|Home | About | Journals | Submit | Contact Us | Français|
Based on the life course perspective and gender differences in stepparental roles, this study examines frequency of social contact between mid- to late-life stepparents and their stepchildren after stepparents’ marriage to their stepchildren’s biological parent has been dissolved through widowhood or divorce.
Using 5 waves of panel data on stepparent–stepchild pairs from the Health and Retirement Study (N = 12,947 stepchild observations on 4,063 stepchildren belonging to 1,663 stepparents) spanning 10 years (1998–2008), I estimate ordered logit multilevel models predicting former stepparent–stepchild contact frequency.
Results indicate that former stepparents have notably less frequent contact with their stepchildren than current stepparents, particularly following divorce. Widowed stepparents’ contact with their stepchildren diminishes gradually following union disruption, whereas divorced stepparents’ contact frequency drops abruptly. Former stepfathers have less contact with their stepchildren than former stepmothers. Finally, I uncover evidence of the moderating role of (step)parents’ marriage length and stepparents’ number of biological children on widowed stepparent–stepchild contact frequency.
Older stepparents’ social contact with their stepchildren is largely conditional on stepparents’ enduring marital bond to their stepchildren’s biological parent. This study contributes to a growing literature portraying relatively weak ties between older adults and their stepchildren.
Increases in the prevalence of divorce and remarriage that took place over the second half of the past century have resulted in unprecedentedly large numbers of stepparents among current mid- to late-life Americans (Ganong, 2008; Silverstein & Giarrusso, 2010; Stewart, 2007). The growing prominence of stepchildren in older adults’ families is projected to accelerate with the aging of the large Baby-Boom cohorts who experienced particularly high rates of stepfamily formation (Fingerman, Pillemer, Silverstein, & Suitor, 2012; Stewart, 2007; Wachter, 1997) along with relatively low fertility levels (Ganong, 2008; Wachter, 1998).
Relative to intact families, obligations between members of stepfamilies are less clearly defined and guidelines for role performance less institutionalized (Silverstein & Giarrusso, 2010; van der Pas & van Tilburg, 2010). This has raised concern among researchers and policy makers that older stepparent–stepchild ties may be significantly weaker and less enduring than older parents’ ties to their biological children (Fingerman et al., 2012; Pezzin, Pollak, & Schone, 2008; Silverstein & Giarrusso, 2010; Wachter, 1997). Indeed, a small but growing body of research on aging stepfamilies indicates that, relative to adult biological children, stepchildren have less contact and worse relationship quality with their stepparents (van der Pas & van Tilburg, 2010; Ward, Spitze, & Deane, 2009), are less likely to provide stepparents with care and financial support (Henretta, Soldo, & Van Voorhis, 2011; Pezzin et al., 2008), have a lower likelihood of coresiding with a stepparent (Pezzin et al., 2008), and are less likely to receive social and financial support from their stepparents (Pruett, Calsyn, & Jensen, 1993).
This study contributes to the current debate on the strength of older stepparent–stepchild relations. Based on theoretical insights from the life course perspective and previously established gender differences in stepparental roles, I examine the extent to which stepparent–stepchild ties persist after older stepparents are no longer married to their stepchildren’s biological parent (henceforth referred to as former stepparents). With few exceptions (e.g., Klaus, Nauck, & Steinbach, 2012; Schmeeckle, Giarrusso, Feng, & Bengtson, 2006), prior studies have either focused on older stepparents who are currently married to their stepchildren’s biological parent (e.g., Clawson & Ganong, 2002; Schmeeckle, 2007) or overlooked variation in stepparent–biological parent relationship status (e.g., Coleman, Ganong, & Rothrauff, 2006; van der Pas & van Tilburg, 2010; Ward et al., 2009). By differentiating among older stepparents according to their marital relationship with their stepchildren’s parent, this investigation moves attention to the changing and conditional nature of older stepparents’ ties with their stepchildren. Furthermore, this study expands the literature on aging stepfamilies beyond its current focus on contrasting older adults’ relations to their biological children and stepchildren. I focus on the frequency of social contact between former stepparents and their stepchildren. Intergenerational social contact has been associated with exchanges of support and information between older adults and their children (Grundy & Read, 2012; Mancini & Blieszner, 1989). There is also some evidence that frequency of intergenerational contact constitutes an overall measure of the strength of the older parent–child tie (Cooney & Uhlenberg, 1990) and an indirect indicator of intergenerational solidarity (Bengtson & Roberts, 1991).
