Charting Changes in Commitment: Trajectories of On-again/Off-again Relationships

Article

Charting changes incommitment:Trajectories of on-again/off-again relationships

Rene M. DaileyNicholas BrodyLeah LeFebvreBrittani CrookUniversity of Texas at Austin, USA

AbstractThe present study examined how turning points reported by individuals in on-again/off-again (on–off) relationships reflected relationship trajectories. Participants (N ! 581)completed an online Retrospective Interview Technique asking them to report on upto 10 turning points. Participants indicated their commitment level at each turning point.Based on the variations in commitment across turning points, five trajectories emerged.Trajectory groups were compared regarding relational stability factors. Results suggestthat on–off partners with a steady-low commitment trajectory reported less stabilitythan individuals with steady-high commitment. Additionally, partners in the fluctuatingtrajectory, which would seemingly represent less stability, reported moderate percep-tions of their relationships, faring better than the low-steady commitment group. Over-all, findings add to an understanding of how to best characterize on–off relationships.

KeywordsBreakups, commitment, on-again/off-again relationships, relational trajectories, relationaltransitions, relational stability, renewals, turning points

Corresponding author:Rene M. Dailey, Department of Communication Studies, University of Texas at Austin, 1 University StationA1105, Austin, TX 78712-0115, USA.Email: [email protected]

J S P R

Journal of Social andPersonal Relationships

30(8) 1020–1044ª The Author(s) 2013

Reprints and permissions:sagepub.co.uk/journalsPermissions.nav

DOI: 10.1177/0265407513480284spr.sagepub.com

http://spr.sagepub.com

http://www.sagepub.co.uk/journalsPermissions.nav

A major goal in understanding different paths or trajectories of romantic relationships isto better predict and explain the dynamics as well as the ultimate outcomes of rela-tionships (i.e., whether couples will stay together or break up). This is especiallyimportant in on-again/off-again (on–off) relationships given that defining or measuringrelational stability is more difficult (e.g., even if a couple is broken up, the relationshipmay not be permanently terminated). Hence, the trajectory or progression of the relation-ship may be more revealing about the nature of the relationship as well as the predictiveof the final outcomes than overarching relational characteristics (e.g., trust, satisfaction)assessed at one time. Further, given that two thirds of adults have experienced an on–offrelationship and one third are currently in one (Dailey, 2012) as well as that these rela-tionships vary in dynamics and relational quality (Dailey, McCracken, Jin, Rossetto, &Green, in press; Dailey, Middleton, & Green, 2012; Dailey, Pfiester, Jin, Beck, & Clark,2009), it is important to identify the developmental trajectories of these relationships tobetter predict which couples will achieve relative stability, permanently dissolve, andcontinue to cycle. Research regarding on–off relationships is moving toward this goal(Dailey, Middleton, et al., 2012) by identifying the types of on–off relationships (Daileyet al., in press) and how dimensions based on these types were associated with relationaldynamics (Dailey, Jin, Brody, & McCracken, 2012).

Previous on–off research has, however, restricted analyses of turning points tobreakups and renewals (e.g., Dailey et al., in press). Greater insights can be gained byassessing all turning points the partners perceive in their relationship (Baxter & Bullis,1986; Mongeau & Teubner, 2002), and on–off relationships may encompass certainunique turning points that other relationships do not experience. Additionally, assessingon–off trajectories as well as differences in these trajectories may reveal greater nuancesabout the nature of these relationships and their ultimate outcomes. Hence, the currentresearch more broadly assesses all turning points that the partners identify and how theseturning points reflect different trajectories in on–off relationships.

Turning points

Turning points perspective is a process-oriented view of relationships—relationshipsprogress and regress over time through critical events and experiences. Baxter and Bullis(1986) defined a turning point as ‘‘any event or occurrence that is associated with changein a relationship’’ (p. 470). Given the change or an opportunity for change, turning pointsare also conceptualized as transformative events (Conville, 1987) and as such are typi-cally assessed as changes in commitment to the relationship or chance of marriage (e.g.,Huston, Surra, Fitzgerald, & Cate, 1981; Koenig Kellas, Bean, Cunningham, & Cheng,2008; Surra 1985, 1987). Given this definition, breakups and renewals are likely the sig-nificant turning points in on–off relationships (Dailey et al., in press) as transformationor change is implied in these relational transitions. Yet, even relational transitions likelyinvolve multiple turning points (Mongeau & Teubner, 2002). Further, similar to otherdating relationships, there are likely many other types of turning points that on–off part-ners experience such as quality time or physical separation found in Baxter and Bullis(1986) or sexual encounters and waning feelings as found in Koenig Kellas et al.(2008). Hence, assessing all turning points that on–off partners perceive throughout their

Dailey et al. 1021

relationship would provide a richer picture of on–off relationship trajectories. As such,our first research question is:

RQ1: What turning points do partners in on–off relationships report?

Relationship trajectories

A larger goal of research examining turning points is identifying the types of relationshiptrajectories (Huston et al., 1981; Koenig Kellas et al., 2008; Surra 1985, 1987). Based onthe holistic impressions of the graphical representations of turning points over time orspecific qualities of the graphs (e.g., size of increases or decreases in commitment as wellas consistency of the direction of change), researchers have extracted certain trajectories.For example, Huston et al. and Surra identified four trajectories from the reports ofmarital partners on their dating relationships: accelerated (i.e., rapid progression to highprobability of marriage), accelerated–arrested (i.e., accelerated with a point of backwardprogression), intermediate (i.e., moderate progression), and prolonged (i.e., slower andfluctuating path) trajectories. Surra and colleagues also identified two types of datingrelationships: relationship-driven (i.e., steady and gradual increases in commitment) andevent-driven (i.e., more extreme fluctuations in commitment) (Surra & Gray, 2000;Surra & Hughes, 1997). Thus, we add the following research question:

RQ2: What are the different relational trajectories of on–off relationships?

Describing the trajectories

Beyond identifying trajectories of on–off relationships, we sought to understand howdifferent groups vary regarding a variety of dynamics associated with romantic rela-tionships. Ultimately, this will aid in predicting the outcomes of these relationships andwhether the couple reaches relative stability, permanently dissolves the relationship, orcontinues to cycle between being together and broken up. Because trajectories aretypically based on changes in commitment or chance for a long-term relationship (e.g.,marriage), they are essentially centered on relationship stability. Hence, factors associ-ated with relational stability should reveal the greatest differences in the trajectories andprovide greater understanding of the different paths on–off partners traverse.

Based on the reviews of relational stability research (e.g., Cate, Levin, & Richmond,2002; Le, Dove, Agnew, Korn, & Mutso, 2010), the turning points literature (Baxter &Bullis, 1986), and the factors that have been associated with different types of on–offrelationships (Dailey et al., in press), we chose multiple dimensions of romantic rela-tionships to assess in relation to the identified trajectories. We focus on several generaldimensions: relationship dynamics, which are perceptions, states, or feelings about therelationship (e.g., closeness, trust, love, and sexual satisfaction); communicationdynamics, which are behaviors exhibited within the relationship (e.g., affection, open-ness, and conflict); and structural factors of the relationship (e.g., the ratio of relationshiplength to turning points). The specific relationship and communication dynamicsselected were stronger or more consistent predictors of stability across the research

1022 Journal of Social and Personal Relationships 30(8)

(excluding commitment) or features that have emerged as significant in on–off rela-tionships. Furthermore, we specifically highlight communication factors (compare Cateet al. and Le et al. who include these under relationship or dyadic factors) as different typesof on–off relationships may vary in how they communicate but have similar feelings aboutthe relationship (e.g., love and closeness). The specific structural characteristics includedblend factors that have been salient in previous research (e.g., proportion of negativeturning points; Baxter & Bullis, 1986) and features unique to on–off relationships enablingan examination of characteristics such as the proportion of turning points that are breakupsand renewals and ratio of relationship length to renewals. Combined, these factors shouldprovide a fuller understanding of how relational trajectories might vary. Importantly,theory building will differ if we find the trajectories varying primarily by relationshipdynamics, for example, as compared to varying on all these dimensions.

