Determinants of self-care arrangements among school-age children

Determinants of self-care arrangements among school-age children

t#W Pergamon Children and Youth Services Review. Vol. 2 1. No. 6. DD. 497-520. I999 Co&right 0 1999 Ekvier Science Ltd Printed in the USA. All rights...

1MB Sizes 0 Downloads 39 Views

t#W Pergamon

Children and Youth Services Review. Vol. 2 1. No. 6. DD. 497-520. I999 Co&right 0 1999 Ekvier Science Ltd Printed in the USA. All rights reserved 0190-7409/99/Gsee front matter

PI1 so19o-74o!q!x9ooo35-3

Determinants of Self-Care Arrangements among School-Age Children Peter David Brandon University of Massachusetts-Amherst This study investigates determinants of self-care arrangements among school-aged children. The study finds that mothers’ hours of work increase the likelihood of self-care arrangements. However, accompanying results suggest that maternal work and use of self-care does not imply school children are unsupervised. Effects of kin, older siblings, the cost of care and use of only maternal care previously suggest that self-care is chosen only when special conditions are satisfied. When those stipulations are meet, self-care has a higher likelihood of use.

With the rise in the number of mothers working outside the home, many fear that the number of school-aged children caring for themselves is growing. This concern has prompted public agencies, as well as researchers interested in childcare, to estimate how many school-aged children are left home alone. Bruno (1987) and Cain and Hofferth (1989). for instance, using data from the Current Population Survey, estimated that over two million children between the ages of five and thirteen are left unsupervised at home at some time during the day. Although estimates for self-care arrangements differ, the numbers do indeed suggest that a significant number of school-aged children are caring for themselves during non-school hours. For many adults knowing that a large group of school-aged children care for themselves-some in situations they perhaps cannot handle-is distressing. Yet, with the figures indicating that many parents are The author wishes to thank the Searle Fellowship Fund for supporting this research and Robert Michael, Robert Willis, Lindsay Chase-Lansdale, Judith Seltzer and the anonymous referees for helpful comments and suggestions. Suggestions from workshop participants at the University of Chicago and the University of WisconsinMadison are appreciated as well. Request for reprints should be sent to Peter David Brandon, Department of Sociology, University of Massachusetts-Amherst, MA 01003 [[email protected]].

497

Brandon

498

letting their children be home alone, certain factors must make selfcare arrangements a childcare alternative. Surely, a positive set of attributes-rather than negative ones, like benign neglect or ignorance of children’s needs-better explain why large numbers of parents find self-care arrangements a potential childcare mode. The aim of this paper, therefore, is to better understand those familial characteristics and other factors leading parents to choose self-care arrangements, i.e., arrangements in which children care for themselves unsupervised by an adult during non-school hours. The results contained in this study suggest that there are certain family traits associated with greater use of self-care, like accessible kin and availability of older siblings to supervise youngsters. These findings suggest that families using self-care arrangements choose it for reasons other than a lack of resources, social isolation. or indifference to other forms of childcare.

Background Little is known about the factors determining self-care arrangements, despite surveys suggesting that many children care for themselves. ’ Beyond Rodman and Pratt0 (1987) and Cain and Hofferth’s (1989) works, data on factors leading mothers to allow children to care for themselves is scant. The Rodman and Pratt0 (1987) study investigated factors associated with the time children spend in self-care. Their main findings were that ages of children and mothers’ hours of work were significantly related to greater use of child self-care arrangements. On the other hand, they also reported that maternal religiosity led to less use of child self-care arrangements. These results, however, are based on a flawed study design. First, Rodman and Pratt0 (1987) use a nonrandom sample of employed women, thereby compromising the generalizability of results. Second, their methodology neglected variables that could influence self-care arrangements. Thus, other 1

Little is known about the effects of self-care on children’s emotional development. Some studies argue that children in self-care arrangements are more likely than other children to have behavioral problems (Long and Long, 1982; Rodman et al., 1985; Steinberg, 1986.)

