Parental Nonstandard Work Schedules,Parent–Child Communication,and Adolescent Substance Use

Abstract

The purpose of this study was to examine the cumulative impact of parental nonstandard work schedules (NWS) on adolescent alcohol and cigarette use, with a focus on the mediating role of parent–child communication. Using the National Longitudinal Survey of Youth 1979 and its Child Supplement, our path analyses revealed that (a) parental NWS affected adolescent alcohol and cigarette use via the openness of parent–child communication rather than the frequency of parent–child communication and (b) the pattern and directionality of the mediating effects differed by who worked NWS, when parents worked NWS, and what types of NWS parents worked. Implications and directions for future studies are discussed.

Keywords

nonstandard work schedules parent–child communication substance use cigarette use alcohol use shift work

With the rise of a 24/7 economy, a large number of Americans are working nonstandard work schedules (NWS), defined as schedules that lie outside of the typical daytime schedule between 6 a.m. and 6 p.m. (Presser, 2003). A national survey revealed that one fifth of all employed Americans worked evening, late night, or rotating shifts (McMenamin, 2007), and another recent study found that nearly 90% of Americans have worked a nonstandard schedule over the course of their work life (Presser & Ward, 2011). Several studies have found that NWS may have adverse effects on adults’ health and social well-being (Akerstedt, Fredlund, Gillberg, & Jansson, 2002; Gold, 1992; Simon, 1990). Recently, this line of research has been extended to examine the impact of parental work schedules (PWS) on child development and behavior (Barnett & Gareis, 2007; Han, 2008; Han & Waldfogel, 2007; Joshi & Bogen, 2007; Strazdins, Clements, Korda, Broom, & D’Souza, 2006).

However, most research has focused on examining the impact of parental NWS on young children, with a limited understanding of the impact on adolescents (Han, Miller, & Waldfogel, 2010). Many adolescents face difficulties in adjusting to normative changes, such as the onset of puberty and the pursuit of more autonomy, that often lead to increased engagement in risk behavior, such as alcohol and cigarette use (Johnston, O’Malley, Bachman, & Schulenberg, 2011; Smetana, Campione-Barr, & Metzger, 2006). The 2011 Substance Abuse and Mental Health Services Administration (SAMHSA) report revealed that 13.3% of adolescents (ages 12-17) reported alcohol use and 10% reported cigarette use (SAMHSA, 2012). Given that adolescent alcohol and cigarette use are associated with a number of adverse consequences in adolescence and adulthood, such as HIV-risk behaviors, unemployment, divorce, and continued substance use (Bachman, Wadsworth, O’Malley, Johnston, & Schulenberg, 1997; Kiene, Barta, Tennen, & Armeli, 2009), it is important to examine whether parental NWS affects adolescent substance use.

It is also important to examine mechanisms through which parental NWS affect adolescent substance use (Han & Waldfogel, 2007). Research has consistently found that parent–child interaction, particularly parent–child communication (PCC), is a protective factor against adolescent substance use (Jaccard, Dodge, & Dittus, 2002). PCC has the potential to change adolescents’ attitude toward substance use (Ackard, Neumark-Sztainer, Story, & Perry, 2006; Guilamo-Ramos, Jaccard, Dittus, & Bouris, 2006). Given that frequent and appropriate parent–child interaction could be a function of PWS (Täht & Mills, 2012), PCC may serve as a mediator linking PWS with adolescent substance use. To our knowledge, no study to date has examined PCC as a pathway through which PWS affect adolescent substance use. To address these limitations, using a national longitudinal survey data set, this study examines whether parental NWS affect adolescent alcohol and cigarette use, with a focus on the mediating role of PCC.

Literature Review

Parental Nonstandard Work Schedules and Children’s Behavior

Parental NWS have often been conceptualized as a risk factor for negative child behavior (Han, 2008; Heymann, 2000). NWS may restrict the amount of time parents and children spend together; therefore, such schedules may interfere with parents’ ability to maintain healthy interactions with children (Gassman-Pines, 2011). Evidence that shift work results in deleterious health outcomes, such as sleep disorders, chronic fatigue, and mental health issues (Akerstedt et al., 2002; Gold, 1992; Simon, 1990), also suggests that parents might have less physical and mental energy for healthy interactions with their child. The combination of shift work and challenges around spending quality time with children has potential to increase the probability of children engaging in risky behaviors. However, given that NWS might be a preferable choice for some parents (Garey, 1999; Presser, 1995), such schedules may also allow parents to have more time with their children during the day (Joshi & Bogen, 2007). In particular, researchers have found that some parents, especially those with young children, may use a tag-team parenting strategy that desynchronizes their work schedules so that at least one parent is available for child care (Hattery, 2001; Täht & Mills, 2012). Thus, NWS may allow parents with young children to have more parent–child interactions that result in better behavioral outcomes for children.

Previous studies on the impact of NWS on child behavior have produced inconsistent results. Joshi and Bogen (2007), in their study on low-income children, found that 2- to 4-year-old children whose mothers worked non–day shifts had higher problem behaviors because of their mothers’ high levels of parenting stress. Han (2008) also reported a positive association between maternal shift work and behavioral problems in children ages 4 to 10. Similarly, Strazdins et al. (2006) found that parental shift work was associated with children’s (ages 4-10) property offenses, highlighting that such an association might be explained by parental depression and ineffective parenting style. However, other researchers have found contrasting evidence that point to insignificant or positive effects of PWS on children’s behavior. Phillips (2002) reported that parental NWS were not associated with low-income children’s positive behavior. Dunifon, Kalil, and Bajracharya (2005) found insignificant effects of maternal NWS on behavioral outcomes for low-income school-age children. Barnett and Gareis (2007), who examined dual-earner couples with children age 8 to 14, found that maternal evening shift was inversely associated with children’s behavioral problems and risk-taking behavior.

These varied findings might be due to the fact that some studies measured PWS as a dichotomous variable (standard vs. nonstandard). The type of NWS, such as evening shift (between 2 p.m. and midnight), night shift (between 9 p.m. and 8 a.m.), rotating shifts (i.e., alternating between day, evening, or night shifts, but on a fixed schedule), and irregular hours (i.e., some evening or night hours but not on a fixed schedule) affect family processes and children differently (Han & Waldfogel, 2007; Presser, 2003). For instance, evening or night shift work makes it difficult for parents to participate in activities with their children. However, at the same time, this shift may allow parents to provide supervision during the day. Parents who work rotating or irregular shifts might have a difficultly in securing time to spend with their children, although it is possible that such schedules, in particular irregular shifts, might help parents respond to child care needs more flexibly.

These inconsistent findings might also be due to the fact that previous studies have paid little attention to the timing and the cumulative length of PWS. As Han et al. (2010) argued, the duration of past PWS may have cumulative effects on child development. Furthermore, given that children’s tasks and parental influence differ by developmental stages (Shonkoff & Phillips, 2000), these mixed findings may reflect the differential impacts of PWS depending on children’s age. Han et al. (2010) found that children (ages 13-14) were more likely to engage in risk behavior (e.g., substance use) when their mothers worked more years at night or in the evening compared with other types of nonstandard schedules. Additionally, the effects of maternal night shift on adolescent risk behaviors were more salient if mothers worked such schedules when children were younger.

Another source for the inconsistent findings is the exclusive focus on mothers’ work schedules and their role in parenting, with little focus on fathers. As the concept of tag-team parenting suggests, the decision about work schedules is often jointly made by fathers and mothers who share familial responsibilities (Han, 2004). Furthermore, previous studies (Brayfield, 1995; Han, 2004) have found that fathers play a more active role in parenting when mothers have a NWS, suggesting that fathers’ parenting is associated with both mother’s and father’s schedules. Therefore, the potential (positive) effects of mothers’ NWS on children through father’s parenting may be masked unless both maternal and paternal schedules and their roles in parenting are considered. Barnett and Gareis (2007), who modeled both maternal and paternal schedules, found beneficial effects of maternal NWS on children and pointed out that such positive effects were due to fathers spending more time with children and being aware of the children’s activities when wives worked NWS.