Previous research does suggest that different groups of on–off partners vary in theirreports of factors similar to those based on the perceptions of their relationship’s stability(Dailey, Middleton, et al., 2012). For example, those who perceived greater stability intheir on–off relationship reported more maintenance and positive communication pat-terns, those who perceived their relationship as permanently terminated reported lessmaintenance and more negative communication, and those who anticipated more rela-tional transitions (i.e., greater instability) reported moderate levels of these factors.Instead of grouping on–off partners based on their perceived stability, the current studywill assess how different trajectories vary with regard to the factors associated withstability. We thus add the following research question:

RQ3: How do the on–off trajectories differ with regard to the factors associated withrelational stability?

Triangulating types of on–off relationships

Establishing trajectories is one means of understanding how on–off relationships differ.With a similar goal, Dailey et al. (in press) extracted five types of on–off relationshipsfrom how individuals reported negotiating the transitions across their relationship. Thehabitual type reflected relationships in which partners tended to fall back into therelationship without much negotiation regarding the transitions. These partners renewedtheir relationships because it was relatively convenient and easy. For many, the on–offrelationship provided companionship and comfort until one or both of the partnersbecame interested in someone else. The mismatched type encompassed relationships inwhich the cyclical nature was due to the partners being mismatched regarding issuesinternal to their relationship (i.e., personalities and desires) or to external factors (e.g.,geographic distance and different schedules). For partners in the capitalized on transi-tions type, relational transitions were used to test their relationship, manage relationalproblems, or create opportunities to improve the relationship. This category representscouples in which the on–off nature facilitated positive change in the relationship,whether it was strategic or fortuitous. Partners in the gradual separators type eventuallyrealized that the relationship was not going to work or that they were no longer interestedin continuing the relationship. Their relationships gradually came to a more defined

Dailey et al. 1023

ending, and as such, these partners reported having more closure about the ending of therelationship. In the controlling partner type, one partner tended to be more in control ofthe progression of the relationship or more persistent in continuing the relationship. Theuse of manipulation and control tactics was common.

Finding overlaps between these types and commitment trajectories may reveal strongerdimensions or types of on–off relationships. For example, a recent examination of thetypes as dimensions revealed that capitalizing on transitions and gradual separation facetsof on–off relationships were more prominent in how they were associated with the aspectsof relational quality (Dailey, Jin, et al., 2012). Hence, exploring the association betweenthe previously established types and the trajectories may highlight factors that betterdistinguish on–off types. Thus, our final research question is:

RQ4: Are the on–off commitment trajectories related to the types of on–offrelationships?

Method

Participants

A total of 581 participants who were currently in or had recently experienced an on–offrelationship were recruited from the Amazon.com Mechanical Turk Web site.1 Parti-cipants were eligible for participation if they were currently in or had been in an on–offrelationship within the last 5 years (with ‘‘on–off’’ defined to participants as a rela-tionship that had been broken up and renewed at least once) and were at least 18 years ofage. The data were collected in two rounds of approximately equal numbers (n1 ! 298;n2 ! 283); the second round was initiated to increase the sample sizes of the trajectoriesidentified for analyses (see below). The two samples did not significantly differ in termsof sex, age, ethnicity, current status, or relationship length. Preliminary analyses alsoshowed that data collection round was not a significant covariate in any of the analysesand was thus excluded.

A majority of the sample was female (n ! 350, 60.2%), and participants’ ages rangedfrom 18 to 61 years (M ! 27.98, SD ! 9.03). Ethnicities of the sample included: 396(68.2%) Caucasian, 46 (7.9%) Asian or Pacific Islander, 46 (7.9%) Black or AfricanAmerican, 35 (6.0%) Latino or Hispanic, and 56 (9.6%) other or multiple ethnicities; twoparticipants declined to report ethnicity. Most participants reported having a hetero-sexual relationship (94.5%).

Length of the relationship ranged from less than 1 to 386 months (M ! 41.06,SD ! 33.70, Mdn ! 36). Participants reported reconciling an average of 2.44 times(SD ! 2.02, range was 1–12 times), excluding 12 participants who reported 20–100renewals. At the time of participation, 306 (52.7%) participants were currently datingtheir on–off partner (17 declined to report current status). About half of the sample hadcohabitated with their on–off partner at some point (46.5%); and at the time of partic-ipation, half of these cohabitors (54.1%) were currently living with their on–off partner.Having cohabitated did not significantly affect the results, and thus, this factor was notincluded as a control variable.


Procedure

Participants completed an online questionnaire by reporting on their current or mostrecent on–off relationship. Given that previous research using the RetrospectiveInterview Technique (RIT) showed participants reported 5–10 turning points on anaverage (e.g., Baxter & Bullis, 1986; Koenig Kellas et al., 2008; Surra, 1985), we askedthe participants to identify up to 10 turning points that occurred during their relationship.Turning points were defined to the participants as events that ‘‘change the level ofcommitment you have towards the relationship (i.e., the intent to maintain the rela-tionship), could be either positive or negative events, and can, but does not have toinclude breakups or renewals.’’ We asked the participants to report all the turning pointsin their relationship, regardless of whether they occurred during times they were togetheror times they were broken up. To capture on–off turning points, we employed a modifiedversion of the RIT (Huston et al., 1981; Koenig Kellas et al., 2008). The original RITasked the participants to graph turning points across the course of the relationship(usually measured in months on the horizontal axis) and to specify how the turning pointaffected their commitment to the relationship (usually measured by a percentage ofcommitment from 0 to 100% on the vertical axis) (Baxter & Bullis, 1986; Koenig Kellaset al., 2008). In the current study, we modified the RIT techniques to comply with an onlinecompletionwhere the participantswere asked, throughwritten instructions, to describe eachturning point. In attempting to make the online version as similar to the original RIT aspossible, we designed the survey to be interactive in that the graph of their commitmenttrajectory was updated after each turning point was reported (see Figure 1 for an example).As such, participants saw how their graphs changed with each additional turning point,similar to how an interviewer would modify their trajectory during the face-to-face methodof the RIT.

For each turning point, participants were asked to (1) label the turning point (e.g.,major conflict, moving in together, breakup), (2) indicate their commitment at that pointin the relationship on a scale of 0–100 (0 being no commitment and 100 being completecommitment2), (3) describe why the event resulted in a change in commitment to therelationship, and (4) answer a variety of quantitative items regarding the relationshipduring the time period in which the turning point occurred. Participants reported between1 and 10 turning points in their on–off relationships (M ! 2.94, SD ! 2.38).3 Given thaton–off relationships are unique in the experience of multiple relational transitions (i.e.,breakups and renewals), we calculated the proportion of turning points that were rela-tional transitions: one third of the turning points reported were a breakup or a renewal(M ! .33, SD ! .35).4

Turning point coding

We identified types of turning points with the first sample of data. All authors readthrough the turning point descriptions and independently developed categories. We thenmet to collapse, integrate, and finalize a coding scheme. In doing so, we discussed thesimilarities and differences among the categories, and referenced previously establishedturning point categorizations (Baxter & Bullis, 1986; Koenig Kellas et al., 2008). This

Dailey et al. 1025

resulted in 16 supratypes and 36 subtypes (see Table 1). Two authors subsequently coded20% of the data separately. Although reliability was high (k! .73), the coders felt slightmodifications to certain categories and an addition of the ‘‘realization about therelationship’’ category was needed. The coding manual was refined, and another 20% ofthe turning points were coded. Reliability was again high (k ! .74). Discrepancies wereresolved through discussion, and one researcher then coded the remaining data. Toensure that bias was not present in the authors’ coding, the same 20% of the turningpoints were coded by an outside coder, unfamiliar with the turning points literature andblind to the study’s purposes, also yielding acceptable reliability (k ! .63). Thecategories of turning points are described in the Results section.