The purpose of this study is threefold: First, I contrast former and current stepparents’ frequency of contact with their stepchildren. Of particular interest is an examination of whether former widowed and divorced stepparents differ in their contact frequency with their stepchildren. A second purpose of this research is to investigate the effect of duration since (step)parents’ marriage dissolution on former stepparent–stepchild contact. This allows me to determine whether changes in contact frequency following marriage termination take place gradually or abruptly. Finally, I examine the role of four characteristics identified by prior research as having the potential to moderate the relationship between former stepparenthood and contact with stepchildren: (a) the stepchild’s age at (step)parental marriage formation (Ganong & Coleman, 2006a), (b) the duration of (step)parents’ marriage (Aquilino, 2005; Klaus et al., 2012; Orchard & Solberg, 1999; Stewart, 2005), (c) the number of biological children of the stepparent (Orchard & Solberg, 1999; Pruett et al., 1993), and (d) stepparent’s gender (Schmeeckle, 2007). The analyses reported here are based on longitudinal data on stepparent–stepchildren dyads from the Health and Retirement Study (HRS) extending over a 10-year period (1998–2008).
A life course perspective recognizes family relations as dynamic over the life course (Elder, Johnson, & Crosnoe, 2003; Settersten, 2003). Transitions such as entry into widowhood and divorce typically involve a modification of social roles and personal identity, thus creating opportunities for alteration of behaviors and kin relations (Elder & Johnson, 2003; Guiaux, Van Tilburg, & Broese Van Groenou, 2007; Hagestad, 2003; Macmillan & Copher, 2005). Further, a life course perspective directs attention to heterogeneity in social realities and life histories. Such diversity has the potential to play a moderating role in the effect of life course transitions on family ties (Macmillan & Copher, 2005; Settersten, 2003). I suggest hypotheses pertaining to the frequency of former stepparent–stepchild contact based on four fundamental principles of the life course theoretical framework: (a) linked lives, (b) duration, (c) timing, and (d) role configurations.
The concept of linked lives posits that individuals’ lives are interconnected with that of others (Elder et al., 2003; Macmillan & Copher, 2005) and linked across generations by bonds of kinship and processes of intergenerational transmission (Hagestad, 2003). Prior work suggests that dissolution of older (step)parents’ marriage shapes stepchildren’s opportunity structure for interaction with their stepparents (Hans, Ganong, & Coleman, 2009). Relative to current stepparents, older former stepparents are notably less likely to be viewed as family members (Schmeeckle et al., 2006). Former stepfathers also have lower levels of joint activities and material transfers as well as weaker emotional bonds to their grown stepchildren (Klaus et al., 2012). Stepchildren perceive of their continued relationship with former stepparents as voluntary (Clawson & Ganong, 2002; Ganong & Coleman, 2006b; Ganong, Coleman, McDaniel, & Killian, 1998). These findings suggest strong dependency of the older stepparent–stepchild relationship on stepparents’ marital status (Ganong & Coleman, 2006a). Married biological parents, particularly older mothers, may be directly involved in creating and maintaining ties between their grown children and their spouse (McGraw & Walker, 2004). Biological parents’ mediating role may also be indirect, as stepchildren perceive of their relationship with their stepparent as a way to fulfill family obligations to their married parent (Ganong & Coleman, 2006a, 2006b; Ganong et al., 1998). Thus, prior research suggests that former stepparents have less frequent contact with their stepchildren than current stepparents.
Given divorce’s intentional and often conflictual nature, it is likely to result in a greater reduction in stepparent–stepchild contact than the unintended transition to widowhood (Curran, McLanahan, & Knab, 1998). Consistent with this prediction, prior research indicates that weakening of older parent–child ties is notably more pronounced following divorce than following spousal loss, particularly for fathers (Kalmijn, 2007; Shapiro, 2012). Stepchildren perceive of (step)parents’ divorce as a strong motive to sever ties with stepparents (Ganong & Coleman, 2006b; Hans et al., 2009). Finally, stepchildren’s loyalty to their parent (Clawson & Ganong, 2002; Ganong & Coleman, 2006b) may prevent them from maintaining contact with a divorced stepparent more than with a widowed stepparent.
