Relationship Churning, Physical Violence, and Verbal Abuse in Young Adult Relationships

Abstract

Young adults’ romantic relationships are often unstable, commonly including breakup-reconcile patterns. From the developmental perspective of emerging adulthood exploration, such relationship “churning” is expected; however, minor conflicts are more common in churning relationships. Using TARS data (N = 792), we test whether relationship churning is associated with more serious conflict: physical violence and verbal abuse. Those who are stably broken up (breakup only – no reconciliation) are similar to those who are stably together in their conflict experiences. In contrast, churners (i.e., those involved in on-off relationships) are twice as likely as those who are stably together or stably broken up to report physical violence and half again as likely to report the presence of verbal abuse in their relationships; this association between churning and conflict holds net of a host of demographic, personal, and relationship characteristics. These findings have implications for better understanding unhealthy relationship behaviors.

Young adults’ romantic relationships are quite fluid and often unstable, with more than four in ten young adults experiencing both a breakup and reconciliation in their present or most recent relationship (Dailey, Pfiester, Jin, Beck, & Clark, 2009; Halpern-Meekin, Manning, Giordano, & Longmore, 2010). From the developmental perspective of emerging adulthood, this kind of “churning” in relationships would be expected, as these years are meant to be spent exploring one’s identity and intimate relationships (Arnett, 2000, 2007a, 2007b).

In this study we ask whether relationship churning is associated with physical violence and verbal abuse. We examine this potential association because there are indications that minor conflicts, like arguing, are more common in relationships characterized by this churning pattern (Dailey et al., 2009; Dailey, Rossetto, Pfiester, & Surra, 2009; Halpern-Meekin et al., 2010). However, it remains unclear whether more serious forms of conflict, namely physical and verbal abuse, are also related to the churning pattern. We do not posit a particular causal relationship between churning and abuse; rather, we argue that both arise as part of a package of dynamics centered on an inability to properly manage conflict and prevent escalation in the context of a relationship with perceived benefits, such as bonds formed through intimate self-disclosure.

Background

Emerging Adulthood

The period of emerging adulthood, extending from the late teens to the mid-to-late twenties, has been defined as a time of exploration, with young people relatively free of both the rules of childhood and the responsibilities of adulthood. During this time in life romantic relationships begin to take center stage, as young adults figure out what they want in a long-term partner and who will fill this role (Arnett, 2000; Meeus, Branje, van der Valk, & de Wied, 2007). This allows young people to develop the relationship skills to successfully take on such long-term commitments (Arnett, 2007a, 2007b). However, the norm is not for young people to settle quickly into committed relationships (Settersen, Furstenberg, & Rumbaut, 2005); rather, emerging adulthood is characterized as a time of instability and exploration, particularly when it comes to intimate relationships (Arnett, 2005). From the perspective of the emerging adulthood literature, relationship instability is appropriate. We may question this assumption, however, if relationship instability, like churning, is associated with physical violence and verbal abuse.

Relationship Churning

Recently researchers have begun to explore various forms of instability in young adult relationships beyond breaking up. Dailey and colleagues (Dailey et al., 2009a, 2009b; Dailey, Hampel, & Roberts, 2010; Dailey, Jin, Pfiester, & Beck, 2011; Dailey, Middleton, & Green, 2012) have found that among college students, compared with stable relationships, on/off relationships were characterized by longer durations, but less satisfaction, commitment, and passion. One factor underlying the lower quality and unstable nature of these relationships was the presence of conflict (such as problems with communication), and the lower likelihood of engaging in relationship maintenance behaviors, like being cooperative, patient, and polite during discussions. Using a broader sample of young adults, Halpern-Meekin et al. (2010) showed that churning relationships were more likely to be characterized by arguing and lower commitment, but also by positive features of the union, such as intimate self-disclosure among partners, perhaps suggesting why couples might separate and reunite. Based on the association between churning and negative relationship qualities, we hypothesize that relationship churning is associated with more major forms of conflict, such as physical violence and verbal abuse.

Conflict

Intimate partner violence is relatively frequent among young adults (Brown & Bulanda, 2008), with estimates ranging from one quarter to more than half of men and women reporting aggressive behavior towards their partners, with mutual violence being most common (Giordano, Soto, Manning, & Longmore, 2010; Whitaker, Haileyesus, Swahn, & Saltzman, 2007).

