https://orcid.org/0000-0002-9240-7187
Abstract
Young adult romantic relationships have undergone significant changes in recent decades, resulting in such distinct demographic trends as rising rates of relationship dissolution. Union dissolution during young adulthood can influence future relationship stability, financial well-being, and health. Reproductive experiences are an important factor that can shape relationship stability. Much of past research, however, has focused on the impact of childbearing on relationship stability while less attention has been paid to other reproductive experiences that could also shape relationship stability, such as pregnancy scares. A pregnancy scare is when a woman suspects she has an undesired pregnancy but later discovers she is not pregnant. This experience might increase or decrease relationship stability. Drawing on data collected from young women in the Relationship Dynamics and Social Life study, this analysis examines the association between pregnancy scares and union dissolution. Results suggest that pregnancy scares are negatively associated with union dissolution, and this relationship persists over time; however, this association varies by relationship type, with serious dating relationships experiencing the most protective benefits.
Introduction
Contemporary young adult romantic relationships differ substantially from those in previous decades. Unlike cohorts born before 1969, most U.S. women now delay marriage until late young adulthood (Bloome and Ang 2020). Instead, young women often cohabit (Manning 2020), even though most millennial women still plan to get married (Manning et al. 2019). Shifts in romantic relationship patterns have resulted in unique demographic trends, including a rise in relationship dissolution among recent cohorts (Eickmeyer 2019; Eickmeyer and Manning 2018). Because of this elevated dissolution rate, young adults are likely to experience multiple relationships prior to their first marriage (Eickmeyer 2019). Previous relationships—given their tendency for dissolution—can affect future relationship stability (Young et al. 2011), financial well-being, and mental health (Avellar and Smock 2005; Halpern-Meekin and Turney 2023; Kamp Dush 2013). Family demographers still do not fully understand the various factors that influence relationship stability and their potential variations across different relationship types, including marriages, cohabitations, and casual dating relationships (Manning 2020; Rosenfeld and Roesler 2019; Sassler et al. 2018).
Reproductive experiences can shape the stability of relationships (Steele et al. 2005; Yabiku and Gager 2009), even such nonmarital relationships as cohabitations (Manning 2004; Wu 1995). Children can also influence a couple's decision to transition to a greater commitment level, such as from cohabitation to marriage (Ter Kuile et al. 2021); however, as rates of nonmarital childbearing rise (Musick 2007; Sweeney and Raley 2014), the role of children in relationship stability might decrease. Unmarried individuals might feel less obligated to stay in a relationship they can easily exit. Conversely, childbearing might still be an investment in the relationship, contributing to romantic union stability, even in young adulthood.
Most past research on reproductive experiences and union stability has focused on the period during pregnancy or after a child's birth. Less attention has been given to other reproductive experiences, such as a pregnancy scare—that is, when a woman suspects that she has an undesired pregnancy but later finds out that she is not pregnant (Barber et al. 2021; Gatny et al. 2014).1 Although far less research has concentrated on pregnancy scares than on pregnancies or births, a national survey revealed that about half of all U.S. women have experienced a scare (The National Campaign to Prevent Teen and Unplanned Pregnancy 2013). While undesired pregnancies, particularly those occurring too soon, have decreased in recent years (Kost et al. 2023), pregnancy scares remain a distinct aspect of this reproductive experience and might not have diminished in frequency. Unlike childbirth, this experience is transient and consistently characterized as undesired. Pregnancy scares typically last for a few days (Nettleman et al. 2009), although in some cases they can last up to eight weeks (Swanson et al. 2014).
Studying pregnancy scares is important because their impact on relationships might differ from that of confirmed pregnancies. Couples often weigh the costs and benefits of having a child in terms of their relationship and personal desires (Emerson 1976; Poortman and Mills 2012). Research on relationship stability and pregnancy suggests that relationships often intensify after a pregnancy is confirmed (Guzzo and Hayford 2014; Lichter et al. 2016; Sassler et al. 2009); however, these relationships are at a high risk of dissolution, especially after birth of a child, if the pregnancy is undesired (Guzzo and Hayford 2014; Lichter et al. 2016). When confronted with a pregnancy scare, couples might respond differently than they would to a confirmed pregnancy. Couples who experience a pregnancy scare might reap the relationship benefits associated with pregnancy without rushing into a highly committed union they are unprepared for. In contrast, they might terminate a relationship they perceive as unsuitable for raising children.
