Mood or Closeness 1 Abstract This study examined the effects a person’s prior mood state (happy/sad) has on his or her ability to perceive deception in friends or strangers (N=208). This study makes predictions based on the Cognitive Functional Model (Nabi, 1999) and Interpersonal Deception Theory (Buller & Burgoon, 1996). Prior mood was manipulated via television viewing. Results indicated that interviewers in the happy mood were significantly less likely to detect deception compared to those in the sad mood. Results showed no differences in the ability to detect deception based on relational closeness. The discussion highlights the importance of mood context in deception detection and on message processing. Implications and plans for future study are discussed. Emotions or Closeness 2 Keywords: Interpersonal Deception, Affect, Mood, ELM, HSM, Negative State Relief, and Cognitive Functional Model. Mood State or Relational Closeness: Explaining the Impact of Mood on the Ability to Detect Deception in Friends and Strangers In the 20th century deception has played an important role in politics, media, business, and relationships (Miller & Stiff, 1993). Likewise, deception and deception detection has been studied extensively in the late 1960’s and beyond. Zuckerman, Depaulo, and Rosenthal (1981) provide a comprehensive overview of literature related to verbal and nonverbal communication deception. Zuckerman et al. (1980) defines deception based on the notion that deception is the communication of specific (verbal and nonverbal) information. The wealth of deception research reviewed by Zuckerman et al. (1981) focused on the typology of specific verbal and nonverbal cues and was largely atheoretical. Other researchers have spent considerable time and effort exploring the physiological cues related to deception and how deceiving impacts the deceiver (Waid & Orne, 1981). McCornack (1992) argues that early applications of deception research operated from a “primitive” perspective propagating the “myth” that deception is characterized by specific verbal and nonverbal attributes. In recent years, Buller and Burgoon’s (1996) Interpersonal Deception Theory (IDT) offered the contention that language, contextual, and relational elements rather than purely verbal and nonverbal cues explain deception. Even more recently, Bond and Depaulo (2008) conducted a meta-analysis of 247 deception studies and found that Mood or Closeness 3 contextual moderators and individual difference variables only accounted for minute differences in peoples ability to detect deception and that peoples ability to detect deception is only slightly better than chance. In the real world, deceptive contexts are often filled with affect. For example, a person might be angry and try to catch a relational partner in a lie regarding his or her whereabouts one evening. A person might be happy and have no clue as their relational partner lies about the quality of their relationship and discover a deception after weeks or months (Park, Levine, McCornack, Morrison, & Ferrara, 2002). In recent years, scholars have asked for more research attempting to understand the effects moods have on persuasion (e.g., Dillard, 1993). Nabi (1999) contends that little effort has been made to understand the role of mood in persuasion. As Miller and Stiff (1993) note, deception is a persuasive act. With these concepts in mind, we offer the current study. This study examines the role that relational closeness plays on the ability to detect deception. More importantly, this study examines the role the mood state of the receiver (of a deceptive message) plays on a receiver’s ability to detect deception. Prior deception research has tended to focus on the affect state of the deceiver (not the deceived), ignoring the transactional nature of deceptive interactions (Buller & Burgoon, 1996). We believe that there is a significant gap in the literature on mood and deception detection. Of the 247 studies examined by Bond and Depaulo (2008), none focused on mood as potential independent variable account for ability to detect deception. In order to better explicate our predictions and rationale, this study reviews the findings related to IDT and discusses the literature on traditional message processing models (e.g., Elaboration Likelihood Model or ELM & Heuristic-Systematic Model or HSM), Nabi’s (1999) Cognitive Functional Model (CFM), and the Negative State Relief Model (Cialdini, Darby, & Vincent, 1973). This investigation starts with a review of the IDT literature. IDT Emotions or Closeness 4 Considerable debate was started when Buller and Burgoon (1996) offered IDT as an alternative theory to explain deception. Buller and Burgoon (1996b & 1996c) argued that McCornack’s (1992) Information Manipulation Theory (IMT) was not a theoretical explanation of deception at all and at best was a taxonomy because IMT failed to “…explain how, why, and under what circumstances deception works” (p.93). To better meet the standard of a theory, Burgoon and Buller (1996) offer a detailed explanation of IDT’s. Buller and Burgoon (1996) note that IDT is grounded in interpersonal communication. Based on this assumption, they explain the deception is complex and multifaceted in nature. Moreover, they assume that the interpersonal process of deception is interactive and requires both cognitive effort and physical effort that extends beyond the effort levels of normal conversation. They posit that deception is grounded in social norms and expectations; particularly the normative presumption that others are being truthful. Lastly, they offer that there are affect reactions of arousal and negative affect associated with deception. This last assumption is of paramount importance to the current investigation Tying these assumptions together, Buller and Burgoon (1996) offer the 18 propositions to explain IDT. Only a few of these propositions are relevant to the current investigation: (1) What people in a deception context think and behave like varies based on how interpersonal the relationship is; (2) People who are more motivated display more leakage and a respondent’s initial reactions depend on importance of relationship and initial suspicion; (3) Relational familiarity causes deceivers to be more fearful and have more nonverbal leakage; (4) The deceiver’s credibility is positively correlated to interactivity, truth bias, and their communication skills, but is reduced when the deceiver communicates in an unexpected manner. Numerous investigations have found varying degrees of support for IDT (Buller, Burgoon, Buslig, & Roiger, 1996; Burgoon & Floyd, 2000; Burgoon, Buller, Dillman, & Walther, 1995; Burgoon, Buller, Ebesu, White, & Rockwell, 1996; Burgoon, Buller, & Floyd, 2001; Burgoon, Buller, Guerrero, Afifi, & Feldman, 1996; Dunbar, Ramirez, & Burgoon, 2003; White & Burgoon, 