Parent–Child Communication and Adolescent Substance Use

It is well established that PCC affects adolescents’ decision to engage in risk behavior (Guilamo-Ramos, Bouris, Dittus, & Jaccard, 2007; Luk, Farhat, Iannotti, & Simons-Morton, 2010; Otten, Harakeh, Vermulst, Van den Eijnden, & Engels, 2007). PCC can foster healthy parent–child bonds and protect children from risks, in that conversations can be value-based, tailored to the child’s age, and the impact of such communication can be sustained over time (Jaccard et al., 2002). Among the various aspects of PCC, the openness and frequency of PCC have been found to play an important role in reducing adolescent substance use (Miller-Day & Kam, 2010). Open communication, as an aspect of the quality of PCC, is when one feels like their opinion is heard and when ideas and decisions are openly shared while the frequency of communication is the number of times the open communication occurs between a parents and a child (Miller-Day & Kam, 2010).

The impact of PCC on child substance use varies by parent and by the openness and frequency of the communication. Otten et al. (2007) found that, in their sample of 428 families, only open PCC, and not frequent communication, was longitudinally associated with children’s attitudes toward smoking, which ultimately affected smoking behavior. The frequency of communication has been found to produce inconsistent effects on children. Van der Vorst, Engels, Meeus, Deković, and Van Leeuwe (2005), in their sample of 428 children ages 13 to 16, found that the frequency of PCC about alcohol actually increased adolescents’ alcohol use, whereas Ennett, Bauman, Foshee, Pemberton, and Hicks (2001) found that frequent discussions were significantly associated with lower adolescent smoking rates. Additionally, scholars have found a differential impact on child substance use by parent (Luk et al., 2010). In their study, with a national sample of 10th graders, Luk et al. (2010) found that mother–son communication was inversely related to cigarette use, and father–son communication was inversely related to their son’s marijuana use. On the other hand, White (2012) found that the frequency of only father–child communication and not mother–child was significantly associated with reduced smoking experimentation. Given these inconsistent results, it is important to conceptually understand and analyze both the differential impact of mother and father communication and the openness and frequency of communication independently.

Research Questions and Hypotheses

Our literature review suggests that both the frequency and openness of mother–child and father–child communication have the potential to mediate the effects of parental work schedules on adolescent substance use. Given the differential effects on child behavior, we take the various types of NWS and both paternal and maternal schedules into account in our analysis. Specifically, using the cumulative length of parental NWS by children’s developmental stages may be more appropriate than the categorical measure of their contemporary schedules (Han et al., 2010). Thus, we examined the impact of the years a mother and a father worked evening shifts, night shifts, and other types of nonstandard schedules that occurred at different phases of their child’s life (preschool age, middle-childhood, and early adolescence) on adolescents’ cigarette and alcohol use, through the openness and frequency of mother–child or father–child communication.

With regard to the effects of PCC on substance use, we expect that more frequent and open PCC will decrease adolescent cigarette and alcohol use. Regarding the effects of parental NWS on the frequency and openness of PCC, it is difficult to anticipate the direction by types of NWS (i.e., evening, night, and other shifts) because each work situation has dual-potential to constrain or facilitate PCC, as discussed earlier. However, tag-team parenting strategies are more actively used when children are younger (Hattery, 2001), and the positive effects of NWS have been found to be salient when NWS were used by parents with young children (Han et al., 2010). Thus, we anticipate that the beneficial effects of all NWS on PCC are likely to occur when parents worked NWS during children’s preschool years, whereas the adverse effects of NWS on PCC would be more dominant when parents worked NWS during children’s middle childhood and adolescence. In addition, given that parental NWS may affect child behavior even via its impacts on the interaction between spouse and children (Barnett & Gareis, 2007), we anticipate that the effects of maternal and parental NWS are mediated by both mother–child and father–child communication.

In sum, we expect that the years a mother worked evening shifts, night shifts, and other NWS, respectively, during children’s preschool years will decrease adolescent alcohol and cigarette use via its positive effects on the frequency and openness of mother child-communication (HM1-FMCC; HM1-OMCC)¹ as well as via its positive effects on the frequency and openness of father–child communication (HM1-FFCC; HM1-OFCC). On the other hand, we expect that the years a mother worked evening shifts, night shifts, and other NWS during their child’s middle-childhood and early adolescence would increase adolescent alcohol and cigarette use via its negative effects on the frequency and openness of mother–child communication (HM2-FMCC and HM3-FMCC; HM2-OMCC and HM3-OMCC) and via its negative effects on the frequency and openness of father–child communication (HM2-FFCC and HM3-FFCC; HM2-OFCC and HM3-OFCC).

Similarly, we expect that the years a father worked evening shifts, night shifts, and other NWS, respectively, during children’s preschool years will decrease adolescent alcohol and cigarette use via its positive effects on the frequency and openness of mother–child communication (HF1-FMCC; HF1-OMCC) as well as via its positive effects on the frequency and openness of father–child communication (HF1-FFCC; HF1-OFCC). On the other hand, we expect that the years a father worked evening shifts, night shifts, and other NWS during middle-childhood and early adolescence would increase adolescent alcohol and cigarette use via its negative effects on the frequency and openness of mother–child communication (HF2-FMCC and HF3-FMCC; HF2-OMCC and HF3-OMCC) and via its negative effects on the frequency and openness of father–child communication (HF2-FFCC and HF3-FFCC; HF2-OFCC and HF3-OFCC).

Method

Data

This study used the National Longitudinal Survey of Youth 1979 (NLSY79) and its Child Supplement (NLSY-CS). Beginning in 1979, the NLSY79 has followed a nationally representative sample of 6,403 men and 6,283 women who were aged 14 to 21 in 1979. In 1986, a separate biennial Child Supplement (NLSY-CS) began to follow all children born to the NLSY female sample. Linking the NLSY79 and the NLSY-CS provides a unique opportunity to examine the intergenerational effects of PWS on PCC and adolescent substance use.

Our sample included all children who had been followed for a 13- to 14-year period since birth to the NLSY female sample, with valid substance use outcomes measured at age 13 or 14. In other words, we pooled children who had been followed for the same period of time (from birth to age 13 or 14) with varying birth years. Pooling in this manner helps us attain a substantial sample size for adolescents who have information on his or her PWS over the course of children’s life, and enhances the heterogeneity of the sample that ensures greater generalizability (Center for Human Resource Research, 2009). Our sample had 3,030 children born between 1984 and 1995 who were followed from birth to age 13 or 14 in 1998 to 2008.

Measures

Dependent Variables

This study used two dichotomous variables of substance use (alcohol and cigarette) measured at age 13 or 14, indicating whether the children had ever used cigarette or alcohol, respectively (1 = Yes and 0 = No).

Independent Variables

Information on maternal and paternal schedules on the primary job was collected using a categorical variable with (a) a category of standard (a shift from 6 a.m. or later to 6 p.m.); (b) evening shift (a shift from 2 p.m. or later to 12 a.m.); (c) night shift (a shift from 9 p.m. or later and to 6 a.m.); (d) other, including split-shift, rotating shift, and irregular hours; and (e) not working.² Han et al. (2010), however, argued that a categorical measure of contemporary parental work schedules may fail to capture the effects of parental work schedules in the past as well as the duration of specific NWS. Therefore, using the categorical measure of the contemporary PWS at each survey year, we created a PWS variables representing the number of years a parent (mother and father, respectively) worked (a) evening, (b) night, or (c) other nonstandard schedules, respectively, by three developmental stage of children—(a) preschool years (age 0 to 4), (b) middle childhood (ages 5 to 10), and (c) early adolescence (ages 11 to 14). For example, to create a variable, the number of years a mother worked evening shifts during children’s preschool years, we counted the number of years a mother had worked evening shift since a child’s birth until the child’s age 4, making the variable represent the sum of the years a mother had worked such schedule between children’s age 0 to 4, with a minimum of 0 and a maximum of 4 years.