Exploration of pregnancy scares as a distinct area of study allows for the determination of whether experiences related to pregnancy are directly associated with relationship stability beyond the mere presence of a child. Pregnancy scares remain an underexamined aspect of relationship dissolution, primarily because studying such short-term events is challenging. In this article, we use high-frequency longitudinal data to overcome this methodological barrier and examine how pregnancy scares are related to union dissolution in a sample of young women. We concentrate on young adult relationships because of their current elevated dissolution rates (Eickmeyer 2019; Eickmeyer and Manning 2018) and the trend toward delayed childbearing (Kost et al. 2023). These trends prompt the inquiry into whether reproductive experiences beyond childbearing are related to union stability in this demographic. Applying a social exchange framework, we discuss how pregnancy scares might incite couples to assess the costs and benefits of their relationship. We investigate how pregnancy scares and their timing are associated with relationship dissolution using discrete-time hazard models. Furthermore, we explore how the level of relationship commitment modifies the link between pregnancy scares and the risk of dissolution. This study makes theoretical contributions by incorporating pregnancy scares into existing frameworks of relationship stability and adds to the knowledge of relationship dissolution in young adulthood.
Background
Social Exchange, Childbearing, and Relationship Stability
Social exchange theory provides a theoretical perspective for understanding how childbearing is related to relationship stability (Emerson 1976). People remain in relationships because of their attraction to the union, barriers to dissolution, and the availability of alternative partners (Levinger 1976). During relationships, individuals continually evaluate the costs, benefits, and barriers to leaving and incentives to stay (Luke et al. 2011; Yabiku and Gager 2009). Couples must balance both dedication (reasons for staying in a relationship) and constraint commitment (the costs—economic or social—associated with relationship dissolution) (Kelmer et al. 2013; Stanley et al. 2010).
Children can influence both the dedication and constraint commitment in a relationship, thereby affecting the stability of the union. Research suggests children generally increase relationship stability or accelerate the transition to a highly committed relationship (Coppola and Di Cesare 2008; Guzzo and Hayford 2012; Steele et al. 2005; Wright 2019). Childbearing can shape relationship stability through two distinct pathways: discussions and planning related to having a child and the actual birth of a child. Discussions surrounding childbearing can increase relationship stability because of joint decision-making and the shared investment couples have in choosing to have a child together (Duvander et al. 2020; Poortman and Mills 2012). Under social exchange theory, children can be viewed as relationship-specific capital (Emerson 1976; Poortman and Mills 2012). Children can also represent substantial obstacles to leaving a relationship (Levinger 1976). Importantly, whether the child is from a present union matters, as children from previous relationships can destabilize current relationships (Amato 2010; Guzzo 2017). Childbearing might also enhance relationship stability by contributing to increased well-being and happiness after a birth, particularly for first or second births (Kohler et al. 2005; Myrskylä and Margolis 2014). These two different pathways of how childbearing shapes relationship stability beg the question of what is most important: the experiences surrounding the birth of the child, such as discussions and planning, or the presence of the child?
Beyond couple-level reproductive experiences, various relationship- and individual-level factors shape relationship stability. Multiple relationship-level characteristics are related to dissolution, including union duration (Felmlee 1995; Szepsenwol et al. 2015), relationship quality (Arriaga 2001), division of labor (Arpino et al. 2022), and commitment to the relationship (Lau 2012). Various individual-level factors also influence the risk of dissolution, such as race and ethnicity, education (Lundberg et al. 2016; Raley et al. 2015), parental background (Amato 1996; Lyngstad and Jalovaara 2010; Wolfinger 1999; Yabiku and Newmyer 2022), and health (Wagner 2019).
Pregnancy Scares
Contemporary young women often delay childbearing until late young adulthood (Sweeney and Raley 2014; Tillman et al. 2019). This delay highlights the need to consider other early reproductive experiences—such as pregnancy scares—that might also contribute to relationship stability. Additionally, a primary focus on fertility minimizes the importance that reproductive planning and discussions might have for relationship stability.
Research on pregnancy scares is sparse, largely because capturing this experience in data collection is challenging. Pregnancy scares depend on how women perceive the possibility of a pregnancy, with only those who do not desire pregnancy at that time at risk.2 Young women who are racialized as Black; those with unmarried parents, low educational attainment, or a history of low use of contraception; and those who are cohabiting face the highest risk of experiencing pregnancy scares (Gatny et al. 2014). Previous studies on pregnancy scares have focused primarily on contraceptive use (Evans et al. 1976; Gatny et al. 2018), undesired pregnancy (Gatny et al. 2014), and reproductive attitudes (Barber et al. 2021; Newmyer and Yabiku 2022). Women who experience pregnancy scares typically do not change their contraceptive behavior afterward (Evans et al. 1976; Gatny et al. 2018), potentially increasing their risk of undesired pregnancies (Gatny et al. 2014). Pregnancy scares are also linked to attitudes toward abortion (Newmyer and Yabiku 2022) and pregnancy desire (Barber et al. 2021).
