Understanding Children’s Television Exposure From a Life Logistics Perspective: A Longitudinal Study of the Association Between Mothers’ Working Hours and Young Children’s Television Time

The Role of Mothers’ Parenting Time Pressure in Children’s Television Exposure

A possible mechanism linking mothers’ working hours with children’s television exposure is mothers’ parenting time pressure. The theoretical grounds for these relationships lie in the stress process model (Pearlin, 1989), border theory (Clark, 2000), and time conflict theory (Eby, Casper, Lockwood, Bordeaux, & Brinley, 2005). An explanation for the relationship between mothers’ working hours and parenting time pressure can be found in the stress process model (Pearlin, 1989) and border theory (Clark, 2000). The stress process model identifies relationships between stressors, such as role demands (Wheaton, 1999), and stress outcomes. According to the stress process model, primary stressors create secondary stressors, in a process known as stress proliferation. Border theory (Clark, 2000) posits that demands in one sphere may influence stressors in another sphere. Building on the notions of the stress process model and border theory, mothers’ demands in the work sphere (primary stressor) may create time pressure in the parenting sphere (secondary stressor). Moreover, because working parents in modern Western societies increasingly face demands in both their work and family life (van der Lippe, 2007), research grounded in the work-family literature argues for a role-overload hypothesis, stating that separate roles with different demands may conflict (Moen et al., 1995).

Consistent with the assumptions of time conflict theory (Eby et al., 2005; Greenhaus & Beutell, 1985), the amount of time parents work has frequently been identified as an important work stressor contributing to work-family conflict and cutting into parents’ time and energy (Huston & Rosenkrantz Aronson, 2005). A specific type of work-family conflict is parenting time pressure, which is defined as a feeling of time deficit, feeling rushed and worried about time when executing the parenting role (Roxburgh, 2006, 2012). Empirical evidence exists for the relationship between stressful aspects in the work domain and stressful circumstances related to parenting, particularly between long working hours and parenting time pressure (Roxburgh, 2006, 2012).

Goode (1960) predicted that when people experience conflict when juggling different role demands, they will try to cope by adjusting their roles or the role-related behavior. Mothers who experience parenting time pressure because of their long working hours may try to cope with this time pressure by entertaining their child with television or using television as a babysitter for their child, so that it provides a break from parenting and creates time for themselves (Evans et al., 2011).

Based on the premises of the stress process model (Pearlin, 1989), border theory (Clark, 2000), and time conflict theory (Eby et al., 2005; Greenhaus & Beutell, 1985), we expect that the relationship between maternal working hours and children’s television viewing time is partially mediated by parenting time pressure. Therefore, we propose the following hypothesis:

The Role of Mothers’ Well-Being in Children’s Television Exposure

Another possible mechanism linking mothers’ working hours with children’s television exposure is mothers’ well-being. Support for a relationship between mothers’ working hours and their well-being can be found in studies that support the role-enhancement perspective (Moen et al., 1995; Rozario, Morrow-Howell, & Hinterlong, 2004). According to this perspective, being employed improves one’s well-being, because working people experience more social contact, increased self-fulfillment, and have more resources and other benefits (Moen et al., 1995). This is supported by studies showing that employed women (Glass & Fujimoto, 1994), and employed mothers in particular (Baruch & Barnett, 1986; Lavee, Sharlin, & Katz, 1996), are less depressed than non-employed women and mothers.

Furthermore, indications exist that mothers’ well-being influences children’s television viewing. More specifically, mothers with lower well-being prefer to use television to occupy or entertain their child in order to cope with their poor well-being (Potts & Sanchez, 1994). For instance, mothers with poor well-being may use television to occupy their child so that they may attend to their personal needs, because these mothers may believe that watching television is more beneficial for their child than engaging in mother-child interactions or because these mothers neither have the motivation nor the energy to engage in interactions with their child (Conners et al., 2007; Lovejoy, Graczyk, O’Hare, & Neuman, 2000; Potts & Sanchez, 1994). This is supported by empirical evidence demonstrating that children of mothers with lower well-being watch more television (Bank et al., 2012; Bickham et al., 2003; Conners et al., 2007; Thompson & Christakis, 2007).