Trajectory coding

After the first sample of data was collected, trajectories were identified by visuallyassessing the graphs constructed from the commitment ratings for each turning point. Tobe able to assess a trajectory, only participants with three or more turning points wereincluded (n ! 123). Using a method similar to Koenig Kellas et al. (2008), the authorsmet to develop a coding scheme based on visual representations of the commitmenttrajectories.5 We collectively determined the criteria for grouping visually similartrajectory graphs. To begin with, we grouped a random selection of 30 graphs intocategories according to the similarities in trajectory patterns. This resulted in three mainpatterns, two with two subtypes. Two authors then independently coded a randomsample of approximately half of the participants into the five trajectory categories.Intercoder reliability was established with k ! .74. Any discrepancies were discusseduntil one trajectory category was agreed upon and assigned. One coder then indepen-dently coded the remaining data.

To increase the sample sizes for the analyses of RQs 3 and 4, we collected the secondround of data. Visual graphs were produced including only those with three or more

Figure 1. Example of turning points graph seen by participants during the survey (updated aftereach turning point reported).


turning points (n ! 122). The same two researchers independently coded a randomsample of the second round (n! 30). Reliability was again high (k! .91) and one of theresearchers coded the remaining trajectories. Descriptions of the five trajectories arepresented in the Results section.

Table 1. Turning points reported by on-again/off-again partners.

Supratypes Frequency (%) Subtypes

1. Get-to-know you time 38 (4.1) First date/meetingInitial activity time

2. Relational escalation 234 (25.1) Expressions of feelingsDiscussion of relationshipSerious commitmentMutual investmentRelational reconciliationSubsequent quality time

3. Physical experiences 41 (4.4) First physical encounterSubsequent physical encounterNegative physical encounter

4. Pregnancy/having children 29 (3.1)5. Relational de-escalation 215 (23.1) Relational dissolution

Discussion leading to less commitmentConflictLetting goQuality timeMoving

6. External relational threats 55 (5.9) New potential threatsPrevious partner

7. Social networks 22 (2.4) Meeting family and/or friendsNetwork disapproval

8. Positive/helpful gestures 18 (1.9) Instrumental supportSocial support

9. Serendipity 15 (1.6) Circumstantial meetingSimilar vicinity

10. Special event(s) 37 (4.0) Holiday/celebrationsInvitation/partiesRenewed contact/visitationsVacations

11. Career-related obstacles 49 (5.3) StatusRelocationEmployment demands

12. Negative external behavior 21 (2.3)13. Relational transgression 81 (8.7) General transgression

InfidelityLyingViolence/abuse

14. Personality characteristic 17 (1.8)15. Realization about the relationship 38 (4.1)16. Miscellaneous/uncodable 22 (2.4)

Dailey et al. 1027

Measures

After reporting on their turning points, participants were asked to complete measures ontheir general relational qualities. Specifically, for most of the measures, participantswere asked to think across their relationship when completing these measures. Thefollowing scales were assessed on a Likert scale (1 ! strongly disagree; 7 ! stronglyagree) unless otherwise noted. Means were used to create the overall scores, with higherscores indicating higher levels of each variable. See Table 2 for the overall means andSDs for the variables. All variables were normally distributed.

Relationship factors. We assessed similarity using three items that were created for thepurpose of this study based on previous romantic relationship research (e.g., Surra &Longstreth, 1990). Items asked participants how similar they felt they and their partnerwere in: attitudes and beliefs, interests and activities, and personalities. Responses weresolicited on a 7-point Likert-type scale (1! not at all similar; 7! very similar). Reliability(Cronbach’s a) across all items was .76. Love and ambivalencewere measured using itemsadapted from Braiker and Kelley (1979). Seven items were in the love subscale (e.g., ‘‘Towhat extent have you felt that your relationship is special compared with others you hadbeen in?’’; a ! .92) and five items were in the ambivalence subscale (e.g., ‘‘How muchhave you thought or worried about losing some of your independence by being involvedwith your partner?’’; a ! .84). Responses were solicited on a 7-point Likert-type scale(1 ! not at all; 7 ! a great deal), and phrasing of the items was modified to reflect theentirety of the relationship. To measure closeness, we utilized Aron, Aron, and Smollan’s(1992) one-item Inclusion of Other in Self measure to assess closeness at the time of par-ticipation. Response was solicited on seven Venn diagrams (1! least closeness; 7! mostcloseness). The trustmeasure includes eight items from the Dyadic Trust Scale (Larzelere& Huston, 1980) to assess the degree to which the participants felt they could rely ontheir partners. Example items include: ‘‘I feel that I can trust my partner completely’’ and‘‘My partner is primarily interested in his/her own welfare’’ (reverse-scored; a ! .92).Sexual satisfaction was measured using a two-item index (Sprecher, 2002). One item wasa global assessment, ‘‘How sexually satisfying is the relationship to you?’’ on a 7-pointLikert-type scale (1 ! not at all; 7 ! very), while the other asked ‘‘How unrewardingor rewarding are your partner’s contributions in the area of sex (e.g., meets your needs andpreferences)?’’ on a 7-point Likert-type scale (1 ! very rewarding; 7 ! very rewarding).The correlation of these two items was strong (r ! .84, p < .001). The measure of rela-tional stress included six items from Dailey, Middleton, et al. (2012) that reflected threedimensions of relational stress prevalently found in on–off relationships (a ! .82): rela-tional uncertainty and ambivalence (e.g., ‘‘I have had a lot of mixed feelings about thisrelationship’’), emotional frustration (e.g., ‘‘This relationship has made me feel like I wason an emotional roller-coaster’’), and doubt or disappointment (‘‘I have often been disap-pointed about things in our relationship’’).

Communication factors. Affection was measured using eight items from the AffectionateCommunication Index (Floyd & Morman, 1998), which assess the amount of affectionparticipants communicated to their on–off partner through verbal expressions (e.g., saying


Table

2.Raw

means

(and

SDs)

forstability

characteristicsby

trajectory.

Earlydip

Laterdip

Steady-high

Steady-low

Fluctuating

Total

Relationshipfactors

Similarity

4.93

(1.01)

5.13

(1.44)

5.03

(1.18)

4.10

(1.36)

4.64

(1.50)

4.73

(1.42)

Love

6.16

(.79

)6.02

(.97

)6.32

(.67

)4.86

(1.56)

5.68

(1.15)

5.79

(1.20)

Ambivalence

4.32

(1.44)

3.88

(1.60)

3.83

(1.78)

4.92

(1.42)

4.23

(1.57)

4.21

(1.61)

Closeness

4.20

(1.97)

5.17

(1.85)

4.88

(2.00)

2.13

(1.47)

3.15

(2.07)

3.72

(2.20)

Trust

4.52

(1.44)

4.65

(1.72)

4.31

(1.57)

3.14

(1.63)

3.50

(1.79)

3.86

(1.77)

Sexu

alsatisfaction

5.60

(1.21)

5.17

(2.00)

5.28

(1.81)

4.40

(1.85)

5.09

(1.73)

5.05

(1.80)

Stress

4.92

(1.33)

4.49

(1.48)

4.72

(1.53)

5.05

(1.21)

5.08

(1.18)

4.89

(1.33)

Commun

icationand

interactiondynamics

Affection

5.92

(.97

)5.86

(1.02)

5.84

(.93

)4.98

(1.17)

5.39

(1.27)

5.52

(1.17)

Openn

ess

6.17

(.82

)5.78

(1.17)

5.86

(1.29)

5.08

(1.46)

5.52

(1.42)

5.60

(1.35)

Maintenance

5.53

(.83

)5.40

(1.23)

5.70

(.97

)4.66

(1.56)

5.00

(1.33)

5.18

(1.31)

Conflict

4.63

(1.27)

4.50

(1.40)

4.68

(1.18)

4.75

(1.61)

4.59

(1.48)

4.62

(1.41)

Work

onrelationship

5.55

(1.11)

5.48

(1.32)

5.85

(1.07)

4.51

(1.40)

