Personalizing Conflict in Different Interpersonal Relationship Types Taking conflict personally (TCP) occurs when a person experiences interpersonal conflict as a punishing, hurtful activity that others might even have initiated in order to harm self (Hample & Dallinger, 1995; for literature reviews see Hample, 1999, and Hample & Cionea, 2010). Personalization is a complex of specific feelings, but also has several cognitive elements, including estimates of how conflicts affect relationship prospects. The base feeling of personalization is typically accompanied by stress reactions, persecution feelings, pessimism about the effects of conflicts on personal relationships, and negative valence for the general activity of participating in conflicts (Hample & Cionea, 2010). The bulk of TCP research has treated personalization as a trait, and has connected it to a number of other enduring individual characteristics, including argumentativeness, verbal aggressiveness, neuroticism, masculinity, femininity, psychological reactance, argument frames, and a variety of other measures in the U.S. (Hample & Cionea, 2010). Some welcome attention to trait TCP as experienced in other cultures is also underway (Avtgis & Rancer, 2004; Hample & Anagondahalli, 2014; Kim, Yamaguchi, Kim, & Miyahara, 2015), with some leading U.S. results being replicated. From the beginning, however, TCP was also understood to have the possible character of a state, and a limited number of studies have explored people’s levels of personalization in regard to a serial argument (Hample & Richards, in press) or a just-completed interaction (Hample, Dallinger, & Fofano, 1995; Hample, Dallinger, & Nelson, 1995). In several of these studies, state and trait measures of TCP correlated substantially and positively, but not at such levels as to indicate that trait and state levels of TCP were essentially the same within each person. Standing next to conceptions of TCP as either a trait or a state, this investigation re-introduces an intermediate idea, the possibility that TCP levels might be relationship-specific. The general hypothesis is that a person might have one level of personalization for her marriage, another at work, and yet another for arguing with her father. An unpublished paper by Dallinger and Hample (2001) explored this idea, but had a sample that was limited both in size (N = 98, with a maximum of 75 and a minimum of 14 participating in each of the relationships studied) and in the fact that all respondents were undergraduates. They found an overall significant effect of relationship type on TCP level. The relationship type that produced the highest personalization scores was conflict with parents, followed by romantic partners and coworkers. Conflicts with siblings, best friends, and acquaintances were least punishing. The current study pursues these findings with larger and more varied samples and provides a theoretical frame with which to interpret the findings. Relationship Types and Conflicts Why should we expect relationship type to make any difference in the degree to which people personalize their conflicts? Our thinking begins with social exchange theory (Homans, 1958; Roloff, 1981). Relationships develop as people offer and receive various resources, expressions, and activities due to their connections with one another. As time goes on, the values implicit in these exchanges increase, change their character (e.g., from a pleasant greeting to a declaration of love), and, one hopes, become more dependable. The general idea of social exchange has been pointedly applied to interpersonal relationships by interdependence theory (Kelley, Holmes, Kerr, Reis, Rusbult, & van Lange, 2003; Kelley & Thibaut, 1978; Thibaut & Kelley, 1959). This theory holds that a key feature of a situation is the sort of interdependence the two participants have in that moment: Does one person dictate the other’s outcomes? Must they act jointly in order for either to have positive experiences? Notice that this conception is sensitive to the possibility that a given situation might involve conflict, competition, cooperation, and/or mixed motives during the interactions. Paired people of course participate in interpersonal situations together, and those situations can cumulate, acquire their own trajectories, enable their own re-occurrence, and become familiar, so that ongoing interpersonal relationships can also be characterized as often having one sort of interdependence or another. Over time people in a relationship must coordinate and exchange their behaviors. Their outcomes might typically be shared or contrary, and their dependence can be unilateral or joint. A person’s approach to a situation will certainly be affected by the nature of the situation and the personal characteristics that influence the person’s interpretation of the immediate circumstances (van Lange & Rusbult, 2012). However, relational life affords the near-certainty that various patterns of interaction and understanding will occur repeatedly and characteristically in the interactions between the two related people (Gottman, 1998; Watzlawick, Beavin, & Jackson, 1967). Therefore, dyads tend to communicate, and presumably argue, in ways typical to the dyad, with its characteristic patterns based on types of interdependence. Cionea (2011) has recently proposed that Walton’s (1998) theory of dialogues be used to