Mediators

The frequency of mother–child and father–child communication was measured by a single item asking children (age 13 or 14) about: “How often each of the parents talks over important decisions with the adolescents” (1 = often, 2 = sometimes, and 3 = hardly ever; with reverse coding 1 = hardly ever, 2 = sometimes, and 3 = often). The openness of mother–child and father–child communication was also measured by a single item asking children (age 13 or 14) about: “How well the adolescents and each of parents share ideas or talk about things that really matter” (1 = not very well, 2 = fairly well, 3 = quite well, and 4 = extremely well).

Control Variables

We included a rich set of control variables representing parental and adolescent characteristics that had potential to be correlated with both parental schedules and adolescents’ substance use. The control variables included (a) child characteristics, such as gender, race/ethnicity, and birth order; (b) parental characteristics, such as age, education level at children’s ages 13 to 14, and whether a mother had smoked a cigarette or drank alcohol at pregnancy; and (c) family characteristics, such as the number of family members, residential area, and the number of years the adolescent has lived in single-parenthood at children’s age 13 to 14. In addition, we controlled for parents’ weekly work hours at children’s ages of 13 to 14 and inflation-adjusted family income (constant dollar in year 2000) between children’ age 10 to 14 in that part-timers and low-wage parents are more likely to work nonstandard hours (Presser, 2003). Finally, given that the study sample consisted of children with different birth years, we included birth year dummy variables with a reference category of 1984-1985.

Analytic Strategy

In estimating the relationships among PWS, PCC, and adolescent substance use, we used Mplus 6 and conducted path analyses within a structural equation modeling (SEM) framework. Although this study did not take a latent variable approach, the SEM framework makes it possible to simultaneously estimate the direct effects of PWS on adolescent substance use and their indirect effects on substance use via PCC.

We estimated path models in which (a) the dependent variable (i.e., adolescent substance use) was regressed on the mediators (i.e., the openness and frequency of mother–child and father–child communication) and the independent variables (i.e., the number of years a parent worked three types of NWS, respectively, by three development stages) and control variables and (b) the mediators were also regressed on the independent and control variables (see Figure 1). Because the residuals of mediators were allowed to correlate with each other (Preacher & Hayes, 2008), the path models were fully saturated (i.e., just-identified models), and thus these models have a perfect model fit, representing no difference between the observed covariance matrix and model-implied covariance matrix. Although this approach may increase model complexity, it helps us obtain estimates close to the causal relationship among observed variables, which is particularly important when using observational data.

Given the nature of the pooled children sample, it is possible that multiple children came from the same family, which raised concerns about the non-independence resulting from the clustering. To deal with this issue, we used maximum likelihood estimation with Huber–White standard error (MLR) estimation (Freedman, 2006), which also helped deal with the nonnormality issue with ordinal and dichotomous observed variables here (Muthén & Muthén, 1998-2012). When the MLR estimator is used, Mplus uses linear regression for the mediators and logistic regression for dichotomous substance use outcomes. We tested the indirect effects, based on the product of the linear regression coefficients (the effects of PWS on PCC) and logistic regression coefficients (the effects of PCC on substance use), because the product-based indirect effects are more accurate than the difference-based indirect effects (i.e., total effect − direct effect) and are not susceptible to the scaling problem with linear regression coefficients and logistic regression coefficients (MacKinnon, Lockwood, Brown, Wang, & Hoffman, 2007).³

Missing data were handled by Full Information Maximum Likelihood method, which provides unbiased estimates under the assumption of missing at random and missing completely at random (Enders, 2010). We did not use sampling weights for the regression analyses; however, the weights were used for descriptive statistics. Winship and Radbill (1994) demonstrated that if variables used to select samples are included in a regression model as a control, estimates are unbiased and more efficient without weighting. In our analyses, variables used for oversampling on the NLSY79, such as race/ethnicity and income, were included as controls; moreover, we experimented with the sampling weights in the analyses and found that results were not qualitatively different.

Results

Descriptive Statistics

Descriptive statistics (Table 1) revealed that nonstandard schedules were a prevalent work option for both mothers and fathers. About 60% of mothers and 50% of fathers had worked nonstandard schedules by the time their children were age 13 or 14. Mothers worked nonstandard schedules for an average of 2.71 years and fathers for an average of 2.87 years by the time their children were age 13 or 14. Mothers spent more time working “other” shifts, including rotating shifts, split-shift, and hours vary, than evening and night shifts, regardless of children’s developmental stages. Although fathers also spent most time working other shifts over the children’s life, they spent more time working night shifts than evening shifts as the children got older, unlike mothers who always spent more time working evening shifts than night shifts. 17% of adolescents reported alcohol use at age 13 or 14, whereas 15% reported cigarette use. The prevalence rates were slightly higher than other national estimates (SAMHSA, 2012), and high enough to raise concern even though a minority of adolescents engaged in substance use. With regard to PCC, mothers had better open communication with their children than fathers. The average score for the openness of mother–child communication was 3.11, indicating that the quality was between Quite well and Extremely well, whereas the average score for the openness of father–child communication was 2.70, which was between Fairly well and Quite well. Similarly, mothers had more frequent communication with their children than fathers (2.34 vs. 2.02). Finally, regarding children’s demographics, about 78% of children were non-Black and non-Hispanic, 14% were Black, 8% were Hispanic, and about half of the sample was boys.

Table 1.

Weighted Descriptive Statistics of Study Variables (N = 3,030).

Variables	Mean or %	SD	Variables	Mean or %	SD
Substance use			M-C communication
Alcohol (Yes,%)	16.97		Openness	3.11	0.75
Cigarette (Yes,%)	15.43		Frequency	2.34	0.58
Maternal NWS			F-C communication
Ever worked NWS (%)	59.43		Openness	2.70	0.92
# of years during (P)			Frequency	2.02	0.66
Evening shifts	0.27	0.74
Night shifts	0.06	0.33	Demographics
Other shifts	0.46	0.87	Race/ethnicity (%)
# of years during (M)			Hispanic	7.99
Evening shifts	0.31	0.92	Black	14.13
Night shifts	0.13	0.64	Non-Black, Non-Hispanic	77.87
Other shifts	0.72	1.26	Gender (%)
# of years during (A)			Boy	51.37
Evening shifts	0.16	0.64	Girl	48.63
Night shifts	0.11	0.57	Residence (%)
Other shifts	0.49	1.01	Rural	33.14
Paternal NWS			Urban	66.86
Ever worked NWS (%)	51.54		Birth year (%)
# of years during (P)			1984-1985	19.35
Evening shifts	0.15	0.58	1986-1987	20.14
Night shifts	0.11	0.50	1988-1989	21.60
Other shifts	0.66	1.12	1990-1991	16.11
# of years during (M)			1992-1993	13.16
Evening shifts	0.17	0.72	1994-1995	9.65
Night shifts	0.22	0.89	Family income ($)	97,671	88,005
Other shifts	0.78	1.47	Mother’s education (year)	13.39	2.57
# of years during (A)			Father’s education (year)	13.83	2.83
Evening shifts	0.10	0.56	Mother’s age	41.09	3.66
Night shifts	0.16	0.71	Father’s age	43.28	5.62
Other shifts	0.52	1.16	Single parenthood (year)	2.77	5.00
Weekly work hours			Family size	4.31	1.44
Mother	29.79	16.98
Father	43.09	15.48

Note. (P) = children’s preschool years (age 0-4); (M) = children’s middle-childhood (age 5-10); (A) = children’s early adolescence (age 11-14); M-C communication = Mother-Child communication; F-C communication = Father-Child communication.