Based on these assumptions, we expect that mothers’ well-being mediates the relationship between mothers’ working hours and children’s television time. Accordingly, we will investigate the following hypothesis:

The Role of Mothers’ Parenting Time Pressure and Well-Being in Children’s Television Exposure

Mothers’ working hours may relate to children’s television time in another way as well. The relationship between mothers’ working hours and well-being (Moen et al., 1995), which is associated with children’s television time (Bank et al., 2012) as explained above, may be partially mediated by parenting time pressure. According to the stress process model (Pearlin, 1989), stressors, such as role demands, are related to stress outcomes, such as anxiety and depression. In this respect, research has found an association of feelings of time pressure with depression (Roxburgh, 2012).

Applying the notions of the stress process model to this study, long working hours (primary stressor) may create parenting time pressure (secondary stressor), which may result in lower well-being (stress outcome). Research has supported this assumption by demonstrating that working longer hours creates parenting time pressure and, consequently, decreases mothers’ well-being (Roxburgh, 2012). Therefore, we expect that parenting time pressure partially mediates the relationship between mothers’ working hours and their mental well-being, which, in turn, negatively predicts children’s television time. Hence, we hypothesize as follows:

Alternative Models for Understanding Children’s Television Exposure

In order to investigate whether the life logistics model withstands controls for classical factors that were already known to be associated with children’s television exposure, the present study will investigate two alternative models.

The first alternative model will include mothers’ television attitudes, in particular mothers’ attitudes toward the educational value of television, because attitudes are a major recurring factor that contributes to children’s television viewing. Research has demonstrated that maternal attitudes toward television predict children’s television use, showing that children of parents with more positive views of television watch more television (Vandewater et al., 2005). Therefore, the following research question is posited:

The second alternative model will include mothers’ television viewing, because it may (1) predict children’s television viewing, because parents are important role models for their children’s television viewing (Bleakley, Jordan, & Hennessy, 2013; Rideout & Hamel, 2006), and (2) mediate the relationship of mothers’ parenting time pressure and well-being with children’s television viewing (Conners et al., 2007), because adults experiencing stress (Anderson, Collins, Schmitt, & Jacobvitz, 1996) and low well-being (Sidney et al., 1996) watch more television. In this line, we expect that mothers with low well-being and with parenting time pressure watch more television. Subsequently, because children’s television viewing is influenced by parental modeling (Bickham et al., 2003; Bleakley et al., 2013; Rideout & Hamel, 2006), we expect that children watch more television when their mothers watch high amounts of television. Thus, mothers’ stress and well-being may influence their own television exposure, thereby influencing their children’s television exposure (Conners et al., 2007). The present study will test these assumptions in the following research question:

Control Variables

Previous studies have shown that parents’ level of education (Bickham et al., 2003; Certain & Kahn, 2002), the number of children in a family (Barr, Danziger, Hilliard, Andolina, & Ruskis, 2010), and children’s age (Barr, Danziger, et al., 2010; Certain & Kahn, 2002) are associated with children’s television viewing time. These variables may play a role in the relationship between mothers’ life logistics and children’s television exposure. Therefore, these variables are included as control variables in our analyses.

