferences) Category 1 (Personal History) Category 2 ( Desire) Category 1(Family Composition) Category 2 (Opinion) Category 2 (preference) Category 1(past experience) . 3. Core Layer (Category 3) This layer involves information that is highly intimate in nature. These include highly personal aspects related to basic values, fears, needs, and self - concept. This information discloses highly vulnerable aspects about self. O pinions about highly sensitive topics such as religion and politics should be coded as core layer. - concept) (self - concept) - concept) - concept) 80 - concept) NOTE: Mostly this information has a negative valence but not always. 81 Appendix G : Identifying Breadth of Self - Disclosures 1. Demographic a. Name b. Age c. Gender d. Marital Status e. Marriage Date f. Zodiac g. Education i. Major ii. Minor iii. College iv. Class/Course title v. Class description vi. Duration in school vii. Graduation date h. Classification i. Freshman/sophomore/junior/senior i. Job/ Internship Title j. Job/ Internship Description k. Job Location l. Family Composition m. Languages spoken n. Hometown o. Birth Place p. Current Location q. Future Location 82 2. Biographic a. Personal Experiences from the past and present outside of school, work and job b. Plans for the future except for school, work and jobs c. Aspects of self that subject likes/dislikes d. Feelings that subject has difficulty expressing or controlling e. Things that subj ect feels ashamed or guilty about f. Things that make subject feel furious, depressed, worried, anxious, afraid, hurt g. Deeply, proud, full of self - esteem, full of self - respect, etc. 3. Hobbies, tastes, interests, preferences, and opinions a. Favorite foods and beverages b. Likes and dislikes in music, movies, tv shows, books, clothing, recreation , food, etc. c. Taste in social gatherings d. Favorite ways to spend free time e. What participant would appreciate most in a present a. Hobbies and interests one has participated in Examples: - NOTE: Distinction betwe en 2 and 3 (Experience and Hobbies/interest/etc) A lot of times these two categories can get confusing. The best way to think about the distinction between experience and hobbies is to look out for verbs and adjectives. If you find words that indicate some sort of action (VERBS), i.e., it involves the participant actually DOING something, then it is an experience and coded as 2. However, if you find words which indicate 83 ent on a particular activity then it is a hobby/interest/opinion etc. and coded as 3. Example: ----------- LIKE (adjective) hence 3 Versus ----------- PLAY (verb) hence 2 EXCEP TION: While discussing hobbies, sometimes participants use verbs to describe them. For tion needs to be coded as 3 even though it has the verb LISTEN in it. This is because, it is understandable that a person will not listen to a music they do not like. Same goes for sequences such as --- Category 3. 4. School / Work a. Reason for academic and career choices b. Pressures, strains at work/school c. What is boring, unsatisfying, exciting, satisfying about work/school d. Shortcomings, handicaps, qualifications, and strengths e. Ambitions and goals f. Feelings about salary and rewards g. How participant feels work is appreciated by others h. How participant feels about choice of career or major i. School assignments, course load, classroom experiences j. School related skills (math, reading, typing) k. School events (football games, the do rm lottery) NOTE: Job title is demographic, as is academic major Therefore, if participant discloses 84 But (school/work) IMPORTANT: 1. SUBJECTIVE about work and school, not OBJECTIVE FACTUAL information. responses that are factual and objective in nature). If however, participants are - disclosures need to be coded as 4 (because these questions generate responses that are subjective in nature, i.e., it can vary from person to person). 2. Also, if a participant discloses about his/her pre ference of future location without referencing any sort of work ambition, then code it as 1. Else code it as 4. 5. Love and relationships with others a. Relationship with family / people at work / roommate / dating partners / colleagues. b. How participant feels a bout people works for and with/takes classes from and with c. How participant feels about family members, dating partners, roommates, and others. NOTE: Asking about the structure of the family or work place, that is coded as category 01, but if participant asks about feelings t oward family members it is coded as 05. 