4.89

(1.37)

5.15

(1.37)

Structuralfactors

Propo

rtionofpo

sitive

turningpo

ints

.58(.16

).55(.19

).52(.32

).42(.30

).50(.15

).50(.23

)Propo

rtionofnegative

turningpo

ints

.39(.15

).39(.17

).33(.31

).56(.29

).47(.16

).44(.23

)Relationshiplength/turning

points

9.80

(6.48)

8.40

(5.90)

11.86(7.37)

9.67

(7.21)

7.15

(5.61)

8.78

(6.51)

Relationshiplength/renew

als

30.83(19.53

)18

.76(16.97

)20

.14(14.90

)16

.63(17.58

)17

.82(15.95

)19

.15(16.65

)Propo

rtionofturningpo

ints

that

werebreakups/renew

als

.48(.24

).39(.22

).39(.26

).33(.24

).36(.20

).37(.22

)

1029

‘‘I love you’’) and nonverbal gestures (e.g., hugging). Reliability (Cronbach’s a) across allitems was .85. Openness was measured with the 10-item Self-Disclosure Index (Miller,Berg, & Archer, 1983). Responses were solicited on a 7-point Likert-type scale(1! have not discussed at all; 7! have discussed fully and completely). Participants wereasked to rate the degree to which they had disclosed about certain areas (e.g., ‘‘my personalhabits’’ and ‘‘what I like and dislike about myself’’). Cronbach’s a across all items was .94.Maintenance (‘‘To what extent have you tried to change your behavior to help solve cer-tain problems between you and your partner?’’; a ! .83) and conflict (e.g., ‘‘How oftenhave you and your partner argued with each other?’’; a ! .84) were measured throughBraiker and Kelley’s (1979) five-item subscales. The relationship work measure includedfour items created for the purpose of this survey assessing participants’ proactive actionstoward the relationship to resolve or make changes created for the purpose of this study(e.g., ‘‘We worked to resolve the issues in our relationship’’ and ‘‘We tried to make ourrelationship better between the breakup(s) and renewal(s)’’; a ! .85).

Structural factors. Given that a few participants noted turning points that did not changetheir commitment level,we calculated theproportion of positive turning points (i.e., numberof turning points that increased commitment divided by the total number of turning pointsreported) and proportion of negative turning points (i.e., number of turning points thatdecreased commitment divided by the total number of turning points) separately. We alsoassessed the ratios of relationship length to the total number of turning points, relationshiplength to number of renewals, and number of breakups to number of renewals.

On–off types. To assess on–off types for RQ4, we employed the paragraphs created byDailey et al. (in press), which reflect the characteristics described above, and asked theparticipants to choose which paragraph best described their relationship.We also asked theparticipants to rate how well the paragraph chosen reflected their relationship on a 7-pointscale (1 ! does not describe our relationship well; 7 ! very much describes our relation-ship). In terms of frequencies, 35 participants (14.3%) selected the habitual paragraph,46 participants (18.8%) selected mismatched, 54 participants (22.0%) selected capitalizedon transitions, 27 participants (11.0%) selected gradual separators, and 65 participants(26.5%) selected controlling, while 18 participants (7.3%) did not select any paragraph.

Results

Types of turning points

RQ 1 asked what types of turning points partners in on–off relationships report. A total of932 turning points were identified from the first round of data collection (summarized inTable 1). Several turning points provided further insight into the process of developing anddissolving an on–off relationship. For instance, the most frequently reported turning pointswere the relational escalation and relational de-escalation supratypes. This finding makesintuitive sense, given that both escalating and de-escalating transitions are defining featuresof on–off relationships. Yet, although many of the other supratypes (e.g., get to know yourtime, physical experiences, external threats, social networks, negative external behavior,


and personality characteristics) overlapped to some degree with the previous turningpoints research (e.g., Baxter & Bullis, 1986; Koenig Kellas et al., 2008), this more diversesample (e.g., in age) yielded additional relational experiences that can affect the develop-mental process of romantic relationships. The following description focuses on these moreunique categories of turning points that were salient to these on–off relationships.

The pregnancy or having children supratype involved participants discussing the ideaof having children or the experience of preparing for and raising child(ren). Participantreports of pregnancy or having children often coincided with increases in commitmentand/or a renewal in the relationship. Because more women reported this category thanmen, future research may consider potential sex differences on how these turning pointsand on–off relationships are experienced. Conversely, a typically mitigating influence oncommitment was career-related obstacles, which involved turning points surrounding orinvolving employment based on three distinct issues: change of status, relocation, anddemands. The status subtype involved participants choosing new employment, selectinga different occupation, or becoming unemployed that impacted the relationship. Therelocation subtype indicated separation caused by employment, military obligations, orcollege attendance, while the employment demands subtype indicated changes in therelationship due to external occupational demands placed on one or both partners (e.g.,working for long hours, taking the bar exam, and graduate school stressors).

Other turning points also emerged at the onset of certain special events. The specialevents supratype often led participants to re-evaluate their relationships based on theirinclusion or exclusion in events. Specifically, subtypes included special events, invita-tions/parties, vacations, and renewed contact/visitations. These occasions either createdmore relational strength through invitations to events or exposure to partner’s social net-works. Or the rejection (e.g., refusing invitation to attend a family affair) or rebuffing(e.g., neglecting to buy a holiday present) acted as a relational barometer to show lessinterest or involvement in the relationship.

Additionally, circumstantial events created new opportunities to renew throughserendipitous occasions and similar proximities. Similar to a circumstantial meeting byKoenig Kellas et al. (2008), the serendipity supratype classified participants’ turningpoints that were born out of accident or coincidence (e.g., attending a mutual friend’sparty or funeral). Another opportunity for relationship renewal appeared when partnerswere again in similar vicinities (e.g., living in nearby proximity again, returning tocollege, or back from the military deployment).

Relational transgressions also emerged more explicitly in this on–off sample. Manywere serious infractions of relational rules such as infidelity, lying, and less prevalently,violence or abuse. Other smaller violations fell under a general category such as beingstood up for a date. As would be expected, these transgressions were associated withdecreases in commitment.

On–off commitment trajectories

RQ2 asked what types of trajectories are revealed by on–off partners’ changes incommitment across their reported turning points. The five trajectories identified arepresented in Figure 2. The first trajectory type, ‘‘dip’’ in commitment (n ! 57, 23.3%),

Dailey et al. 1031

represents those relationships where one or two times in the relationship there was adecrease in commitment, often followed by a trend toward increased commitmentafterward. Participants within this group also experienced relatively high, or average,levels of commitment overall. The dip in commitment group was subdivided into twosubtypes, early dip and later dip. The early dip group (n ! 15, 6.1%) tended to expe-rience a downward turn in commitment at the beginning of the relationship and be morestable at the end. The later dip group (n ! 42, 17.1%) showed a downward turn incommitment toward the middle or at the end of the relationship. This group as a wholemay resemble the accelerated–arrested type found by Surra (1987).

The second trajectory type, steady commitment (n ! 86, 35.1%), illustrated a rela-tively unchanging, flat, linear progression of the on–off relationship over time.

Figure 2. On-again/off-again trajectory examples: (a) early dip in commitment; (b) later dip incommitment. (c) steady-high; (d) steady-low and (e) fluctuating.


Participants in this group experienced small changes with the trend demonstrating eitherincreases or decreases to commitment. As such, this type was divided into twosubcategories; steady-high (n ! 43, 17.6%) representing those trajectories with higherlevels of commitment (averaging approximately 60% or higher across turning points),and steady-low (n ! 43, 17.6%) represents those trajectories with moderate, low ordecreasing commitment (60% or lower on average).

The final and most frequent trajectory type (n ! 102, 41.6%), fluctuating commit-ment, showed larger, more frequent changes in the on–off relationship across time.Commitment could be either high or low at the end of the relationship and turning pointswere often more evenly spaced across time. Because of the often drastic rise and fall ofcommitment levels across the on–off relationship, on average, participants in thiscategory tended to have moderate levels of commitment overall.