Method

Results

Descriptive Statistics

At baseline, the mothers worked an average of 29.39 hours per week (SD = 12.57). This average is comparable, t(390) = −.69, p = .49, with the average weekly work hours of mothers with young children in Belgium reported in previous research (Abendroth, van der Lippe, & Maas, 2012). One tenth of the mothers did not engage in paid employment (10%), which is representative of the overall population of Belgian mothers with young children (10.75%; Child and Family, 2012). Most mothers agreed that they have time to enjoy their child (76.1%). However, one fourth of the mothers (24.4%) agreed that they worry about having sufficient time to care for their child, and approximately one in six mothers (15.4%) agreed that they feel rushed when they are with their child. The mothers’ well-being could be conceived as moderate because the mean level of mental well-being was slightly above the cutoff point of being at risk of mood disorders (M = 22.68, SD = 3.57). On a weekly basis, children spent 6 hours, on average, watching television (M = 5.71, SD = 5.44). The zero-order correlation matrix of all key variables of interest and control variables is provided in Table 1.

OPEN IN VIEWER

Table 1.

Zero-Order Correlations.

	Mothers’ working hours W1	Mothers’ parenting time pressure W1	Mothers’ well-being W1	Children’s age	Mothers’ education	No. of children in family	Mothers’ attitudes toward television	Mothers’ television time	Children’s television time W1	Children’s television time W2
Mothers’ working hours W1	1	.12*	.08	−.08	.16**	−.08	−.09 †	−.07	−.09 †	−.11 †
Mothers’ parenting time pressure W1		1	−.36***	.02	.01	.13**	−.00	−.06	−.03	−.02
Mothers’ well-being W1			1	−.06	.13*	−.03	−.03	.05	−.03	−.13*
Children’s age				1	−.09 †	.16***	−.08	−.03	.40***	.25***
Mothers’ education					1	−.01	−.06	−.28***	−.32***	−.32***
No. of children in family						1	.05	−.13*	.02	−.04
Mothers’ attitudes toward television							1	.07	.15**	.15**
Mothers’ television time								1	.34***	.27***
Children’s television time W1									1	.70***
Children’s television time W2										1

†

p < .10. *p < .05. **p < .01. ***p < .001.

Testing the Life Logistics Model

Confirmatory factor analysis (CFA) showed that the measurement model of the life logistics model yielded a good fit, χ²(14) = 32.67, χ²/df = 2.33, p < .01, CFI = .98, RMSEA = .06, GFI = .98, AGFI = .95. Structural equation modeling showed that the life logistics model (controlling for mothers’ education, number of children, children’s age, and television exposure at Wave 1) yielded a good fit, χ²(61) = 114.62, χ²/df = 1.88, p < .001, CFI = .97, RMSEA = .05, GFI = .96, AGFI = .93. As shown in Figure 2, mothers’ working hours were a positive predictor of mothers’ parenting time pressure, β = .15, B = 0.01, SE = 0.00, p < .01. As expected, when mothers worked longer hours, they felt more time pressure during parenting. However, mothers’ parenting time pressure did not predict children’s television time, β = −.07, B = −0.50, SE = .39, p = .20. Thus, H1, which predicted that longer working hours increase mothers’ parenting time pressure, thereby increasing children’s television viewing time, was not supported.

OPEN IN VIEWER

Figure 2.

Observed life logistics model.

The analyses further revealed that mothers’ working hours marginally increased their well-being, β = .09, B = 0.00, SE = 0.00, p < .10. When mothers worked longer hours, they had better mental well-being. Mothers’ well-being negatively predicted children’s television time, β = −.16, B = −1.72, SE = 0.49, p < .001, even when controlling for children’s age, number of children in the family, mothers’ education, and children’s television time at Wave 1. As we expected, the better mothers’ well-being was, the less television children watched. Sobel’s z test indicated that the indirect effect of working hours on television viewing through well-being was marginally significant (−.01 = .09 x −.16; z = −1.74, p < .10). This indicates that children watched less television when their mothers worked longer because the longer working hours improved their mothers’ well-being. Thus, weak support was found for H2.