characterize the argumentative patterns in close relationships. Walton analyzed the goals and nature of several kinds of argumentative dialogues (persuasion, inquiry, negotiation, information-seeking, deliberation, and eristic). Cionea used this system to suggest that continuing relationships might well have a typical way of dealing with disagreement. One couple might handle disagreements by trying to defeat the relational partner at any cost (eristic); another pair might tend to view disagreements as temporary inconveniences caused by the partners not having the same information and so might try to obtain and share information (information-seeking); another dyad might consistently ignore principles and simply try to bargain out some sort of middle arrangement (negotiation); and so forth. These ways of characteristically dealing with disagreement reflect, reproduce, and even possibly cause the sort of interdependence that characterizes the relationship. One study showed that the dialogues have a preference order that roughly corresponds to their niceness or regard for the other person (Cionea & Hample, 2014). In that sample of American adults (mean age of 32), persuasion was most preferred, followed by information giving. The deliberation, inquiry, negotiation, and information-seeking dialogues were next, but with no preference distinctions among them. By far the least preferred was eristic. An obvious question might be, if two people find themselves consistently fighting and personally punishing one another with eristic behaviors, why would they stay in the relationship? Possibly the couple has so few disagreements that an occasional fight makes little difference to their relational lives. But a more theoretically interesting answer is that the people may have so much investment in the relationship that they feel obliged to persist in it (Rusbult, 1983; Rusbult, Agnew, & Arriaga, 2012; Rusbult, Zembrodt, & Gunn, 1982). Rusbult’s investment model, a variant of interdependence theory, is that persistence in a relationship is immediately predicted by participants’ level of commitment to it, and commitment is jointly predicted by relational satisfaction, quality of alternative relationships, and investment size (Rusbult, Agnew, & Arriaga, 2012; Rusbult, Martz, & Agnew, 1998). An implication of this line of research is that persistence in a relationship can make it more and more central to one’s life, thus increasing one’s investment in it and so raising one’s commitment level. The circularity of this (life recursiveness, not theoretical tautology) encourages us to consider what sorts of relationships might involve the highest life stakes or importance for their participants. Supposing that a person might personalize conflicts in any setting, what sort of relationships could he or she easily leave, and in what sorts might the person persist in the face of continuing feelings of hurt during conflict? Highly interdependent relationships such as marriages and parent/child relationships would seem to be most consequential, stubborn, and common. Sibling relationships are also difficult to leave, but adults often let these relationships falter to some degree in favor of close friendships, marriages, and parent/child involvements. These might be followed in some order by more ordinary friendships and important professional connections, such as superior/subordinate or coworker, depending on the sort of work the person does. Acquaintances and strangers would fill in the end of the line of commonly experienced relationships. Notice that these are described in rough order of permanence and emotional value. Our reasoning to this point is roughly this. Conflict personalization is punishing and so a person high in TCP will tend to avoid conflictive encounters if possible (Dallinger & Hample, 1995). Relational life inherently affords the risk of disagreement, and so personalization will encourage exit if exit is easy. We should only find high levels of TCP in relationships that can be expected to involve high levels of commitment, investment, satisfaction, and absence of alternatives, since these are the conditions that discourage exit (Rusbult, Zembrodt, & Gunn, 1982). Comparative data of this sort on all the relationships we study here seems to be oddly sparse. Dallinger and Hample (2001), as mentioned above, found the highest levels of conflict personalization in college students’ conflicts with their parents, with romantic partners and coworkers ranked next. Siblings, best friends, and acquaintances produced the least TCP. The low ranking for best friend relationships might reflect the fact that this relationship is voluntary, and might well have been chosen in part for the ease and rarity of conflict discussion. We summarize our thinking with these research objectives: H1: Relationship type will predict TCP levels. We expect TCP levels to vary by relationship because some punishing relationships can easily be left but others cannot. We lack the literature base to predict any specific ordering among relationships. Somewhat more specifically, we consider that some of the relationship types will tend to involve more commitment and perseverance, and that the more satisfying and enduring relationships will either have interaction systems that help TCP to dissolve or that swamp it with high value exchange patterns so that it doesn’t register