Trajectory differences and relational stability characteristics (RQ3)

RQ3 asked how different on–off trajectories differ with regard to the factors associatedwith relational stability. We employed analyses of covariance (ANCOVA) to comparethe on–off trajectories with the relational stability factors, controlling for current statusof the relationship.6 Pairwise comparisons using a Bonferroni adjustment assessed wherethe specific significant differences occurred. Table 2 includes the raw means and SDs ofthe characteristics by trajectory. Table 3 presents the results of the ANCOVAs, estimatedmarginal means, and SEs by trajectory.

Relational dynamics. When controlling for current relational status, two of the sevenrelational dynamics showed differences by trajectory. Those with a steady-low trajectoryreported significantly less love with their on–off partner than all other trajectory types.For closeness, those with the trajectories of later dip and steady-high reported signifi-cantly more closeness than both those with a steady-low trajectory and a fluctuatingtrajectory. No significant differences were found between trajectories and similarity,ambivalence, relational stress, trust, or sexual satisfaction.

Communication and interaction dynamics. Of the five communication factors, three factorsvaried by trajectory. Those with the trajectories of later dip and steady-high reported sig-nificantly more affection than those with a steady-low trajectory. In addition, those witha steady-high trajectory reported significantly more maintenance than those with asteady-low trajectory. Finally, those with a steady-high trajectory reported significantlymore relational work than both those with a steady-low trajectory and a fluctuatingtrajectory. No significant differences were found for openness or conflict.

Structural characteristics. For analyses including relationship length, the two outliers (i.e.,168 and 362 months) were excluded from the analyses. Of the five structural charac-teristics, two characteristics varied by trajectory when controlling for current status.Those with a steady-high trajectory reported significantly lower proportion of negativeturning points than those with a steady-low trajectory. Trajectory varied for the length ofthe relationship in relation to the number of turning points, with the fluctuating group

Dailey et al. 1033

Table

3.ANCOVAresultswithestimated

marginalm

eans

(and

standard

errors)forstability

characteristicsby

trajectory.

Earlydip

Laterdip

Steady-high

Steady-low

Fluctuating

Fdf

Trajectory

partialZ

2

Relationshipfactors

Similarity

4.70

(.35

)4.93

(.22

)4.92

(.21

)4.27

(.23

)4.75

(.14

)1.40

4,22

3.03

Love

5.99

(.29

)a6.06

(.17

)b6.24

(.17

)c4.99

(.18

)abcd

5.76

(.12

)d6.78

***

4,22

2.11

Ambivalence

4.55

(.41

)4.08

(.25

)3.94

(.25

)4.75

(.26

)4.11

(.16

)1.72

4,22

2.03

Closeness

3.62

(.46

)4.65

(.28

)ab

4.60

(.28

)2.57

(.29

)ac

3.41

(.18

)bc

9.16

***

4,22

5.14

Trust

3.99

(.40

)4.17

(.24

)4.05

(.24

)3.55

(.25

)3.75

(.16

)0.98

4,22

5.02

Sexu

alsatisfaction

5.34

(.46

)4.94

(.28

)5.16

(.28

)4.59

(.29

)5.21

(.19

)1.02

4,22

5.02

Stress

5.11

(.34

)4.65

(.21

)4.81

(.21

)4.91

(.21

)5.00

(.14

)0.64

4,22

5.01

Commun

icationandinteractiondynamics

Affection

5.87

(.30

)5.81

(.18

)a5.81

(.18

)b5.02

(.19

)ab

5.42

(.12

)3.13

*4,

223

.05

Openn

ess

6.04

(.35

)5.66

(.21

)5.79

(.21

)5.18

(.22

)5.58

(.14

)1.46

4,22

3.03

Maintenance

5.40

(.33

)5.29

(.20

)5.64

(.20

)a4.76

(.21

)a5.06

(.13

)2.51

*4,

222

.04

Conflict

4.77

(.37

)4.63

(.23

)4.75

(.22

)4.64

(.24

)4.52

(.15

)0.22

4,22

2.00

Work

onrelationship

5.25

(.32

)5.21

(.20

)5.70

(.20

)ab

4.74

(.20

)a5.03

(.13

)b3.10

*4,

225

.05

Structuralfactors

Propo

rtionofpo

sitive

turningpo

ints

.51(.06

).49(.03

).50(.03

).46(.03

).53(.02

).81

4,23

5.01

Propo

rtionofnegative

turningpo

ints

.45(.05

).44(.03

).36(.03

)a.52(.03

)a.45(.02

)2.84

*4,

235

.05

Relationshiplength/turningpo

ints

9.05

(1.64)

7.72

(1.00)

11.52(.98

)a10

.22(1.01)

7.46

(0.64)

a3.94

**4,

233

.06

Relationshiplength/renew

als

28.95(4.25)

16.85(2.80)

19.31(2.55)

18.11(2.88)

18.74(1.76)

1.58

4,20

8.03

Propo

rtionofturningpo

ints

that

were

breakups/renew

als

0.49

(.06

)0.39

(.04

)0.39

(.04

)0.33

(.04

)0.36

(.02

)1.44

4,23

5.02

Note.Samesuperscripts

indicate

significant

differences

inpairwisecomparisons.

*p<.05,

**p<.01,

***p

<.001

.

1034

reporting a smaller ratio of length to turning points than the steady-high group. Nosignificant differences were found between the trajectory groups and the proportion ofpositive turning points, length of relationship in relation to the number of renewals,or the proportion of turning points that were breakups or renewals.

On–off types and trajectories (RQ4)

Analysis of RQ4 showed that commitment trajectories varied by on–off relationshiptype, w2(16) ! 32.29, p < .01.7 The early dip participants were more likely to endorsecontrolling (53.3%), the later dip and steady-high participants were more likely toendorse capitalized on transitions (35.7 and 39.0%, respectively) followed by controlling(23.8 and 26.8%, respectively), the stable-low group was fairly equally likely to endorsehabitual (27.0%), mismatched (21.6%), and controlling (29.7%), and the fluctuatinggroup was more likely to endorse mismatched (29.3%) or controlling (27.2%).

Discussion

Previous research on romantic relationships has emphasized the importance of assessingturning points throughout the life span of a relationship (e.g., Baxter & Bullis, 1986).Researchers have examined trajectories of dating (Surra & Hughes, 1997), marital(Huston et al., 1981), and postdissolution (Koenig Kellas et al., 2008) relationships.However, on–off relationships are in many ways different from noncyclical relationships(Dailey et al., 2009) and are also unique in that turning points may occur during breakupsand renewals as well as other salient relational events. Although previous research hasexamined the breakups and renewals as turning points in on–off relationships (Daileyet al., in press), the present study extends on this by examining all potential turningpoints and the varying trajectories the turning points reflect.

Turning points in on–off relationships

Many of the turning point categories that emerged in this study overlap with the existingliterature. However, perhaps due to both the nature of on–off relationships and thecurrent nonstudent sample, several novel turning points emerged or were more pre-dominant, such as pregnancy/having children, career-related obstacles, special events,serendipity, and relational transgressions.

The pregnancy/having children and career-related obstacles categories representlarger forces that might push a relationship together or pull a relationship apart. Forexample, pregnancy or having children may be acting as an investment or external psy-chological barrier to exiting their relationships (Levinger, 1976, 1999; Rusbult, 1983).Previous research indeed shows that dissolution is less likely when children are present(Kamp Dush, 2012). Conversely, career-related moves or a high dedication to work mayhinder commitment to a relationship. This finding also substantiates the mismatched on–off type, in which external factors such as life stages, bad timing, or geographic distanceplayed a role in the on–off transitions (Dailey et al., in press). In on–off relationships,these larger forces might have influences or implications less considered in research

Dailey et al. 1035

on dating relationships such as financial costs (e.g., traveling to see the relational partner,additional childcare costs), which may lead partners to renew or break off therelationship.