Finally, H3 was supported because mothers who worked longer hours experienced more parenting time pressure, β = .15, B = 0.01, SE = 0.00, p < .01, and had, in turn, lower well-being, β = −.49, B = −0.35, SE = 0.06, p < .001, which resulted in increased television viewing time among their children, β = −.16, B = −1.72, SE = 0.49, p < .001. Thus, mothers’ working hours predicted children’s television viewing time through parenting time pressure and well-being, even when controlling for mothers’ education, number of children, children’s age, and television time at Wave 1. Sobel tests revealed that parenting time pressure significantly mediated the relationship between working hours and well-being (−.07 = .15 x −.49; z = −2.65, p < .01) and that mothers’ well-being significantly mediated the relationship between mothers’ parenting time pressure and children’s television time (.08 = −.49 x −.16; z = 2.98, p < .01).

Analyses further showed that children’s television exposure at Wave 1 positively predicted children’s television time, β = .67, B = 0.68, SE = 0.04, p < .001, and that mothers’ education negatively predicted children’s television time, β = −.09, B = −0.65, SE = 0.27, p < .05. The number of children in a family did not predict children’s television time, β = −.05, B = −0.37, SE = 0.28, p = .18, nor did children’s age, β = −.02, B = −0.01, SE = 0.02, p = .66. Overall, the predictors in the model collectively explained 51% of the variance in television viewing (R² = .51).

Finally, we investigated a potential unmediated relationship between mothers’ working hours and children’s television time and the possibility of a reverse relationship between children’s television viewing and mothers’ well-being. First, the analyses showed that, in a simple structural equation model including only mothers’ working hours and children’s television time, a direct relationship was found showing that the more mothers worked, the less television children watched, β = −.11, B = −0.05, SE = 0.02, p < .05. However, in the life logistics model, no direct relationship was found between mothers’ working hours and children’s television time, β = −.03, B = −0.01, SE = 0.02, p = .49. This indicates that mothers’ parenting time pressure and well-being mediated the relationship. Second, the analyses showed that mothers’ well-being (Wave 2) was not affected by their children’s television viewing (Wave 1), β = −.03, B = −0.00, SE = 0.01, p = .55. Thus, there was no evidence of a reciprocal relationship.

Testing Alternative Models

As shown in Figure 3, the alternative model that included mothers’ television attitudes (RQ1) yielded a good fit, χ²(101) = 174.46, χ²/df = 1.73, p < .001, CFI = .96, RMSEA = .04, GFI = .95, AGFI = .93. The significant paths of the life logistics model remained significant when controlling for mothers’ television attitudes. Moreover, mothers’ attitudes did not predict children’s television viewing time, β = .06, B = 0.65, SE = 0.46, p = .16. As shown in Table 2, adding mothers’ television attitudes to the life logistics model did not improve the fit. The AIC and BIC differences (Raftery, 1995) provided very strong evidence favoring the life logistics model over the alternative model including mothers’ attitudes (ΔAIC_i = 278.46 − 202.62 = 75.84; ΔBIC_i = 486.02 − 378.25 = 107.77). The predictors in this model collectively explained 52% of the variance in television viewing (R² = .52).

OPEN IN VIEWER

Figure 3.

Observed life logistics model including mothers’ television attitudes.

OPEN IN VIEWER

Table 2.

Fit Indices and Model Comparison for the Life Logistics Model and Four Alternative Models.

Model	CFI	RMSEA	GFI	AGFI	AIC	Rank	BIC	Rank	ECVI	Rank
1. Life logistics model	.97	.05	.96	.93	202.62	1	378.25	1	0.51	1
2. Life logistics model with mothers’ television attitudes	.96	.04	.95	.93	278.46	4	486.02	4	0.70	4
3. Life logistics model with mothers’ television time	.94	.06	.95	.91	265.00	3	452.60	3	0.66	3
4. Life logistics model with curvilinear relationship	.97	.04	.96	.93	221.35	2	404.95	2	0.56	2
5. Integrative life logistics model	.96	.04	.95	.92	329.84	5	569.33	5	0.83	5

Note. Rank indicates the order of the size of the AIC, BIC, and ECVI values (1 = smallest value, and 5 = highest value). CFI = comparative fit index; RMSEA = root mean square error of approximation; GFI = goodness of fit index; AGFI = adjusted goodness of fit index; AIC = Akaike information criterion (Akaike, 1987); BIC = Bayesian information criterion (Raftery, 1995); ECVI = expected cross-validation index (Browne & Cudeck, 1993).