saliently. Therefore, we propose H2: Feelings of personalization will be lower in the face of higher relational satisfaction and relational importance. Satisfaction should point toward pleasanter experience with conflict. The satisfaction might encourage constructive conflict behavior or might derive in part from it, and our data cannot distinguish these causalities. Importance may have built in part because conflicts were rewardingly handled, but we are less sure of this portion of the prediction because people may find themselves trapped in an important but unpleasant relationship. Method Procedures Data were collected from two distinct samples. One was undergraduates at a large public mid-Atlantic university in the U.S., and the other was an adult sample living in the U.S. recruited through Amazon’s mTurk system (see Buhrmester, Kwang, & Gosling, 2011; Goodman, Cryder, & Cheema, 2012; Mason & Suri, 2011). The undergraduates received minor course credit for participation and the mTurk workers were paid $0.50 each for their efforts. The online surveys took about 15 minutes to complete. Participants The undergraduate sample (N = 320) had a mean age of 19.4 years (SD = 1.8). Only 28% were male. Most were freshmen (37%) or sophomores (27%). Most (96%) self-reported that they were heterosexual, with four self-identifying as gay or lesbian, and seven as bisexual. Only one respondent was married. 32% described themselves as being in a serious romantic relationship at the time of the survey. Another 10% said they had been in such a relationship previously. Most (56%) described themselves as single. About half the sample (49%) said they were Euro-American, with 11% identifying as Asian-American, 9% as African-American, 7% as Hispanic-American, and the remainder scattered among various nationalities or a combination (8%) of our choices. The mTurk sample (N = 281) was 33.5 years old on average (SD = 11.7). All but one was a U.S. citizen. 56% of respondents were male. 3% of participants were retired, 52% had a full-time job, 21% worked part-time and 21% were unemployed (possibly including current students). 94% self-reported being heterosexual, with six gay or lesbian and nine bisexual. 34% were married, 7% divorced, 1% widowed, 26% in a serious romantic relationship at the time of the survey, 8% had previously experienced a serious romantic relationship, and 25% simply described themselves as single. This sample overwhelmingly (72%) self-identified as Euro-American, with 6% African-American, 5% Hispanic-American, 7% Asian-American, and 2% Native American. Asked about the income of the family unit in which they lived, the main answers were $20,000 or less (19%), $20,000 to $40,000 (26%), $40,000 to $60,000 (25%), and $60,000 to $80,000 (7%). 15% reported a family income of more than $80,000. The modal education level for this sample was “university degree” (N = 96, 34%), with “some university” almost identical in frequency (N = 95, 34%). 13% had attended graduate school, and 18% had a high school degree. Instrumentation With the exception of several demographic questions oriented to the particularities of each sample, all respondents filled out the same instruments. Reliabilities, means, and standard deviations for both samples are in Table 1, and correlations among the measures are in Table 2. Respondents were asked to choose an interpersonal relationship for which they could provide answers. The prompt was: “This study is about your feelings and beliefs about a particular interpersonal relationship that you are in. You will only respond in terms of a single relationship. We would like to have a nice variety of relationships represented in our study. Therefore we have listed possible relationships below, with the most common ones at the bottom. Please choose the first one in the list that you can give responses about.” The order of relationships is as follows, with the percentages of undergraduates and mTurk respondents choosing it shown in parentheses: your child (0.3%, 14%), spouse or serious romantic partner (33%, 50%), a subordinate at work (1%, 1%), a boss at work (4%, 6%), a coworker (1%, 4%), your parent (29%, 10%), a friend (31%, 16%). Participants were requested to respond to all further questions on the survey in light of the relationship they chose. All Likert-style questions used ten point response formats. The taking conflict personally (TCP) scales from Hample and Dallinger (1995) were slightly reworded to refer to “the relationship you have selected” and to arguing or conflict “with this person.” The items resolve into six subscales. Direct personalization is the most general indicant of TCP, and this subscale has seven items (e.g., “I usually take criticisms personally when arguing with this person”). Persecution feelings reflect the belief that others are participating in order to harm the respondent (e.g., “Conflict with this person leaves me feeling victimized”). Stress reactions include self-reports of both physical and psychological tension (e.g., “Stressful discussions with this person make my stomach hurt,” and “After a stressful meeting with this person, my day is usually ruined”). Both positive relational effects and negative relational effects are measured because conflicts can have both or neither of these consequences (e.g., “Sometimes you can discover admirable features in this person when he or she is arguing strongly,” and “Conflict discussions with this person can really jeopardize our friendship”). Finally valence is scored to indicate positive feelings about engaging in conflicts (e.g., “Conflict is an intensely enjoyable kind of interaction when arguing with this person”). The relational satisfaction items from Hample and Richards (in press) were slightly reworded to refer to conflict within the selected relationship. These eight items involve global impressions of how conflicts generally affect the interpersonal relationship (e.g., “As a result of conflicts, the relationship usually becomes better,” “As a result of conflicts, our relationship usually becomes less welcoming,” and “As a result of conflicts, I usually lose respect for the other person”). We also created a five-item measure of relational importance. The items were: “In the most general terms, how important is this relationship to you?” “How important is this relationship to you emotionally?” “How important is this relationship to you in practical terms (e.g., financial, entertainment)?” “How attached are you to this relationship?” and “If this person were to leave your life for some reason, how hard would it be to fill that loss in your life?” All the instruments had acceptable Cronbach’s alphas (see Table 1). Results Preliminary Results Several results help contextualize our findings and may be of interest to researchers pursuing other theoretical interests. First, Table 1 shows that our two samples differed on several of the measures, always in the direction of undergraduates being more optimistic and positive about conflict and relational experiences than the older adults in the mTurk sample (cf. Arnett, 2000). We also examined the possibility of sex differences within the two samples. The patterns of results were fairly comparable for the two groups. For the undergraduates, women rated their relationships as more important (8.37 v. 7.87, t (318) = 2.24, p < .05), but with more direct personalization (6.34 v. 5.51, t (318) = 3.90, p < .001) and stress (5.88 v. 5.36, t (318) = 2.24, p < .05) as well as less positive valence (2.90 v. 4.05, t (318) = 6.02, p < .001). In the mTurk sample, women also rated their relationships as more important (8.70 v. 8.06, t (278.8) = 3.00, p < .01) and reported more direct personalization (6.62 v. 5.78, t (242.5) = 3.46, p < .001) and more stress (6.29 v. 5.77, t (248.7) = 2.02, p < .05), but less positive relational effects (4.64 v. 5.20, t (233.6) = 2.12, p < .05) and less positive valence (2.56 v. 3.29, t (278.7) = 3.50, p < .001). Other comparisons were not statistically significant, and details are available from the authors. Overall, women in both samples felt that relationships were more important, but had less favorable reactions to conflict interactions within them. Finally, we took notice of the associations between relational satisfaction and relational importance. As Table 2 shows, these were positive and substantial in both samples. These results were noticeably smaller than the meta-analytic finding that satisfaction and commitment correlated at r = .68 (Le & Agnew, 2003), suggesting that our measure of relationship importance may have captured a substantively different construct than relational commitment. Hypotheses Our first prediction was that different TCP levels will be found for different relationship types. We undertook a 2 (sample) x 7 (relationship) multivariate analysis of variance, using all six TCP measures as dependent variables. Relationship had a significant main effect using Pillai’s Trace (F (36, 3522) = 4.57, p < .001), but neither sample (F (6, 582) = 0.91, p = .49) nor the relationship by sample interaction (F (36, 3522) = 1.06, p = .37) was significant. We therefore undertook univariate follow-up tests of the effects of relationship on each TCP measure. TCP means categorized by relationship type are in Table 3. All of the analyses produced statistically significant findings. For direct personalization, F (6, 594) = 16.31, p < .001, partial 2 = .14, with subordinate at work having the lowest mean (3.41) and parent (6.65) and spouse/romantic partner (6.67) the highest (Duncan post hoc tests were used). For stress reactions, F (6, 594) = 10.10, p < .001, partial 2 = .09. The greatest stress was reported for spouse/romantic partner (6.33) and the lowest for subordinate (3.74). Persecution feelings were also significant (F (6, 594) = 2.17, p < .05, partial 2 = .02), with subordinates again being least problematic (2.54) and higher values for children (3.87), spouse/romantic partner (3.94), parent (4.17), boss (4.34), and coworker (4.80). Positive relational effects (F (6, 594) = 4.94, p < .001, partial 2 = .05) produced the lowest mean for subordinate (2.93) and the highest for friends (5.39) and spouse/romantic partner (5.46). Negative relational effects (F (6, 594) = 5.25, p < .001, partial 2 = .05) followed a different pattern. The lowest means were for friendship (4.41) and the highest for boss (6.39). Finally valence (F (6, 594) = 8.33, p < .001, partial 2 = .08) distinguished between child (2.66) at the low end and subordinates (4.22) and coworkers (4.26) at the other. These results afford clear support for H1 with reasonable effect sizes. People did, in fact, have different levels and details for personalization of conflict depending on what relationship they had in mind while responding. Relationships with