Additionally, several participants reported turning points relating to special orserendipitous events that often led to renewing the relationship. Although relativelyinfrequent, these types of turning points suggest either that these events happen more inon–off relationships or that partners who enter into on–off relationships are more sus-ceptible to these events. Future research is needed to explore how these events play a rolein the on–off nature of these relationships as compared with other romantic relationships.

Relational transgressions were also reported relatively frequently. These transgres-sions tended to be more severe in nature such as aggression, deception, or infidelity,which can have implications for the stability of the relationship (Afifi, Falato, &Weiner,2001; DeSteno, Bartlett, Braverman, & Salovey, 2002). These on–off partners did notnecessarily permanently dissolve their relationship when faced with a severe relationaltransgression. Or viewed in a different way, the transgressions may have been seriousenough to warrant a breakup, yet the partners found a way to resolve the transgressionand reconcile the relationship. As such, the process of forgiveness in on–off relationshipswould be insightful to examine in future research. For example, perhaps these partnersare more likely to employ a conditional type of forgiveness (e.g., Waldron & Kelley,2005) that facilitates the on–off nature of the relationship.

Trajectories of on–off relationships

Examining relationship trajectories is a natural extension of a turning points approach(Huston et al., 1981; Koenig Kellas et al., 2008; Surra 1985, 1987), and examining howtrajectories differentiate on–off relationships based on a variety of characteristicsprovides a richer understanding of how on–off relationships function. Our analysisrevealed three general trajectories and several subtypes—steady (low and high), dip incommitment (early and later), and fluctuating. One major conclusion across thefindings is that individuals with a steady-low trajectory reported lower levels of close-ness, relationship work, affection, and love. That these differences emerged for a groupwhich is characterized by consistently low levels of relational quality is not particu-larly surprising. However, the significant differences remained even when controllingfor current relational status. In other words, some participants in steady-low relation-ships reported lower levels of certain relationship and communication dynamics thanother trajectories, regardless of whether they were together. This finding somewhatcontrasts with the previous on–off research, which has typically found that a positivecharacteristic (e.g., those who capitalize on transitions enact positive change in a rela-tionship) differentiates types of on–off relationships (e.g., Dailey, Jin, et al., 2012; Dai-ley et al., in press). When considering on–off relationships as a function of theirtrajectories, it appears that some partners remain in relationships with consistently lowcommitment, and these couples tend to be different from other types of on–off relation-ships. Perhaps, as interdependence theory suggests (e.g., Thibault & Kelley, 1959),these are individuals who perceive no greater alternatives and prefer the companion-ship of a less satisfying relationship than to be alone (e.g., Felmlee, Sprecher & Bassin,


1990; Sprecher, 2001). Hence, it is not surprising that this steady-low group was rel-atively more likely to endorse being in the habitual on–off type, a type that repeatedlyreturns to the relationship because of a desire for companionship and with minimal dis-cussion about changing the relationship.

Additionally, although individuals who reported a steady-low trajectory tended toreport fewer positive factors (e.g., love and affection), they did not necessarily reportmore negative factors in their relationship. For example, individuals with a steady-lowtrajectory did not report significantly more stress or conflict, yet they did experience ahigher proportion of negative turning points. Given this, it appears that most on–off rela-tionships are characterized by some negative perceptions or behaviors, regardless of theirtrajectory or turning points.

Interestingly, despite the fluctuating trajectory reflecting oscillations in commitment(e.g., seemingly less stability), individuals with fluctuating trajectories only differed intheir reports of relational closeness, work, and love—typically falling in between thesteady-high and steady-low trajectories. The current fluctuating group resembles rela-tional trajectories that emerged in previous research. Koenig Kellas et al. (2008) reporteda ‘‘turbulent’’ trajectory in postdissolution relationships, and people who reported a tur-bulent trajectory had lower quality relationships than individuals who reported anupward projection trajectory (which is conceptually similar to individuals with asteady-high or dip in commitment trajectories in the present study). Additionally, Surraand Hughes (1997) found evidence for an event-driven pattern of commitment in pre-marital relationships defined as frequent and dramatic changes in commitment due toa focus on singular events similar to the fluctuating trajectory. Also similar to the presentstudy, individuals with event-driven patterns differed in some ways (e.g., stability ofcommitment) from the other trajectory types, but did not differ significantly for factorssuch as perceived alternatives, love, or current status (Surra & Hughes, 1997). Overall,these findings generally parallel those of the present study.

Finally, individuals with dip in commitment and steady-high trajectories did not sig-nificantly differ from one another. Even if partners experience a dip in commitment atsome point in their relationship, they still report similar overall relational quality as indi-viduals with a stable, highly committed trajectory. These trajectories resemble Surra’s(1985) accelerated and accelerated–arrested, Surra and Hughes’ (1997) relationship-driven, and Koenig Kellas et al.’s (2008) upward relational progression trajectories,which reflected more positive relationship dynamics. The extent to which on–off part-ners capitalize on their relational transitions closely relates to relational quality (Dailey,Jin, et al., 2012), so it may be that individuals with only one or two dips in commitmentthroughout their relationship were able to capitalize on the transition to improve theirrelationship. Indeed, as discussed in the following section, individuals with a dip incommitment and steady-high trajectories were more likely to report a capitalizing ontransitions type of relationship.

These patterns in trajectories generally parallel the findings of Dailey, Middleton,et al. (2012) who presented a model of on–off relationships in determining whichrelationships would achieve relative stability, permanently dissolve, or continue to cycle.Those with only a dip in commitment or steady-high commitment are likely those whowill achieve relative stability given their higher relational quality. Those with a

Dailey et al. 1037

consistently low level of commitment are likely those who will eventually permanentlyterminate their relationship. With low levels of relational quality, these partners arelikely waiting for a better alternative and will dissolve the relationship if this alternativematerializes. The fluctuating trajectory likely reflects those who will continue to cycle.In the findings of Dailey, Middleton, et al., those expecting their relationship to continuecycling reported moderate levels of relational quality when compared with the othergroups, which is consistent with the current findings for the fluctuating trajectory. Com-bining the current factors associated with the trajectories with the factors assessed byDailey, Middleton, et al. (e.g., relational uncertainty and relational maintenance) pro-vides more evidence regarding how on–off relationships differ in relational quality; thefluctuating or continually cycling group has more tepid perceptions of their relationship,which may result in not feeling comfortable in fully committing to the relationship orleaving the relationship.

Given these patterns, those with short dips or consistently high commitment levels orsteady, low levels of commitment may be relatively similar to other dating relationships.The fluctuating group, however, may be the most distinct from other dating relation-ships. For example, the current fluctuating group in some ways contrasts with the workof Arriaga (2001; see also Arriaga, Reed, Goodfriend, & Agnew, 2006), which showsthat individuals reporting fluctuations in satisfaction were more likely to dissolve theirrelationships even after controlling the overall level of satisfaction. Controlling forcurrent status, our data revealed that the fluctuating group fared better than those withsteady, low commitment. Perhaps fluctuations in one’s own commitment operate dif-ferently than satisfaction, which may fluctuate on a daily basis as compared to a generalsense of commitment with the relationship. Alternatively, on–off partners may expectand better tolerate fluctuations in their relationships than other types of romantic rela-tionships. Regardless of the explanation, the fluctuating group may be the most revealingabout the nature of on–off relationships.

More generally, in looking at the findings across the three dimensions of relation-ships, no clear patterns emerged in terms of which factors differentiated the trajectoriesbest. Although more of the communication dynamics showed significant differencesamong the types, the relationship dynamics yielded the largest effect sizes. Perhaps thestrongest conclusion is that the structural factors least differentiated the trajectories. Thisis consistent with the previous on–off research suggesting that relational and commu-nication dynamics play a larger role in the varying natures of on–off relationships,whereas structural features play a minimal role (Dailey, Middleton, et al., 2012; Daileyet al., 2009; Dailey et al., in press). Hence, on–off partners appear to have differingexperiences based on their perceptions of their relationship and interaction patternsrather than on whether they have more or less turning points or renewals relative to theirrelationship length. As such, future research and theory building should focus on thesedynamic qualities of relationships rather than the structural features. We should also notethat less than half of the factors we assessed yielded significant differences. Yet, whencurrent relational status is not controlled, most do yield differences by trajectory.Although there are valid reasons for controlling the current status, this is somewhatlimiting in the assessment of on–off relationships given that relational status oftenfluctuates.