According to the principle of duration, social and personal meaning are attached to the length of time spent in various stages of the life course (Elder et al., 2003). Therefore, in addition to one’s social status as former stepparent, duration in the former stepparent role may have implications for stepparent–stepchild contact (Elder & Johnson, 2003). Consistent with this view, divorced fathers experience a gradual reduction in contact with their children (Seltzer, 1991) and have a greater likelihood of losing regular contact with at least one adult child as duration since divorce increases (Cooney & Uhlenberg, 1990). Longitudinal evidence on changes in intergenerational contact following spousal loss is scarce, but one study reports a short-term increase in contact, followed by a progressive decreasing trend (Guiaux et al., 2007). Thus, I anticipated that former stepparents’ frequency of contact with their stepchildren diminishes with duration since stepparents’ divorce and transition to widowhood. In addition, I expected that increases in the duration of (step)parents’ marriage mitigate reductions in stepparent–stepchild contact frequency after (step)parents’ union dissolution. Within the life course framework, the notion of duration is linked to individuals’ embededness in their social environment (Elder et al., 2003). Although life course transitions often lead to altered behavior, increased duration in a given social role results in an accumulation of the “forces of habituation and obligation” (Elder & Johnson, 2003, pp. 55) and in a greater likelihood of behavioral continuity over the life course (Elder et al., 2003). No prior study has contrasted the strength of stepparent–stepchildren ties in newly formed and long-term stepfamilies. However, there exists some evidence that family boundary ambiguity in stepfamilies declines with (step)parents’ union duration, likely reflecting greater clarity of social roles and increased family integration as stepfamilies become more established (Henderson & Taylor, 1999; Stewart, 2005). Increases in (step)parents’ union duration could therefore promote the establishment of norms for stepfamily relations, thereby offering stepparents and stepchildren greater opportunity for close and enduring intergenerational ties (Aquilino, 2005; Klaus et al., 2012; Orchard & Solberg, 1999).
The principle of timing stipulates that life transitions have different social and personal meanings depending on their timing in individuals’ life course (Elder et al., 2003; Settersten, 2003). Transition into the stepchild role at a young age has the potential to enhance opportunities for stepparent–stepchild coresidence and prolonged contact over time (Ganong, 2008; Stewart, 2007). Stepparents and stepchildren who were young at (step)parent union formation are also more likely to have developed shared activities and interests, and to have built a sense of attachment, and shared family history (Clawson & Ganong, 2002). Conversely, it is likely that older stepparents’ bonds with stepchildren who were adults at (step)parental marriage are weaker because older stepparents have had less time to spend with their stepchildren, limiting opportunities to develop an enduring relationship (Ganong & Coleman, 2006a).
The life course perspective recognizes that the meaning of any given social role is shaped by the existence of other roles (Macmillan & Copher, 2005; Settersten, 2003). The notion of role configurations refers to individuals’ multiple social roles that shape the meaning and experience of each component role (Macmillan & Eliason, 2003). Consistent with this view, stepparents have lower relationship quality with their stepchildren (Pruett et al., 1993) and reduced expectations for physical and emotional involvement with their stepchildren (Orchard & Solberg, 1999) in the presence of own biological children. These findings suggests that when older stepparents are also biological parents, the latter role may take precedence over the former (Clawson & Ganong, 2002; Schmeeckle, 2007; Stewart, 2007). Thus, I expected that stepparents’ biological children exacerbate reductions in stepparent–stepchild contact following divorce and entry into widowhood.
Women, rather than men, typically act as kin-keepers within older intact families, often facilitating and maintaining contact between generations (McGraw & Walker, 2004). Despite negative images of stepmothers and biological fathers’ longstanding relation to their children, prior research documents a similar division of kin-keeping activities within older stepfamilies (Ganong & Coleman, 2004; Vinick & Lanspery, 2000). Schmeeckle (2007) reports that older stepmothers are instrumental in maintaining family ties with their spouse’s children, often mediating previously divorced husbands’ reengagement with grown children. In contrast, stepfathers typically take the lead in responding to stepchildren’s instrumental and financial needs. Such gendered division of roles within older stepfamilies is consistent with broad cultural expectations that women are responsible for the well-being of family members (Ganong & Coleman, 2004), whereas men’s conventional role is that of provider (McGraw & Walker, 2004).
The earlier discussion leads me to postulate the following seven hypotheses:
- Hypothesis 1: Mid- to late-life former stepparents have less frequent contact with their stepchildren than mid- to late-life current stepparents.
- Hypothesis 2: Stepparents who are divorced from their stepchildren’s biological parent have less frequent contact with their stepchildren than stepparents who are widowed from their stepchildren’s biological parent.
- Hypothesis 3: Former stepparents’ frequency of contact with their stepchildren diminishes with duration since stepparents’ spousal loss and divorce.
- Hypothesis 4: Stepchildren who were minors at (step)parents’ marriage formation have more frequent contact with their former divorced and widowed stepparents than stepchildren who were adults when their (step)parents were married.
- Hypothesis 5: Increases in (step)parents’ duration of marriage to their stepchildren’s biological parent mitigate the negative effect of stepparents’ widowhood and divorce on stepparent–stepchild contact.