Pregnancy scares are important experiences for women, and potentially for couples. Research is just beginning to investigate how pregnancy scares might impact various domains of individuals’ and couples’ lives. Relationship stability is one domain likely influenced by pregnancy scares. During a pregnancy scare, couples might undergo an experience similar to that of an undesired pregnancy but without the possibility of having a child. Research on undesired pregnancies has suggested that young adult relationships often undergo an intensification in commitment but may subsequently dissolve following childbirth (Guzzo and Hayford 2014; Lichter et al. 2016). Pregnancy scares might affect relationship stability similarly or differently than undesired pregnancies.
Pregnancy Scares and Relationship Dissolution
According to social exchange theory, childbearing can increase attraction to a relationship and act as a barrier to dissolution, thereby promoting relationship stability. Pregnancy scares, however, are markedly different from a confirmed pregnancy or having a newborn. Although a pregnancy scare obviously does not lead to childbirth, it might encourage couples to discuss having children, thereby increasing their commitment to the relationship. Such discussions could contribute to relationship stability by increasing mutual attraction to the relationship and desire for a child (Barber 2001). Therefore, a pregnancy scare might reduce the likelihood of a breakup by spurring consideration of a desirable shared future with relationship-specific capital. In addition, a pregnancy scare could motivate couples to enter a stable union, such as transitioning from dating to cohabiting, creating a new barrier to dissolution (Barber et al. 2019; Musick 2007). Given that pregnancy scares have been linked to a heightened pregnancy desire (Barber et al. 2021), a woman might also opt to prolong her relationship to fulfill this personal goal, regardless of whether her partner has the same desire.
Alternatively, pregnancy scares could increase the risk of dissolution. A scare might decrease the attraction individuals feel toward their partner if they cannot envision having a child together (Rijken and Liefbroer 2009; Rijken and Thomson 2011), and reduced attraction could increase the likelihood of relationship dissolution. Furthermore, couples who experience a pregnancy scare may encounter significant stress akin to an undesired pregnancy, potentially diminishing relationship quality and encouraging dissolution (Bouchard 2005).
A third possibility is that pregnancy scares are unrelated to union dissolution. Given that a pregnancy scare is a temporary event and does not result in childbirth, it might not act as a barrier preventing the dissolution of an unhappy relationship. Additionally, because pregnancy scares are undesired events, the union might not reap the protective benefits associated with a couple's joint reproductive planning (Duvander et al. 2020). Prior research on pregnancy scares and attitudes found a temporary association (Newmyer and Yabiku 2022), suggesting that any protective effect of pregnancy scares on union dissolution may not endure over time.
Finally, unlike most pregnancies and births, pregnancy scares may not always be shared with a partner (Guzzo and Hayford 2012, 2014). Relationship quality likely influences whether a scare is disclosed to a partner. If the scare is not communicated to the partner, it is not likely to be viewed as relationship-specific capital and, therefore, might not shape relationship stability; however, a woman may nevertheless make decisions about her relationship on the basis of her pregnancy scare experience. Research has suggested there is a positive association between pregnancy scares and pregnancy desire (Barber et al. 2021), and consequently a woman's changing pregnancy desire could influence her decision to remain in or leave a relationship.
Pregnancy Scares, Relationship Type, and Dissolution
The association between pregnancy scares and union dissolution likely varies with the type of relationship. It is well-established that there are different expectations of commitment, time horizon, and formality according to relationship type (Frisco et al. 2017). For married or cohabiting women, pregnancy scares could protect against dissolution by encouraging family planning discussions, thereby increasing attraction to the relationship. These couples might not perceive increased constraint commitment through a child as negative. Instead, they might be excited about its potential benefits for the relationship. Compared with individuals in less stable unions, cohabitors are more inclined to have children (Poortman and Mills 2012). Research has shown that couples in highly committed relationships with children are the least likely to separate (Musick and Michelmore 2015), and such relationships generally have a low dissolution rate (Lau 2012).