Results of Path Analyses for Alcohol Use

Figure 2 presents the path analysis results on adolescent alcohol use. Results indicated that maternal NWS, particularly night shift, during children’s early adolescence increased the likelihood of alcohol use by reducing the openness of PCC, whereas maternal night shift during children’s preschool years decreased the likelihood of alcohol use by fostering the openness of PCC. Specifically, the number of years mothers worked night shift during children’s early adolescence was negatively associated with the openness of mother–child communication (b = −.082, p < .05), which was negatively associated with adolescent alcohol use (b = −.309, p < .05). The resulting indirect effect was significant (b = .025, p < .05), suggesting HM3-OMCC holds for night shift. In contrast, as expected in HM1-OFCC, the number of years mothers worked night shifts during children’s preschool years was positively associated with the openness of father–child communication (b = .116, p < .05), which was related to the lower likelihood of adolescent alcohol use (b = −.146, p < .05), although the associated indirect effect was not statistically significant (b = −.017, p > .05). The remaining hypotheses regarding the effects of maternal NWS via the openness of PCC (i.e., HM1-OMCC, HM2-OMCC, HM2-OFCC, and HM3-OFCC) were not supported by our results. Moreover, maternal NWS were not significantly associated with the frequency of mother–child and father–child communication, and the frequency of mother–child and father–child communication was not significantly associated with adolescent alcohol use, which led to reject all hypotheses regarding the frequency of PCC (i.e., HM1-FMCC, HM1-FFCC, HM2-FMCC, HM2-FFCC, HM3-FMCC, and HM3-FFCC).

With regard to paternal NWS, consistent with HF2-OMCC, we found that the number of years fathers worked evening shifts during children’s middle-childhood was negatively associated with the openness of mother–child communication (b = −.087, p < .05), which was negatively associated with adolescent alcohol use too (b = −.309, p < .05). However, the associated indirect effect was not statistically significant (b = −.027, p > .05). Other hypotheses regarding the indirect effects of paternal NWS via the openness of PCC (i.e., HF1-OMCC, HF1-OFCC, HF2-OFCC, HF3-OMCC, and HF3-OFCC) were not supported by our results either. In addition, although the number of years fathers worked evening and night shifts during children’s middle-childhood significantly decreased the frequency of father–child communication (b = −.078, p < .05, and b = −.059, p < .05, respectively), the associated indirect effects were not statistically significant because the frequency of father–child communication had no significant association with adolescent alcohol use. The frequency of mother–child communication also had no significant effects on adolescent alcohol use. As a result, all hypotheses regarding the frequency of PCC (i.e., HF1-FMCC, HF1-FFCC, HF2-FMCC, HF2-FFCC, HF3-FMCC, and HF3-FFCC) were not supported (see Table 2).

Table 2.

Path Coefficients (Dependent Variable: Alcohol Use; N = 3,030).

	Alcohol			Open. of M-C Comm.			Open. of F-C Comm.			Freq. of M-C Comm.			Freq. of F-C Comm.
	b	SE	B	b	SE	B	b	SE	B	b	SE	B	b	SE	B
Parent–child communication
Open. of M-C Comm.	−.309***	.075	−.120
Open. of F-C Comm.	−.146*	.074	−.070
Freq. of M-C Comm.	−.036	.099	−.011
Freq. of F-C Comm.	.028	.103	.010
Materanl NWS (# of years)
Evening shifts (P)	.003	.080	.001	.000	.022	.000	−.027	.027	−.022	.016	.017	.021	.031	.019	.035
Night shifts (P)	−.151	.190	−.026	−.014	.046	−.006	.116*	.055	.042	.013	.034	.007	−.002	.043	−.001
Other shifts (P)	−.049	.067	−.022	−.012	.018	−.014	−.029	.022	−.027	−.013	.014	−.020	−.013	.015	−.017
Evening shifts (M)	.023	.069	.011	−.022	.018	−.027	−.021	.023	−.021	−.012	.014	−.020	−.020	.016	−.028
Nght shifts (M)	.040	.095	.013	.033	.025	.028	−.014	.035	−.010	.007	.020	.008	.009	.020	.009
Other shifts (M)	−.013	.051	−.008	.005	.013	.008	.023	.015	.031	−.003	.010	−.006	.008	.011	.016
Evening shifts (A)	.039	.085	.013	−.014	.026	−.012	−.047	.029	−.032	.018	.019	.019	−.024	.022	−.024
Night shifts (A)	−.010	.100	−.003	−.082**	.026	−.062	−.065	.033	−.040	−.005	.021	−.004	.009	.024	.008
Other shifts (A)	−.015	.055	−.008	.003	.015	.005	.001	.017	.001	.020	.011	.034	.012	.013	.019
Paternal NWS (# of years)
Evening shifts (P)	.074	.149	.023	.070	.040	.054	.034	.051	.021	.030	.031	.030	.046	.035	.041
Night shifts (P)	−.135	.248	−.035	.069	.051	.046	.093	.061	.050	−.030	.037	−.026	.017	.043	.013
Other shifts (P)	.032	.081	.019	−.003	.022	−.005	.006	.026	.007	−.025	.016	−.048	−.003	.019	−.006
Evening shifts (M)	−.073	.151	−.027	−.087*	.041	−.084	−.085	.048	−.066	−.035	.029	−.044	−.078*	.035	−.086
Night shifts (M)	.056	.128	.026	−.039	.036	−.047	−.066	.039	−.063	−.012	.024	−.018	−.059*	.028	−.079
Other shifts (M)	.104	.077	.080	.005	.020	.011	−.015	.025	−.024	.019	.015	.048	−.001	.018	−.002
Evening shifts (A)	.093	.163	.027	.063	.041	.047	.077	.048	.047	.001	.029	.001	−.006	.034	−.005
Night shifts (A)	−.332	.180	−.123	.024	.035	.023	.016	.036	.012	.018	.023	.022	.041	.028	.044
Other shifts (A)	.021	.079	.013	−.014	.023	−.022	.019	.028	.024	−.015	.017	−.029	.006	.020	.011
Child’s sex (ref. = male)	.153	.105	.040	.045	.027	.030	−.260***	.033	−.141	.091***	.021	.078	−.173***	.024	−.131
Child’s race/ethnicity
(ref.= Non-Black, Non-Hispanic)
Black	−.296	.155	−.069	.100*	.041	.060	.046	.049	.022	.124***	.031	.095	.148***	.037	.101
Hispanic	.159	.145	.033	−.005	.040	−.003	.112*	.047	.049	.026	.031	.018	.128***	.035	.078
Family size	−.174***	.044	−.130	−.005	.011	−.009	−.021	.014	−.033	.000	.009	−.001	−.013	.011	−.028
Birth order	.115*	.054	.070	−.027	.014	−.042	−.023	.018	−.030	−.035**	.011	−.070	−.020	.013	−.036
Family income (logged)	.076	.051	.054	.014	.017	.026	.011	.019	.016	.005	.010	.011	−.004	.014	−.008
Mother’s drinking at pregnancy	.196	.111	.051	−.030	.030	−.020	−.047	.036	−.025	−.030	.023	−.026	−.023	.026	−.018
Residence (ref. = Rural)	.113	.125	.026	−.007	.031	−.004	.015	.039	.007	−.012	.025	−.009	.010	.028	.007
Mother’s education	−.072*	.029	−.096	.014*	.007	.049	−.007	.009	−.020	.011*	.005	.048	−.012	.007	−.048
Father’s education	.029	.033	.044	−.019*	.007	−.073	.023*	.010	.073	−.010	.006	−.051	.029***	.007	.125
Mother’s work hour	.010**	.003	.085	−.002*	.001	−.046	−.002	.001	−.031	−.001*	.001	−.041	−.001	.001	−.022
Father’s work hour	−.007	.005	−.057	.002	.001	.032	−.004*	.002	−.063	.001	.001	.016	−.002	.001	−.049
Mother’s age	−.042	.027	−.079	.004	.007	.021	−.010	.009	−.040	.002	.006	.015	−.010	.006	−.055
Father’s age	.001	.015	.002	.001	.004	.010	.003	.005	.017	.002	.003	.024	.000	.003	.004
Years of single-parenthood	.017	.016	.045	−.007	.005	−.045	−.052***	.006	−.283	.005	.003	.040	−.026***	.004	−.196
Birth year (Ref: 1984-1985)
1986-87	.079	.177	.016	−.002	.046	−.001	.041	.055	.017	−.085*	.035	−.056	−.024	.040	−.014
1988-89	.329	.180	.072	−.089	.048	−.050	−.002	.058	−.001	−.078*	.037	−.056	.004	.043	.002
1990-91	.541*	.220	.101	−.075	.057	−.036	.013	.071	.005	−.123**	.046	−.076	.159**	.052	.086
1992-93	.534*	.251	.093	−.080	.066	−.036	−.008	.080	−.003	−.093	.054	−.053	.171**	.059	.086
1994-95	.507	.298	.076	−.209*	.081	−.080	−.051	.097	−.016	−.132*	.061	−.065	.131	.071	.057
R ²		.110			.027			.113			.039			.081
AIC								273,862
BIC								279,806