As shown in Figure 4, the model that included mothers’ television time as a mediator between mothers’ parenting time pressure and well-being and children’s television viewing time (RQ2) yielded a good fit, χ²(70) = 127.93, χ²/df = 1.83, p < .001, CFI = .94, RMSEA = .06, GFI = .95, AGFI = .91. The significant paths of the life logistics model remained significant when mothers’ viewing time was added to the model. However, adding mothers’ television time to the life logistics model did not improve the fit (see Table 2). Very strong evidence was found favoring the life logistics model over the model including mothers’ television time (ΔAIC_i = 265.00 − 202.62 = 62.38; ΔBIC_i = 452.60 − 378.25 = 74.35). The analyses indicated that mothers’ television use was not predicted by their well-being, β = .05, B = 0.81, SE = 1.03, p = .43, or parenting time pressure, β = −.06, B = −0.71, SE = 0.83, p = .39, and that mothers’ television time did not predict children’s television time, β = .02, B = 0.01, SE = 0.02, p = .61. Thus, mothers’ television use did not explain the relationship between mothers’ well-being and children’s television exposure. Overall, the predictors in this model collectively explained 51% of the variance in children’s television viewing (R² = .51).

OPEN IN VIEWER

Figure 4.

Observed life logistics model including mothers’ television time.

To explore the marginally significant and positive relationship between mothers’ working hours and well-being more deeply, we examined whether the relationship was curvilinear, by inserting the polynomial of working hours in our model (Kline, 2011). In order to solve possible collinearity problems, we created a residualized squared term using orthogonalization (Little, 2013; Little, Bovaird, & Widaman, 2006) by regressing the squared term on the main variable. The model yielded a good fit, χ² (74) = 129.35, χ²/df = 1.75, p < .001, CFI = .97, RMSEA = .04, GFI = .96, AGFI = .93. The squared working hours term marginally predicted mothers’ well-being in a positive way, β = .09, B = 0.00, SE = 0.00, p < .10, reflecting a U-shaped relationship, indicating that well-being is lowest for mothers with mean working hours and highest for mothers with low and high working hours. The significant paths of the life logistics model remained significant. However, the model did not perform better than the main life logistics model (ΔAIC_i = 221.35 − 202.62 = 18.73; ΔBIC_i = 404.95 − 378.25 = 26.70).

Finally, to complement our findings, we also tested an integrative life logistics model that encompasses the life logistics model and mothers’ television attitudes, mothers’ television viewing, and the curvilinear relationship between working hours and well-being. This integrative model resulted in a good fit, χ²(130) = 209.84, χ²/df = 1.61, p < .001, CFI = .96, RMSEA = .04, GFI = .95, AGFI = .92. However, there was strong evidence that the integrative model did not perform better than the basic life logistics model (ΔAIC_i = 329.84 − 202.62 = 127.22; ΔBIC_i = 569.33 − 378.25 = 191.08).

Discussion

This study investigated the influence of mothers’ structural life circumstances on children’s television time using a two-wave panel survey. Adopting a life logistics perspective and building upon insights from the stress process model (Pearlin, 1989), border theory (Clark, 2000), and time conflict theory (Eby et al., 2005; Greenhaus & Beutell, 1985), this study sought to contribute to the existing body of knowledge on young children’s television use. As an answer to scholars’ (e.g., Thompson & Christakis, 2007) call to examine the reasons why parents allow their children to watch television, this study showed that life logistics, particularly mothers’ working hours, parenting time pressure, and well-being, are important predictors of children’s television time. The findings highlight the importance of adopting a life logistics perspective in studying children’s television use and offer several distinct contributions to the literature.