Conversely, women in unions with low commitment, such as casual dating relationships, might view a pregnancy scare as a catalyst for a breakup, because they fear a child might complicate exiting a relationship perceived as short term. For these women, an increase in constraint commitment through a child would likely be a negative consequence. Following a pregnancy scare, they might feel relieved about avoiding an actual increase in constraint commitment, which could have hindered union dissolution. Similarly, if a pregnancy scare increases a woman's pregnancy desire (Barber et al. 2021), she might decide to dissolve a relationship she deems insufficiently stable for raising children.
Methods
Data
Our analysis uses data from the Relationship Dynamics and Social Life (RDSL) study. The RDSL's unique high-frequency measurement enables the examination of weekly changes in pregnancy scares and relationships. In 2008, the RDSL sampled a group of young women who resided in Genesee County, Michigan. The study followed these women weekly for 2.5 years. The baseline interview included 1,003 women aged 18–19, 99% of whom agreed to weekly follow-ups. Three quarters of this sample participated in the study for a minimum of 18 months;3 the weekly surveys yielded a total of 58,594 observations.
During the study, women were asked to report on their romantic relationships and reproductive experiences weekly, which allows us to examine comprehensive data regarding both pregnancy scares and relationship dissolution. These women, in their early adulthood (ages 18–22), faced a high risk of undesired pregnancies (Finer and Zolna 2016; Kost et al. 2023) and pregnancy scares. The majority of women in the study experienced at least one type of relationship during the sampling period (Kusunoki and Barber 2019).
Dependent Variable: Relationship Dissolution
We use discrete-time hazard models estimated through logistic regression to assess the association between pregnancy scares and the weekly likelihood of relationship dissolution. We adopt a single-failure approach, focusing on the first observed relationship breakup, given that 80% of observations are from first relationship spells.4 We use robust standard errors to account for women having multiple relationships during the study. Because our data contain less than 1% missing values, we employ listwise deletion in our analysis, which yields an analytic sample of 28,975 observations from 934 women.
Women become at risk of dissolution when they begin a romantic relationship. At the baseline assessment, women were asked a series of questions, such as “Are you currently engaged to be married?” (as well as less formal arrangements) to determine their relationship status. Subsequently, they were asked weekly whether they remained in the same relationship or if it had changed in its commitment level, such as from dating to engaged. If women reported a breakup, they were asked when this occurred. Relationship dissolution occurs when a woman reports a breakup or no longer identifies the same person as her partner. Because we use discrete-time hazard models, the unit of analysis is a person-week. The dependent variable is a binary variable coded 1 if dissolution occurs and 0 otherwise. Once a woman experiences dissolution, that specific relationship no longer contributes person-weeks to the analysis. If a woman enters a new relationship, it is included in the sample and is removed upon its dissolution or if it is censored.
Women who change the commitment of their relationship while remaining with the same partner are considered as being in an intact relationship: for example, the transition from cohabitation to marriage is not considered dissolution. In our models, we parameterize the baseline hazard using the relationship's duration (measured in weeks) and its square, which reflects the curvilinear relationship between duration and the risk of dissolution in our data. As a robustness check, we also parameterized the baseline hazard with a series of dummy variables, which produced similar results. Women in relationships at the baseline assessment also contribute person-weeks for each week they are observed. Because the RDSL collected data on when each relationship existing at the baseline assessment had begun, we can calculate the duration of these relationships. We conducted an additional analysis that excluded all relationships existing at the baseline assessment, which led to similar conclusions as those presented.