6. Physical appearance/body a. Physical measurements or description b. Feelings about parts of own body c. Feelings of attractiveness to opposite sex d. Worries about appearance in past 85 e. Health problems f. Long - range health concerns g. Past record of illness/treatment h. Plans, attempts, or actions related to exercise i. Information about diet 7. Money a. Debt, debtors, savings, investments, financial worth b. Gambling habits c. Sources of income d. Subject's most pressing need for money e. How subject budgets money 8. Task Related a. This category would include statements and questions related to the study itself or the expe riment activities (becoming acquainted; getting recruited) 9. Current and historical (no n - experiential) events a. How participant feels about government policies, world affairs, national affairs b. How participant feels in general about current affairs such as employment c. Events they are just reporting on , they were not physically involved in the event - 10. Sex a. First sexual experience of participant b. Preferences and feelings toward sexual activity c. speaks of. 86 Appendix H : Identifying Communicated Similarity Communicated similarity is defined as instances in conversation where participants talk about similarities between each other, or they communicate agreement with each other, or find common ground with each other. When two individuals are similar it means t hat they are alike in a variety of ways. Similarity can pertain to similarity in factual elements of oneself such as same name, birthdate, place of origin, current location, seniority in school, and so on. These can be summed up as exact similarity . Simil arity can also mean that they have similar tastes in movies, music, tv shows, books, politics, religion, values of life, and so on. Similarity can also be articulated in agreement with the other person. This means that if the other person shares his/her op inion about something (or there is a mutuality of feeling) and the other person agrees with him/her, then that is also a similarity between them. These can be summed up as attitudinal similarity . Lastly, similarity can be based on common ground . Common gro und is defined as: (1) Similarity in past experiences (worked at the same place or went to similar vacation spots), present experiences (affiliation to the same university or taking the same classes), and future experiences (planning to take same classes o r going to the same vacation spot) (2) Similarity in social circle (having someone in social circle who has the same educational background as partner, or having siblings); (3) Affiliation to the same community or group in the past, present, or future (sa me major, same seniority in college, taking same number of classes, same professional goals, sharing an associated with a place or company); (4) Shared knowledge: Things both of the participants are aware of. Hence, Communicated similarity = exact simil arity + attitudinal similarity + common ground Instructions for coding: 1. You will be provided hard copies of chat transcripts between participants. 2. You will be reading the transcript in its entirety to look for instances of communicated similarity between p articipants. 3. When you have identified an utterance that shows exact similarity, highlight it in BLUE; an utterance of attitudinal similarity should be highlighted in PINK; and common ground should be highlighted in ORANGE. 4. After identifying instances of s imilarity, ONLY count the topics on which they were similar, NOT the utterances. This means that, if there are 4 utterances over which participant 1 and 2 have established their similarity in their opinion about a certain movie, then these 4 87 utterances sho uld be counted as 1 based on the fact that they are similar on one topic their opinion about the movie. 