Family conflict theory argues that a process of escalation leads from verbal abuse to physical violence (Holtzworth-Monroe & Stuart, 1994; Whitaker et al., 2007). Previous research postulated that situational couple violence was the result of poorly managed relational conflict, with couples unable to stem their negative escalation behaviors (Follette and Alexander, 1992; Riggs, O’Leary, and Breslin, 1990). Those with weaker social skills, particularly intimate relationship skills, are at greater risk of conflict escalating to the point of violence, especially during times of stress. Thus, problematic conflict typically is part of a package of troubled relationship behaviors, arising from a process of escalation. For example, those who report intimate partner violence are more likely to describe relationships marked by jealousy and verbal conflict (Giordano et al., 2010). A couple with poor conflict management skills may see small issues turn into big fights, escalating to verbal and physical lashing out.

Churning and Conflict

As is the case with relationship conflict, the instability of churning relationships could arise from a process of escalation and poor relationship skills, as opposed to developmentally appropriate relationship exploration. Previous research on married couples indicated that partners’ abilities to cope with relationship problems were predictive of their later divorce risk (Rogge & Bradbury, 1999). Couples with stronger communication skills are better able to negotiate stressful life events, even emerging feeling closer to one another, rather than seeing their unions erode under pressure (Freedman, Low, Markman, & Stanley, 2002). Couples may break up during a fight or times of stress, with partners lacking the skills to work through their problems or using exit as a bargaining chip to get what they want, yet reunite because breaking up was not actually their desired outcome. Churners report significantly higher levels of intimate self-disclosure and longer-lasting unions than those in stable relationships (Halpern-Meekin, et al., 2010); that is, the perceived benefits of the relationship may draw them back together, creating churning despite the presence of conflict.

Previous research has detailed the common occurrence of situational couple violence in community samples (Halpern, et al., 2001; Johnson and Ferraro, 2000); however, we know less about the relationship context in which these incidents occur (Giordano, et al., 2010). This research has not yet adequately recognized the fluidity of relationship status for many couples, focusing instead either on violence within relationships or around the stay/leave decision (c.f., Anderson, 2007; Anderson & Saunders, 2003; Choice & Lamke, 1997). In the present study we propose that couple violence may, for some couples, be part of a more general inability to manage relational conflict in a healthy manner. However, even if the frequency of churning was similar among those with and without conflict in their relationships, the churning would still be of particular concern among those who engage in conflict since they are returning to relationships that can carry serious consequences.

There are a range of other factors that may mediate the association between churning and conflict. Parental education and childhood family structure are associated with later relationship violence experiences (Foshee, Bauman, & Linder, 1999; Halpern, et al., 2001; Heyman & Slep 2002; Simons, Lin, & Gordon, 1998), and age, gender, and race are also associated with the risks of exposure to intimate partner violence (Brown & Bulanda, 2008; Tjaden & Thoennes, 2000). Negative parent-child interactions and violence in the home as a child are predictive of later couple violence and conflict management skills (Kim, Conger, Lorenz, & Elder, 2001; Linder & Collins, 2005), as is aggressive behavior in childhood (Loeber & Hay, 1997). Self-esteem can mediate people’s experiences and perceptions of their intimate relationships (Longmore & DeMaris, 1997) and is related to engagement in aggressive behaviors (Baumeister, Smart, & Boden, 1996); sense of control is key to one’s ability to handle stressors, such as relationship problems (Mirowsky & Ross, 2003). Finally, relationships marked by churning and conflict are distinctive in both their positive and negative characteristics (Dailey et al., 2009a, 2010; Giordano et al., 2010; Halpern-Meekin et al., 2010).

Given indications in previous research that relationship churning and conflict may be related, and that both may result from processes of escalation and an inability to regulate conflict, we ask: (1) what is the frequency of physical conflict and verbal abuse among those who are stably together, stably broken up, and churners, and (2) is the association between conflict and churning robust to the inclusion of controls for demographic, background, and relationship characteristics? Answering these two questions gives us insight into the nature of on/off relationships, the relational context of intimate partner violence and verbal abuse, and the characteristics of romantic relationships in emerging adulthood.