Independent Variables
Pregnancy Scare
We analyze weekly data to determine whether a woman experienced a pregnancy scare. In this article—and in the wider literature—a pregnancy scare is conceptualized as a situation in which a woman who is not intending or desiring to be pregnant believes she is pregnant but later discovers she is not. Pregnancy scares differ from miscarriages because they are always defined as unintended or undesired and do not involve an actual pregnancy.5
We use multiple questions to assess pregnancy scares, guided by previous research that used the RDSL to investigate such experiences (Gatny et al. 2018; Gatny et al. 2014). Each week, women were asked about their desire to avoid pregnancy or to get pregnant in the next month, to which they responded using a five-point Likert scale. Following the guidance created by Gatny et al. (2018) and Gatny et al. (2014), we code respondents into four categories on the basis of their responses to these questions: strongly antinatal; moderately antinatal; ambivalent, indifferent, or moderately pronatal; and strongly pronatal. Each week, women were asked, “Do you think there might be a chance that you are pregnant right now?” They could respond yes or no. A woman who believes that there is a chance she might be pregnant, has not had a positive pregnancy test, and is not strongly pronatal is considered as experiencing a pregnancy scare. Cases where women fit the above criteria but later test positive for a pregnancy within a month are classified as not having experienced a pregnancy scare.6 As noted in the conceptualization of pregnancy scares, we exclude women who are considered strongly pronatal because they would likely experience thinking they are pregnant differently than women who do not strongly desire a pregnancy. Because pregnancy scares are defined as an undesired experience, we include women who are moderately pronatal in our criteria for the experience because they might still consider a pregnancy at the time to be mistimed or undesired, particularly compared with women who are strongly pronatal.7
Additionally, because the association between pregnancy scares and relationship dissolution may fluctuate with the timing of the scare, we examine multiple time-varying measures. First, we include a binary measure of whether a woman is currently experiencing a pregnancy scare. Given that pregnancy scares often resolve quickly, women's experience of this state is brief. Furthermore, this measure assumes that any association between pregnancy scares and dissolution disappears once the scare is resolved. To investigate the possibility of a lasting association, we also consider the time since a pregnancy scare occurred, coded using a time-varying categorical variable that indicates the following exclusive categories: 0 = a woman has not yet had a pregnancy scare, 1 = a scare in the past 31 days, 2 = a scare in the past 32–60 days, and 3 = a scare in the past 61 or more days. If a woman experiences multiple pregnancy scares throughout her relationship, the measure resets to category one (a scare in the past 31 days) with each new scare and the categories then increase on the basis of this latest scare. Last, we create a “sticky” variable to indicate whether a woman has ever experienced a pregnancy scare. This binary variable is coded as 0 until a woman has a pregnancy scare, after which it is set to 1 for all subsequent observations.
Our analysis also considers the importance of a past pregnancy scare by examining whether these experiences might differ if they occurred in a previous versus current relationship. Although a pregnancy scare is an unexpected event, its impact might vary depending on whether a woman has experienced one before. We create a measure of whether a respondent had a pregnancy scare in a previous relationship during the RDSL study. This binary measure is coded as 1 if the respondent had a pregnancy scare in a prior relationship and as 0 if not.
Among women who had a pregnancy scare, 66% experienced one, while 19% had two. In a sensitivity analysis, we excluded respondents who had a high number (>8) of pregnancy scares because of their rarity; the results of these tests were consistent with the conclusions presented. Women who experienced a scare typically reported their first scare around 55 weeks into their relationship.
Relationship Characteristics
Transitions between relationship categories represent important time-varying indicators of relationship types. Each week, respondents were surveyed on previously identified relationships and asked a series of questions to identify new ones. To begin the identification of a partner, women were asked, “In the past [days since last interview], have you had physical or emotional contact, such as kissing, dating, spending time together, sex, or other activities with a partner?” Women were then asked a series of questions to identify a partner so they could be inquired about in future interviews. This survey design enables the capture of even very short-term relationships, such as “hookups.” The RDSL defines relationship types: married or engaged, cohabiting, a special romantic relationship, and a physical/emotional relationship. A relationship is defined as a special romantic relationship if a woman reports being “currently in a special romantic relationship with anyone”; a physical/emotional relationship indicates that a woman is not in a special romantic relationship but, instead, reports being in “any type of relationship that involves physical or emotional contact, such as kissing, dating, spending time together, sex, or other activities with a partner.” Although we retain these categories, which are based on different levels of relationship commitment, we rename them for clarity as “married or engaged,” “cohabiting,” “serious relationship,” and “casual relationship.” The relationship variable is treated as time-varying, meaning relationship types can change during the analysis.8
We also control for various time-varying variables at the partner level. We include a measure of the partner's age and a measure of the proportion of weekly surveys in which a respondent reports having had sex with her current partner; this latter measure is important, as it may indicate relationship quality (Gager and Yabiku 2010) and the likelihood of experiencing a pregnancy scare. In addition, we control for the proportion of surveys in which contraceptive use is not reported. This measure could address an important selection mechanism: women who are ambivalent about pregnancy may not practice birth control consistently (Higgins et al. 2012). Without this control, results might be driven by underlying selection biases instead of the experience of the pregnancy scare itself. Last, we include binary measures for having had a prior pregnancy or a child with the current partner.9
Sociodemographic Variables
We account for multiple static and dynamic sociodemographic variables. Our non-time-varying variables are sourced from the baseline assessment and include government aid receipt, biological parents’ marital status, racial identity, and high school GPA. Because our sample of women is young (ages 18–22), self-reported income may not accurately reflect socioeconomic status. Therefore, we use a binary measure of government aid receipt as an indicator of socioeconomic status. This variable indicates government aid receipt if a woman received any of the following: food stamps; cash welfare; Women, Infants, and Children Program support; or Family Independence Program support. We create a binary measure to indicate whether a woman's biological parents were still married (1 = yes, 0 = no). Racial identity is also a binary variable of non-Hispanic White or not (coded 1 or 0, respectively).10 High school GPA is controlled for as a measure of academic potential, because these young women may not yet have completed their highest level of education.11
We also consider the importance of multiple time-varying sociodemographic variables, assessed at baseline and updated throughout the longitudinal survey. These include the respondent's age, current school enrollment,12 number of prior pregnancies, and number of sexual partners. Age is measured as a continuous variable. School enrollment is coded as a binary variable (1 = yes, 0 = no). Both the number of prior pregnancies and the number of sexual partners are treated as count variables.