- Hypothesis 6: Former stepparents’ biological children amplify the negative effect of stepparents’ widowhood and divorce on stepparent–stepchild contact.
- Hypothesis 7: Former stepfathers have less contact with their stepchildren than former stepmothers.
This study uses pooled longitudinal data from five waves (1998, 2000, 2002, 2004, and 2008) of the HRS spanning a 10-year period (Juster & Suzman, 1995). The HRS is an ongoing biennial panel study of mid- to late-life Americans that began in 1992 (N = 12,654 respondents from N = 7,704 households at initial interview). Spouses (and partners) of the HRS respondents were also interviewed. The 1998 HRS is nationally representative of the noninstitutionalized population aged 51 and older. The 2006 wave was omitted from the pooled data because it collected information on intergenerational contact frequency only from a subset of respondents. However, data from the 2006 HRS were used in constructing marital histories over the observation period. Separate information on each one of a respondent’s and their spouse’s (step)children is reported by one family respondent in each household (usually the wife for married couples), not by the child. In each wave, information is sought about each stepchild who had been reported at the previous interview, regardless of the stepparent’s current relationship status with their stepchild’s biological parent. When available, information was drawn from a reconstructed user-friendly longitudinal file developed by the RAND Center for the Study of Aging (St. Clair et al., 2011).
Records for age-eligible respondents were included in the analytical sample if, in 1998, they (a) were married and (b) had at least one stepchild, defined as their spouse’s (but not their own) biological child (N = 1,882). The analytical sample was further restricted to respondents who were interviewed at least once during the four subsequent waves considered (N = 1,705). Each separate record in the analytical data set consists of a stepparent–stepchild dyad at one of the 2000, 2002, 2004, or 2008 waves. Time-invariant explanatory variables were measured in 1998. Stepparent–stepchild contact and other time-varying variables were measured at each of the later waves a stepparent was interviewed and his or her stepchild was listed. Because married stepparents identified in 1998 died during the observation period or left the study, some stepparents were observed at fewer than four occasions. Depending on their number of stepchildren and their continued participation to the HRS between 2000 and 2008, each sampled stepparent contributes a different number of records to the pooled data. Similarly, each stepchild contributes a maximum of four records to the pooled analytical sample. After deletion of observations with missing data (N = 42 stepparents), the total sample consisted of 12,947 stepchild observations on 4,063 stepchildren belonging to 1,663 stepparents who either remained married or experienced a marital transition over the 10-year observation period.
The outcome variable is derived from family respondents’ reports on the number of times in the previous year that they (and their spouse) had had contact—in person, by phone, or by mail—with each one of their own and their spouse’s nonresident stepchildren. Because the distribution of the original detailed variable was highly skewed toward low frequencies (Hox, 2010), the analyses rely on a derived ordered categorical measure of contact frequency. Following Bucx and colleagues (2008), I considered three categories of stepparent–stepchild contact frequency: 1 = less than weekly, 2 = at least once a week but less than daily, and 3 = daily. Daily contact was assumed for coresident stepchildren as the HRS did not seek information on their contact frequency (see Bucx et al., 2008; Tomassini et al., 2004 for a similar approach).
Stepparents’ relation to their stepchildren’s biological parent is captured by four time-varying indicators of marital status. The reference category is a dummy variable equal to 1 for continuously married stepparents (i.e., current stepparents) and 0 otherwise. The currently widowed and currently divorced variables are coded 1 for stepparents who are widowed or divorced from their stepchildren’s biological parent, respectively, and 0 otherwise. The fourth variable is a dummy indicator coded 1 for stepparents who experienced marriage disruption and are either in a new (non)marital union or have experienced dissolution of this new union, and 0 otherwise.
Two variables ascertain the temporal proximity of the interview from a transition to widowhood or a divorce. Respondents who experienced a divorce since their previous interview were asked to report the month and year of this event. Widowhood dates were derived from the spouse’s recorded death month and year or from linked National Death Index data when the latter dates were not available. From this information, decimal years of transition to widowhood and divorce were calculated. Respondents with a missing month or year of divorce (n = 4) or widowhood (n = 10) were assigned the mean decimal year of disruption reported by sampled respondents with valid dates who experienced a corresponding marital disruption in the same 2-year interval. Finally, variables measuring duration since entry into widowhood and divorce were calculated for each wave by subtracting the decimal year of widowhood or divorce from the decimal year of the interview. Respondents who remained married to their stepchildren’s biological parent were coded 0 on both measures of time since union disruption.