Note. (P) = preschooler (age 0-4); (M) = middle-childhood (age 5-10); (A) = adolescence (age 11-14); b = unstandardized coefficient; B = standardized coefficient; Open. of M-C Comm. = openness of mother–child communication; Open. of F-C Comm. = openness of father–child communication; Freq. of M-C Comm.= frequency of mother-child communication; Freq. of F-C Comm.= frequency of father-child communication; Coefficients on alcohol use indicate logistic regression coefficient.

p < .05. **p < .01. ***p < .001.

Results of Path Analyses for Cigarette Use

As with adolescent alcohol use, our results for cigarette use (see Figure 3) indicated that maternal night shifts during children’s early adolescence increased the likelihood of cigarette use by lowering the openness of PCC, while such a (maternal) schedule during children’s preschool years decreased the likelihood by promoting the openness of PCC. That is, the number of years mothers worked night shifts during children’s early adolescence was negatively associated with the openness of mother–child communication (b = −.077, p < .05), and the openness of mother–child communication was negatively associated with adolescent cigarette use (b = −.363, p < .05). Therefore, the associated indirect effect was statistically significant (b = .028, p < .05), which supports HM3-OMCC with night shifts. In contrast, the years of maternal night shifts during children’s preschool years was positively associated with the openness of father–child communication (b = .120, p < .05), which in turn decreased the likelihood of adolescent cigarette use (b = −.172, p < .05), supporting HM1-OFCC. However, the associated indirect effect was not statistically significant (b = −.021, p > .05). In addition, similar to the path model for alcohol use, other hypotheses regarding the effects of maternal NWS on cigarette use via the openness of PCC (i.e., HM1-OMCC, HM2-OMCC, HM2-OFCC, and HM3-OFCC) were not supported by our results. All hypotheses regarding the frequency of PCC (i.e., HM1-FMCC, HM1-FFCC, HM2-FMCC, HM2-FFCC, HM3-FMCC, and HM3-FFCC) were rejected as well. Finally, it is worth noting that, unlike the path model for alcohol use, even after controlling for PCC, there were significant direct effects of (a) the years of maternal night shifts during children’s middle-childhood and (b) the years of other types of maternal shifts during the early adolescence, such that the former increased the likelihood of adolescent cigarette use (b = .179, p < .05), whereas the latter lowered it (b = −.128, p < .05).

Paternal NWS also had the similar pattern of indirect effects, given that paternal evening shifts during middle-childhood decreased, but paternal night shift during children’s preschool years increased the likelihood of adolescent cigarette use through the openness of PCC. Consistent with HF2-OMCC for evening shifts, the number of years fathers worked evening shifts during children’s middle childhood was negatively associated with the openness of mother–child communication (b = −.091, p < .05), which is negatively associated with adolescent alcohol use (b = −.363, p < .05). The associated indirect effect was statistically significant too (b = .033, p < .05). In addition, consistent with HF1-OFCC (but was not supported with alcohol use), we found that the years of paternal night shifts during children’s preschool years was positively associated with the openness of father–child communication (b = .122, p < .05), which was negatively associated with the likelihood of adolescent cigarette use (b = −.172, p < .05), although the associated indirect effects was not statistically significant. Other hypotheses regarding the indirect effects of paternal NWS via the openness of PCC (i.e., HF1-OMCC, HF2-OFCC, HF3-OMCC, and HF3-OFCC) were not supported, and all hypotheses regarding the frequency of PCC (i.e., HF1-FMCC, HF1-FFCC, HF2-FMCC, HF2-FFCC, HF3-FMCC, and HF3-FFCC) were not supported, as with alcohol use (see Table 3).

Table 3.

Path Coefficients (Dependent Variable: Cigarette Use; N = 3,030).