The relationship between mothers’ working hours and children’s television time arises from two different processes. First, to the extent that work and family demands do not conflict, children’s television viewing does not increase. Although mothers’ working hours only marginally predicted their well-being, the findings suggest that being employed has a positive impact on mothers’ well-being, which, consequently, decreases children’s television time. In this sense, our study supports the role-enhancement perspective (Moen et al., 1995), which starts from the premise that people who are employed have more social contact, more resources, increased self-fulfillment, and other benefits that improve one’s well-being. Additional analyses found a trend toward a U-shaped curve for the relationship between mothers’ working hours and well-being, suggesting that mothers with lower and higher than average working hours have better well-being.

Second, also a contrasting process was found to explain the relationship between mothers’ working hours and children’s television time. The findings indicate that when work and family demands conflict, children’s television use increases. More specifically, children’s television viewing increases because, in line with time conflict theory (Eby et al., 2005; Greenhaus & Beutell, 1985) and studies in support of a role-overload hypothesis (Goode, 1960; Moen et al., 1995), working longer hours increases mothers’ parenting time pressures and, consequently, causes mental distress. Our findings thus indicate that when mothers’ work demands conflict with their parenting demands, children tend to watch more television.

Our life logistics model showed that mothers’ well-being is a fundamental mechanism in the chain of events linking mothers’ working hours with children’s television use. The fact that children watch more television when their mothers have lower well-being confirms the results of previous research (Thompson & Christakis, 2007). However, compared with other studies, the mothers in our study appeared to have relatively low well-being. According to the cutoff point used to identify mothers at risk of mood disorders (Rumpf et al., 2001), more than one third of the mothers (35.4%) we studied could be considered to have mental distress. This proportion is significantly higher, χ²(1) = 54.12, p < .001, than the proportion reported in a previous study of young children’s television viewing using Berwick et al.’s (1991) Mental Health Inventory, in which 21% of the mothers were identified as having mental distress (Thompson & Christakis, 2007), but significantly lower, χ²(1) = 7.686, p < .01, than the percentage of 42% reported by Olson and DiBrigida (1994).

The fact that one third of the mothers in our study have mental distress is of particular concern considering that a substantial proportion of children potentially watch high amounts of television because of their mothers’ mood disorders. This highlights the importance of investigating mothers’ life circumstances to understand children’s television time and the relevance of well-being to identify children who watch high amounts of television. Further research is needed to determine whether watching more television because a parent suffers from a mental illness is necessarily unfavorable. Consistent with what Thompson and Christakis (2007) have argued, it is plausible that watching television, given that it is high-quality content, is more stimulating for a child than interacting with a parent who experiences mental distress. Watching television may offset the negative impact of parental distress on children’s social and cognitive development.

The findings of the present study are of theoretical importance. This study contributes to the existing literature by showing that a life logistics approach is a promising direction for research investigating children’s television use. For instance, media effects research may benefit greatly from adopting a life logistics perspective because structural life circumstance factors may explain why some children are more influenced by television than others. Mothers’ working hours and time pressures may be moderators that help identify children at particular risk of the impact of television viewing and may explain children’s differential susceptibility to television effects (cf. Valkenburg & Peter, 2013). The impact of television on children’s social, cognitive, and physical outcomes (e.g., Dennison & Edmunds, 2008; Thompson & Christakis, 2005; Zimmerman & Christakis, 2007) may be different for children who watch television because they themselves take the initiative to watch, compared with children who watch television due to parents’ structural life circumstances, for instance because parents try to entertain or occupy their child by allowing him or her to watch television when parents feel time pressured or experience mental distress.

In this respect, future research should also consider the impact of other life circumstances and other work characteristics. For instance, studies have demonstrated that nonstandard work schedules are associated with lower involvement in family activities (Presser, 2000) and less cognitively stimulating parent-child interactions (Gassman-Pines, 2011), which may encourage children to watch more television. Also, future research would benefit from exploring additional factors that are linked to levels of child television viewing, such that we increase the empirical validation for the factors tested in this study, which have been tested in a relatively narrow model.