subordinates at work tended to be regarded as generating the least punishing conflicts and the greatest problems tended to be associated with the more obviously intense relationships, particularly spouse/romantic partner. Hypothesis two proposed that feelings of personalization will be lower when relational satisfaction and relational importance are higher. Table 2 supplies the correlations relevant to the prediction. Readers should notice that four of the TCP scales (direct personalization, stress reaction, persecution feelings, and negative relational effects) most immediately indicate personalization, and the other two (positive relational effects and positive valence) suggest the opposite. The correlations in Table 2 give immediate support for the relational satisfaction portion of the hypothesis. The higher a person’s satisfaction in both samples, the lower were his/her estimates of the negative TCP measures and the higher were the estimates of the positive subscales. But for relational importance, results were quite inconsistent. Some correlations were in the predicted direction, others were non-significant, and some were in the opposite direction. Given the noticeable correlations between relational satisfaction and importance, the better statistical procedure is multiple regression. This permits us to see the effects of one variable without confusing those outcomes with any due to common variance with the other predictor. We combined the two samples and predicted each TCP scale with mean-centered satisfaction and importance. These results were quite consistent. The negatively toned TCP scales always had a negative relationship with satisfaction and a positive relationship with importance. For the positively scored TCP scales, the results had the opposite pattern. Satisfaction reduced personalization, and importance increased it. Here are the details, first for the subscales scored to reflect negative feelings about conflict. Direct personalization was predicted (R = .42, p < .001) by satisfaction (unstandardized b= -.37, p < .001) and importance (b= .37, p < .001). Persecution feelings were also predicted (R = .55, p < .001) by satisfaction (b= -.56, p < .001) and importance (b= .11, p < .05). The same pattern appeared for stress reactions (R = .44), which had significant relationships with both satisfaction (b= -.43, p < .001) and importance (b= .34, p < .001). Negative relational effects also produced the same pattern as for the other negatively-scored measures: R = .68, with satisfaction having a negative relationship (b= -.76) and importance a positive one (b= .20, p < .001). The two measures indicating a positive orientation to conflict gave the opposite pattern. Positive relational effects was predicted (R = .70, p < .001) by satisfaction (b= .69, p < .001) and almost by importance (b= -.06, p = .09). Valence (R = .39) was predicted by satisfaction (b= .25, p < .001) and importance (b= -.36, p < .001). Our prediction that TCP would be lower in the face of higher relational satisfaction was supported. However, our parallel expectation that relational importance would have the same effect was quite thoroughly rejected. In fact, all of our predictions about relational importance were wrong: Higher relational importance was associated with higher TCP. Discussion Our conceptual mistake was to be carried away by the common polarities of relational satisfaction and relational importance. We supposed that a good relationship would be satisfying and important (thus, to be approached) and that a bad relationship would be unsatisfying and unimportant (to be avoided). We were not entirely wrong in this, because satisfaction and importance correlated at r = .45 for the undergraduate sample and r = .37 for the adults. But although it occurred to us, we did not give enough weight to the possibilities that a person might experience an unsatisfying relationship that had to be endured because it was important, or might have satisfying relationships of small consequence. The multiple regressions were the procedures that revealed our error. This statistical analysis shows how one predictor asserts itself when the other one has no effect (that is, its regression weight is zero). The regressions showed the effects of satisfaction when importance did not vary from its mean, and the effects of importance when satisfaction took on a constant mean value. So in every case, holding importance constant and uninfluential, satisfaction always reduced feelings of personalization, as we predicted. And in every case, when satisfaction was statistically held to its constant mean value, importance increased personalization, whereas we had expected the opposite effect. Therefore the regression results showed what happened when relationships of equal satisfaction still had variance in importance. With mean relational satisfaction, very high importance is out of kilter and indicates a relationship that is more important that it is satisfying. This case pointed toward more negative experiences of conflict. And with mean relational satisfaction, very low importance is also out of sync (low satisfaction would be more usual), and in this case insufficient importance made conflict less rewarding. As we reported, relational importance did not have the same level of association with relational satisfaction that commitment normally does. Importance and commitment