Stepparents’ exact decimal date of marriage was calculated based on a reconstituted 1998 measure of marriage length. For respondents with intact marriages in 2008, at each wave, marriage duration represents the number of decimal years between their marriage date and their interview date. The same procedure was employed for respondents who experienced a marital disruption at interview waves preceding the dissolution. At interview waves following the disruption, marriage duration is the total decimal years of marriage at marital disruption, derived as the exact marital dissolution date minus the decimal year of marriage.
Stepparents’ demographic characteristics include age (in years), gender, education (in years), and race/ethnicity (non-Hispanic White, African American, Hispanic, or other). Because family size is negatively related to the strength of each child’s parental relation (Henretta et al., 2011), the models include counts of stepparents’ number of stepchildren and number of biological children. I rely on three indicators to account for stepparents’ health and disability status: self-rated health, number of activities of daily living (ADL) difficulties, and number of instrumental ADL (IADL) difficulties. The poor health variable equals 1 for respondents in fair or poor health (0 = excellent, very good, or good). ADL limitations were assessed for six tasks (walking, dressing, eating, bathing, toileting, and transferring in and out of bed) and IADL limitations were evaluated for five tasks (managing money, preparing meals, getting groceries, using the telephone, and taking medications). Responses were coded dichotomously (1 = difficulty) and summed to create an indicator of ADL limitations ranging from 0 to 6 and an indicator of IADL limitations ranging from 0 to 5. Finally, a measure of economic well-being captures stepparents’ total wealth (sum of all equity, savings, stocks, and investments minus debts). To adjust for skewed distribution and accommodate the presence of negative net worth values, the original variable was transformed by taking the natural logarithm of the raw amount in 10 thousands plus 100.
Stepchildren’s demographic characteristics include gender, age (in years), and partnership status (1 = married or cohabiting, 0 = unpartnered). A dummy variable indicates whether the stepchild was younger than 18 at the time of (step)parental marriage formation. Stepchildren’s age at (step)parental marriage was derived by subtracting (step)parents’ union duration in 1998 from stepchildren’s age in 1998. The models also account for each stepchild’s geographic proximity to their (step)parent (0 = lives farther than 10 miles away, 1 = lives within 10 miles). Finally, indicators for the year of observation (reference is 2000) are included to account for unmeasured changes over time in stepparent–stepchild contact.
I estimated ordered logit multilevel models (Goldstein, 2010) where stepchild observations (Level 1) are nested within stepchildren (Level 2) and stepchildren are nested within stepparents (Level 3). Unlike standard regression models, multilevel models divide residual variance into three levels: variation within each stepchild over study waves, variation between different stepchildren belonging to a same stepparent, and variation between stepparents. Because observations in the analytical sample are clustered into higher-level units, they are not independent, thereby violating a fundamental assumption underlying conventional regression models. By explicitly recognizing the data’s hierarchical structure, multilevel modeling adjusts for biases in parameter estimates and provides correct standard errors and significance tests (Guo & Zhao, 2000). In addition, while accommodating unbalanced data, multilevel modeling allows for the inclusion of covariates measured at all levels of the hierarchical structure.
In the multilevel ordered logit model, Zijk represents the observed ordinal contact frequency outcome at repeated observation i, for stepchild j, belonging to stepparent k. The three-category ordinal outcome is characterized in terms of two cumulative response probabilities defined as:
The three-level ordered logit model is specified as:
where the γc (c = 1, 2) are estimated threshold parameters specifying the observation-invariant link between a latent unobserved continuous variable, , and the observed categorical outcome, Zijk as follows:
In these equations, follows a logistic distribution with constant variance and γ1 < γ2.
The Xk, , Xjk, and terms represent time-invariant stepparent, time-varying stepparent, time-invariant stepchild, and time-varying stepchild variables, respectively. The category-invariant β1, β2, β3, and β4 regression coefficients capture these variables’ relationship with the cumulative logits. A positive coefficient for an independent variable indicates that, as values of the predictor increase, so do the log-odds that the contact frequency response is greater than or equal to c, for any c = 1, 2. The stepchild level random effect (ujk) and the stepparent level random effect (vk) are assumed to be normally distributed with expected values of 0 and variances and , respectively. To insure model identification, the constant term 0 is set to 0.
The results reported subsequently are maximum likelihood estimates using the Generalized Linear Latent and Mixed Model procedure’s adaptive quadrature integration method in Stata (Rabe-Hesketh, Skrondal, & Pickles, 2004).