5. After looking through the whole transcript and counting the topics of similarity, please indicate the final count on the top left corner of the transc ript. GUIDELINES FOR IDENTIFYING COMMUNICATED SIMILARITY INSTANCES There are certain phrases that help in you identifying instances of similarity between participants. Some of these are listed below: me too same here so did I as well also so is mine I hear ya I feel the same way exactly I agree true I thought so too a. In order to identify instances of common ground, look for i. Shared experiences, i.e., they went to the same vacation spot, they had the same classes. E ii. They have someone in their social network who share something in common with their conversation partner. ner iii. Common or shared knowledge, i.e., they both know about the same place or they are both aware of a certain event. Please make sure that you identify these instances of shared knowledge with precision. This is because when knowledge is shared, people tend to avoid stating that explicitly. For instance, if both of you know about Transformers and then one of you starts talking about the characters Megatron and Decepticon without explicitly mentioning that both of you have watched/know about Transformers , it is easy to miss identifying this instance of common ground. stances happens when the subject on which the opinion is being g iven is not a mutual link, i.e., it is not shared means that if participant 2 had or has a similar experience of taking 5 classes and has felt the same way, then only is it considered similarity. Hence, having a mutual link or a shared experience is of cardinal importance when coding for similarity. 89 Appendix I : Identifying Act ual Similarity through profiles General Instructions In this task, you will be responsible for counting exact matches of information between two precision and your careful evaluation. Be as objective as possible and make a conscious decision to avoid personal reactions to influence your judgment. When in doubt, always refer to the manual or contact Aditi for further clarification. You will be matching information under the following categories 1) Education i. Present 1. Majo r, 2. University 3. Year of Graduation ii. Past 1. High School 2. Year of Graduation 3. Other university/community college if applicable 2) Work i. Position ii. Company iii. Duration of job (Month Year to Month Year) 3) Current Location i. City ii. State iii. Country (if applicable) 4) Hometown (Might be more than one) i. City iv. In Between Cities (Move in move out date) ii. State iii. Country (if applicable) 5) Sports i. Teams ii. Athletes iii. Cheerleading Squads 6) Music i. Bands iv. Concert Venue ii. Artists v. Record Label iii. Genre 7) Movies i. Series ii. Actors/Actresses/Directors 8) TV Shows i. S eries ii. Actors/Actresses/Directors 9) Books 90 i. Authors ii. Series You will be counting EXACT MATCHES OF INFORMATION Between the Facebook profiles of the participants The rule is, when you find two information pieces that are an EXACT MATCH in both profiles, give both of the participants a score of 1. We will call this the SIMILARITY SCORE . Things to be very careful about 1. Make sure the information pieces you are matching belong to the same category . For example, i) If you find that Person A has Michigan State University listed in his/her Education and Person B has also listed Michigan State University listed in his/her Educa tion, then both of them will have a similarity score of 1. a. This is only when Michigan State University is listed under the same category, that is, either as current university or past university. b. They will not st university is State University. ii) re of city do not match (East Lansing vs. Lansing) but the current state of Person A and Person B match (Michigan). ** NOTE: 1. Facebook changed its profile format during the course of this study. Therefore in some profiles, you will notice that the Work, Education, Current Location, and Hometown are in different formats. Please be extra careful of this when coding. 2. For most of the profiles, participants have provided their informatio n in the sequence: a. Sports b. Music c. Movies d. TV Shows and e. Books However, some participants have not maintained this order. This means that, they have provided their information on TV shows first, then sports, and so on. 91 When comparing information in the same categories, please be careful about the inconsistency in the structure of the profiles. 3. Watch out for similar looking pieces of information . Sometimes information may look the same, but they are different, and should not be given Similarity Scores. F or example, i. If Participant A has liked MSU Athletics and Participant B has liked MSU Basketball, the similarity score must be 0. j. If Participant A has liked Harry Potter and the Goblet of Fire and Participant e, the similarity score must be 0. k. If Participant A has liked J.K. Rowling and Participant B has liked Harry Potter series, the similarity score is 0. l. If Participant A has liked Harry Potter Series and Participant B has liked ALL the books of the Harry Po tter series separately, that is a similarity score of 