2001). Mood or Closeness 5 Based on the literature reviewed, it is our contention that deception is much more than the sum of various verbal and nonverbal components. Buller and Burgoon (1996) illuminated the notions of deception being relational and contextual. These notions have demonstrated great heuristic value and heated academic discussion. Although the underlying processes remain unclear, interactivity, expectations (both verbal and nonverbal), interpersonal communication competence, and relationship type have been shown to have strong impacts on both the perceptions of the deceiver and the deceived. Affect and IDT Affect play an important role in propositions of IDT. In particular, two propositions directly address the role of fear. Specifically, those propositions posit that the fear level of the deceiver will be higher the closer the relationship and inhibit the deceivers’ ability to deceive effectively due to increased nonverbal leakage by the deceiver (caused by the affect state). IDT’s propositions fail to address the potential emotional states of the deceived in the deception process. The remaining review of literature address the impact that the affect state of the receiver has on their ability to be persuaded and process persuasive messages (deceptive messages). Review of Past Research on Affect and Message Processing Various studies have determined that positive moods are linked with peripheral or heuristic processing, whereas negative moods are linked with systematic or cognitive processing (see Bohner & Schwarz, 1993; Isen, 1987; Schwarz, 1990; and Schwarz, Bless, & Bohner, 1991 for reviews). The majority of these studies employed the dual-process models of persuasion, such as Petty and Cacioppo’s (1986) elaboration likelihood model (ELM) and Chaiken, Liberman, and Eagly’s (1989) heuristicsystematic model (HSM). A review of how these models have been linked to affect is warranted. Affect and dual-process models of persuasion. The basic assumption shared by both models is that message acceptance is dependent on a person’s mental reaction to the message. Chaiken et al. (1989) express the existence of two modes of message processing in the HSM: systematic and heuristic. Petty Emotions or Closeness 6 and Cacioppo (1986) propose two similar modes of message processing in their ELM: central and peripheral routes. Systematic or central route processing involves careful, analytic, and effortful evaluation of the message (Chaiken et al., 1989; Petty & Cacioppo, 1986). Heuristic or peripheral route processing involves the use of a simple rule by an individual to determine his or her attitude toward the persuasive message (Chaiken et al., 1989; Petty & Cacioppo, 1986). The main difference between the two models is that in the ELM, individuals are said to engage in either central or peripheral route processing whereas in the HSM, both cognitive and heuristic processes may occur concurrently. According to both models, affect is argued to differentially influence how people interpret and respond to a message following biased systematic processing or heuristic processing of that message. When biased systematic processing happens, affect is said to influence the valence of cognitions occurring in response to a persuasive message. For example, Petty, Gleicher, and Baker (1991) found message acceptance to be a direct function of proportion of positive thoughts reported by individuals (i.e., those induced with a positive mood). Furthermore, affect may also function as information for individuals and serve as a heuristic cue (Schwarz & Clore, 1988). Schwarz and Clore (1988) contend, a person who feels good will readily accept persuasive messages because positive affect is an indication that everything is alright. Conversely, negative affect signals that there is a problem and is likely to motivate individuals to scrutinize the message more before accepting it. A brief overview of discrete emotions will be presented, followed by a review of past studies on discrete emotions and persuasion. Discrete emotions. Discrete emotions possess three related but distinct features: (1) signal value, (2) function, and (3) action tendency (Dillard et al., 1996). The signal value of a discrete emotion describes the state of the environment. For example, happiness is typically associated with goal achievement or progress whereas sadness is usually associated with loss (Dillard et al., 1996). Discrete emotions can also be looked at in terms of their functions for individuals. According to Dillard et al. (1996), the function of happiness is to help people conserve their resources while the function of sadness Mood or Closeness 7 is to help people adjust to loss. Finally, every discrete emotion has its own distinctive action tendency designed to arouse, maintain, and direct both cognitive activity and overt behaviors (Lazarus, 1991). For happiness, the action tendency is for people to maintain their positive feelings by focusing inward, and engage in heuristic processing of persuasive messages (Mitchell, Brown, Morris-Villagran, & Villagran, 2001). According to Cialdini, Darby, and Vincent’s (1973) negative state relief model (NSRM), happy people are not motivated to critically evaluate persuasive content because their focus is on other stimuli that allow them to maintain their positive emotional state. Thus, it is likely that happy people will engage in biased processing of messages, only attending to those that are positive (i.e., selective exposure) or processing only positive aspects of messages, and ignoring the negative aspects (i.e., selective attention). For sadness, the action tendency is for people to focus outward to minimize their negative feelings, and engage in more systematic processing of persuasive messages (Mitchell et al., 2001). Cialdini et al.’s (1973) NSRM explains that sad people want to relieve their bad feelings. One way to accomplish this is to divert themselves from negative thoughts by engaging in other “distracting” activities (Cialdini et al., 1973). Thus, it is likely that sad people are expected to “scrutinize a persuasive message and elaborate its message content” (Mitchell et al., 2001, p. 349). Taken together, it is argued that the discrete emotions of happiness and sadness help people determine the information-processing style to adopt in evaluating a persuasive message. Discrete emotions and message processing. According to Nabi (1999), discrete emotions, with the exception of fear, have virtually been ignored in persuasion research. Specifically, there has been a dearth of research examining the effects of discrete emotions on different aspects of the persuasion process (e.g., message processing, attitude change). One exception is Dillard et al.’s (1996) investigation of the effects of various discrete emotions on message acceptance of several AIDS