must therefore be distinguished conceptually as well. However, it seems likely that the most highly committed relationships will also be the most important, and that people will systematically have less commitment to their less important social connections. Future research should therefore test whether the sort of outcome we found here for importance may also occur for commitment. For instance, with satisfaction held constant, does higher commitment still imply more persistence in the relationship? This fundamental finding (Rusbult, Agnew, & Arriaga, 2012) deserves to be tested under stringent conditions: perhaps there is something peculiar about high commitment in the face of average satisfaction or low commitment when the relationship is ordinarily satisfying. That is the sort of oddity we found here. Our other main result was the finding that conflict personalization varied according to its contextualizing relationship. Prior work has established that TCP has sensible correlates when operationalized as either a trait or a state, but little work had addressed the intermediate possibility that we explored here. Focusing on the relationships with the highest subsample sizes in Table 3, friendships were noticeably associated with easier conflict feelings. Compared to romantic partners/spouses, children, and parents, friendships implied the lowest ratings for the negative TCP measures (direct personalization, stress, persecution, and negative relational effects) and either the highest or second highest for the positive measures (valence and positive relational effects). Romantic partners and spouses seemed to involve the most personalization since they were first or second highest on the negative measures, but they also held out promise because they were also first or second on the positive subscales. Conflict with parents called out high scores for the negative measures and low scores for the positive ones. Children tended to have moderate scores across the board. Not all of these differences were statistically significant, but it is interesting that the involuntary relationships (parent and child) had middling levels of personalization while romance had the most friction (offering high levels of both positive and negative orientations) and friendship was most pleasant. These results were consistent with those in Dallinger and Hample (2001). These findings – showing that personalization is at least partly relationship-centered – support Cionea’s (2011) suggestion that particular dyads may have characteristic methods of dealing with disagreement. We did not collect any behavioral data here, but the fact that people have clearly different orientations to conflict in their various relationships implies that they may behave differently in them as well. TCP has predicted levels of challenge and accommodation behaviors in prior work (Hample, Dallinger, & Fofano, 1995; Hample, Dallinger, & Nelson, 1995), and has also been associated with differing conflict styles and motivations (Avtgis & Rancer, 2004; Dallinger & Hample, 1995; Kim, Yamaguchi, Kim, & Miyahara, 2015). Cionea’s proposal that some couples may be information-sharers, others may be negotiators, some may be fighters, and so forth, is very energizing and points the way toward connecting particular sorts of conflict dialogues with particular relationships. It may prove possible to distinguish the influence of interdependence from communication patterns in typifying how dyads conduct themselves under disagreement. As with all research, this study had a few limitations. First, people self-selected the relationship on which they reported according to the prioritized list provided (child first, spouse or romantic partner second, subordinate at work third, etc.). It is possible that differences that led people to select one relationship type over another, rather than differences in TCP experienced within the relationship types themselves, accounted for the results. For example, we found that conflict with spouses and romantic partners elicited more direct personalization than conflict with children. One alternative explanation is that people with children, compared to spouses and those in romantic relationships without children, may be less likely to personalize conflict across all relationship types due to changes in perspective associated with parenting. Alternatively, people who are inclined toward parenthood may be naturally predisposed to having lower TCP. Since participants were asked to report on a child relationship before a spousal relationship if they were able, lower direct personalization scores reported with a child may not be unique to that relationship type, but instead due to differences in trait TCP for parents versus nonparents in committed relationships. Future research would do well to control for individual differences that may complicate findings about relationship type. Second, although we made efforts to collect a broad range of participants, our sample was limited to college students and users of mTurk. We found no significant differences in how these two groups experienced TCP, which offers some confidence in the generalizability of these findings. However, people that fall outside these groups may experience TCP in relationships differently from the participants in this study. Finally, we sought to compare differences in TCP across relationship types. Implicit