Method

Data

The Toledo Adolescent Relationships Study (TARS) is a stratified, random sample of 1,321 students registered for the 7^th, 9^th, and 11^th grades in Lucas County, Ohio, a metropolitan area largely consisting of the city of Toledo. Incorporating oversamples of black and Hispanic youths, the initial sample was devised by the National Opinion Research Center and was drawn from the enrollment records of 62 schools from seven school districts. Respondents completed interview questionnaires at home using laptop computers, and school attendance was not a requirement for inclusion in the sample; in addition, parents completed questionnaires about their respondent children at Wave 1. The retention rate from Wave 1 to Wave 4 of the TARS sample was 82.8%. The attrition was due largely to our inability to locate the respondents as well as two respondents who passed away since the initial interview. Only nine respondents refused to participate in the follow-up survey. Attrition analyses indicated that participation at Wave 4 was not related to most characteristics assessed at Wave 1. However, the follow-up sample was slightly more likely to be female and slightly younger (average age at Wave 1 of 15.2 for those who participated at Wave 4 compared to an average age of 15.3 for the full Wave 1 sample). Census data indicated that this sample shared similar sociodemographic characteristics with the Toledo metropolitan area in terms of education, median family income, marital status, and racial distribution; in addition, the Toledo metropolitan area was similar to the nation as a whole in terms of sociodemographic composition along the lines of race, education, median family income, and marital status (Center for Family and Demographic Research, 2011).

In the present study, we relied on the data from Wave 4 respondents who were interviewed in 2006 when they were 17-24 years old. Our data were cross-sectional; as such, our goal was to explore the association between relationship churning and conflict, not to make causal statements about the relationship between the two.

The analytic sample was comprised of those who were currently or had recently (within the last two years) been in a dating or cohabiting relationship (N=792). In TARS, there were 1,092 respondents in the wave 4 sample. We excluded all those who had not dated anyone in the past two years (remaining n=987; 90.4% of original wave 4 sample). We excluded all those who were married (n=65) or who were only interested in same-sex romantic relationships (n=8) (remaining n=914; 83.7% of original wave 4 sample). Finally, we excluded those who were missing information about the number of break-ups they experienced in their present or most recent relationship (remaining n=792; 72.5% of original wave 4 sample). The analytic sample is similar to the entire Wave 4 sample, although the analytic sample is slightly more male (50.34% versus 46.98%), more black (25.55% versus 23.72%), and less Hispanic (3.41% versus 5.59%). Respondents reported on their current or most recent focal relationship. We chose to include both those reporting on current and previous relationships because prior research had shown that the boundaries defining the end of a relationship were quite fluid (Dailey et al., 2009a; Halpern-Meekin et al., 2010).

Measures

We focused on two primary variables of interest, relationship churning and conflict. For those reporting on a current relationship, we coded respondents as having experienced churning (that is, a disruption followed by a reunion) if they reported ever having broken up with this partner. For those who were reporting on a previous relationship, we coded respondents as having experienced churning if they reported having broken up with their ex more than once. Respondents were coded as stably together if they were currently in a relationship and had never broken up with this partner. Respondents were coded as stably broken up if they were reporting on a previous relationship and only broke up once with this partner.

We considered two types of conflict behaviors in the present study, physical violence and verbal abuse; both represent conflict behaviors that have escalated to problematic levels. Our conflict measures examined physical violence and verbal abuse as characteristics of relationships, as opposed to focusing on which partner acts as perpetrator and victim. The standard we used for identifying the presence of conflict for each measure was if the respondent reported these behaviors having ever occurred (regardless of who was victim/perpetrator); only those who reported that none of the verbal abuse and none of the physical violence behaviors had occurred in their relationship received a 0 on the respective conflict measures (1 = any physical violence/verbal abuse, 0 = none). To capture verbal abuse respondents were asked two questions: first, “During this relationship, how many times has [boy/girlfriend]: ridiculed or criticized your values or beliefs; put down your physical appearance; and put you down in front of other people.” Second, “During this relationship, how many times have you: ridiculed or criticized [boy/girlfriend’s] values or beliefs; put down his/her physical appearance; and put him/her down in front of other people” (α = 0.84). Our measures of physical violence also asked the respondent to report both on being a victim and being a perpetrator. First, “During this relationship, how many times has [boy/girlfriend]: thrown something at you; pushed, shoved, or grabbed you; slapped you in the face or the head with an open hand; and hit you.” The second set of questions asked “During this relationship, how many times have you: thrown something at [boy/girlfriend]; pushed, shoved, or grabbed him/her; slapped him/her in the face or the head with an open hand; and hit him/her” (α = 0.89).