Descriptive information on stepparents and stepchildren is presented in Table 1. The vast majority of sample observations (87.6%) correspond to current stepparent–stepchild pairs. Roughly 11% and 1% of all observations pertain to widowed and divorced stepparents, respectively. During the observation period, 368 stepparents were widowed and 59 divorced. In the pooled data set, 1,415 records are contributed by widowed stepparent–stepchildren pairs and 110 observations by divorced stepparent–stepchildren dyads. Reflecting the large proportion of observations on continuously married stepparents, the overall mean durations in widowhood and since divorce are 0.3 and 0 years, respectively. Mean durations in widowhood and since divorce across observations on widowed stepparents and divorced stepparents are 3.1 and 3 years, respectively (data not shown). Stepparents’ mean marriage duration is 22.5 years across all records. Across observations corresponding to stepparents who remained married, were widowed and divorced, the mean lengths of marriage are 21.6, 25.3, and 12.8 years, respectively (data not shown). Non-Hispanic White stepparents make up 82.1% of all observations, followed by African Americans (11.9%), Hispanics (4.5%), and other races (1.5%). Approximately two thirds of observations in the sample are of stepfather–stepchild dyads and 45% are of stepchildren who were younger than 18 at the time their (step)parents’ were married. Mean stepparent age across observations is 67.7, whereas mean stepchild age is 43.3.
Observation frequencies on sampled stepparents and stepchildren are shown at the bottom of Table 1. Most stepparents (57.4%) were interviewed at four occasions. Among stepparents who survived the 10-year observation period (78.3% of the sample), 87.2% were interviewed in all four waves (data not shown). The proportions of sampled stepparents observed in one, two, and three wave(s) are 10.6%, 11.8%, and 20.2%, respectively. The mean number of waves a stepparent was observed is 3.2. Of all sampled stepchildren, 10.9%, 13.0%, 22.6%, and 53.5% were observed in one, two, three, and four waves, respectively. The average stepparent and stepchild contributed 7.8 and 3.2 records to the pooled data, respectively.
Table 2 presents the distribution of the outcome measure of stepparent–stepchild contact frequency, by stepparents’ marital status. Overall, 56.1%, 32.6%, and 11.3% of observations were of stepchildren who had less than weekly, weekly, and daily contact with their stepparent, respectively. Table 2 suggests less frequent stepchild contact among former than among current stepparents. For instance, 52.5% of stepchildren in current stepparent–stepchild pairs have less than weekly contact. This compares with 81.2% (p ≤ .01) of stepchildren in widowed stepparents–stepchildren dyads and 89.1% (p ≤ .01) in divorced stepparents–stepchildren pairs. Furthermore, the significant difference (p ≤ .05) in the proportions of widowed stepparents–stepchildren pairs and divorced stepparents–stepchildren pairs with less than weekly contact suggests greater contact frequency following (step)parents’ transition to widowhood than following (step)parents’ divorce.
Table 3 presents results of two ordered logit multilevel models predicting stepparent–stepchild contact frequency. Findings from the main model, which includes all independent variables described earlier, are used to evaluate Hypotheses 1, 2, and 3. Consistent with my expectations, the coefficients on the widowed variable (−0.961, p ≤ .01) and the divorced variable (−3.527, p ≤ .01) are negative and significant. These results indicate that stepparents’ widowhood and divorce from their stepchildren’s biological parent are associated with important reductions in the frequency of stepparent–stepchild contact. Hence, I find strong evidence in support of Hypothesis 1 predicting less frequent contact between former stepparent–stepchild pairs than between current stepparent–stepchild dyads. To assess whether divorced stepparents have less frequent contact with their stepchildren than widowed stepparents (Hypothesis 2), I performed a Wald test (using the covariance matrix of regression coefficients) of equality between the coefficients for the widowed and divorced variables. Test results lead to a rejection of the assumption of coefficient equality (χ2 = 19.6, 1 df, p ≤ .01), thereby offering strong support for Hypothesis 2. The negative coefficient on the variable measuring duration since entry into widowhood (−0.286, p ≤ .01) indicates that widowed stepparents gradually lose contact with their stepchildren as length of time since the death of their stepchildren’s biological parent increases. Conversely, the nonsignificant point estimate on the time since divorce variable (−0.020, p = .854) suggests an abrupt drop in divorced stepparent–stepchild contact as there is no negative influence of the passing of time since divorce. Thus, my results offer partial support for Hypothesis 3 pertaining to reductions in former stepparent–stepchild contact with duration since stepparents’ union dissolution.