PSAs. Dillard et al. (1996) found that the discrete emotions of fear, surprise, and sadness positively predicted message Emotions or Closeness 8 acceptance, whereas puzzlement negatively predicted message acceptance of the PSAs. Although discrete emotions were examined in that study, Dillard et al. (1996) only focused on how discrete emotions induced by the PSAs affected processing of those same messages (i.e., the PSAs). Lacking is an understanding of how discrete emotions as the dominant affect influence the information-processing style individuals adopt in approaching a subsequent message. In light of this, the present study specifically focuses on investigating the effects of two discrete emotions (i.e., happiness and sadness) as determinants of people’s information-processing style in approaching a potential deception context. Although there are a wide range of discrete positive (e.g., joy, happiness) and negative emotions (e.g., fear, anger, sadness) to choose from, the discrete emotions of happiness and sadness were selected because they are likely to be the most prevalent positive and negative emotional states experienced by individuals respectively. Izard (1977) noted that sadness is the most commonly experienced negative emotion. Although no such claim was made regarding happiness, it is speculated that feelings of happiness as a positive emotion are likely to be pervasive as well. The next section describes Nabi’s (1999) CFM which explains how discrete emotions direct people to either heuristically or cognitively process messages. Cognitive-Functional Model According to Nabi (1999), a person experiencing a discrete emotion will be motivated to engage simultaneously in two processes: (1) motivated attention and (2) motivated processing. With regards to motivated attention, certain emotions cause us to avoid or approach the emotion-inducing stimulus. As Nabi (1999) explains, sadness is an approach-based emotion. Happiness is also considered an approachbased emotion (Dillard & Peck, 2001). An approach-based emotion is one which leads a person to attend to the emotion’s source or the emotion-inducing situation. Specifically, Nabi (1999) proposes that an approach-based emotion is likely to increase an individual’s motivation to attend to subsequent messages. Mood or Closeness 9 Although both happiness and sadness are expected to motivate individuals to attend to subsequent messages, the style of processing (e.g., cognitive/content-based, heuristic/sensory-based) may differ depending on the discrete emotion experienced. Specifically, Nabi (1999) asserts that with approachbased emotions, cognitive processing is expected provided the individual expects that such processing will help him or her satisfy the emotion-induced goal. In the case of sadness, the emotion-induced goal is to alleviate the negative feelings. The NSRM suggests that one way to do this is to engage in a distracting activity to divert attention away from negative thoughts. And so, a message that effectively serves this distraction function for sad individuals should be cognitively processed. In the present study, the potential deception context may serve as an effective distraction for sad individuals. Processing of the deceptive messages is likely to facilitate achievement of the emotion-induced goal for sadness. On the other hand, the emotion-induced goal of happiness is to maintain the positive emotional state. As Nabi (1999) explains, for those individuals who do not expect message processing to help satisfy the emotion-induced goal, heuristic processing will occur. In the present study, processing of the deceptive message is likely to inhibit achievement of the emotion-induced goal for happiness. The reason being that finding out someone is lying to you may threaten a person’s positive emotional state (i.e., they may get angry or disappointed). Thus, it was expected that interviewers induced to feel sad would be more motivated to attend to the interviewees’ responses (as a way of relieving negative affect), and be more aware of potential leakage cues, resulting in greater likelihood of detecting deception. Conversely, interviewers induced to feel happy would be less motivated to consider the interviewees’ responses as deceptive (given their goal of maintaining happiness), and as a result, should report less deception on the part of the interviewee. Taken together, we offer the following hypothesis: H1: Deceived participants in the sad emotional condition would be more likely to perceive deception than deceived participants in the happy emotional condition. Emotions or Closeness 10 Additionally, according to IDT when people are more relationally close and expectancy violations occur, those violations are typically associated with deception. Specifically, IDT argues that the closer individuals are to each other, when deception occurs, the more likely the deceived will report greater expectancy violations by the deceiver (who’ll engage in more unexpected behaviors given their increased arousal due to relational closeness). Such expectancy violations in turn, are likely to be construed by the deceived as signs (i.e., leakage cues) that the other person is lying. Thus we propose the following two predictions: H2: As relational closeness increases, the level of expectancy violations perceived by the deceived participant also increases. H3: As the level of expectancy violations perceived by the deceived participant increases, the level of perceived deception also increases. In considering the complex roles that the emotional state of the deceived, relational closeness, and expectancy violations play in deception detection, it raises two interesting research questions: R1: Does relational closeness directly predict perceived deception? R2: When controlling for relational closeness and expectancy violations, does mood state still account for a significant amount of variance in perceived deception? At the heart of both of these questions is whether relational closeness or emotional state serves as the better predictor of perceived deception. To address these hypotheses and research questions we offer the following methods. Methods Participants A total of 208 students from basic communication courses at a large southeastern university participated in the study. The ethnic breakdown of the sample was primarily European-America/white Mood or Closeness 11 (88.9%). The average age of the participants was 20, (M = 19.77, SD= 1.79) and the participants were principally female (63.5%). Students were able to use their participation in the study to fulfill the research requirement assigned to those in the department’s speech communication courses. Extra credit for participation in the study was also offered to participants at their instructor’s discretion. Design A single-factor experimental design exploring the effects