Self-Care Arrangements

499

unaccounted factors could have generated the reported significant associations.’ A more sophisticated multivariate analysis of a nationally representative sample of households by Cain and Hofferth (1989) found, contrary to Rodman and Pratt0 (1987), that mothers’ hours of work were insignificant to use of self-care. Instead, they found that the most important factor predicting increased use of self-care was the age of the child. And the important factor predicting decreased use of selfcare was the availability of other coresiding adults. Contrary to what they expected, mothers’ occupation and educational levels had no discernible effect on the likelihood of using self-care. They also found that there is no difference between self-care by children in one- and two-parent families... (1989:75). Cain and Hofferth (1989) also found that self-care arrangements among school-aged children were more common among middle class children than among poorer children and more likely among higherincome families than among lower-income families. And, controlling for income and family structure, white children were more likely than black children. to have self-care arrangements. These findings contradict the prevailing view that self-care arrangements are correlated with female-headed families and inner city poverty. Cain and Hofferth’s (1989) study contributed to understanding the correlates of self-care arrangements. My study builds upon theirs by allowing the dependent variable in my models to have more than two possible outcomes. The dependent variable in Cain and Hofferth’s (1989) models had collapsed all the alternatives to self-care arrangements into one broader category, non-parental care. The subsequent results that their models yielded rested on the assumption that parents’ childcare choices were revealed by specifying only these two childcare alternatives and parents then choosing one. But, the dichotomous dependent variable Cain and Hofferth (1989) created, self-care versus non-parental care, masks information about parents’ childcare choice behavior. To retrieve that information, I create a dependent variable that contains a broader set of possible childcare modes from which parents choose one. Thus, this dependent variable permits an estimation strategy that shed information about 2The authors fail to address survey non-response rates, item non-response, and selection bias.

Brandon

500

patterns in parents’ childcare choices for school-aged children. I also include additional explanatory variables that were previously omitted from other studies. Cain and Hofferth’s (1989) paper left questions about effects of childcare costs, relationships among sibling ages, and kin networks unanswered. Regression models here, however, take such factors into account, thereby leading to a richer specification of the determinants of self-care and new information, Overall, this study adds to the meager literature on determinants of self-care and informs the debate about childcare for school-aged children.

Theoretical

Perspectives

and Hypotheses

My theoretical positions are drawn from findings in the literature suggesting that parents make age-appropriate childcare choices for their preschool-aged children (Leibowitz, Waite, & Witsberger, 1988). There seems no reason why parents would stop making those types of choices just because their children start school. For many working parents, childcare during non-school hours is a concern (Hayes, Palmer, & Zaslow, 1990). A reasonable proposition is that parents continue making age-appropriate childcare choices throughout the childhood of children. Those choices reflect the developmental needs of children and the resources parents have to meet those needs. Hence, I argue that parents choose self-care arrangements because those arrangements meet their children’s needs and provide a level of care equivalent to care that kin or market providers could offer. Thus, evaluating the gains from use of self-care relative to its alternatives is just like past evaluations parents made about the childcare services their children received when they were pre-school aged. Results produced by testing two hypotheses should provide evidence to support the contention that self-care arrangements are ageappropriate and belong to the set of possible childcare modes for school-age children. The first is that if self-care is a substitute for other modes of care, parental choice of self-care should show a predictable pattern of use that is consistent with the literature. The second proposition is that availability of extended family and family members for childcare provision should affect the likelihood that selfcare is chosen. That is, a special configuration of conditions within