Analytic Strategy
Table 1 presents the descriptive statistics for our analytic sample; online appendix C provides further descriptives, segmented by experiences of pregnancy scares. Table 2 presents the results of our logistic regression discrete-time hazard models predicting relationship dissolution. Each model accounts for different timings of pregnancy scares: Model 1 investigates current scares, Model 2 delves into the specific timing of scares, and Model 3 examines having ever experienced a scare. Models control for all sociodemographic and other important variables discussed earlier.
Next, we investigate how pregnancy scares might shape the risk of dissolution differently across various relationship types. In these models, we interact the relationship variable with the ever pregnancy scare variable, while also accounting for all sociodemographic and other critical variables. We then compute predicted probabilities to better understand this interaction, given that it is a nonlinear model (Mize 2019). We present the results of these interactions in Figure 1.
Results
Descriptive Statistics
Given the brief average length of pregnancy scares (approximately two weeks), our sample infrequently reports a current scare, about 2.4% of the weekly observations or 707 observations. However, ever having experienced a pregnancy scare is more common, occurring in 17.9% of weekly observations. Our categorical variable for time since a pregnancy scare indicates that most weekly observations (82.1%) are of women who have not had a scare, followed by those who experienced a scare 61 or more days ago (11.7%), within the past 31 days (4.4%), and 32–60 days ago (1.9%). Across observations, women are predominantly in a serious relationship (55.0% of weekly observations), followed by marriage or engagement (22.3%), cohabitation (17.2%), and casual relationship (5.6%). The average duration of these relationships is 82.2 weeks, or about a year and a half.
Discrete-Time Hazard Models
Table 2 shows the results of our discrete-time hazard models predicting romantic relationship dissolution. The association of pregnancy scares with the log odds of relationship dissolution is dependent on the timing of a pregnancy scare. Model 1 shows that although there is a negative association between currently experiencing a pregnancy scare and dissolution, it is not significant, suggesting that relationship stability likely remains unchanged during the brief period of a current pregnancy scare.
However, Model 2 suggests there is a relationship between pregnancy scares and dissolution over time. Similar to Model 1, these results suggest that having a pregnancy scare in the past 31 days is associated with a decrease in the likelihood of dissolution, although this association is only marginally significant (p < .10). However, pregnancy scares appear to have a delayed association with relationship stability. There is a significant negative association between dissolution and experiencing a scare in the past 32–60 days (β = –.544, p < .05), compared with periods without a scare. This result suggests that women are less likely to break up with their partner in the month or two following a scare. These protective benefits of a pregnancy scare might persist with time. Having a scare in the past 61 or more days—compared with no scare—has a significant negative association with dissolution (β = –.550, p < .001). This result implies that the likelihood of a woman breaking up with her partner decreases as more time passes following a pregnancy scare. Note that we control for the baseline hazard (duration) of relationship dissolution, so these associations are net of any inherent association between relationship duration and dissolution.
Model 3 further indicates that the association between a pregnancy scare and dissolution may persist over time. There is a significant negative association between ever experiencing a pregnancy scare and dissolution (β = –.456, p < .001), implying that ever experiencing a scare in a current relationship may reduce the likelihood of that relationship ending.
Across Models 1, 2, and 3, our results suggest a different association between dissolution and having experienced a pregnancy scare with a previous partner than with a pregnancy scare in a current relationship. We found a significant positive association between having a scare before the current relationship and the dissolution of the present relationship. This finding suggests that women who had a pregnancy scare before their current partnership are at an elevated risk of dissolution.
The results are largely consistent when considering relationship type and control variables. As expected, there is a significant association between relationship type and dissolution. Serious and casual relationships have a higher likelihood of dissolution than marriages or engagements; the log odds of dissolution for cohabiting relationships are not significantly different than those for marriages or engagements.