This study provided a robust test of the life logistics model. Our model withstood critical examination, providing evidence that the impact of structural life circumstances on children’s television exposure persists over time, even when controlling for children’s baseline television viewing and variables that have previously been shown to be strong predictors of children’s television use. Compared with mothers’ attitudes about the educational value of television for children, mothers’ structural life circumstances have a stronger influence on children’s television use. Mothers’ television attitudes did relate to children’s television viewing, but not when they were included in the life logistics model. This indicates that mothers’ life logistics outweigh their attitudes toward the educational value of television as predictors of children’s television viewing time.

It is important to note, however, that whereas the amount of television exposure is primarily predicted by life logistics, the content that children watch may be primarily under attitudinal control. In this respect, research showed that although parental attitudes toward television did not predict the time children spent watching television, they did predict the content (Barr, Danziger, et al., 2010). Also, several studies have indicated that attitudes about television are a strong predictor of parental mediation of children’s television viewing (Nathanson, 2001; Warren, 2005).

Although mothers’ television use correlated significantly with children’s television exposure (r = .27, p < .001), parental modeling of television viewing did not play a role in children’s viewing when it was integrated in the life logistics model. The assumption that mothers with low well-being would watch more television, thereby modeling the television behavior for their children so that their children would watch more television, as suggested by others (Conners et al., 2007), was not supported in this study. A possible explanation might be that mothers may do much of their viewing late at night, after having taken care of the children, when the children can no longer observe it such that parental modeling cannot take place. This is possible as mothers’ television viewing is influenced by structural factors as well. For instance, working mothers may have less time for viewing (Hochschild & Machung, 2012; Juster & Stafford, 1985).

Future studies are needed to further disentangle the relationship between maternal well-being and children’s television use. For instance, it is possible that watching television displaces mother-child interactions because depressed mothers have poorer parenting abilities (Downey & Coyne, 1990) and exhibit fewer interactions with their children (Lovejoy et al., 2000). In this view, scholars have argued that depressed mothers may be more likely to use television as an electronic babysitter (Conners et al., 2007). Also, research is needed that includes children’s well-being to investigate whether young children might be learning to use television to entertain themselves as a way to cope with stress. Research among older children already showed that they frequently watch television to cope with stress (Chen & Kennedy, 2005).

The longitudinal design of our study showed that the relations in the life logistics model hold longitudinally, even while controlling for children’s baseline television viewing, thereby documenting changes in children’s television viewing over time. Furthermore, the two-wave design also allowed us to establish the temporal order of the relations (Kline, 2011), showing that mothers’ structural circumstances predict children’s television viewing, but not vice versa. In this sense, our study moves beyond cross-sectional research by complying with the temporal precedence condition of causal order (Kline, 2011). Following other scholars (Lee et al., 2009), we encourage more research to investigate longitudinal relationships when examining predictors of children’s television viewing.

The present study is also of practical importance. The findings indicate that mothers’ life logistics make it challenging to monitor young children’s television exposure. Because mothers’ structural life circumstances are associated with increased television viewing among children when it creates mental distress, future interventions aimed at reducing children’s television time should target mothers’ life logistics. We agree with Thompson and Christakis (2007) that health practitioners should screen for maternal mental distress. Policy makers should provide opportunities for families that create a healthy work-family balance so that parents are able to better attend to the needs of their children and limit children’s television viewing time.