of emotional state (i.e., happy/sad) on deception detection was utilized. Participants were randomly assigned to one of two conditions: (a) sad or (b) happy. The relationship between relational closeness and deception detection was also examined as part of the overall study. Procedures Half the participants in the study were randomly asked to bring a close personal friend with them and the other half were asked to come alone. This step was conducted to facilitate the friend versus stranger (relational closeness) independent variable. Participants that came alone were randomly paired with another solo participant that was attending the study at the same research time. While participants that were asked to bring a friend were paired with their friend. The research sign-ups for participants informed them that the study examined interpersonal interview activities. The two participants were taken to two different rooms and given informed consent forms to sign. Once the consent forms were completed, the researchers commenced with the study. In one room, a participant was told s/he would be viewing 5 minutes of television programming and at the conclusion of the programming would conduct an interview of another person (the friend or a stranger), and at the conclusion of the interview s/he would respond to a questionnaire about the interview. After describing the study, the researcher started to play the television programming (happy-Friends or sad-Law & Order). The television programming was viewed on a 15-inch color video monitor from 6 feet away and the volume was set at the same level Emotions or Closeness 12 throughout the data collection. A prior study (Wong & Householder, 2004), found that 5 minutes of television viewing caused significant differences in self-reported emotional state. Dillard et al. (1996) found that watching a 30 second or 1-minute public service announcement significantly altered the affect state of participants. Participants in this study were randomly assigned to either the happy or sad affect manipulation. Participants in the other room (upon completion of an informed consent form), were informed by the researcher that the study was not an interpersonal interview study but an interpersonal deception study. These participants were asked to be confederates in the study. As confederates, participants were told that the study took the form of an interview, the person in the next room would ask them a set of 10 questions, and that for the first five (or last five) questions of the interview all your answers should be deceptive and not true. A few of the interview questions were, “If you could own any car, what would it be?” and “What are some of the qualities of your ideal relational partner?” Additionally, the confederates were informed that the interviewer could ask follow-up questions and that all follow-up responses to the first five (or last five) questions should also be dishonest. For the last five (or first five) questions in the interview confederates were instructed to be completely honest. Participants were randomly assigned to either lie on the first five questions in the interview or the last five questions in the interview (no order effects were found). The confederates were also told that there were consequences for being caught. The consequence for being caught required the confederate to fill out a much longer and boring questionnaire. This step was just a bluff to encourage participants to lie in a way as not to get caught. No confederate actually received a longer questionnaire. Finally, the confederates were instructed to not discuss the interview with the interviewer and return to the confederate room for filling out the long or short questionnaire. Once the television programming concluded, the participants were brought back together. The interviewer was given a list of 10 interview questions and instructed to state the number of the question Mood or Closeness 13 prior to asking each one (as a way of helping interviewees keep track). Also, they were told to feel free to ask probing or follow-up questions. Additionally, both were informed not to talk about the interview after its culmination and that the interviewee should leave the room after responding to the final question. Both participants were told at this juncture that the study examined their impressions of their interaction during the interview. Upon the completion of the interview, both participants were separated and the interviewer was given a questionnaire to complete while the confederate waited in the other room. The confederate also filled out a short follow-up questionnaire. When both participants completed their questionnaires, they were brought back together for a final time and were fully debriefed about the study. Researchers assured the interviewers that the deception was part of the study and not an indication of the interviewee’s personality or character. Materials Media. This study used two media artifacts: an excerpt of the hit primetime comedy Friends, and an excerpt from the hit primetime drama Law & Order. The episode of Friends depicted cast members playing a personalized version of Jeopardy. The Friends excerpt had a total running time of five-minutes. The Law & Order excerpt depicted cast members investigating a murder and interviewing a sad young boy who had witnessed the brutal homicide. This excerpt also had a five-minute running time. Measures The following measures were gathered in identical ways for both affect conditions. Affect items. To measure the interviewer’s affective state (i.e. level of happiness, etc.) after watching the television program segment, they were asked to complete a 10-item measure, rating the following statements for level of agreement from 1-strongly disagree to 7-strongly agree (e.g., I liked the program, I felt sad after watching the program, I felt happy after watching the program). Several of the items were reverse-coded so that higher scores indicate greater feelings of positive affective state. Overall, this measure proved to be highly reliable (=.95). Emotions or Closeness 14 Honesty items. To measure the interviewer’s perceptions regarding the honesty of the other person’s responses during the interpersonal interview, a 4-item Likert-type scale was adopted from Levine et al. (2000), and asked interviewers to rate their interviewees’ responses from 1-completely deceptive to 7-completely truthful. Example items included, “During this interview, the person being interviewed was being,” and “The answers that the other person provided were.” Higher scores indicate greater attributions of honesty made about the interviewee. Overall, this measure proved to be highly reliable (=.97). Expectancy items. To measure the interviewer’s perceptions regarding the expectancy of the interviewee’s behaviors during the interview, four 