in this analysis is the assumption that commonalities exist within distinct types of relationships. However, we acknowledge that relationships of the same type can not only differ in satisfaction and importance, but also in other variables that likely affect the experience of TCP. For example, power, external social support, and cultural norms associated with conflict, all things that can affect TCP, can vary for a single person even across different relationships of the same type. A person may have a respectful relationship with one coworker but a disrespectful relationship with another. The particular relationships that participants identified within each category may have biased these results if there were systematic reasons to report one relationship over another (e.g., people were more likely think of their disrespectful, rather than respectful, coworkers). Subsequent studies should consider this issue when soliciting specific relationships on which participants report. This project helped to fill out our theoretical understanding of what it means to take conflict personally. The extensive research connecting TCP-as-trait to other enduring individual differences has included the finding that TCP is associated with several supertraits, mainly neuroticism and extraversion (Hample & Cionea, 2010). Bouchard (2004) has shown that genetic influence on these supertraits is about 50%. So the trait qualities of TCP are well established. Fewer studies have examined state TCP, but those investigations showed that (a) state and trait TCP are associated but not identical, and (b) that state TCP was more predictive of immediate behaviors and impressions than was trait TCP. So the idea that personalization can be a state, a reaction to a particular experience, is also clear. Here we have supported a third line of thought about TCP – that it can be a reliable (and perhaps enduring) orientation to conflict on a relationship by relationship basis.  References Arnett, J. J. (2000). Emerging adulthood: A theory of development from the late teens through the twenties. American Psychologist, 55, 469-480. doi: 10.1037//0003-066X.55.5.469 Avtgis, T. A., & Rancer, A. S. (2004). Personalization of conflict across cultures: A comparison among the United States, New Zealand and Australia. Journal of Intercultural Communication Research, 33, 109-118.10.1037//0003-066X.55.5 Bouchard, T. J., Jr. (2004). Genetic influence on human psychological traits: A survey. Current Directions in Psychological Science, 13, 148-151. doi: 10.1111/j.0963-7214.2004.00295.x Buhrmester, M., Kwang, T., & Gosling, S. D. (2011). Amazon’s Mechanical Turk: A new source of inexpensive, yet high-quality data? Perspectives on Psychological Science, 6, 3-5. doi: 10.1177/1745691610393980 Cionea, I. A. (2011). Dialogue and interpersonal communication: How informal logic can enhance our understanding of the dynamics of close relationships. Cogency, 3, 93-105. Cionea, I. A., & Hample, D. (2014, July). Dialogue orientations and argumentative behaviors. Paper presented at the meeting of the International Society for the Study of Argumentation, Amsterdam, Netherlands. Dallinger, J. M., & Hample, D. (1995). Personalizing and managing conflict. International Journal of Conflict Management, 6, 273-289. http://dx.doi.org/10.1108/eb022766 Dallinger, J. M., & Hample, D. (2001, November). Taking conflict personally: Trait and state measures, and the effects of relationship type, sex, and self-monitoring. Paper presented to the annual meeting of the National Communication Association, Atlanta, GA. Goodman, J. K., Cryder, C. E., & Cheema, A. (2012). Data collection in a flat world: The strengths and weaknesses of Mechanical Turk samples. Journal of Behavioral Decision Making, 26, 213-224. doi: 10.1002/bdm.1753 Gottman, J. M. (1998). Psychology and the study of marital processes. Annual Review of Psychology, 49, 169-197. doi: 10.1146/annurev.psych.49.1.169 Hample, D. (1999). The life space of personalized conflicts. Communication Yearbook, 22, 171-208. Hample, D., & Anagondahalli, D. (2014, November). Understandings of arguing in India: Argument frames, personalization of conflict, argumentativeness, and verbal aggressiveness. Paper presented to the annual meeting of the National Communication Association, Chicago, IL. Hample, D., & Cionea, I. A. (2010). Taking conflict personally and its connections with aggressiveness. In T. A. Avtgis & A. S. Rancer (Eds.), Arguments, aggression, and conflict: New directions in theory and research (pp. 372-387). New York, NY: Routledge, Taylor, and Francis. Hample, D., & Dallinger, J. M. (1995). A Lewinian perspective on taking conflict personally: Revision, refinement, and validation of the instrument. Communication Quarterly, 43, 297-319. doi: 10.1080/01463379509369978 Hample, D., Dallinger, J. M., & Fofano, J. (1995). Perceiving and predicting the tendency to personalize arguments. In S. Jackson (Ed.), Argumentation and values (pp. 434-438). Annandale, VA: Speech Communication Association. Hample, D., Dallinger, J. M., & Nelson, G. K. (1995). Aggressive, argumentative, and maintenance arguing behaviors, and their relationship to taking conflict personally. In F. H. van Eemeren, R. Grootendorst, J. A. Blair, & C. A. Willard (Eds.), Proceedings of the third International Society for the Study of Argumentation conference on argumentation, vol. III: Reconstruction and application (pp. 238-250). Amsterdam, the Netherlands: SicSat. Hample, D., & Richards, A. S. (in press). Attachment style, serial argument, and taking conflict personally. Journal of Argumentation in Context. Homans, G. (1958). Social behavior as exchange. American Journal of Sociology, 63, 597-606. doi: 10.1086/222355 Kelley, H. H., Holmes, J. G., Kerr, N. L., Reis, H. T., Rusbult, C. E., & van Lange, P. A. M. (2003). An atlas of interpersonal situations. Cambridge: Cambridge University Press. Kelley, H. H., & Thibaut, J. W. (1978). Interpersonal relations: A theory of interdependence. New York: John Wiley. Kim, E. j., Yamaguchi, A., Kim, M.