1. m. If Participant A has liked the series House of Cards, and Participant B has liked House of Cards (Season 1), code is as a similarity score of 1. n. If Participant A has liked the series House of Cards, and Participant B has liked House of Cards (Season 1) and House of Cards (Season 2), then the similarity score is 1. DO NOT SCORE LOCATION OF SAME WORK AND EDUCATION LOCATION For example, i. If Participant A has Kroger, Inc. at East Lansing, MI and Particip ant B has Kroger, Inc. at East Lansing, MI, then the similarity score is going to be 1. It is not going to be 3 because you will not be counting the city and state of the location of the job. 92 Appendix J : Counting Amount of Information share d in profiles You will be counting information under the following categories 10) Education i. Present (Major, University, Year of Graduation) ii. Past (High School, Year of Graduation, other university/community college if applicable) 11) Work i. Position ii. Company iii. Location of c ompany iv. Duration of job (Month Year to Month Year) 12) Current Location i. City ii. State iii. Country (if applicable) 13) Hometown (Might be more than one) i. City iv. In Between Cities (Move in move out date) ii. State iii. Country (if applicable) 14) Sports i. Teams ii. Athletes iii. Cheerleading Squads 15) Music i. Bands iv. Concert Venue ii. Artists v. Record Label iii. Genre 16) Movies i. Series ii. Actors/Actresses/Directors 17) TV Shows i. Series ii. Actors/Actresses/Directors 18) Books i. Authors ii. Series 93 MAIN TASK You will be counting UNIQUE PIECES OF INFORMATION in each of the abo ve categories. MUST DO Before starting to count, do a thorough scan of the Facebook profile screenshot to find out any repetitive information (especially under sports/music/movies/tv shows/books). You might overlook repetitive information if you get busy just blindly counting the information. WATCH OUT FOR SIMILAR LOOKING items Sometimes, you will find participants liking different movies/books/music by the same auth or or artist. Then you must count them as separate units of information. For instance, i. A participant might like Harry Potter and the Deathly Hallows, Harry Potter and the Order of the Phoenix. Code them as separate pieces of information . ii. TRICKY SPORTS CAT EGORY In some profiles, you might see participants liked MSU Athletics and also liked something more specific like MSU FOOTBALL or MSU BASKETBALL. Code them as separate pieces of information. WATCH OUT FOR SPECIFICS of information In the Work section, you will find some participants being very specific about their work details. For instance, i. If a participant has mentioned that he/she worked as a Chef in Wonders Dining Hall at MSU from July 2000 - August 2002, then this will be counted as 4 pieces of inform ation, namely: 1. Chef, 2. Wonders Hall, 3. MSU, 4. July 2000 - 2002. Some participants preferred to provide no choices in some of the sections like sports/music/movies/tv shows/books. Therefore, in some profiles you mig ht encounter picture in the icon). Code the m as a one piece of information. What NOT to count: 1. DO NOT count information that is repetitive in the same category. In the work/education/sports/music/movies/books/tv shows categories, if a participant has the same team, artist, movie, book mentioned more than once, count the information as one piece. 94 For instance, i. If a participant has liked Taylor Swift multiple times, co de it as one single piece of information. ii. If a participant has mentioned that he/she attended MSU multiple times, code it as one single piece of information. 95 Appendix K : Procedures in the Paul and Morrison (2014) study 89 participants (39 males and 50 females) were recruited at a large public university in midwestern United States. Participants either received monetary compensation or extra credit for their participation. The only requirement to participate in the study was that participants needed to h ave a Facebook profile . The participation process for the study was divided into two steps. In the first step, the researcher sent out an email to the participants asking for the URL of their public Facebook profiles after making certain temporary changes to the privacy and content of their profiles. r, and, school and work information are considered as public information. Together with these fields, participants were also asked to provide access to the following fields: their current location, hometown, high school, college, work, favorite music, movi es, television