7-point semantic differential-type items were adopted from Levine et al. (2000) (e.g., the behaviors exhibited by the other person during the interview were: (1) unexpected—expected, (2) not anticipated—anticipated, (3) very surprising—not very surprising, and (4) not what I predicted—what I predicted). Higher scores indicate lower expectancy violations on the interviewer’s part. Overall, this measure proved to be highly reliable (=.94). Relational closeness items. To measure both the interviewer’s and interviewee’s perceptions regarding the level of relational closeness they felt towards one another, they were asked to respond to four Likert-type items (e.g., how close is this relationship, how well do you and the interviewer know each other, how well do you and the interviewer understand each other’s feelings), from 1-not at all to 7very well. For the question on relationship commitment, the anchor used was from 1-not at all to 7-very committed. Overall, this measure proved to be highly reliable (=.98). Manipulation check items. To ensure that the interviewee’s responses were not completely truthful (i.e., half of them were true, half were lies), four items were created to check the deception manipulation. Two items asked interviewee’s about the truthfulness of responses prior to question #6 on the 10-item interview (e.g., during this interview, prior to question #6, I was being ____ in my responses) from 1-completely deceptive and 7-completely truthful. These two items yielded a high reliability (= .98). Two items also asked interviewees about the truthfulness of responses after question #5 on the 10item interview (e.g., during this interview, following question #5, I was being ____ in my responses), from 1-completely deceptive and 7-completely truthful). And these two items also yielded a high Emotions or Closeness 15 reliability (=.99). Additionally, one item asked the interviewee to assess how successful they felt at deceiving the interviewer (e.g., I felt I was very successful at deceiving the other person during the interview), with 1-strongly disagree and 7-strongly agree. Results Affect Manipulation Check The effectiveness of the affective induction was tested with a one-way ANOVA with the two affect inducement conditions (i.e., sad and happy) as the grouping variable, and the positive affective measure as the dependent variable. Interviewers who were initially primed to experience positive affect (i.e., watch 5 minutes of Friends) reported more positive affect (M = 6.32, s.d. = .48) than those initially primed to experience negative affect (i.e., watch 5 minutes of Law and Order) (M = 3.80, s.d.=.88). A Bonferroni adjusted contrast was conducted between the two conditions yielding a significant difference, F(1, 102) = 329.73, p<.001, η2 =.764. The affective conditions accounted for 76.4% of the variance for the positive affective measure. Thus, the affective manipulation was highly effective. Deception Manipulation Check The effectiveness of the deception manipulation for the interviewees was assessed by examining the means of their self-reported honesty ratings based on the order in which they were to lie (i.e., be deceptive on the first 5 questions or be deceptive on the last 5 questions). For those who were instructed to lie on the first 5 questions, the mean honesty rating for questions 1-5 (M = 4.33, SD = .44) were lower than the mean honesty rating for questions 6-10 (M = 4.72, SD = 1.10). As for those who were instructed Emotions or Closeness 16 to lie on the last 5 questions, the mean honesty rating for questions 1-5 (M = 4.38, SD = .55) were higher than the mean honesty rating for questions 6-10 (M = 4.18, SD = 1.89). Together, the data suggested that the interviewees did a fairly good job of following instructions and that the deception manipulation worked. Moreover, on average, the interviewees felt they were somewhat successful at deceiving the other person (M = 4.81, SD = 1.64). For your convenience, the means and standard deviations associated with the two manipulation checks are presented in Table 1. Order Effects In this study, because interviewees were instructed to either lie during the first half of the interview, or the second half of the interview, it was important to rule out possible order effects for explanation of findings. Two one-way ANOVAs were performed on the data set with honesty ratings of interviewee (for questions 1-5 and questions 6-10) as the dependent variable, and order of deception (i.e., first 5 deceptive, second 5 deceptive) as the grouping factor. Regarding honesty ratings for questions 1-5, no significant order effect was found, F(1,102) = .28, p = .60. Similarly, regarding honesty ratings for questions 6-10, no significant order effect was found, F(1, 102) = 2.89, p = .09. Taken together, the results suggest that there were no order effects for the study. Main Analyses In order to test the hypotheses, and answer the research questions posed in our study, a number of analyses were performed on the data set. First, a one-way ANOVA was performed with interviewers’ honesty ratings of the interviewee as the dependent measure, and affective condition as the grouping factor (H1). Moreover, two linear regressions were also performed, regressing interviewers’ expectancy ratings on relational closeness (H2), and interviewers’ honesty ratings on their expectancy ratings (H3). Emotions or Closeness 17 Also, a linear regression was performed regressing honesty ratings on relational closeness (R1). Finally, an ANCOVA was performed with honesty ratings as the dependent measure, affective condition as the grouping factor, and expectancy ratings and relational closeness as covariates (R2). Effects of Affect on Interviewers’ Ratings of Honesty for Interviewees Hypothesis one posited that deceived participants in the sad mood condition are more likely to perceive deception than deceived participants in the happy mood condition. Results indicate that participants in the sad affect condition (M = 4.64, s.d. =2.02), compared to those in the happy affect condition (M = 6.04, s.d. = 1.01), did significantly report more deception, F (1, 102) = 19.97, p<.001, η2 = .164. Thus, hypothesis one was supported. Hypothesis two posited that as relational closeness increases, the level of perceived expectancy violation by the interviewer also increases. Consistent with IDT, relational closeness was a significant predictor of deceived participants’ expectancy ratings of the deceiver (R = -.194), F(1, 102) = 3.97, p<.05, R2 = .037, adjusted for shrinkage, R2 = .028. Specifically, the higher the level of relational closeness between the interview pair, the more likely the deceived participant reported the deceiver’s behaviors as an expectancy violation. Hypothesis three posited that as the level of perceived expectancy violation by the interviewer increases, the