Self-Care Arrangements

501

families and between families and their kin must occur before parents choose self-care arrangements from among the available alternatives. Dealing with the latter hypothesis on the effects of intrafamily attributes on use of self-care first, I contend that parents let their children provide their own home care when they are much older, or if the eldest child in the family can essentially act as the childcare provider. However, if kin are available, or the family has a history of home care exclusively provided by the mother, i.e., maternal care only, then parents will not let their children care for themselves. For families with school-aged children, it seems a reasonable extension, since research already shows that kin are potential providers of childcare for preschool-aged children (Hogan, Hoa, & Parish, 1990), to view older siblings, as well as kin, as possible providers of informal childcare services. If true, that the eldest child can care for their younger siblings, then I should find a direct link between the age of the eldest child and the occurrence of self-care. In fact, until a certain threshold is reached in the difference in ages, selfcare should happen more frequently as the age of the eldest sibling increases; basically, as they mature, their supervision becomes a close substitute for adult supervision. However, eldest siblings are not always a source of low-cost childcare services. If they ‘are too much older than their youngest sibling, the more likely their own activities compete for their time during non-school hours. According to my theory, parents encourage their eldest child’s involvement in school projects and extra-curricular activities, even if this necessitates alternative childcare for their youngest child. I assume that the eldest child’s pursuits matter as tiuch to parents as wanting quality care for their youngest child, an assumption certainly compatible with my theory. So, when the gap in ages between eldest and youngest is too large, the eldest child is essentially unavailable for childcare. In the empirical analyses predicting use of self-care, too large an age gap should reduce the probability of parents choosing self-care. My theory, which is partly based on the-link between kin networks and the provision of childcare, predicts that the presence of kin influences childcare decisions for older school-aged children too. I argue that self-care arrangements will not occur when there is a wellestablished kin network, a network containing various mixtures of

502

Brandon

coresident kin and nearby kin. Relatives, coresiding or nearby, act as low-cost informal providers and their availability should reduce the use of self-care.3 Thus, available kin and eldest siblings are potential low-cost informal childcare providers. Yet their effects on the choice of selfcare should behave in opposite directions. Availability of kin should lower the likelihood that self-care is chosen whereas availability of siblings should increase the chances that self-care is chosen. The reason why the effects of siblings and kin on use of self-care should be polar opposites of each other is that siblings, although they do not constitute adult supervision of children, essentially duplicate it. Because eldest siblings act as surrogate low-cost adult care providers, while coresiding with the younger siblings needing care, empirical models should disclose a negative correlation between presence of an older sibling and choice of self-care. Earlier, I stated that if self-care is considered along with other modes of care, then self-care should also reveal a predictable pattern of use, analogous to patterns of use already documented in the literature. Estimated effects for maternal traits and economic factors, for instance, should parallel findings reported in previous studies. Effects of maternal characteristics on current childcare decisions should operate similarly to their effects on past childcare decisions, though some maternal traits could have changed over time. Better educated mothers, those working longer hours, and those living in the South should still prefer market alternatives for their children, like they did when they were pre-school aged (Lehrer, 1983). Likewise, no compelling reasons exist to think that black mothers and unmarried mothers no longer prefer kin to other modes of care (Stack, 1977); or, that mothers working in blue-collar jobs stop sharing childcare duties with their spouses and start using other modes (Presser, 1986). Nor do grounds exist to think that mothers no longer prefer close substitutes to maternal care when it is needed (Heckman, 1974; Ribar, 1989). Self-care, is a poor proxy for maternal care and this poor 3

A kin network does not perform like a surveillance

device for children who care for

themselves. It seems illogical that parents would confine their available relatives just to this role when by definition their checking on children in self-care must imply they have time to provide child care. It is more realistic to think of parents asking kin to provide the care if those kin have said that they can monitor the children.

503

Self-Care Arrangements

substitutability is indirectly testable. The test involves examining in one child families the correlation between exclusive use of maternal care in the past,4 when the child was of pre-school age, and use of selfcare now. A strong negative correlation between prior use of only maternal care and current use of self-care must hold. If not, then a theory stating that mothers still prefer close substitutes for their own care of their school-aged children during non-school hours lacks verification. My view that self-care is an age-appropriate arrangement would also lack a persuasive piece of evidence. Data show that there is market-provided childcare services during non-school hours for school-age children (Hayes, Palmer, & Zaslow, 1990). Hence, the demand for school-aged childcare should be sensitive to changes in the level of income and to variations in the price of market childcare. The demand for school-aged childcare should behave like the demand for pre-school aged childcare: as the price of market childcare increases, parents substitute away from market care into self-care, a mode typified by low-cost supervision by kin and elder siblings. In addition, income levels should affect the demand for self-care. Higher levels of income lead parents to demand higher quality care, which I assume comes through market providers. The opposite case, where higher income levels increase the quality of self-care seems infeasible_ That means that better quality self-care is purchased, implying added investments in either children or the self-care setting. Both cases seem unlikely.’