An increase in the proportion of surveys in which a woman is sexually active with her partner is significantly associated with a decrease in the log odds of relationship dissolution (p < .001). Our results also highlight the importance of past reproductive behavior, as there is a significant negative association between having been pregnant with a current partner and dissolution (p < .01). However, we observe a significant positive association between having a child with a partner and the relationship dissolving (p < .01).
Our results also reveal that relationship dissolution is significantly associated with racial identity and high school GPA. Specifically, non-Hispanic White women and women with a high GPA have the lowest likelihood of ending their relationship. We also see a significant positive association between prior pregnancies and the log odds of dissolution. Last, relationship duration has a significant negative association with dissolution.
Relationship Type and Pregnancy Scares
Next, to better understand whether the association between ever experiencing a pregnancy scare and relationship dissolution varies across relationship types, in Figure 1 we present findings based on models estimated with an interaction between these variables. Overall, the figure highlights that casual relationships have a significantly higher probability of dissolution than married/engaged, cohabiting, or serious relationships; married/engaged and cohabiting relationships have the lowest probability. Compared with women in serious relationships who have never experienced a pregnancy scare, those in such a relationship who have had a scare have a lower probability of dissolution.
Sensitivity Analysis
Besides controlling for having experienced a pregnancy scare in a previous relationship, we also considered the importance of having experienced multiple pregnancy scares within the current relationship. In sensitivity analyses, we measured pregnancy scares as a cumulative count of the number of scares a woman experienced in her current relationship. We defined a pregnancy scare spell as the duration from when a woman suspects she is pregnant until it is confirmed that she is not. Results support those previously presented, as they show a significant negative association between the frequency of pregnancy scare spells within a relationship and dissolution (see online appendix D for the full model output).
Discussion
Understanding the contributors to young adult relationship stability is crucial, particularly in an era characterized by declining fertility rates and increased rates of nonmarital childbearing. This study investigated the association between young women's relationship dissolution and an understudied reproductive event: a pregnancy scare. To better understand this relationship, we examined how the timing of pregnancy scares and relationship type affect this association. Overall, our study suggests a positive relationship between pregnancy scares and relationship stability; however, the nature of the association varies with the timing of the scare. The strongest evidence of a negative association between pregnancy scare and relationship dissolution is if the scare occurred at least a month ago or if a woman had experienced a previous pregnancy scare in her current relationship. Because women typically take a pregnancy test no later than eight weeks after suspecting a pregnancy (Swanson et al. 2014), this time period likely corresponds to when a woman has confirmed that she is not pregnant. Other research on pregnancy scares using the RDSL data found that scares were associated with increased pregnancy desire (Barber et al. 2021). An increase in pregnancy desire following a scare might elevate a woman's likelihood of maintaining her relationship to fulfill this desire. A couple may also be less likely to end their relationship following a scare, possibly because of active attempts to conceive or because they now positively envision having a child together.
Pregnancy scares might also differentially influence the rate of dissolution according to the type of relationship. Earlier research focused on childbearing found that having children can increase the stability of marriages (Thornton 1977) and cohabiting unions (Manning 2004; Wu 1995). Couples in committed unions are also the most likely to decide to have children (Poortman and Mills 2012). Our results suggest that serious relationships in which a pregnancy scare has occurred have a lower likelihood of dissolution than other serious relationships. Married and engaged couples already have substantial relationship-specific capital, and the significance of pregnancy scares may be diminished in these relationships as there is already a high number of existing barriers to ending it. Cohabiting relationships may also be unresponsive to pregnancy scares because the couples already have shared social and economic investments through living together. Furthermore, the impact of pregnancy scares on the stability of casual relationships may be minimal given the high dissolution rate of this union type.
This study builds on social exchange theory by demonstrating how experiences and perceived barriers, such as pregnancy scares, can discourage relationship dissolution by enhancing a couple's perception of their relationship's rewards and potential for future childbearing and success. The results highlight how couples may reevaluate the importance of their relationship when faced with joint uncertainties and investments. A pregnancy scare experience might serve as a barrier preventing dissolution.
We identify multiple implications for understanding the role of reproductive experiences in young adult relationships. Although past research has assessed fluctuations in young adult relationships (Arriaga 2001; De Goede et al. 2012; Szepsenwol et al. 2015) and how relationships respond to pregnancy (Guzzo and Hayford 2012, 2014; Newmyer 2024), our study shows that other reproductive experiences, such as pregnancy scares, can also shape relationship dynamics. Like childbearing experiences, pregnancy scares might enhance relationship stability because couples contemplate parenthood together (Guzzo and Hayford 2014; Ter Kuile et al. 2021), or scares could strengthen the union through a shared experience. Our results also suggest that there is a negative association between experiencing a pregnancy with a partner and relationship dissolution, mirroring what we see in our pregnancy scare results. However, we found a positive relationship between having a child with a partner and dissolution. These contrasting associations could be attributed to the challenges of early childbearing, which can destabilize relationships (Edin and Tach 2011). These varied results emphasize the distinct effects that experiencing a pregnancy versus having a child can have on young adult relationships.