shows, books, and sports teams. After the participants made the required changes to the accessibility of information and emailed their Facebook URL, the researcher took a screen - shot of the profiles and saved the screenshot as a jpeg file i n a jump drive to be used later during the experiment. Any extra information captured from the profile was removed from the jpeg file. After this, the researcher proceeded to the final step of the recruitment process by sending an online link to the partic ipants where they signed up for a one hour time slot with another person to participate in the experiment. After the dyads of participants were formed, they were randomly assigned to one of the two online conversation conditions: chat - then - profile (CTP) or profile - then - chat (PTC). In the CTP condition, participants chatted with their previously unacquainted partner for fifteen minutes and then looked at their Facebook profile. On the other hand, in the PTC condition, participants looked at the Facebook pro file of their previously unacquainted partner and then chatted with them for fifteen minutes. There were 43 participants in the CTP condition and 46 participants in the PTC condition. here the researcher greeted them and ushered them to a separate room with a Windows - based laptop in it. The researcher made sure that the participants did not speak to each other or see each other before the start of the experiment. The researcher asked fo name of their conversation partners to check if they knew each other. If the participants indicated that they knew each other prior to the experiment their data were excluded from the study. The participants then filled out the consent forms and received unique subject ID numbers. After this, the researcher gave them subsequent instructions depending on the experimental condition to which they were assigned. In both the conditions, PTC and CTP, the researcher told the participants that their task today was to get to know their partner as much as possible. In the PTC condition, after this profile and were given 3 - 5 minu tes to view the profile. After they had viewed the profile they were instructed to answer the first survey based on the Facebook profile they just saw. Following the survey, the researcher informed the participants that they were now going to engage in a 1 5 minute online text - based conversation on the chat portal Chatzy. com with the person whose 96 Facebook profile they just viewed, and that their task was the same to get to know their partner as much as possible. At the end of the chat, the researcher came in and administered the final survey and instructed the participants to fill up the survey bas ed on the profile they saw as well as the conversation they had with their partners.In the CTP condition, similar instructions were provided but the order of the instructions provided was reversed because the sequence followed in this condition was the rev erse of the PTC condition. In the CTP condition, the participants chatted with their partners, then took the first survey, following which they looked at each In both the conditions, the first and the second surveys contained questions that provided measurements for uncertainty reduction, similarity, and liking. Together with these items, the second survey in both the conditions consisted of additional demographic variables. In order to avoid fa tigue and order effects that come with repeated items measurement, the ordering of uncertainty reduction, liking, and similarity measurement items was changed for the second survey. The average time for filling up each of the surveys in both the conditions was around 2 - 3 minutes. The average chat time varied between 15 - 17 minutes. At the end of the second survey, study. In addition to this, they were also asked to provide their permission to use their chat transcripts for post - conversation analyses. 