Methods and Data This section explains the methods and data used to test my predictions that choosing self-care is affected by: (1) availability of kin and the eldest sibling (2) income levels and the price of market childcare and (3) mothers’ preferences for close substitutes for their own care. 4

Exclusive use of maternal care means that mothers neither work nor study while car-

pg full-time for their child; mothering was the main activity. Stemberg (1986) does, however, note that some parents invest in home items so that children caring for themselves are safe and can contact their parents.

504

Brandon

My theoretical model emphasizes the role of parental preferences and substitution in determining childcare choices. Issues of quality are put aside since my data only identify the presence of particular modes of school-age childcare. The model must assume that the decision making process captures the quality of care used. The parent is assumed to choose among several modes of childcare. To explain the choice of self-care I assume that the cumulative distribution function of F is best described by the cumulative logistic function. Hence I use the multinomial logit formulation of this probabilistic relation to test the hypotheses about use of self-care that were discussed previously.6 The data I use to empirically test my predictions come from the National Longitudinal Study of the High School Class of 1972 (NLS’72). The NLS’72 is a national probability sample of over 22,000 high school seniors in 1972. The fifth follow-up survey, which is the data for this study, was delivered to an unequal probability subsample of 14,489 of the original respondent (Spencer et al., 1987). The comprehensiveness of the data gathered on childcare makes the NLS’72 suited to analyzing determinants of self-care arrangements. Precise questions collected data on the types of care used, the cost of each type, and the hours utilized in each. The question about self-care specifically asked whether parents’ school-age child(ren) cared for themselves without supervision for some period of time during nonschool hours.7 The data on modes of care chosen enabled me to construct an unordered categorical variable Care Choice. This variable, which has four categories of childcare (1 = self-care; 2 = parental care; 3 = kin care; 4 = market care), is the dependent variable for the analyses. The NLS’72 provides much data on demographic traits of mothers, their employment histories, and birth histories of their children. These data allow testing the predictions that the age structure between

‘%he conditional multinomial logit can suffer from a property known as the independence of irrelevant alternatives (Judge et al., 1985). The property results from assuming that the errors are independent when in fact they are not. To guard against this problem, I explored the data ensuring that the child care alternatives were distinct. ‘Responses to this survey item identify mothers who solely use self-care and those who use it in conjunction with other modes of care. Unfortunately, though, activities these children pursue, where they go, and when they care for themselves is unknown.

Self-Care Arrangements

505

youngest and eldest siblings affects use of self-care, as does mothers’ preferences for substitutes for their own care. I test the latter prediction by exploiting the employment and education histories that were collected on all females in the fifth follow-up survey. I measure (in one child families only) the elapsed time between the birth of that only child and the mother’s entry (or reentry) into the labor market.* Those mothers with longer spans of time spent full-time mothering, (i.e., they did not work or attend school), are considered the mothers most likely to prefer close substitutes for their own care when their child reaches school-age and needs alternative care during non-school hours. Creating variables to test the former prediction about the effect of the age gap between siblings was straightforward. The survey module on birth histories furnishes each child’s age. Once these ages are obtained, the gap between the eldest and youngest child is easily generated. I include variables on mothers’ traits, such as completed education, marital status, race, and occupation, along with regional controls, so that I can identify the effects of work effort on the likelihood of self-care use. The NLS’72 also provides detailed information on mothers’ kin networks. The survey enumerates the number of coresiding kin, establishes proximity of respondents’ parents, and tabulates the frequency of visits to respondents. I use this data to construct two measures of kin availability. One measure indicates the number of coresiding kin. It is recoded to a binary variable that simply records the presence of coresident kin. The second measure indicates the closeness of a parent. A parent is coded as close if they are within 10 miles and visit more than once a week, I view these two explicit measures of kin availability as tests of my theory. As the data also contain information on the cost of care chosen, ordinary least squares are used to predict childcare prices for competing modes of childcare. This technique is used’ as geographic identifiers in the survey allow me to append county-level data to