	Cigarette			Open. of M-C Comm.			Open. of F-C Comm.			Freq. of M-C Comm.			Freq. of F-C Comm.
	b	SE	B	b	SE	B	b	SE	B	b	SE	B	b	SE	B
Parent–child communication
Open. of M-C Comm.	−.363***	.082	−.136
Open. of F-C Comm.	−.172*	.082	−.080
Freq. of M-C Comm.	.147	.110	.043
Freq. of F-C Comm.	−.017	.113	−.005
Materanl NWS (# of years)
Evening shifts (P)	−.096	.081	−.035	.005	.023	.005	−.023	.027	−.018	.020	.017	.025	.031	.020	.035
Night shifts (P)	−.094	.175	−.016	−.010	.047	−.005	.120*	.054	.043	.013	.034	.008	−.003	.043	−.002
Other shifts (P)	−.124	.080	−.054	−.005	.018	−.006	−.024	.022	−.023	−.010	.014	−.014	−.011	.015	−.015
Evening shifts (M)	.052	.064	.024	−.023	.018	−.028	−.023	.022	−.023	−.013	.014	−.021	−.023	.016	−.032
Nght shifts (M)	.179*	.087	.058	.030	.025	.025	−.012	.033	−.008	.006	.020	.007	.011	.021	.011
Other shifts (M)	.095	.049	.060	−.001	.013	−.001	.015	.015	.021	−.006	.010	−.013	.004	.011	.007
Evening shifts (A)	.021	.085	.007	−.013	.026	−.011	−.048	.029	−.033	.021	.020	.022	−.022	.022	−.021
Night shifts (A)	−.122	.097	−.035	−.077**	.026	−.058	−.060	.034	−.037	.000	.021	.000	.009	.024	.008
Other shifts (A)	−.128*	.060	−.065	.007	.015	.010	.007	.017	.008	.022	.012	.038	.015	.013	.023
Paternal NWS (# of years)
Evening shifts (P)	.037	.168	.011	.075	.040	.059	.018	.052	.012	.030	.031	.031	.037	.036	.033
Night shifts (P)	.082	.185	.020	.070	.051	.047	.122*	.061	.066	−.031	.038	−.027	.028	.043	.021
Other shifts (P)	.080	.088	.046	−.001	.022	−.002	−.004	.027	−.005	−.023	.017	−.046	−.012	.020	−.021
Evening shifts (M)	.106	.148	.039	−.091*	.041	−.088	−.085	.049	−.067	−.035	.030	−.044	−.078*	.035	−.085
Nght shifts (M)	.091	.121	.040	−.028	.036	−.033	−.067	.040	−.064	−.012	.024	−.018	−.055*	.028	−.073
Other shifts (M)	.068	.083	.050	.006	.021	.012	−.011	.026	−.017	.020	.016	.049	.003	.018	.007
Evening shifts (A)	.047	.149	.013	.068	.041	.051	.082	.049	.050	−.008	.030	−.008	−.002	.034	−.002
Night shifts (A)	−.138	.129	−.049	.014	.035	.014	.012	.037	.009	.016	.024	.020	.037	.029	.040
Other shifts (A)	.049	.085	.028	−.020	.023	−.031	.016	.028	.020	−.020	.018	−.040	.005	.019	.008
Child’s sex (ref. = male)	.153	.110	.038	.045	.027	.030	−.260***	.033	−.140	.091***	.021	.078	−.173***	.024	−.131
Child’s race/ethnicity
(ref.= Non-Black, Non-Hispanic)
Black	−.737***	.168	−.166	.104**	.040	.062	.053	.049	.026	.134***	.031	.102	.150***	.036	.102
Hispanic	−.317*	.160	−.064	−.005	.040	−.003	.107*	.047	.047	.028	.031	.019	.127***	.035	.077
Family size	−.119*	.049	−.086	−.004	.011	−.007	−.017	.014	−.026	.002	.009	.004	−.011	.011	−.025
Birth order	.069	.055	.040	−.023	.014	−.036	−.018	.018	−.023	−.032**	.011	−.064	−.016	.014	−.029
Family income (logged)	−.030	.046	−.021	.014	.016	.025	.010	.019	.015	.004	.011	.009	−.001	.014	−.002
Mother’s smoking at pregnancy	.689	.376	.118	−.122	.081	−.056	−.159	.102	−.059	−.102	.070	−.060	−.085	.077	−.044
Residence (ref. = Rural)	−.099	.124	−.022	−.007	.031	−.004	.010	.039	.005	−.013	.025	−.010	.007	.028	.004
Mother’s education	−.057	.031	−.074	.011	.007	.038	−.012	.010	−.034	.008	.006	.036	−.015*	.007	−.058
Father’s education	−.042	.030	−.060	−.021**	.007	−.081	.024*	.010	.075	−.012*	.006	−.060	.029***	.007	.125
Mother’s work hour	.007	.004	.060	−.002*	.001	−.048	−.001	.001	−.027	−.001*	.001	−.040	−.001	.001	−.018
Father’s work hour	−.006	.004	−.047	.002	.001	.038	−.004*	.002	−.067	.000	.001	.012	−.003*	.001	−.061
Mother’s age	−.064*	.027	−.118	.005	.007	.023	−.011	.009	−.043	.002	.006	.014	−.010	.006	−.056
Father’s age	.030*	.014	.085	.000	.004	.003	.003	.005	.021	.002	.003	.022	.000	.003	.003
Years of single-parenthood	.036*	.016	.090	−.005	.004	−.033	−.050***	.006	−.272	.005	.003	.040	−.025***	.004	−.189
Birth year (Ref: 1984-1985)
1986-87	−.078	.167	−.015	−.001	.046	−.001	.035	.055	.014	−.086*	.035	−.056	−.029	.040	−.016
1988-89	−.116	.182	−.024	−.096*	.049	−.054	−.013	.059	−.006	−.084*	.037	−.060	−.006	.044	−.004
1990-91	−.236	.235	−.042	−.079	.058	−.038	−.001	.072	.000	−.126**	.047	−.077	.151**	.053	.082
1992-93	−.546	.284	−.091	−.080	.066	−.036	−.022	.081	−.008	−.095	.054	−.054	.161**	.059	.081
1994-95	.006	.308	.001	−.215**	.082	−.083	−.067	.100	−.021	−.139*	.063	−.069	.117	.072	.051
R ²		.172			.030			.117			.041			.083
AIC								270,189
BIC								276,134

Note. (P) = preschooler (age 0-4); (M) = middle-childhood (age 5-10); (A) = adolescence (age 11-14); b = unstandardized coefficient; B = standardized coefficient; Open. of M-C Comm. = openness of mother–child communication; Open. of F-C Comm. = openness of father–child communication; Freq. of M-C Comm.= frequency of mother-child communication; Freq. of F-C Comm.= frequency of father-child communication; Coefficients on cigarette use indicate logistic regression coefficient.

p < .05. **p < .01. ***p < .001.

Discussion

We examined whether parental evening shifts, night shifts, and other types of NWS had cumulative effects on adolescent alcohol and cigarette use, through the openness and frequency of PCC. The major findings of our study are as follows: (a) parental NWS affect adolescent alcohol and cigarette use through the openness of PCC reflecting one aspect of PCC quality rather than the frequency of PCC and (b) the pattern and directionality of the mediating effects differ by who worked NWS, when parents worked NWS, and what types of NWS parents worked.

Our finding that the openness of PCC, rather than the frequency, serves as a protective factor against adolescent alcohol and cigarette use is consistent with research on the impact of PCC on adolescent substance use (Otten et al., 2007). Specifically, the frequency of mother–child communication was not affected by maternal NWS, although some paternal NWS (i.e., evening and night shifts during children’s middle childhood) affected the frequency of father–child communication. These differential effects of NWS on the frequency by parent suggests that, as Bianchi (2000) and Barnett and Gareis (2007) pointed out, mothers’ NWS, unlike fathers’ NWS, do not reduce mother’s total time with children because mothers tend to sacrifice their leisure time or spend more time with children when they are not at work. However, it should be noted that even if mothers make an effort to maintain the frequency of PCC, their NWS still affected the quality of PCC (e.g., openness), suggesting that it is important to focus on how parental NWS affect the openness of PCC that in turn affects adolescent substance use.

Our study indicated that the effects of maternal NWS on the openness of PCC differ by the types of NWS and when they worked such schedules. Notably, the negative effect of maternal NWS was salient when mothers worked NWS during children’s adolescence. Moreover, this association particularly held for night shifts, in that the number of years mothers worked night shifts during children’s early adolescence increased the likelihood of adolescent alcohol and cigarette use through its negative association with the openness of mother–child communication. This evidence suggests that night shifts (beginning at 9 p.m. or later) make it more difficult for parents to maintain family routines (e.g., having dinner together) in which parent–child communication often takes place (Täht & Mills, 2012). This also implies that mothers’ night shift work may impose a toll on mothers’ physical and mental energy, making it difficult to have conversations with their children during the daytime.

In contrast, the number of years mothers worked night shifts when children were young had the potential to foster the openness of PCC and thus decrease adolescent alcohol and cigarette use. We found that mothers’ years of night shift work during children’s preschool years increased the openness of father–child communication, which ultimately played a preventive role in adolescents’ cigarette and alcohol use, although the associated indirect effects were relatively weak in magnitude. This may be a result of parents, especially those with young children, using a tag-team parenting (Hattery, 2001). When tag-team parenting is used, fathers’ involvement in child care tends to increase because such couples often shift child care from formal center-based care to fathers while mothers are working (Han, 2004). Coupled with our findings and evidence from previous studies (Barnett & Gareis, 2007; Brayfield, 1995; Han, 2004), it can be implied that mothers, when working night shifts, might use tag-team parenting during children’s preschool years, and such parenting choices may increase parent–child interaction, especially for fathers.

The effects of paternal NWS on the openness of PCC also differ by children’s developmental stage. Like those of maternal NWS, the years of paternal night shifts during children’s preschool years decreased adolescent cigarette use through its positive effect on the openness of father–child communication. Although the associated path held true only for cigarette use, this evidence is in line with the positive effects of NWS produced by tag-team parenting with younger children. However, paternal evening shifts, if worked when children were older, had the potential to increase the likelihood of adolescent alcohol and cigarette use because it hindered the openness of mother–child communication. It is not surprising to find that parental NWS affected the interaction between spouse and child, as shown with maternal night shift (during preschool years), resulting in the positive effects on father–child communication. Unlike maternal night shifts, however, paternal evening shifts had a negative effect on mother–child communication. One potential reason for this pattern might be that fathers’ evening shift work has detrimental effects on marital quality (e.g., marital happiness and stability; White & Keith, 1990; Presser, 2000), and such negative marital interaction may adversely affect mother–child interactions (Kitzmann, 2000; Sobolewski & Amato, 2007). Another potential reason may be due to ceiling effects, meaning that most mothers have a high level of open communication with their children, as shown in the descriptive statistics. Therefore, there is little room for paternal NWS to increase the openness of mother–child communication, making it more likely to find negative (or insignificant) effects. In any case, given the lack of studies on the association between paternal NWS and mothers’ interaction with children, additional analyses are needed to better understand the nature of this negative association between paternal NWS and mother–child communication.