Our study leverages unique longitudinal data, but future research might capitalize on new data and methods to deepen understanding of the link between pregnancy scares and relationship dissolution. Because the RDSL data are longitudinal, they are affected by attrition. The demographic traits of the women least likely to complete 18 months in the study align with some factors that increase the risk of a pregnancy scare (Gatny et al. 2014). If these women had completed the study, the observed association between pregnancy scares and dissolution might have been slightly larger in magnitude; however, these predictions are speculative without direct observation of these women's behavior. More research is also needed on other understudied reproductive experiences, such as miscarriage, and how they relate to relationship stability. Although the RDSL provides detailed data on relationships, we are unable to identify the sex of partners, which prevents us from examining variation between same- and different-sex couples. This limitation also means that some couples are possibly included in the data but are not at risk of experiencing a pregnancy scare. Future research should also consider why pregnancy scares are related with relationship dissolution by examining factors such as relationship conflict, childbearing beliefs, and other family attitudes. Another subject of potential inquiry is how pregnancy scares blur the distinction between mistimed and undesired pregnancies and whether these factors shape relationship dissolution.
This research underscores the continued importance of reproductive experiences in young adult relationships, despite the rising trend of delayed first births and nonmarital births. Pregnancy scares can contribute to relationship stability, a factor that varies depending on the timing of the scare and the relationship type. Future research should continue to explore how reproductive experiences, beyond just childbearing, influence relationship dissolution risk, particularly if current demographic patterns persist.
Acknowledgments
This research was supported in part by the Center for Family and Demographic Research, Bowling Green State University, which has core funding from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD; R24HD050959-08), and the Population Research Institute at Penn State University, which is supported by an infrastructure grant from NICHD (P2CHD041025). A previous version of this manuscript was presented at the 2022 annual meeting of the American Sociological Association.
Notes
We use the term “undesired” instead of “unintended” throughout our manuscript because most fertility preference measures assess pregnancy desires and not intentions, including the fertility preference measure used in our analysis.
This factor means there might be different selection patterns of those who are at risk of having a pregnancy scare, which likely vary across the life course.
Of the 25% who did not complete at least 18 months, some patterns emerged: compared with individuals who stayed in the study longer, they were less likely to be racialized as non-Hispanic White and, at the baseline survey, more likely to receive governmental aid and to not have completed high school.
In our sample, 11.40% of observations are from a second spell. In total, there was a maximum of 18 relationship spells in the data. Sensitivity analyses including all relationship spells yield results consistent with those presented.
Miscarriages are not defined by an individual’s fertility preference and involve an actual pregnancy. Research has shown that miscarriages are unique reproductive experiences (Bommaraju et al. 2016; Broen et al. 2004). Because earlier research has highlighted the distinction between miscarriages and all other reproductive experiences, and because of the very small number of miscarriages in our analytic sample (n = 24), we do not analyze this experience.
There are 34 observations from 22 women that fit this criterion.
We conducted multiple sensitivity analyses (shown in online appendix A) to assess the robustness of this measurement, which suggest similar conclusions as those presented.
There are 348 observations in which a relationship decreases in commitment, and 801 observations showing an increase in commitment.
Because of the high correlation between these two variables (>.5), we also ran analyses including just one, which yielded similar results.
Our sample of women is predominantly non-Hispanic White (n = 542). Our racial measure is a binary measure of non-Hispanic White versus other races and ethnicities because of the small sample sizes of other racial/ethnic groups. Two hundred and ninety-two women identified as Black or African American and 22 as either American Indian/Alaska Native, Asian, or Native Hawaiian or Pacific Islander. Our sample of woman who are of Hispanic origin is also limited (n = 78), thus not allowing us to meaningfully analyze Hispanic ethnicity. See online appendix B for additional sensitivity analyses for our measurement of race and ethnicity.
We impute a total of 1,145 missing observations for GPA as one standard deviation below the mean (3.153485 – 0.6067488).
This measure assesses current enrollment in either a high school, two-year junior or community college, four-year college, vocational school, or another type of school.