level of perceived deception also increases. As predicted, expectancy ratings was a significant predictor of deceived participants’ honesty ratings of the deceiver (R = .624), F(1, 102) = 65.06, p<.001, R2 = .389, adjusted for shrinkage, R2 = .383. Specifically, the higher the expectancy violation ratings interviewers gave for their interviewees’ behaviors, the higher their ratings of the interviewees’ deception during the interview. Taken together, the results for hypotheses two and three suggest that relational closeness should be a significant predictor of perceived deception on the part of the interviewer. That is, the closer the Emotions or Closeness 18 interviewer is to his/her interviewee, the more likely he/she will report deception on the part of the interviewee (due to greater awareness of expectancy violations). To test this possibility, a research question was posed. Research question one asked whether or not relational closeness was a direct predictor of perceived deception. A linear regression was performed on the data set, regressing interviewers’ honesty ratings of the interviewee onto relational closeness. Contrary to what was expected, relational closeness was not a significant predictor of interviewers’ honesty ratings of the interviewee (R = -.105), F(1, 102) = 1.14, p=.287, R2 = .011, adjusted for shrinkage, R2 = .001. And so, perceptions of deception by the interviewer were consistent across the study, regardless of whether the interviewee was a friend/stranger. Given this finding, it raises the question of whether or not relational closeness matters when it comes to predicting deception. Perhaps it is based solely on perceived expectancy ratings or on the affective state of the deceiver. And so, we had asked another research question. Research question two asked after controlling for relational closeness and expectancy ratings, does emotional state still account for a significant amount of variance in perceived deception. An ANCOVA was performed with interviewers’ honesty ratings of the interviewee as the dependent variable, affect condition as the independent variable, and relational closeness and expectancy ratings as covariates. Interestingly, interviewers’ emotional state did not account for a significant portion of the variance for their honesty ratings of the interviewee, F (1, 100) = 2.51, p=.116, η2 = .025, after controlling for relational closeness and expectancy ratings. In fact, a closer examination of the ANCOVA revealed that only expectancy ratings accounted for a significant amount of variance in honesty ratings, F (1, 100) = 40.05, p<.001, η2 = .286. Thus, the data suggests that expectancy ratings (not affective state or relational closeness) accounts for the most amount of variance for interviewers’ honesty ratings. Post-Hoc Analysis Emotions or Closeness 19 Given the current results, we were interested in looking at whether or not interviewers in the two affective conditions significantly differed in their expectancy ratings for their interviewees. Perhaps the reason why we found significant differences between the two groups on perceived deception is because one group picked up on expectancy violations committed by the interviewee better than the other group. Based on the emotions and message processing literature, we speculated that interviewers in a negative affective state will report more expectancy violations for their interviewees’ behaviors compared to those in a positive affective state. Thus, to test this speculation, a post-hoc analysis was done. An ANCOVA was performed with interviewers’ expectancy ratings of the interviewee as the dependent variable, affect condition as the grouping factor, and relational closeness as a covariate. The results indicate that interviewers induced to feel sadness reported significantly lower expectancy ratings (i.e., greater expectancy violations) for their interviewees’ behaviors than interviewers induced to feel happiness, F (1, 101) = 27.70, p<.001, η2 = .215. The means for the two groups were (M = 4.05) and (M = 5.74) respectively. For your convenience, the means and standard deviations associated with H1, RQ2, and the posthoc analysis are summarized in Table 2. And the means and standard deviations associated with H2, H3, and RQ1 are summarized in Table 3. Discussion The objective of this study was to determine the impact that the deceived persons emotional state and level of relational closeness to the deceiver have on their ability to perceive deceptive messages. The study found strong support for the prediction based on CFM and NSRM. Conversely, the study found mixed support for predictions based on IDT. These findings are further discussed in relation to message processing and future suggestions for IDT are proposed. Emotions or Closeness 20 The results of this study provide strong support for the argument that the emotional state of the person being deceived plays a vital role in deceptive contexts. In both conditions, interviewers were exposed to deceptive interactions. Yet, we found significant differences in the interviewers’ ability to detect deception based on our mood manipulations. As our methods indicate, the affect state of the interviewer was induced and measured prior to the onset of the interview. As our regression analyses illustrate, as level of happiness of the interviewer increases, perceived deceptiveness decrees. Conversely, as the level of sadness of the interviewer increases, perceptions of deceptiveness also increase. One strong explanation for these findings is that participants in the happy mood state failed to process those cues that would indicate deception. On the other hand, their counterparts in the sad emotional state did process or pick up the deceptive cues. Based on the CFM (Nabi, 1999), the goal of people in a happy l state is to stay happy. If processing deceptive cues fails to fulfill one’s goal, people will turn to biased or heuristic processing modes. In the interview context, it may be that because participants were ‘having fun’ with the interview, they didn’t want to consider the fact that the other person was lying to them (i.e., this would jeopardize their happy affect state). As a result of negative state avoidance, happy interviewers apparently chose not to cognitively process the deceptive messages and so deceivers were perceived to be more honest than dishonest. Conversely, people seeking negative state relief were able to more accurately detect deceptive cues. Based on the NSRM, people in a negative affect state are looking for any information that might help them get out of the negative state (e.g., focus on distractions). In this study, the interview served as an effective distraction to focus on. And so, interviewers in a negative affect state (due to their heightened focus on the interview) were more likely to be aware of