The positive coefficient on the male variable in the main model is consistent with the predominance of current stepparent–stepchild pairs in the sample and with stepfathers’ tendency to benefit from their wife’s kin-keeping toward her biological children (Kalmijn, 2007). Hispanic stepparents have notably more contact with their stepchildren than their non-Hispanic White counterparts. Stepparents’ biological children and the length of (step)parents’ marriage are negatively associated with stepparent–stepchild contact, whereas stepparents’ number of IADLs and net worth are positively related to their contact with stepchildren. Generally, coefficients on the variables measuring stepchildren characteristics are in the expected direction. Stepparent–stepchild contact is more frequent with stepdaughters and stepchildren living in proximity to their stepparents. Conversely, stepchildren’s age is negatively related to their frequency of contact with their stepparents. However, I do not find evidence of differences in stepparent–stepchild contact according to whether stepchildren were minors at the time of their (step)parents’ marriage.
To evaluate Hypotheses 4, 5, 6, and 7 pertaining to moderators of the relationship between former stepparenthood and stepparent–stepchild contact frequency, I estimated a second model that adds eight interaction terms to the main model (last two columns of Table 3). Included are interactions between stepparents’ widowed and divorced status and the indicator of whether the stepchild was younger than 18 at (step)parent’s marriage formation. The nonsignificant coefficients on both these interaction terms (−0.208, p = .499 and 1.425, p = .161 for the widowhood and divorce terms, respectively) lead me to reject Hypothesis 4, pertaining to the mitigating effect of stepchild’s young age at (step)parents’ marriage formation on reductions in stepparent–stepchild contact after (step)parents’ marital dissolution. In contrast, among widowed stepparents, I find support for Hypothesis 5. Although the mitigating effect of (step)parents’ longer marriage duration is small, the coefficient on the widowed × marriage length interaction term is strongly significant (0.004, p ≤ .01). However, I do not find support for Hypothesis 5 among divorced stepparents (0.006, p = .371). Consistent with Hypothesis 6, the coefficient on the widowed × number of biological children interaction term is negative (−0.189, p ≤ .01). Therefore, I find evidence that biological children amplify reductions in widowed stepparent–stepchild contact frequency. In contrast, the nonsignificant coefficient on the divorced × number of biological children interaction term (0.024, p = .939) leads me to reject Hypothesis 6 among divorced stepparents. Finally, I examined gender variations in the effect of being a former stepparent on stepparent–stepchild contact frequency. I uncover strong evidence in support of Hypothesis 7 that widowed stepfathers (−1.217, p ≤ .01) and divorced stepfathers (−2.606, p ≤ .05) have notably less contact with their stepchildren than widowed stepmothers and divorced stepmothers, respectively.
Guided by theoretical insights from the life course perspective and based on previously established gender differences in stepparental roles, I examined social contact between mid- to late-life stepparents and their stepchildren after stepparents’ marriage to their stepchildren’s biological parent has been dissolved. In agreement with my expectation based on the principle of linked lives, study results demonstrate that both former widowed and divorced stepparents have significantly less frequent contact with their stepchildren than current stepparents. This finding is consistent with prior evidence showing that former stepparents are less likely to be perceived as family members and have worse ties to their grown stepchildren than current stepparents (Klaus et al., 2012; Schmeeckle et al., 2006). Therefore, I uncover evidence that contact between older stepparents and their stepchildren is, to a large extent, conditional on stepparents’ contemporaneous marital bond. My research makes an important contribution to a growing literature portraying relatively weak ties between older stepparents and their stepchildren, while directing attention to the changing nature of the stepparent–stepchild relation across stepparents’ life course.
The analysis provides evidence that divorce is more detrimental to stepparent–stepchild contact frequency than widowhood. In this sense, an intentional termination of the marital bond such as divorce carries with it greater potential to reduce stepparent–stepchild contact than the unintended transition to widowhood. An alternative interpretation is that my result reflects selection of stepparents with weak ties to their stepchildren into divorce. To test this hypothesis, in ancillary analyses, I reestimated the main model presented in Table 3, adding a dummy indicator of predivorce stepparents (results available upon request). Given the nonsignificant coefficient on this variable (0.468, p = .153), it is improbable that my result reflects a selection effect. (Step)parental divorce is likely to become a prominent dimension of older stepparent–stepchild relations as later-life divorce rates continue to increase (Brown & Lin, 2012; Shapiro, 2003).
I also examined the influence of two temporal dimensions of older stepparents’ marital life course. In accordance with the principle of duration, the data support my hypothesis concerning the negative effect of time since union disruption among widowed stepparent–stepchild pairs. Whereas widowed stepparents gradually lose contact with their stepchildren, divorced stepparents appear to do so abruptly. Moreover, reductions in widowed stepparents’ contact frequency with stepchildren are moderately mitigated by increases in stepparents’ marriage length. I find no corresponding effect among divorced stepparents.