The individual indicators included gender (male = 1), respondent’s age, and race/ethnicity (White, Black, Hispanic, and other/mixed race). The family structure the respondent lived in as a teenager at Wave 1 was based on a four-category measure (two-parent, single-parent, stepparent, or an alternative arrangement with no parents). Because many young adults have not completed their educations or launched their careers, we used the respondent’s parent’s level of education as a proxy for family socioeconomic status (parents were classified as having less than a high school degree, a high school degree, some college, or a college degree); this was measured by parental self-reports at Wave 1.

We made use of the parent survey, collected at Wave 1, to control for previous experiences with conflict. The conflict history scale – parent included two measures, with the parent reporting how often in the previous month she threatened to physically hurt her child and how often she pushed, grabbed, slapped, or hit her child; answer options ranged from never to every day (possible score range = 2-10; correlation = 0.54). The conflict history scale – child included two measures, with the parent reporting how often in the previous month her child threatened to physically hurt her and how often her child pushed, grabbed, slapped or hit her; answer options ranged from never to every day (possible score range = 2-10; correlation = 0.58). Higher scores on both conflict history scales indicated more frequent conflict. In addition, since we theorized that an inability to manage conflict may underlie both the experiences of conflict and relationship churning, we included a measure of adolescent temper as reported by the parent at Wave 1, indicating how easily the child lost his or her temper on a five-point scale, with higher scores indicating a more volatile temper.

We used the respondent’s report of the length of the intimate relationship to capture duration; responses ranged from 1-8, from less than a week to a year or more. A relationship status measure indicated whether the respondent was in a cohabiting or dating relationship. We included two social psychological measures, self-esteem and sense of control. The measure of self-esteem was a scale of six items (possible score range = 6-30; α = 0.75): “I am able to do things as well as other people;” I feel that I have a number of good qualities;” “I feel I do not have much to be proud of” (reverse coded); “At times I think I am no good at all” (reverse coded); “I feel that I am a person of worth, at least on an equal basis with others;” and “I take a positive attitude toward myself.” Sense of control was constructed following Mirowsky and Ross’ (1990) formulation; this is not a simple scale, therefore no alpha value is available (see Mirowsky & Ross, 1990, for further information). Respondents rated their agreement (strongly disagree to strongly agree) to the following items: “I can do just about anything I really set my mind to;” “I have little control over the bad things that happen to me;” “My misfortunes are the result of mistakes I have made;” “I am responsible for my failures;” “The really good things that happen to me are mostly luck;” “There’s no sense in planning a lot – if something good is going to happen it will;” “Most of my problems are due to bad breaks;” and “I am responsible for my own success.” A higher score indicated a stronger sense of personal control.

We examined both negative and positive aspects of the relationship. A scale of two items captured receiving validation from a partner: partner makes me feel attractive and partner makes me feel good about myself (possible score range = 2-10; correlation = .68); a higher score indicated receiving more validation. Intimate self-disclosure was measured by how often (never to very often) respondents reported talking about the following topics with their partners: something really bad that happened; home life and family; private thoughts and feelings; and the future (possible score range = 4-20; α = .91); a higher score indicated more frequent disclosure. Love was a measure of how strongly respondents reported loving their partners on a five-point scale, with higher scores indicating stronger feelings of love. Commitment was measured by how strongly on a five-point scale respondents agreed that they “may not want to be with [partner] a few years from now” for those currently in a relationship or that they “didn’t want to be with [partner] long term” for those reporting on an ended relationship; we coded the variable so that a higher score indicated stronger commitment. Communication skills was a self-reported measure on a five-point scale, indicating how strongly respondents agreed that they and their partner had “the communication skills a couple needs to make a relationship work.” Mistrust of partner was measured on a five-point scale by how strongly respondents agreed that there were times when their partners could not be trusted. We measured need asymmetry on a five-point scale based on the strength of agreement with the statement “I need [my partner] more than she/he needs me.” Asymmetry in doing for partner was measured by the strength of agreement on a five-point scale that “I do more for [my partner] than she/he does for me.” Relationship alternatives captured the respondents’ beliefs that they could find another partner who was as good as the focal partner; it was measured on a five-point scale.