their deceivers’ leakage cues, perceiving the deceivers as more dishonest than honest. Emotions or Closeness 21 As for the predictions based on IDT, as relational closeness between interviewer and interviewee increased, the level of expectancy violation perceived by the interviewer also increased. Similarly, as perceived expectancy violations increased, perceived deception also increased. However, contrary to what IDT would predict, relational closeness was not a significant predictor of perceived deception. Thus, the results offer mixed support for IDT. Taken together, the results suggest that interviewers only judged their interviewees as deceptive if they acted in an unexpected manner, regardless of how close they were to the person, or how they felt during the interview. However, it is important to note that both the interviewers’ level of relational closeness to the interviewee, and their affective state during the interview did seem to significantly impact their perceptions of expectancy regarding the other person’s behaviors (see results for H2 and post-hoc analysis). Thus, two different models for predicting deception detection may be proposed, one based on relational closeness, the other based on affect. Future study can be done to test these two models against each other to determine which offers a better fit. Bringing the mood of the deceived to IDT These findings present interesting implications for IDT and deception research. We agree with Buller and Burgoon (1996) that deception is more than a collection of verbal and nonverbal cues. Moreover, we agree with IDT proponents that contexts and relationships are important to deception detection. We argue that IDT is lacking propositions that address the mood state of the deceived party in a deception interaction and that the research related to message processing better explains deception detection. Moreover, we posit that in interpersonal interactions, moods will play a more important role than in impersonal contexts. For example, if applying these findings to real world contexts like confrontations about infidelity or cheating on an exam, it is easy to recognize that these situations are often affect charged for both the deceiver and the deceived. And so, for these highly affect-laden Emotions or Closeness 22 situations, the deceived individuals’ mood state may have a stronger impact on how effective they will be at detecting deception by the other person, more so than relational closeness. It may be the case that for less affect-laden situations (e.g., situations where there is little at stake for getting caught lying), relational closeness may play a more pivotal role in affecting a person’s ability to detect deception. Limitations and Directions for Future Research This study has two significant limitations. One major limitation of this study was the lack of a no deception control group. As a result of this limitation, occurrences of false positives for perceived deception as a result of a receiver’s emotional state cannot be ruled out as function of an affect heuristic. The addition of a no deception control group would have controlled for the potential confounding factor of an affect heuristic. Since our predictions focused on the relationship between the two different affect conditions, we felt the control group was not necessary, but future studies would benefit from the additional data. Additionally, we have concerns regarding the programming type used to create our negative affect condition. Our emotion manipulation was conducted via the use of a crime detective drama. It is possible that the programming type used activated a detective interaction approach cognitive schema for the interviewers. As a result, it is possible that the activation of detective interaction approach schema could account for the increased ability to detect deception in our interviewers and not mood. The study was also limited in the control and measurement of few important factors. For example, the study did not control for time of interview or measure amount of probing questions interviewers used. Since the study was done under the guise of an interpersonal interview, we felt it best to allow some freedom to the interactions. Future research would benefit greatly from controlling and/or measuring these important factors. Most importantly, this study has strong heuristic value and a number of opportunities for future Emotions or Closeness 23 research are opened up as a result of this study. First, researchers can examine the impact that other discrete emotions have on the ability to detect deception. Since this study only examined happy and sad mood states, other affect conditions like anger and frustration could be important to interpersonal deception. Also, future study can examine the impact that discrete emotions (other than fear) have on the ability of the deceiver to mask leakage cues or at picking up on the deceived’s growing suspicions during an interaction. Future study can also consider if the relationship between different emotions and the ability to detect deception in a linear relationship. For example, at what point does one’s negative affect state inhibit (rather than facilitate) cognitive processing and cause one to revert to heuristic processing? Future study can examine the interactions of mixed versus matching emotional states of deceiver and deceived, on the ability to deceive and detect deception. Finally, as the Bond and Depaulo (2008) metaanalysis indicate, there is a lack of support for individual difference variables linking to ones ability to detect deception. We feel prior deception research might be over looking important individual differences link to mood. Future studies could look at depression and other affective disorders and there relationship with deception detection. Emotions or Closeness 24 References Bohner, G., & Schwarz, N. (1993). Mood states influence the production of persuasive arguments. Communication Research, 20(5), 696-722. Bond, C.F., & Depaulo, B.M. (2008). Individual differences in judging deception: accuracy and bias. Psychological Bulletin, 134 (4), 477-492. Buller, D. B., & Burgoon, J. K. (1996). Interpersonal deception theory. Communication Theory, 6, 203–242. Buller, D. B., & Burgoon, J. K. (1996b). Another look at information management: A rejoinder to McCornack, Levine, Morrison, and Lapinski. Communication Monographs, 63, 92–98. Buller, D. B., & Burgoon, J. K. (1996c).Reflections on the nature of theory building and the theoretical status of interpersonal deception theory. Communication Theory, 6, 311-328. Buller, D. B., & Burgoon, J. K., Busling, A., & Roiger, J. (1996). Testing interpersonal deception theory: The language of interpersonal deception. Communication Theory, 6, 268-289. Burgoon, J. K., Buller, D. B., Dillman, L., & Walther, J.B. (1995). Interpersonal Deception