97 BIBLIOGRAPHY 98 BIBLIOGRAPHY Altman, I., & Taylor, D. A. (1973). Social penetration: The development of interpersonal relationships . New York, NY: Holt, Rinehart & Winston. Andersen, P., & Wang, H. (2005). Self - disclosure in long - distance friendships: A comparison between face - to - face and computer - mediated communication. Paper presented at the conference of the Internati onal Communication Association, New York, NY. Antheunis, M. L., Peter, J., Schouten, A. P., & Valkenburg, P. M. (2009). Intervening processes between computer - mediated communication and interpersonal attraction: An experimental comparison . In Annual meeting of the International Communication Association, Chicago, Illinois. Antheunis, M. L., Valkenburg, P. M., & Peter, J. (2010). Getting acquainted through social network sites: Testing a model of uncertainty reduction and social attraction. Computers in Human Behavior , 26 , 100 - 109. Bell, R. A., & Daly, J. A. (1984). The affinity seeking function of communication. Communications Monographs , 51 , 91 - 115. Berger, C. R. (1979). Beyond initial interaction: Uncertainty, understanding, and the development of interpersonal relationships. In H. Giles (Ed.), Language and Social Psychology (pp. 122 144). London, UK: Basil Blackwell. Berger, C. R., & Bradac, J. J . (1982). Language and social knowledge: Uncertainty in interpersonal relationships . London: Edward Arnold. Berger, C. R., & Calabrese, R. J. (1975). Some explorations in initial interaction and beyond: Toward a developmental theory of interpersonal commu nication. Human Communication Research , 1 , 99 - 112. Berger, C. R., Gardner, R. R., Clatterbuck, G. W., & Schulman, L. S. (1976). Perceptions of information sequencing in relationship development. Human Communication Research , 3 , 29 - 46. Berger, C. R., & Ke llermann, K. A. (1983). To ask or not to ask: Is that a question? Communication yearbook , 7 , 342 - 368. Berscheid, E., & Reis, H. T. (1998). Attraction and close relationships. The handbook of social psychology , 2 , 193 - 281. boyd, d. m., & Ellison, N. B. (2007). Social network sites: Definition, history, and scholarship. Journal of Computer - Mediated Communication , 13 , 210 230. 99 Byrne, D. (1971). The attraction paradigm . New York: Academic Press. Byrne, D., Ervin, C. R., & Lamberth, J. (197 0). Continuity between the experimental study of attraction and real - life computer dating. Journal of Personality and Social psychology , 16 , 157. between atti tude similarity and attraction. Communications Monographs , 59 , 180 - 189. Clark, H. H. (1996). Using language . Cambridge: Cambridge university press. Clatterbuck, G. W. (1979). Attributional confidence and uncertainty in initial interaction. Human Communication Research , 5 , 147 - 157. Cody, M. J., Pratt, L., Wendt, P. F., & Wiseman, R. L. (1999). Interrogative strategies and information exchange in computer - mediated communication. Communication Quarterly , 47 , 46 - 66. Cohen, J. (1988). Statistical powe r analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Coleman, L. H., Paternite, C. E., & Sherman, R. C. (1999). A re - examination of deindividuation in synchronous computer - mediated communication. Computers in Human Behavior , 15 , 51 - 65. Condon, J. W., & Crano, W. D. (1988). Inferred evaluation and the relation between attitude similarity and interpersonal attraction. Journal of Personality and Social Psychology , 54 , 789 - 797. Cozby, P. C. (1973). Self - disclosure: A literature review. Psychological bulletin , 79 , 73 - 91. Dindia, K. (1983). Reciprocity of self - disclosure: A sequential analysis. In M. Burgoon (Ed.), Communication yearbook 6 (pp. 506 528). Beverly Hills, CA: Sage Dindia, K., Fitzpatrick, M. A., & Kenny, D. A. (1997 ). Self Disclosure in spouse and stranger interaction a social relations analysis. Human Communication Research , 23 , 388 - 412. Dipboye, R. L., Fontenelle, G. A., & Garner, K. (1984). Effects of previewing the application on interview process and outcomes. Journal of Applied Psychology , 69 , 118 - 128. Dillard, J. P., Segrin, C., & Harden, J. M. (1989). Primary and secondary goals in the production of interpersonal influence messages. Communications Monographs , 56 , 19 - 38. www.usatoday.com/tech/webguide/internetlife/2007 03 - 07 - facebook - stalking_N.htm 100 Dollard, J., & Mowrer, O. (1947). A method of measuring tension in written documents. The Journal of Abno rmal and Social Psychology , 42 , 323 - 332. Douglas, W. (1990). Uncertainty, information seeking, and liking during initial interaction. Western Journal of Speech Communication , 54 , 66 81. Douglas, W. (1991). Expectations about initial interaction: An examination of the effects of global uncertainty. Human Communication Research , 17 , 355 384. Douglas, W. (1994). The acquaintanceship process: An examination of uncertainty, information seeki ng, and social attraction during initial conversation. Communication Research , 21 , 154 - 176. Jo urnal of Computer Mediated Communication , 12 , 1143 - 1168. Facebook Mission Statement. (2004, February 4). Retrieved from