Analytic Strategy

We first provide the descriptive characteristics of the sample and then describe the extent to which the two types of conflict are experienced among those who are stably together, stably broken up, and churners. We next present logistic regressions that predict physical violence and verbal abuse experience. Model 1 predicts conflict experiences by relationship type (stably together, stably broken up, and churners); Model 2 adds the demographic and background controls; and Model 3 adds the relationship characteristic controls. We do not assume a causal relationship between the independent and dependent variables because our theory does not assert there necessarily is one; in addition, the cross-sectional nature of our data means the measures of relationship status and relationship characteristics are potentially endogenous to the relationship’s churning and conflict status. Therefore, the analyses allow us to test our hypothesis that churning and conflict are associated, net of demographic, social psychological, and relationship quality characteristics.

Results

Table 1 displays the descriptive characteristics of the sample. Approximately forty percent of respondents reported physical violence and just over half reported verbal abuse, with nearly one-third reporting both types of conflict (not shown). The most common relationship status was churning, with 44.16% having experienced churning in their present or most recent relationship, followed by 40.97% who were stably together, and 14.87% who were stably broken up. Present churners made up 69.3% of the total population of churners (results not shown).

Table 1.

Demographic and relationship characteristics of young adult daters and cohabitors (% or M) (N = 792)

Variables	% or M (SD)
Physical conflict	39.84%
Verbal abuse	52.51%
Stably together	40.97%
Stably broken up	14.87%
Churning	44.16%
Age	20.300 (0.061)
Male	50.34%
Female	49.66%
Race
White	65.43%
Black	25.55%
Hispanic	3.41%
Other race	5.47%
Family structure
Two parent	49.53%
Single parent	26.26%
Stepparent	18.06%
Other living arrangement	5.06%
Parental education
Less than high school degree	11.38%
High school degree	30.98%
Some college	32.90%
College or more	23.26%
Conflict history scale – parent	2.477 (0.062)
Conflict history scale – child	2.188 (0.048)
Adolescent temper	2.816 (0.054)
Cohabitation	23.69%
Dating	76.31%
Relationship duration	6.746 (0.060)
Self-esteem	24.450 (0.126)
Sense of control	0.652 (0.015)
Intimate self-disclosure	15.612 (0.134)
Love	4.175 (0.041)
Validation	8.035 (0.057)
Commitment	3.642 (0.043)
Asymmetry – need	2.308 (0.035)
Asymmetry – do for partner	2.411 (0.039)
Relationship alternatives	2.820 (0.043)
Communication skills	3.767 (0.036)
Mistrust of partner	2.200 (0.043)

Note: Data are weighted to reflect population of Toledo.

Table 2 shows how common experiences with conflict were for each relationship status group. Approximately one-quarter of those who were stably together and stably broken up reported physical conflict experience, compared to over half of churners did so. Verbal abuse was more common than physical conflict across relationship groups, and was more frequent among churners than those who were stably together or broken up (64% versus 42-45%). Fewer than one in five respondents who were stably together or stably broken up reported experiencing both types of conflict, compared to nearly half of churners who did so. The similarities in conflict experiences between the stably together and the stably broken up is notable, as is the far more common conflict experience among churners.

Table 2.

Conflict experience by relationship types (N = 792)

	Physical conflict	Verbal abuse	Both physical & verbal conflict
Stably together	26.85%	42.48%	19.75%
Stably broken up	26.55%	45.13%	18.58%
Churning	57.18%	64.23%	48.17%

To further clarify these results, we re-examined the distributions in Table 2 with the conflict measures divided into mutual conflict (both partners act as perpetrators and victims), female-victim-only conflict, and male-victim-only conflict (results not shown). The results for physical violence indicated that churners were more likely than the other two groups to engage in mutual, female-victim-only, and male-victim-only conflict. Churners were also more likely than the other two groups to engage in mutual verbal abuse, but were not more likely to engage in female- or male-victim-only conflict. Therefore, variation between churners and the other groups for verbal conflict in Table 2 appeared to be driven by those who were engaging in mutual abuse.