It is important to note that the data for this study were collected in Belgium, a Western European country. Whereas studies of children’s television viewing primarily use data collected in U.S. samples (e.g., Bank et al., 2012; Bickham et al., 2003; Conners et al., 2007; Thompson & Christakis, 2007; Warren, 2005), this study adds a European perspective to the literature by using data collected in a European, non-U.S. sample. However, while the average working hours are representative of Belgian mothers with young children and comparable with the average reported by European employed mothers (Abendroth et al., 2012), they are relatively low compared with the average working hours of U.S. mothers (e.g., Roxburgh, 2006). Also, children’s television exposure tends to be higher in U.S. samples (e.g., Barr, Lauricella, Zack, & Calvert, 2010; Bickham et al., 2003; Thompson & Christakis, 2007) than in our sample and other European samples (e.g., Beyens & Eggermont, 2014; Kourlaba et al., 2009; van Stralen et al., 2012). Future studies should therefore investigate whether U.S. mothers’ life logistics would have a stronger impact. The higher working hours and higher viewing time suggest that the life logistics model might be more powerful in a U.S. context.

Furthermore, some selection bias might be inherent to the recruitment procedure of this study, in which children were recruited from daycare centers and kindergarten. People with different backgrounds, either ethnic, social, or other, might have different approaches to child raising and the use of child care provision. It might be possible that people with a particular background are more likely to use kindergarten and daycare centers. Therefore, White respondents are over-represented in our study as well as working mothers. In order to increase our understanding of the relationships examined in this study, more research is needed in samples with different ethnic, social, and other backgrounds. Also, research should look at fathers’ role and investigate the role of fathers’ structural circumstances. In order to increase the validity of the life logistics model, studies should test the model in samples of fathers and people from different ethnic, cultural, and social backgrounds. Furthermore, other socioeconomic factors should be included in future studies. For instance, while mothers’ educational level was used as an indicator of the socioeconomic status and controlled for in our study, other studies might also include other socioeconomic factors as a control variable, such as parents’ income.

It should also be noted that the parenting time pressure scale used in this study had a relatively low reliability. However, Cronbach’s alpha (.69) was slightly below the generally accepted level of .70 (Shelby, 2011) and even slightly higher than in the original study (.67; Roxburgh, 2006). Moreover, lower alphas are less of a problem with scales that encompass fewer items (Cortina, 1993). Little (2013) argues that underdeveloped measures are allowed if the theoretical propositions being tested are worthy of analysis. Future research may improve the parenting time pressure scale and consequently improve the quality of the structural circumstance factors included in the life logistics model.

Also, future studies should look at the predictive value of other attitude dimensions in the life logistics model. The attitudes toward television measure used in our study asked only about the educational value of television for children and may therefore not sufficiently test mothers’ attitudes toward television as a predictor in the life logistics model. This is not a global attitude construct, which would tap into other relevant dimensions of attitude as well, as it is a multidimensional construct (Warren, 2003). Thus, while our study shows that the second, alternative model finds no relationship between the attitude construct and children’s television viewing, it is still possible that other dimensions of mothers’ attitudes would be predictive. For instance, parents’ attitudes toward the value of television for parenting (e.g., occupying the child so the parent can do household chores, or using television as a reward for good behavior or as a punishment for bad behavior) may be more powerful predictors of children’s television viewing in our life logistics model.

Conclusion

This study highlights the importance of adopting a life logistics perspective in research investigating children’s television use. The life logistics model revealed important mechanisms linking mothers’ structural life circumstances with children’s television use. Overall, our study indicates that to the extent that mothers’ work demands are beneficial for their well-being, children tend to watch less television. However, when mothers’ work demands create parenting time pressure and thereby decrease their well-being, children watch more television.

Children’s television exposure seems to be primarily determined by structural circumstances and not so much by their mothers’ television attitudes or modeling of television use when integrated in the life logistics model. It is possible that the impact of parents’ attitudes and television use observed in prior research (Barr, Danziger, et al., 2010; Bleakley et al., 2013; Vandewater et al., 2005) may be confounded by life logistics not investigated in previous studies.

This study offers an important step to advance our empirical understanding of the role of mothers’ life circumstances in children’s television viewing and to build theoretical grounds for studying young children’s media use. Continued research is needed on these and other predictors of children’s television exposure.