https://www.facebook.com/facebook/info?tab=page_info Fiore, A. T. & Donath, J. S. (2005). Homophily in online dating: When do you like someone like yourself? Computer - Human Interaction 2005, 1371 - 1374. Gibbs, J. L., Ellison, N. B., & Lai, C. H. (2011). First comes love, then comes Google: An investigation of uncertai nty reduction strategies and self - disclosure in online dating. Communication Research , 38 , 70 - 100. Goldenson, R. M. (1984). Longman dictionary of psychology and psychiatry . New York, NY: Longman. Hancock, J. T., Toma, C. L., & Fenner, K. (2008). I know something you don't: The use of asymmetric personal information for interpersonal advantage. In Proceedings of the 2008 ACM conference on computer supported cooperative work (pp. 413 - 416). San Diego, CA. Harvey, J. H., Wells, G. L., & Alvarez. M. D. (19 78). Attribution in the context of conflict and separation in close relationships. In J. H. Harvey, W. Ickes. & R. F. Kidd (Eds.), New directions in attribution research . Hillsdale, N.J.: Erlbaum. High, A. C., & Caplan, S. E. (2009). Social anxiety and co mputer - mediated communication during initial interactions: Implications for the hyperpersonal perspective. Computers in Human Behavior , 25 , 475 - 482. Hoyle, R. H. (1993). Interpersonal attraction in the absence of explicit attitudinal information. Social Cognition , 11, 309 - 320. 101 Jablin, F. M., & Miller, V. D. (1980). Interviewer and applicant questioning behavior in employment interviews. Management Communication Quarterly , 4 , 51 - 86. Jourard, S. M., & Lasakow, P. (1958). Some factors in self - disclosure. The Journal of Abnormal and Social Psychology , 56 , 91 - 98. Kaplan, A. M., & Haenlein, M. (2010). Users of the world, unite! The challenges and opportunities of Social Media. Business horizons , 53 , 59 - 68. Kellerman, K. A., & Berger C. R. (1984). Affect and the acquisition of social information: Sit back, relax, and tell me about yourself. In R. N. Bostrom (Ed.), Communication Yearbook 8 (pp. 412 - 45). Newbury Park, CA: Sage. Kellermann, K., & Lim, T. (1989). Conversational ac quaintance: The flexibility of routinized Rethinking Communication (pp. 172 - 187). Newbury Park, CA: Sage Publications, Inc. Kellerman, K.A., & Reynolds, R. S. (1990). When ignoran ce is bliss: The role of motivation to reduce uncertainty in uncertainty reduction theory. Human Communication Research , 17 , 5 - 75. Knapp, M. L. (1978). Social intercourse: From greeting to goodbye. Boston: Allyn and Bacon. Korolija, N., & Linell, P. (1996). Episodes: coding and analyzing coherence in multiparty conversation. Linguistics , 34 , 799 - 832. Krasnova, H., Spiekermann, S., Koroleva, K., & Hildebrand, T. (2010). Online social networks: why we disclose. Journal of Information Technology , 25 , 109 - 125. Maynard, D. W., & Zimmerman, D. H. (1984). Topical talk, ritual and the social organization of relationships. Social Psychology Quarterly , 47, 301 - 316. McCroskey, J. C., & McCain, T. A. (1974). The measurement of interpersonal at traction. Speech Monographs , 41, 261 - 266. McCroskey, L. L., McCroskey, J. C., & Richmond, V. P. (2006). Analysis and improvement of the measurement of interpersonal attraction and homophily. Communication Quarterly , 54 , 1 - 31. McCroskey, J. C., Richmond, V. P., & Daly, J. A. (1975). The development of a measure of perceived homophily in interpersonal communication. Human Communication Research , 1 , 323 - 332. McKenna, K. Y., Green, A. S., & Gleason, M. E. (2002). Relationship formation on the Internet: Journal of Social Issues , 58 , 9 - 31. McLaughlin, M. L., Cody, M. J., & Rosenstein, N. E. (1983). Account sequences in conversations between strangers. Communication Monographs , 50 , 102 125. 102 Miller, V. D., & Jablin, F. M. (1991). Information seeking during organizational entry: Influences, tactics, and a model of the process. Academy of Management Review , 16 , 92 - 120. Mongeau, P. A., Jacobsen, J. L., & Donnerstein, C. (2007). Generalizing from undergraduates to single adults: Defining dates and first date goals. Communication Research , 34 , 526 - 547. Montoya, R. M., Horton, R. S., & Kirchner, J. (2008). Is actual similarity necessary for attraction? A meta - analysis of actual and perceived similarity. Journal of Social and Personal Relationships , 25 , 889 - 922. Morris, W. (Ed.). (1976). The American heritage dictionary of the English l anguage. Boston: Houghton Mifflin. Paul, A., & Morrison, K. (2014). Examining the unique and combined effects of passive and interactive