-S., & Miyahara, A. (2015). Effects of taking conflict personally on conflict management styles across cultures. Personality and Individual Differences, 72, 143-149. doi: 10.1016/j.paid.2014.08.004 van Lange, P.A.M., & Rusbult, C. E. (2012). Interdependence theory. In P.A.M. van Lange, A.W. Kruglanski, & E. T. Higgins (Eds.), Handbook of theories of social psychology, vol. 2 (pp. 251-272). Los Angeles, CA: Sage. Le, B., & Agnew, C. R. (2003). Commitment and its theorized determinants: A meta-analysis of the investment model. Personal Relationships, 10, 37-57. Doi: 10.1111/1475-6811.00035 Mason, W., & Suri, S. (2011). Conducting behavioral research on Amazon’s Mechanical Turk. Behavior Research Methods, 44, 1-23. doi: 10.3758/s13428-011-0124-6 Roloff, M. E. (1981). Interpersonal communication: The social exchange approach. Beverly Hills, CA: Sage. Rusbult, C. E. (1983). A longitudinal test of the investment model: The development (and deterioration) of satisfaction and commitment in hetrosexual involvements. Journal of Personality and social Psychology, 45, 101-117. doi: 10.1037/0022-3514.45.1.101 Rusbult, C.E., Agnew, C. R., & Arriaga, X. B. (2012). The investment model of commitment processes. In P.A.M. van Lange, A.W. Kruglanski, & E. T. Higgins (Eds.), Handbook of theories of social psychology, vol. 2 (pp. 218-231). Los Angeles, CA: Sage. Rusbult, C. E., Martz, J. M., & Agnew, C. R. (1998). The investment model scale: Measuring commitment level, satisfaction level, quality of alternatives, and investment size. Personal Relationships, 5, 357-391. doi: 10.1111/j.1475-6811.1998.tb00177.x Rusbult, C. E., Zembrodt, I. M., & Gunn, L. K. (1982). Exit, voice, loyalty, and neglect: Responses to dissatisfaction in romantic involvements. Journal of Personality and Social Psychology, 43, 1230-1242. doi: 10.1037/0022-3514.43.6.1230 Thibaut, J. W., & Kelley, H. H. (1959). The social psychology of groups. New York: John Wiley. Walton, D. (1998). The new dialectic: Conversational contexts of argument. Toronto: University of Toronto Press. Watzlawick, P., Beavin, J. H., & Jackson, D. D. (1967). Pragmatics of human communication. New York: Norton.  Table 1 Descriptive Statistics Undergraduates mTurk Sample  Mean SD  Mean SD t Direct Personalization .81 6.10 1.76 .87 6.15 2.04 0.32 Stress Reaction .73 5.73 1.88 .79 6.00 2.14 1.65 Persecution Feelings .85 3.72 1.92 .86 4.14 2.17 2.53* Positive Relational Effects .89 5.37 1.92 .91 4.95 2.08 2.59** Negative Relational Effects .87 4.86 2.04 .88 5.48 2.34 3.43*** Positive Valence .84 3.23 1.62 .88 2.97 1.77 1.87 Relational Satisfaction .92 6.53 1.92 .93 6.03 2.27 2.90** Relational Importance .85 8.23 1.80 .84 8.34 1.85 0.77 Note.  is Cronbach’s alpha. t tests compare means from the two samples. Degrees of freedom for t tests were 599 unless an adjustment for unequal variances was made, in which case degrees of freedom were between 553 and 573. * p < .05 ** p < .01 *** p < .001  Table 2 Correlations Among Measures 1 2 3 4 5 6 7 Undergraduate Sample 1 Relational Satisfaction 2 Relational Importance .45 3 Direct Personalization -.23 .18 4 Stress Reaction -.22 .12 .63 5 Persecution Feelings -.49 -.11 .45 .42 6 Positive Relational Effects .68 .26 -.07 -.09 -.15 7 Negative Relational Effects -.64 -.17 .52 .50 .67 -.31 8 Valence .06 -.28 -.48 -.40 .02 .21 -.22 mTurk Sample 1 Relational Satisfaction 2 Relational Importance .37 3 Direct Personalization -.29 .21 4 Stress Reaction -.42 .13 .65 5 Persecution Feelings -.57 -.18 .44 .52 6 Positive Relational Effects .72 .21 -.19 -.34 -.22 7 Negative Relational Effects -.67 -.10 .57 .68 .66 -.49 8 Valence .23 -.26 -.41 -.54 -.07 .40 -.38 Note. For the undergraduate sample, r equal to or greater than |.12| is significant at p < .05, two-tailed. For the mTurk sample, the parallel value is |.13|.  Table 3 TCP Means by Relationship Type Direct Stress Persec PosRel NegRel Valence Your Child 5.65 b,c 5.89 b 3.87 b 4.94 b,c 5.02 a,b,c 2.66 a Spouse/Rom 6.67 c 6.34 c 3.94 b 5.46 c 5.33 a,b,c 2.83 a,b Subordinate 3.41 a 3.74 a 2.55 a 2.94 a 4.49 a,b 4.22 c Boss 5.58 b,c 5.84 b,c 4.34 b 4.03 a,b 6.39 c 3.23 a,b Coworker 5.31 b 5.63 b,c 4.80 b 5.21 b,c 5.81 b,c 4.26 c Your Parent 6.65 c 6.14 b,c 4.17 b 4.83 b,c 5.40 a,b,c 2.79 a,b Your Friend 5.21 b 4.92 b 3.58 a,b 5.39 c 4.41 a 3.79 b,c Note. Letters after each mean indicate what statistical groups contain the means, for each column. So two means sharing a subscript do not differ significantly by Duncan post-hoc tests, but two means with different subscripts are statistically distinguishable. Sample sizes for each relationship are child (40), spouse/romantic partner (245), subordinate at work (7), boss at work (28), coworker (14), parent (123), and friend (144). TCP IN RELATIONSHIPS 24