Table 3 presents the results of the logistic regressions predicting physical and verbal abuse experiences. In Model 1, predicting physical violence, churners were significantly more likely to report physical violence compared to those who were stably together. In Model 2, including demographic and background controls, the significantly greater odds of physical conflict experience for churners remained. In addition, male respondents, those raised in stepparent households, those whose parents did not graduate high school, those whose parents threatened or engaged in more violence towards them, and those with worse tempers in adolescence were more likely to report physical conflict experience. In Model 3, including relationship characteristic controls, the size of the odds ratio for churners was smaller, but remained significant and substantively large. Men, those whose parents did not graduate high school, and those with worse adolescent tempers remained more likely to report physical conflict. Longer relationship duration, greater asymmetry in doing things for one’s partner, and partner mistrust were associated with an increased likelihood of physical conflict, and those with higher self-esteem and sense of control were less likely to report physical conflict.

Table 3.

Logistic regressions predicting physical conflict and verbal abuse (N = 792)

			Physical conflict						Verbal abuse
	Model 1		Model 2		Model 3		Model 1		Model 2		Model 3
	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE	OR	SE
Stably broken  up	0.98	0.24	0.93	0.24	0.76	0.23	1.12	0.25	1.13	0.26	0.77	0.21
Churning	3.64***	0.60	3.27***	0.56	1.99***	0.39	2.45***	0.39	2.46***	0.41	1.42†	0.27
Age			1.02	0.05	0.98	0.05			1.08†	0.05	1.05	0.05
Male			1.31†	0.21	1.57*	0.29			1.48**	0.22	1.63**	0.28
Black			1.35	0.28	1.32	0.30			0.85	0.17	0.72	0.16
Hispanic			1.66	0.56	1.74	0.64			1.17	0.38	1.23	0.43
Other race			1.58	0.53	1.72	0.62			1.11	0.37	1.14	0.40
Single Parent			0.97	0.20	0.82	0.18			0.94	0.18	0.80	0.16
Stepparent			1.63*	0.35	1.44	0.33			1.38	0.29	1.26	0.28
Other living  arrangement			1.12	0.42	1.08	0.43			0.82	0.30	0.82	0.32
Parent – less  than high  school			1.84*	0.48	1.81*	0.52			1.40	0.36	1.38	0.39
Parent – some  college			1.01	0.19	0.96	0.20			1.05	0.19	1.01	0.20
Parent – college  or more			0.94	0.21	1.08	0.26			1.01	0.21	1.15	0.26
Conflict history  scale – parent			1.23†	0.15	1.21	0.16			0.93	0.11	0.87	0.11
Conflict history  scale – child			0.96	0.22	0.90	0.22			1.28	0.34	1.31	0.37
Adolescent  temper			1.18*	0.08	1.13†	0.08			1.13†	0.07	1.09	0.07
Cohabitation					1.39	0.30					1.38	0.28
Relationship  duration					1.19*	0.08					1.15*	0.07
Self-esteem					0.95†	0.03					0.92**	0.03
Sense of control					0.62*	0.14					1.06	0.23
Intimate self-  disclosure					1.01	0.03					1.01	0.23
Love					1.11	0.13					1.08	0.12
Validation					1.02	0.06					0.96	0.06
Commitment					0.92	0.08					0.79**	0.07
Asymmetry–  need					1.14	0.11					1.03	0.10
Asymmetry–  do for partner					1.18†	0.11					1.09	0.10
Relationship  alternatives					1.12	0.10					1.18*	0.10
Communication  skills					0.90	0.10					0.98	0.10
Mistrust of  partner					1.41***	0.12					1.45***	0.12
Pseudo R-  squared	0.07		0.11		0.20		0.03		0.06		0.14

Note: reference categories: stably together, female, White, two-parent family, parent – high school degree, dating

^†p<.10.

^*p < .05.

^**p < .01.

^***p < .001

In predicting verbal abuse in Model 1, churners were significantly more likely to report this type of conflict compared to those who were stably together. This association remained in Model 2, net of the demographic and background controls. In addition, being older, male, and having a worse temper in adolescence were associated with an increased likelihood of verbal abuse experience. In Model 3, the association between churning and verbal abuse remained marginally significant, but was mediated somewhat with the addition of relationship characteristic controls. Being male, longer relationship duration, believing in the availability of relationship alternatives, and partner mistrust were associated with an increased likelihood of verbal abuse experience, whereas self-esteem and commitment were associated with a lower likelihood of this kind of conflict.