information seeking strategies on uncertainty reduction in social networking sites . Unpublished manuscript, Department of Communication, Michigan Stat e University, East Lansing, Michigan, USA. Planalp, S. (1993). Friends' and acquaintances' conversations II: Coded differences. Journal of Social and Personal Relationships , 10 , 339 - 354. Planalp, S., & Benson, A. (1992). Friends' and acquaintances' conversations I: Perceived differences. Journal of Social and Personal Relationships , 9 , 483 - 506. Planalp, S., & Garvin - Doxas, K. (1994). Using mutual knowledge in conversation: Friends as experts on each other. Dynamics of rel ationships , 4 , 1 - 26. Pratt, L., Wiseman, R. L., Cody, M. J., & Wendt, P. F. (1999). Interrogative strategies and information exchange in computer mediated communication. Communication Quarterly , 47 , 46 - 66. Prisbell, M., & Andersen, J. F. (1980). The importance of perceived homophily, level of uncertainty, feeling good, safety, and self disclosure in interpersonal relationships. Communication Quarterly , 28 , 22 - 33. Rook, D. W. (1987). The buying impulse. Journal of Consumer Research , 14, 189 - 199. Ross, L., Greene, D., & House, P. (1977). The false consensus effect: An egocentric bias in social perception and attribution processes . Journal of Experimental Social Psychology , 13 , 279 301. Rubin, R. B. (1979). The effect of context on information seek ing across the span of initial interactions. Communication Quarterly , 27 , 13 - 20. 103 Schouten, A. P., Valkenburg, P. M., & Peter, J. (2007). Precursors and underlying processes of adolescents' online self - - attrib ute - Media Psychology , 10 , 292 - 315. Sermat, V., & Smyth, M. (1973). Content analysis of verbal communication in the development of relationship: Conditions influencing self - disclosure. Journal of Personality and Social Psychology, 26 , 332 - 342. Sillars, A. L. (1985). Interpersonal perception in relationships. In W. Ickes (Ed.), Compatible and incompatible relationships (pp. 277 305). New York: Springer. Stutzman, F., Gross, R., & Acquisti, A. (2012). Silent listeners: The evolution of privacy and disclosure on Facebook. Journal of Privacy and Confidentiality , 4 , 7 - 41. Sunnafrank, M. J., & Miller, G. R. (1981). The role of initial conversations in determining attraction to similar and dissimilar strangers. Human Communication Research , 8 , 16 - 25. Sunnafrank, M., & Ramirez, A. (2004). At first sight: Persistent relational effects of get - acquainted conversations. Journal of Social and Personal Relationships , 21 , 361 - 379. Svennevig, J. (2000). Getting Acquainted in Conversation . Philadelphia: John Benjamin Publishing Company. Tamborini, R., & Westerman, D. (2008). Uncertainty reduction over time in initial stranger interactions: A social information processing theory approach . Paper presented at the conference of the International Communication Association, Montreal, Canada. Taylor, D. A., & Altman, I. (1966). Intimacy - scaled stimuli for use in studies of interpersonal relations. Psychological Reports , 19 , 729 730. Tengler, C. D., & Jablin, F. M. (1983). Effects of question type, orientation, and sequencing in the employment screening interview. Communications Monographs , 50 , 245 - 263. Tidwell, L. C. (1997). Comparison of anticipated interaction, channel, and communication goal onimpression formation, self - disclosure, and relational orientation: getting to know one another a bit at a time (Doctoral dissertation). Retrieved from ProQuest Dissertations and Theses. Tidwell, L. C., & Walther, J. B. (2002). Computer - mediated communi cation effects on disclosure, impressions, and interpersonal evaluations: Getting to know one another a bit at a time. Human Communication Research , 28 , 317 - 348. Tidwell, N. D., Eastwick, P. W., & Finkel, E. J. (2013). Perceived, not actual, similarity p redicts initial attraction in a live romantic context: Evidence from the speed dating paradigm. Personal Relationships , 20 , 199 - 215. 104 Wheeless, L. R. (1978). A follow - up study of the relationships among trust, disclosure, and interpersonal solidarity. Human Communication Research , 4 , 143 - 157. Wheeless, L. R., & Grotz, J. (1976). Conceptualization and measurement of reported self - disclosure. Human Communication Research , 2 , 338 - 346. Won - Doornink, M. J. (1979). On getting to know you: The association betwee n the stage of a relationship and reciprocity of self - disclosure. Journal of Experimental Social Psychology, 15, 229 - 241.