Finally, we found a direct effect of mothers’ night shift work during children’s middle-childhood on adolescent cigarette use, even after accounting for the mediators of PCC. The direct association between the number of years mothers worked other shifts during children’s early adolescence and adolescent cigarette use was also significant. These direct effects may be due to other pathways that were unspecified in our analyses, but may link parental NWS with adolescents’ cigarette use rather than real direct effects. Han et al. (2010), for example, documented that time spent together, parent–child closeness, parental knowledge about of child’s whereabouts, and home environment are potential pathways that connect parental work schedules to adolescent risk behaviors. Consistent with our findings, Han et al. (2010) reported that mothers’ night shift work during children’s middle-childhood led to higher incidents of risk behaviors, because such schedules were associated with less time spent together and an unfavorable home environment. Their finding that mothers’ other shift work during children’s early adolescence was related with more knowledge of children’s whereabouts may also explain the negative association between mothers’ other shift work and adolescent cigarette use that was found in the present study.

Limitations

This study has several limitations that should be noted. First, as in all nonexperimental studies, our analyses are not freed of selection bias or omitted variable bias, and thus it is necessary to make a causal interpretation of our findings with caution. Presser (2003) pointed out that those who are young, African American, poorly educated, and unskilled are more likely than others to work nonstandard schedules, implying that parents working nonstandard schedules may represent distinct groups. Therefore, the association that we found between parental NWS and adolescent substance use could be due to family and child characteristics. However, our analysis controlled for an extensive set of family and child characteristics. We experimented with different sets of control variables by progressively adding control variables and found that our results were robust. Nevertheless, it is still possible that the results might be biased by not including important but unmeasured control variables to our model. In future work, statistical models or research designs that better account for the unobserved characteristics need to be considered.

Second, on the NLSY79, information about work schedules is not detailed enough to capture the full spectrum of the effects of NWS. For example, weekend work shifts are considered a form of NWS as well (Presser, 2003), but such information is not collected on the NLSY79. In addition, given that the NLSY has collected information on work schedules on a yearly basis, our measure of work schedule may have undercounted the incidence of NWS because parents might change their schedules multiple times within each survey year (Han et al., 2010). It is also possible that various types of nonstandard schedules under the “other schedules” category may have differential effects on adolescents’ behaviors.

Third, PCC measures are somewhat crude in part due to the fact that we used secondary data, and thus it is difficult to draw sophisticated conclusions about its mediating roles. For example, Miller-Day and Kam (2010) noted that substance use specific PCC serves as a preventive factor against adolescent risk behavior. However, our measures of PCC reflected the child’s perception of general PCC about important topics, which may result in underestimating the effects of PCC. In addition, our four mediators (i.e., the openness and frequency of mother–child and father–child communication) were measured by a single item that may have considerable measurement errors. However, psychometric properties of the single item is not known in part because single items do not have internal consistency reliability calculated, such as Cronbach’s alpha (Nunnally & Bernstein, 1994). Therefore, the indirect effects via PCC should be interpreted with caution.

Another related issue with the psychometric properties of PCC items is the ability to distinguish between the openness and the frequency. The two single items representing the openness and frequency of mother–child and father–child communication, respectively, could be indicators for one underlying factors (e.g., mother–child communication, father–child communication, respectively), However, the correlations between the openness and the frequency were modest (Pearson correlation between the openness and the frequency of mother–child communication = .30; Pearson correlation between the openness and the frequency of father–child communication = .36), and the associated CFA model had poor model fit (chi-square = 254.8 [1], p < .05; CFI = .796; RMSEA = .359; SRMR = .063). More important, our literature review revealed that the four single items are conceptually distinguishable; therefore, we used the 4 single items separately, despite the potential measurement error.

Despite some limitations, this article makes a meaningful contribution. First, compared with previous studies that relied on a small sample or solely on low-income children, our study has greater generalizability, given that our sample consists of children born to a nationally representative female sample (Center for Human Resource Research, 2009). Second, given the relative lack of studies on the mechanisms through which parental NWS affect children and adolescents, our study adds to the knowledge base by finding that PCC may serve as a mechanism linking parental NWS with adolescent behavior. Third, our findings indicate that the impacts of parental NWS on adolescent substance use are not uniformly deleterious or beneficial. This article confirms that the influences of parental work schedules on children are complicated in that their effects on the mediator (i.e., PCC) differ by the type of nonstandard schedules, and when and how long parents worked. Therefore, this study calls for future studies to consider parental NWS as an opportunity as well as a risk for children’s well-being. Finally, our findings may provide important implications for policies and practices aimed at addressing the movement toward the 24/7 economy. In fact, it is questionable whether the United States is able to implement policies that directly regulate NWS in the labor market, unlike European countries that have such policies (e.g., the European Union’s 1993 Working Time Directive; Presser, Gornick, & Parashar, 2008). NWS can be a work option that parents with young children intentionally use, which may have a positive effect on children as shown in this study. This suggests that the regulatory policy might not be necessary. In this respect, it is, arguably, more feasible and appropriate to develop, for example, workplace interventions that provide working parents with more control over their work schedules (Henly, Shaefer, & Waxman, 2006). In addition, there is a dearth of risk reduction interventions that are specifically tailored for adolescents whose parents work NWS. It is evident from this study as well as the scholarship reviewed in this study that these adolescents face a number of behavioral issues, including substance use (Han, 2008; Han et al., 2010). Thus, there is a need for existing adolescent substance use interventions to target adolescents whose parents work NWS and for such interventions to include content to enhance the quality of PCC.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by grants from the Provost Office of the University of Pennsylvania.

Notes

References

Ackard

D. M.

Neumark-Sztainer

Story

Perry

(2006). Parent-child connectedness and behavioral and emotional health among adolescents. American Journal of Preventive Medicine, 30, 59-66.

Akerstedt

Fredlund

Gillberg

Jansson

(2002). Work load and work hours in relation to disturbed sleep and fatigue in a large representative sample. Journal of Psychosomatic Research, 53, 585-588.

Bachman

J. G.

Wadsworth

K. N.

O’Malley

P. M.

Johnston

L. D.

Schulenberg

J. E.

(1997). Smoking, drinking, and drug use in young adulthood: The impact of new freedoms and responsibilities. Mahwah, NJ: Erlbaum.

Barnett

R. C.

Gareis

K. C.

(2007). Shift work, parenting behaviors, and children’s socioemotional well-being: A within-family study. Journal of Family Issues, 28, 727-748.

Bianchi

S. M.

(2000). Maternal employment and time with children: Dramatic change or surprising continuity? Demography, 37, 401-414.

Brayfield

(1995). Juggling jobs and kids: The impact of employment schedules on father’s caring for children. Journal of Marriage and the Family, 57, 321-332.

Center for Human Resource Research. (2009). NLSY79 child and young adult data users guide: A guide to the 1986-2006 child data and 1994-2006 young adult data. Columbus, OH: Author.

Dunifon

Kalil

Bajracharya

(2005). Maternal working conditions and child well-being in welfare-leaving families. Developmental Psychology, 41, 851-859.

Enders

C. K.

(2010). Applied missing data analysis. New York, NY: Guilford.

10.

Ennett

S. T.

Bauman

K. E.

Foshee

V. A.

Pemberton

Hicks

K. A.

(2001). Parent-child communication about adolescent tobacco and alcohol use: What do parents say and does it affect youth behavior? Journal of Marriage and Family, 63, 48-62.

11.