Neither do the data lend support to the mitigating role of stepchildren’s young age at (step)parents’ marriage formation on reductions in stepparent–stepchild contact following union disruption suggested by the principle of timing. It is possible that shared coresidence and stepparents’ involvement in young stepchildren’s lives rather than young age at (step)parents’ marriage formation are predictive of former stepparent–stepchild contact (Schmeeckle et al., 2006).
Consistent with the principle of role configurations, I uncover evidence that increases in stepparents’ number of biological children exacerbate reductions in stepparent–stepchild contact frequency. This finding implies that stepchildren compete with widowed stepparents’ biological children in their relation to their former stepparent. Consistent with older adults’ stronger ties to their biological kin than to their stepkin, when children are present, widowed stepparents may opt to cultivate relations with them to the detriment of ties to their stepchildren.
Finally, reflecting gendered stepparental roles and consistent with prior evidence of stepmothers’ role as kin-keepers (Schmeeckle, 2007), study results reveal greater frequency of stepchild contact among former stepmothers than among former stepfathers. An examination of variations in former stepparent–stepchild contact with stepchildren’s gender was beyond the scope of this study. Yet, prior evidence that the effects of parental divorce on older parent–child contact differ with children’s gender (Aquilino, 1994) and prior findings regarding the strength of the older biological mother–adult daughter tie (Suitor & Pillemer, 2006) call for future research examining the moderating role of former stepparent–stepchild gender concordance.
The findings reported here are partly consistent with prior evidence on the implications of marital disruption for older biological parents’ relations with their children. For instance, it is well established that parental divorce weakens older parents’ ties to their children (Swartz, 2009) and that divorce is particularly detrimental to older fathers’ contact with their children (Cooney & Uhlenberg, 1990; Daatland, 2007; Lye, Klepinger, Hyle, & Nelson, 1995; Shapiro, 2003; Aquilino, 1994). Furthermore, divorce is generally more harmful to older parent–child ties than is widowhood (Kalmijn, 2007; Shapiro, 2012). Nevertheless, this study also reveals important specificities in the effect of marital dissolution on former stepparent–stepchild contact. In particular, in contrast to the results of this study, prior evidence suggests that spousal death does not generally have strong negative effects on biological parent–child ties (Shapiro, 2012) and that widowhood can lead to increased parent–child contact (Guiaux et al., 2007; Roan & Raley, 1996).
Notwithstanding its important contributions, this study has several limitations. First, although my results point to differences in the nature of widowed and divorced stepparents’ ties to their stepchildren, due to the small number of respondents experiencing divorce in this study, results on the implications of divorce should be interpreted cautiously and replicated with larger samples. A second limitation of this study pertains to its exclusive reliance on (step)parents’ reports of social contact with their stepchildren. Although reports of intergenerational contact often differ across generations (Ganong, 2008; Mandemakers & Dykstra, 2008), data constraints prevented me from examining my research questions from the stepchild’s perspective. Third, although I was able to account for the effect of former stepparents’ biological children, data on the existence and quality of relationships with a broader array of kin will be required to fully understand the mechanisms underlying changes in stepparent–stepchild ties associated with (step)parents’ union dissolution. For example, how does a stepchild’s other biological parent affect their relation with former stepparents? What is the role of stepchildren’s relations with stepsiblings and half-siblings in maintaining ties with former stepparents? Does (step)parents’ marital quality and the quality of the stepparent–stepchild bond matter? Moreover, due to small sample sizes, I was not able to examine the implications of nonmarital union dissolution for older adults’ frequency of contact with a former partner’s children. Given the rising prevalence of cohabitation later in the life course (Brown, Lee, & Bulanda, 2006), future studies of older stepparent–stepchild ties should extend to older cohabitors (Schmeeckle et al., 2006). Finally, although this study focused on changes in stepparent–stepchild relations, we lack an understanding of the implications of older (step)parents’ union dissolution for other family ties, including stepsibling, half-sibling, and stepgrandparent relationships (Ganong, 2008; Stewart, 2007).
As the number of older stepfamilies continues to grow, so does the importance of understanding the nature of the relationship between older stepparents and their stepchildren (Stewart, 2007). This study contributes to this research agenda by describing stepparent–stepchild contact as dynamic overtime and largely conditional on the persistence of (step)parents’ marital bond. Therefore, beyond differences in the strength of older biological and stepparents’ intergenerational ties, it is important that researchers consider differences between stepparents in their relations to stepchildren. Future research can fruitfully build on these insights.
This work was supported by funding from the National Institute on Aging (P30 AG017266) awarded to the Center for the Demography of Health and Aging, University of Wisconsin-Madison.
C. Noël-Miller planned the study, conducted all statistical analyses, and prepared the manuscript. The author would like to thank Bob Hauser for insightful comments and suggestions.