IV. Effects of suspicion on perceived communication and nonverbal behavior dynamics. Human Communication Research, 22, 163-196. Burgoon, J. K., Buller, D. B., Ebesu, A.S., White, C.H., & Rockwell, P.A. (1996). Testing interpersonal Emotions or Closeness 25 deception theory: Effects of suspicion on communication behaviors and deception. Communication Theory, 6, 243-267. Burgoon, J. K., Buller, D. B., & Floyd, K. (2001). Does participation affect deception success? A test of the interactivity principle. Human Communication Research, 27, 503–534. Burgoon, J.K., & Floyd, K. (2000) Testing for the motivation impairment effect during deceptive and truthful interaction. Western Journal of Communication, 64, 243-268. Burgoon, J. K., Buller, D. B., Guerrero, L.K., Afifi, W.A., & Feldman, C.M. (1996). Interpersonal deception: XII. Information management dimensions underlying deceptive and truthful messages. Communication Monographs, 63, 50-69. Chaiken, S., Liberman, A., & Eagly, A. H. (1989). Heuristic and systematic information processing within and beyond the persuasion context. In J. S. Uleman & J. A. Bargh (Eds.), Unintended thought (pp. 212-252). New York: Guilford Press. Cialdini, R. B., Darby, B. L., & Vincent, J. E. (1973). Transgression and altruism: A case for hedonism. Journal of Experimental Social Psychology, 9, 502-516. Dillard, J. P. (1993). Persuasion past and present: Attitudes aren’t what they used to be. Communication Monographs, 60, 90-96. Dillard, J. P., Plotnick, C. A., Godbold, L. C., Freimuth, V. S., & Edgar, T. (1996). The multiple affective outcomes of AIDS PSAs: Fear appeals do more than scare people. Communication Research, 23, 44-72. Dunbar, N.E., Ramirez, Jr., A., & Burgoon, J.K. (2003). The effects of participation on the ability to judge deceit. Communication Reports, 16, 23-34. Isen, A. M. (1987). Positive affect, cognitive processes, and social behavior. Advances in Experimental Social Psychology, 20, 203-253. Emotions or Closeness 26 Izard, C. E. (1977). Human emotions. New York: Plenum Press. Lazarus, R. S. (1991). Emotion and adaptation. New York: Oxford University Press. Levine, T.R., Anders, L.N., Banas, J., Baum, K.L., Endo, K., Hu, A.D.S., et al. (2000). Norms, expectations, and deception: A norm violation model of veracity judgments. Communication Monographs, 66, 123-138. McCornack, S. A. (1992). Information manipulation theory. Communication Monographs, 59, 1-16. Miller, G.R., & Stiff, J.B. (1993). Deceptive communication. Newbury Park: Sage. Mitchell, M. M., Brown, K. M., Morris-Villagran, M., & Villagran, P. D. (2001). The effects of anger, sadness, and happiness on persuasive message processing: A test of the negative state relief model. Communication Monographs, 68(4), 347-359. Nabi, R. L. (1999). A cognitive-functional model for the effects of discrete negative emotions on information processing, attitude change, and recall. Communication Theory, 9(3), 292-320. Park, H.S., Levine, T.R., McCornack, S.A., Morrison, K. & Ferrara, S. (2002). How people really detect lies. Communication Monographs, 69, 144-157. Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 19, pp. 123-205). New York: Academic Press. Petty, R. E., Gleicher, F., & Baker, S. M. (1991). Multiple roles for affect in persuasion. In J. P. Forgas (Ed.), Emotion and social judgments (pp. 181-200). New York: Pergamon Press. Schwarz, N. (1990). Feelings as information: Informational and motivational functions of affective states. In E. T. Higgins & R. M. Sorrentino (Eds.), Handbook of motivation and cognition: Foundations of social behavior (Vol. 2, pp. 527-561). New York: Guilford Press. Emotions or Closeness 27 Schwarz, N., Bless, H., & Bohner, G. (1991). Mood and persuasion: Affective states influence the processing of persuasive communications. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 24, pp. 161-199). San Diego, CA: Academic Press. Schwarz, N., & Clore, G. L. (1988). How do I feel about it? Informative functions of affective states. In K. Fiedler & J. Forgas (Eds.), Affect, cognition, and social behavior (pp. 44-62). Gottingen, Germany: Hogrefe. Stiff, J.B. (1996). Theoretical approaches to the study of deceptive communication: Comments on interpersonal deception theory. Communication Theory, 6, 289-295. Waid, W.M., & Orne, M.T. (1981). Cognitive, social, and personality processing in the physiological detection of deception. Advances in Experimental Social Psychology, 14, 61-106. Wong, N.C.H., & Householder, B. J. (2004). Smoke and mirrors: Explaining the impacts of discrete emotions on processing of an anti-smoking PSA. Paper presented at the annual meeting of the International Communication Association in New Orleans, LA. White, C.H., & Burgoon, J.K. (2001). Adaptation and communication design pattern of interaction in truthful and deceptive conversations. Human Communication Research, 27, 9-37. Zuckerman, M., Depaulo, B.M., & Rosenthal, R. (1981). Verbal and nonverbal communication of deception. Advances in Experimental Social Psychology, 14, 2-59. Emotions or Closeness Table 1. Means and Standard Deviations of Key Dependent Variables for Manipulation Check Items Dependent Variables Mean S.D. N Sad condition 3.81 .88 52 Happy condition 6.33 .48 52 Questions # 1-5 4.33 .45 44 Questions # 6-10 4.72 1.10 60 Questions # 1-5 4.38 .55 60 Questions # 6-10 4.18 1.89 44 4.81 1.64 104 Interviewer’s Emotional State Interviewees’ Honesty Ratings Lie on the first 5 questions Lie on the last 5 questions Interviewees’ Ratings of Success at Deception 28 Emotions or Closeness Note: Higher values indicate more positive affect, honesty, and success. Table 2. Means and Standard Deviations of Key Dependent Variables by Affect Conditions Dependent Variables Mean S.D. N Sad emotional condition a 4.64 2.02 52 Happy emotional condition a 6.04 1.00 52 Sad emotional condition 5.10 .202 52 Happy emotional condition 5.58 .202 52 Sad emotional condition b 4.05 .266 52 Happy emotional condition b 5.74 .266 52 Interviewers’ Honesty Ratings of Interviewees Interviewers’ Honesty Ratings of Interviewee (Rel. closeness and expectancy ratings controlled) Interviewers’ Expectancy Ratings of Interviewee (Rel. closeness controlled) Note: Subscripts a and b represent significant differences between the groups at p<.001. 29 Emotions or Closeness Table 3. Regression Coefficients, T-Test Statistics, and Significance Testing for Expectancy Ratings Regressed Onto Relational Closeness, and Honesty Ratings Regressed Onto Expectancy Ratings and Relational Closeness Dependent Variables B SE B t p -.151 .076 -.194 -1.99 .049 Expectancy ratings .587 .073 .624 8.07 <.001 Relational closeness -.077 .072 -.105 -1.07 .287 Expectancy ratings Relational closeness Honesty Ratings For more information or to obtain press passes, please contact: Mercedes Sanchez. (646) 660-6112, mercedes.sanchez@baruch.cuny.edu 30