Freedman

D. A.

(2006). On the so-called “Huber sandwich estimator” and “robust standard errors.” The American Statistician, 60, 299-302.

12.

Garey

A. I.

(1999). Weaving work and motherhood. Philadelphia, PA: Temple University Press.

13.

Gassman-Pines

(2011). Low-income mothers’ nighttime and weekend work: Daily associations with child behavior, mother-child interactions, and mood. Family Relations, 60, 15-29.

14.

Gold

D. R.

(1992). Rotating shift work, sleep, and accidents related to sleepiness in hospital nurses. American Journal of Public Health, 82, 1011-1014.

15.

Guilamo-Ramos

Bouris

A. M.

Dittus

Jaccard

(2007). Mother-adolescent communication about tobacco use in urban Puerto Rican and Dominican families. Youth & Society, 40, 86-113.

16.

Guilamo-Ramos

Jaccard

Dittus

Bouris

(2006). Parental expertise, trustworthiness, and accessibility: Parent-adolescent communication and adolescent risk behavior. Journal of Marriage and Family, 68, 1229-1246.

17.

Han

W.-J.

(2004). Nonstandard work schedules and child care decisions: Evidence from the NICHD Study of Early Child Care. Early Childhood Research Quarterly, 19, 231-256.

18.

Han

W.-J.

(2008). Shift work and child behavioral outcomes. Work, Employment and Society, 22, 67-87.

19.

Han

W.-J.

Miller

D. P.

Waldfogel

(2010). Parental work schedules and adolescent risky behaviors. Developmental Psychology, 46, 1245-1267.

20.

Han

W.-J.

Waldfogel

(2007). Parental work schedules, family process, and early adolescents’ risky behavior. Children and Youth Services Review, 29, 1249-1266.

21.

Hattery

A. J.

(2001). Tag-team parenting: Cost and benefits of utilizing nonoverlapoing shift work in families with young children. Families in Society, 82, 419-427.

22.

Henly

J. R.

Shaefer

H. L.

Waxman

(2006). Nonstandard work schedules: Employer- and employee-driven flexibility in retail jobs. Social Service Review, 80, 609-634.

23.

Heymann

(2000). The widening gap: Why America’s working families are in jeopardy and what can be done about it. New York, NY: Basic Books.

24.

Jaccard

Dodge

Dittus

(2002). Parent-adolescent communication about sex and birth control: A conceptual framework. In Feldman

Rosenthal

D. A.

(Eds.), Talking sexuality: Parent-adolescent communication (pp. 9-41). San Francisco, CA: Jossey-Baas.

25.

Johnston

L. D.

O’Malley

P. M.

Bachman

J. G.

Schulenberg

J. E.

(2011). Monitoring the Future national results on adolescent drug use: Overview of key findings, 2010. Ann Arbor: Institute for Social Research, the University of Michigan.

26.

Joshi

Bogen

(2007). Nonstandard schedules and young children’s behavioral outcomes among working low-income families. Journal of Marriage and Family, 69, 139-156.

27.

Kiene

S. M.

Barta

W. D.

Tennen

Armeli

(2009). Alcohol, helping young adults to have unprotected sex with casual partners: Findings from a daily diary study of alcohol use and sexual behavior. Journal of Adolescent Health, 44, 73-80.

28.

Kitzmann

K. M.

(2000). Effects of marital conflict on subsequent triadic family interactions and parenting. Developmental Psychology, 36, 3-13.

29.

Luk

J. W.

Farhat

Iannotti

R. J.

Simons-Morton

B. G.

(2010). Parent-child communication and substance use among adolescents: Do father and mother communication play a different role for sons and daughters? Addictive Behaviors, 35, 426-431.

30.

MacKinnon

D. P.

Lockwood

C. M.

Brown

C. H.

Wang

Hoffman

J. M.

(2007). The intermediate endpoint effect in logistic and probit regression. Clinical Trials, 4, 499-513.

31.

McMenamin

T. M.

(2007). A time to work: Recent trends in shift work and flexible schedules. Monthly Labor Review, 130(12), 3-15.

32.

Miller-Day

Kam

J. A.

(2010). More than just openness: Developing and validating a measure of targeted parent-child communication about alcohol. Health Communication, 25, 293-302.

33.

Muthén

L. K.

Muthén

B. O.

(1998-2012). Mplus user’s guide: Statistical analysis with latent variables (7th ed.). Los Angeles, CA: Muthén & Muthén.

34.

Nunnally

J. C.

Bernstein

I. H.

(1994). Psychometric theory (3rd ed.). New York, NY: McGraw-Hill.

35.

Otten

Harakeh

Vermulst

A. A.

Van den Eijnden

R. J.

Engels

R. C.

(2007). Frequency and quality of parental communication as antecedents of adolescent smoking cognitions and smoking onset. Journal of Addictive Behaviors, 21, 1-12.

36.

Phillips

K. R.

(2002). Parent work and child well-being in low-income families (Occasional Paper No. 56). Washington, DC: Urban Institute. Retrieved from http://www.urban.org/Uploadedpdf/310509_OP56.pdf

37.

Preacher

K. J.

Hayes

A. F.

(2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40, 879-891.

38.

Presser

H. B.

(1995). Job, family, and gender: Determinants of nonstandard work schedules among employed Americans in 1991. Demography, 32, 577-598.

39.

Presser

H. B.

(2000). Nonstandard work schedules and marital instability. Journal of Marriage and the Family, 62, 93-110.

40.

Presser

H. B.

(2003). Working in a 24/7 economy: Challenges for American families. New York, NY: Russell Sage Foundation.

41.

Presser

H. B.

Gornick

J. C.

Parashar

(2008). Gender and nonstandard work hours in 12 European countries. Monthly Labor Review, 131(2), 83-103.

42.

Presser

H. B.

Ward

B. W.

(2011). Nonstandard work schedules over the life course: A first look. Monthly Labor Review, 134(7), 3-16.

43.

Shonkoff

J. P.

Phillips

D. A.

(Eds.). (2000). From neurons to neighborhoods: The science of early childhood development. Washington, DC: National Academy Press.

44.

Simon

B. L.

(1990). Impact of shift work on individuals and families. Families in Society, 7, 342-348.

45.

Smetana

J. G.

Campione-Barr

Metzger

(2006). Adolescent development in interpersonal and societal contexts. Annual Review of Psychology, 57, 255-284.

46.

Sobolewski

J. M.

Amato

P. R.

(2007). Parents’ discord and divorce, parent-child relationships, and subjective well-being in early adulthood: Is feeling close to two parents always better than feeling close to one? Social Forces, 85, 1105-1124.

47.

Strazdins

Clements

M. S.

Korda

R. J.

Broom

D. H.

D’Souza

R. M.

(2006). Unsociable work? Nonstandard work schedules, family relationships, and children’s well-being. Journal of Marriage and Family, 68, 394-410.

48.

Substance Abuse and Mental Health Services Administration. (2012). Results from the 2011 National Survey on Drug Use and Health: Summary of national findings (NSDUH Series H-44, HHS Publication No. [SMA] 12-4713). Rockville, MD: Author.

49.

Täht

Mills

(2012). Nonstandard work schedules, couple desynchronization, and parent-child interaction: A mixed-methods analysis. Journal of Family Issues, 33, 1054-1087.

50.

Van der Vorst

Engels

R. C. M. E.

Meeus

Deković

Van Leeuwe

(2005). The role of alcohol-specific socialization in adolescents’ drinking behaviour. Addiction, 100, 1464-1476.

51.

White

(2012). The contribution of parent-child interactions to smoking experimentation in adolescence: implications for prevention. Health Education Research, 27, 46-56.

52.

White

Keith

(1990). The effect of shift work on the quality and stability of marital relations. Journal of Marriage and the Family, 52, 453-462.

53.

Winship

Radbill

(1994). Sampling weights and regression-analysis. Sociological Methods & Research, 23, 230-257.