Triangulating types based on trajectories. The study also investigated whether individualsdiffered on their self-reported on–off relationship type based on their trajectory. Asmentioned in the previous section, individuals with a dip in commitment or steady-high trajectories were more likely to report capitalized on transitions type of relation-ship. This is a chief finding from this analysis and supports that some couples are ableto use the transitions to improve or stabilize the relationship (Dailey et al., in press). Inaddition, individuals with a fluctuating trajectory were more likely to report a mis-matched or controlling type. The controlling type involves one partner persistently con-trolling the progression of the relationship, and mismatched types often have externalfactors, such as different life stages or geographic distance, which keep the relationshipin a fluctuating state (Dailey et al.). In the present study, it appears that individuals whoexperienced frequent, often dramatic changes in commitment over the course of their rela-tionship were also more likely to self-categorize their relationship as being influenced byexternal factors and partner control. Future research would be helpful in determining ifthere are multiple types of fluctuating trajectories in terms of experiences in and outcomesof the relationship.

Despite some overlap, there was not an exact, one-to-one match between trajectoriesand types. Further, a surprising finding was that many of the participants in the threerelatively steady and high commitment trajectories selected the controlling type, whichis highly counterintuitive. The weak level of overlap and counterintuitive findings maybe due to the measure of on–off types. To decrease fatigue effects and to clearly classifyparticipants into one type, we employed paragraph descriptions of each type and askedthe participants to select the one paragraph that described their relationship best.Although most participants reported the paragraph that described their relationship well(M! 5.72 on a 7-point scale), more sophisticated measures may reveal stronger associa-tions between the types and trajectories.

Limitations and future directions

The present study has several strengths, including the extension of the RIT to an onlineformat and assessing the on–off relationships from a developmental perspectiveincluding the recall of all potential turning points and not just the breakups and renewals.However, there are several limitations that should be addressed in future studies. First,the study utilized only half of the sample in identifying the trajectories, as only indi-viduals who reported more than three turning points were included, and those whoreported three or more turning points did differ from those reporting only one or two(e.g., less likely to be currently together and more stressed, but also more affectionateand open). Additionally, an interview-based method of RIT may have provided anopportunity to inquire about smaller turning points. Yet, the potential drawback of theonline version is mitigated by the fact that this methodology allowed for the recruitmentof a larger and more diverse sample. The diversity of the sample could, however, beincreased. The sample was not representative of the U.S. population and important dif-ferences may exist for those of different life stages or ethnicities. In addition, althoughsex differences did not strongly emerge in these data, men and women may experienceon–off relationships differently (e.g., pregnancy as a turning point). Furthermore, we

Dailey et al. 1039

included married individuals in the current sample, as preliminary analyses did not showthey were characteristically different from those who were dating; yet, marital status mayimpact certain dimensions and experiences of on–off relationships.

Coding trajectories with only three turning points may also present validity issuesgiven that the addition of a fourth turning point might have changed the classification.Furthermore, many individuals may experience more turning points after participation,which could change their trajectory. Despite this difficulty, the coders had relatively highagreement when assessing the trajectories. Perhaps this indicates that a minimum rela-tionship length should be employed; yet, the correlation between the relationship lengthand number of turning points was ".02, p ! .597. Excluding trajectories with fewerturning points would also ignore certain nuances that may be important in identifying thetypes of on–off relationships.

Finally, the trajectories did not differ on several relational qualities, includingopenness, work on the relationship, stress, sexual satisfaction, and conflict after con-trolling for current relational status. Yet, perhaps commitment (the central variable in theRIT) is not the optimum variable for examining on–off relationships. For example, in anon–off relationship, in particular, partners may break up but still have high commitmentto their relationship. Future research should investigate whether another variable bettersuits the investigation of on–off relationships such as satisfaction or perceived viabilityof the relationship.

Conclusion

The present study extends the previous research on the trajectories and turning pointsof romantic relationships by obtaining retrospective accounts of major events duringon–off relationships. Several unique turning points (such as having children) haveemerged. In addition to supporting the previous research’s finding that some on–offcouples can utilize their transitions to make positive changes in the relationship, thecurrent study offers new contributions in understanding on–off relationships. Forexample, a stable and low-level of commitment or involvement in the relationship maybe an additional factor to consider; this trajectory was most clearly different than theothers, a finding that has not emerged in previous on–off research. Additionally, part-ners in the fluctuating trajectory, which would seemingly represent less stability, actu-ally reported more moderate perceptions of their relationships, faring better than thesteady-low commitment group. Overall, categorizing on–off partners based on theirrelational trajectories provides further insight into the heterogeneous nature of on–offrelationships.

Authors’ note

Rene M. Dailey (University of California, Santa Barbara) is an associate professor in the Com-munication Studies Department at the University of Texas at Austin where NB (Arizona StateUniversity), LL (University of Wisconsin-Milwaukee), and BC (Texas State University, SanMarcos) are doctoral students. A previous version of this paper was presented at the 2012 Interna-tional Association for Relationship Research conference in Chicago, Illinois, USA.


Notes

1. Mechanical Turk is an online survey service hosted by Amazon, where workers complete the tasks

for nominal fees and provides similar results to other online and traditional recruitment methods

(Buhrmester, Kwang, & Gosling, 2011). Participants were paid $0.50 for completing the survey.

2. Because some couples’ goals may not be marriage and some states do not allow legal same-sex

marriages, we allowed the participant to self-define commitment. The anchors for the commit-

ment scale were: ‘‘0 indicates no commitment; 100 indicates complete commitment.’’

3. Participants reported fewer turning points than expected based on previous research. An advan-

tage of the second round of data collection was that we were able to ask the participants why

they reported the number of turning points that they mentioned. For those reporting five or less

turning points, we provided several options and asked them to check all that applied in terms of

the number of turning points reported. About half of this second sample (45.6%) chose the

option of ‘‘The turning points I reported were the only ones in my relationship’’ and 57.1% also

chose ‘‘Although my relationship had some smaller turning points, I only reported on the most

meaningful ones’’ (over 80% of the sample reported at least one of these). Very few felt con-

strained by being allowed to report on only 10 (4.6%) or that reporting more would have taken

too much time (4.2%). Interestingly, 14.3% preferred not to report all of their turning points.

Overall, although the average number of turning points reported was fewer than previous

research, the vast majority felt they reported the meaningful turning points in their relationship.

4. Given the data we obtained in the first round of data collection, we asked the participants

whether they reported all the breakups and renewals as turning points in the second round.

For those who responded no (n ! 93), we asked them to provide their reason(s) why in an

open-ended question. Two of the authors coded the responses into one of the six categories

(k ! .69, differences were resolved through discussion). The most prevalent reason (23.7%)

was that the transitions were minor suggesting that the breakup or renewal did not significantly

impact the relationship. Other reasons included: some of the transitions were for an insufficient

duration (e.g., last only a couple of days, 16.1%), the transitions were a part of the nature of the

relationship (15.1%), recalling the turning points was too painful or brought up bad memories

(14.0%), transitions were too frequent to report all of them (12.9%), or they reflected a redun-

dant pattern (3.2%). The remaining was coded into a miscellaneous category (15.1%).

5. We first employed cluster analysis using characteristics of the trajectories (e.g., number of

upturns and downturns, average upturns, average downturns, relationships length, length of

time between turning points) based on the previous use of the RIT (e.g., Surra, 1985). However,

the cluster analysis only provided a two-cluster solution, and the resulting categories were less

theoretically grounded than using a similar coding technique as given by Koenig Kellas et al.

(2008). Hence, we identified the trajectories visually.