We re-ran Models 1-3 for both abuse measures with separate independent variables for being a past churner (30.7% of churners) and a present churner (69.3%) to be sure that one group of churners was not driving our results (results not shown). The results were the same for both groups of churners; they were significantly more likely to report both types of conflict relative to those in stable relationships. The only exception to this was for Model 3 predicting verbal conflict; although the association between churning in a past relationship and reporting verbal abuse was marginally significant, the association between being in a present relationship and verbal abuse was no longer significant. These results indicated that the association between churning and conflict was not driven exclusively by either past or present churners.

Conclusion

In the present sample, approximately 4 in 10 unmarried young adults experienced relationship churning, 4 in 10 experienced physical violence, and 5 in 10 experienced verbal abuse in their present or most recent relationship; this likely underestimates the cumulative experience of these events across each respondent’s romantic relationships during this life stage, meaning these experiences of relationship instability and conflict are quite common. Notably, these experiences are likely to occur in conjunction with one another; relationships marked by churning are more likely than others to include physical violence and verbal abuse, net of a host of demographic, personal, and relationship characteristics. These results support our theory that both dynamics arise due to an inability to properly prevent and manage conflict escalation.

Churners are twice as likely as those who are stably together or stably broken up to report physical violence occurring in their relationships, and half again as likely to report the presence of verbal abuse in their relationships. Importantly, we find that the people who are experiencing the most conflict are not those who are choosing to permanently exit their relationships. Those who are stably broken up are remarkably similar to those who are stably together in their conflict experiences; churners are far more likely to report these conflict experiences than those who broke up (and stayed that way). The fact that the association between churning and verbal abuse in Model 3 is reduced to marginal significance is likely driven by the similarities between all three relationship groups in the incidence of female-victim-only and male-victim-only conflict; because the descriptive difference between churners and non-churners in terms of verbal abuse is driven by their higher incidence of mutual conflict, we see the relationship group differences minimized in the model including measures of other relationship characteristics.

From the emerging adulthood perspective the instability of relationship churning would be expected (Arnett, 2000, 2007a, 2007b), but that churning is likely to go hand-in-hand with conflict should raise a red flag about such instability being a developmental norm. Although instability may be expected and common in young adult relationships, such churning behavior could actually be an indicator of distress rather than age-appropriate relationship development.

Widespread experiences with physical conflict and verbal abuse are inherently problematic, but there is also an array of reasons to be concerned about the churning pattern we have focused on in the current investigation. Past studies found that arguing was more common in relationships marked by churning (Dailey et al. 2009a; Dailey et al., 2009b; Halpern-Meekin et al., 2010). The present findings show that the differences in conflict between churning and stable relationships may be far more problematic than simply a greater likelihood of arguing; churning is strongly associated with an increased likelihood of physical violence and verbal abuse experiences, and there are reasons to believe that this bundle of relationship behaviors may be indicative of an inability to properly manage conflict and prevent escalation. Establishing this association between conflict and churning is essential for those studying the relational dynamics around intimate partner violence; stay/leave decisions may be far more complex than they appear at first glance due to the impermanence, or churning, around such decisions. The post-breakup reunification may be driven by the positive features of the relationship, which can be present alongside churning or conflict (Giordano, et al., 2010; Halpern-Meekin, et al., 2010).

Future research should replicate this study with a nationally representative sample, as opposed to the regional sample used in the present project, and could include comparisons to those in marital relationships. A longitudinal data set with shorter time periods between waves – enabling the collection of multiple observations of these relatively short-lived romances -- would enable testing for a causal relationship between churning and conflict. Further, additional research should examine whether the association between churning and conflict in nonmarital relationships is unique to this particular period in the life course, or is also found in adolescence, middle adulthood, and older adulthood. The association between having a relationship marked by churning and conflict in young adulthood and later union outcomes should also be explored, as there are indications that early experiences of conflict and churning are predictive of such experiences in the future (Halpern et al., 2001; O’Leary & Slep, 2003; Wineberg, 1999).

Rather than being developmentally appropriate, instability in emerging adulthood may be predictive of problematic romantic relationships in both the short and long term. The present findings contribute to our understanding of romantic relationships in emerging adulthood, the association between relationship instability and intimate partner violence, and the nature of relationship instability itself; results underline the importance of distinguishing those who are stably broken up from churners, as their relationship experiences, at least in terms of physical violence and verbal abuse, are distinctive. The dynamics of a relationship marked by both churning and conflict seem to be volatile and complex, which has implications for both practitioners and researchers interested in understanding and supporting healthy relationships and their development in young adulthood.