6. Without controlling for relational status, the trajectories varied on most of the stability factors

assessed (the exceptions being stress, sexual satisfaction, and conflict).

7. To increase the validity of the types of classification, we excluded 13 participants who did not

feel the paragraph chosen described their relationship well (three or less on the 7-point scale).

Funding

This research received no specific grant from any funding agency in the public, commercial, ornot-for-profit sectors.

Dailey et al. 1041

References

Afifi, W. A., Falato, W. L., & Weiner, J. L. (2001). Identity concerns following a severe relational

transgression: The role of discovery method for relational outcomes of infidelity. Journal of

Social and Personal Relationships, 18, 291–308.

Aron, A., Aron, E. N., & Smollan, D. (1992). Inclusion of other in the self scale and the

structure of interpersonal closeness. Journal of Personality and Social Psychology, 63,

595–612.

Arriaga, X. B. (2001). The ups and downs of dating: Fluctuations in satisfaction in newly formed

romantic relationships. Journal of Personality and Social Psychology, 80, 754–765. doi:10.

1037//0022-3514.80.5.754

Arriaga, X. B., Reed, J. T., Goodfriend, W., & Agnew, C. R. (2006). Relationship perceptions and

persistence: Do fluctuations in perceived partner commitment undermine dating relationships?

Journal of Personality and Social Psychology, 91, 1045–1065. doi:10.1037/0022-3514.91.6.

1045

Baxter, L. A., & Bullis, C. (1986). Turning points in the development of blended family relation-

ships. Journal of Social and Personal Relationships, 16, 291–313. doi:10.1111!j.1468–2958.

1986.tb00088.x

Braiker, H. B., & Kelley, H. H. (1979). Conflict in the development of close relationships.

In R. L. Burgess & T. L. Huston (Eds.), Social exchange in developing relationships

(pp. 135–168). New York, NY: Academic Press.

Buhrmester, M. D., 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

Cate, R. M., Levin, L. A., & Richmond, L. S. (2002). Premarital relationship stability: A review of

recent research. Journal of Social and Personal Relationships, 19, 261–284. doi:10.1177/

0265407502192005

Conville, R. L. (1987). Relational transitions: Inquiry into their structure and function. Journal of

Social and Personal Relationships, 5, 422–437.

Dailey, R. M. (2012). Prevalence and correlates of on-again/off-again relationships. Unpublished

manuscript. University of Texas at Austin.

Dailey, R. M., Jin, B., Brody, N., &McCracken, A. A. (2012, November). A dimensional approach

to characterizing on-again/off-again dating relationships. Paper presented at the National

Communication Association Convention, Orlando, FL.

Dailey, R. M., McCracken, A. A., Jin, B., Rossetto, K. R., & Green, E. W. (in press). Negotiating

breakups and renewals: Types of on-again/off-again dating relationships. Western Journal of

Communication.

Dailey, R. M., Middleton, A. V., & Green, E. W. (2012). Perceived relational stability in on-again/

off-again relationships. Journal of Social and Personal Relationships, 29, 52–76.

Dailey, R. M., Pfiester, A., Jin, B., Beck, G., & Clark, G. (2009). On-again/off-again dating

relationships: How are they different from other dating relationships? Personal Relationships,

16, 23–47.

DeSteno, D., Bartlett, M., Braverman, J., & Salovey, P. (2002). Sex differences in jealousy: Evolu-

tionary mechanism or artifact of measurement. Journal of Personality and Social Psychology, 83,

1103–1116.


Felmlee, D., Sprecher, S., & Bassin, E. (1990). The dissolution of intimate relationships: A hazard

model. Social Psychology Quarterly 53, 13–30.

Floyd, K., & Morman, M. T. (1998). The measurement of affectionate communication. Commu-

nication Quarterly, 46, 144–162. doi:10.1080/01463379809370092

Huston, T. L., Surra, C., Fitzgerald, N., & Cate, R. (1981). From courtship to marriage: Mate

selection as an interpersonal process. In S. Duck & R. Gilmour (Eds.), The emerging field of

personal relationships (pp. 53–88). New York, NY: Academic Press.

Kamp Dush, C. M. (2012). Relationship-specific investments, family chaos, and cohabitation

dissolution following a nonmarital birth. Family Relations, 60, 586–601.

Koenig Kellas, J., Bean, D., Cunningham, C., & Cheng, K. Y. (2008). The ex-files: Trajectories,

turning points, and adjustment in the development of post-dissolutional relationships. Journal

of Social and Personal Relationships, 25, 23–50. doi:10.1177!0265407507086804

Larzelere, R. E., & Huston, T. L. (1980). The dyadic trust scale: Toward understanding interper-

sonal trust in close relationships. Journal of Marriage and Family, 42, 595–604.

Le, B., Dove, N. L., Agnew, C. R., Korn, M. S., & Mutso, A. A. (2010). Predicting nonmarital

romantic relationship dissolution: A meta-analytic synthesis. Personal Relationships, 17,

377–390. doi:10.111/j.1475-6811.2010.01285.x

Levinger, G. (1976). A social psychological perspective on marital dissolution. Journal of Social

Issues, 32, 21–47. doi: 10.1111/j.1540-4560.1976.tb02478.x

Levinger, G. (1999). Duty toward whom? Reconsidering attractions and barriers as determinants

of commitment in a relationship. In W. H. Jones & J. M. Adams (Eds.), Handbook of interper-

sonal commitment and relationship stability (pp. 37–52). New York, NY: Plenum.

Miller, L. C., Berg, J. H., & Archer, R. L. (1983). Openers: Individuals who elicit intimate

self-disclosure. Journal of Personality and Social Psychology, 44, 1234–1244.

Mongeau, P. A., & Teubner, G. (2002). Romantic relationship transitions. Paper presented to the

National Communication Association, New Orleans, LA.

Rusbult, C. E. (1983). A longitudinal test of the investment model: The development (and dete-

rioration) of satisfaction and commitment in heterosexual involvements. Journal of Personality

and Social Psychology, 45, 101–117.

Sprecher, S. (2001). Equity and social exchange in dating couples: Associations with satisfaction,

commitment, and stability. Journal of Marriage and Family, 63, 599–613.

Sprecher, S. (2002). Sexual satisfaction in premarital relationships: Associations with satisfaction,

love, commitment, and stability. Journal of Sex Research, 39, 190–196. doi:10.1080/

00224490209552141

Surra, C. A. (1985). Courtship types: Variations in interdependence between partners and social

networks. Journal of Personality and Social Psychology, 49, 357–375. doi:10.1037/

0022–3514.49.2.357

Surra, C. A. (1987). Reasons for change in commitment: Variations by courtship type. Journal of

Social and Personal Relationships, 4, 17–33. doi:10.1177/0265407587041002

Surra, C. A., & Gray, C. E. (2000). A typology of processes of commitment to marriage: Why

do partners commit to problematic relationships? In L. J. Waite, C. Backrach, M. Hindin,

E. Thomson, & A. Thornton (Eds.), The ties that bind: Perspectives on marriage and coha-

bitation (pp. 253–280). New York, NY: Aldine de Gruyter.

Surra, C. A., & Hughes, D. K. (1997). Commitment processes in accounts of the development of

premarital relationships. Journal of Marriage and the Family, 59, 5–21. doi:10.2307/353658

Dailey et al. 1043

Surra, C. A., & Longstreth, M. (1990). Similarities in outcomes, interdependence, and conflict in

dating relationships. Journal of Personality and Social Psychology, 59, 501–516.

Thibaut, J. W., & Kelley, H. H. (1959). The social psychology of groups. New York, NY: Wiley.

Waldron, V. R., & Kelley, D. L. (2005). Forgiving communication as a response to relational

transgressions. Journal of Social and Personal Relationships, 22, 723–742.


Charting Changes in Commitment: Trajectories of On-again/Off-again Relationships

Documents

Transcript of Charting Changes in Commitment: Trajectories of On-again/Off-again Relationships