‘See Spencer et al., 1987, for details on the modules for schooling, employment, and fertility. ‘Others have used the modal regional price for each child care mode as a representation of the market price of child care (Stolzenberg & Waite, 1984; Blau & Robins, 1988).

Brandon

506

reported childcare expenses. This permits estimating a predicted price for market childcare under the assumption that adding county-level variables reflects market conditions affecting the price of childcare for school-aged children. A final strength of these data is the inclusion of non-employed women using childcare. Their inclusion allows for increased confidence in estimated effects because childcare decisions are not censored by employment status.” Past studies of childcare demand have lacked this population of childcare users and have had to make sample selection corrections to estimated parameters (Heckman, 1979; Hotz and Kilburn, 1991). Notwithstanding its strengths, a weakness of the survey is that it limits the sample to those who attained the twelfth grade in high school. Hence, the survey is not representative of the national population. This selection bias is a minor issue for white mothers in the sample because high school dropout rates have plummeted for whites (Jaynes and Williams, 1989). But the selection bias is more troublesome for the sample of black mothers. Data indicate that up to 45 percent of blacks in this cohort could have dropped out of high school (Fine, 1986; Jaynes and Williams, 1989). Hence, results pertain only to that subset of the population who reached their senior year in this grade cohort.

Results My findings are based on a sample of 918 mothers who responded to the module on childcare arrangements for school-age children in the fifth follow-up survey and who had children between the ages of 6 and 13.” Of this group, 156 mothers (16.9 percent) used only self-care, 380 mothers (41.4 percent) used only parental care, and 382 mothers (41.4 percent) used childcare from either kin or market providers. Table 1 contains definitions of variables used in the analyses. Before “Other comparable child care data (e.g., SIPP and NLSY) ask child care questions only of working women. “The fifth follow-up data contains 4,021 who specified the form of child care used. The subsample of 1,404 mothers who only had school-aged children are part of that group and 870 of them use some form of non-parental child care. Of the 1,404 mothers, 20.1 percent said that the “child cares for self (without supervision)” during nonschool hours. Missing data reduced the sample to 9 18 mothers.

507

Self-Care Arrangements

discussing multivariate results, however, Table 2 shows differences among this sample of mothers that are evident once they are grouped according to the modes of care chosen. TABLE 1 Summary Definition of Variables Used in Modeling Determinants Choice

Variable

College degree Black Blue-collar worker Chlsage Coresident kin Education Family income Hours in care Hours worked Northeast Midwest South West Grandmother close by Sibgap Price of childcare Unmarried Work experience Number of children Mattime Self-care Parental care Kin care Market care Care choice

of Childcare

Definition

1 if received college degree, 0 otherwise 1 if black, 0 otherwise 1 if works in blue-collar job, 0 otherwise Age of oldest child currently co-residing 1 if respondent lives with kin, 0 otherwise Years of education Income from all sources in 1985 (in thousands of dollars) Number of hours in non-parental care per week Hours worked in the labor market per week 1 if lives in Northeast, 0 otherwise 1 if Iives in Midwest, 0 otherwise 1 if lives in South, 0 otherwise 1 if lives in West, 0 otherwise 1 if mother within 10 miles, 0 otherwise Difference in years between age of eldest school-aged sibling and age of youngest sibling who is also school-aged Predicted hourly price of childcare 1 if unmarried, 0 otherwise Years of full-time work for pay Number of own children in household Amount of time mother spent in full-time mothering after the birth of her only child (in months) 1 if only use self-care, 0 otherwise 1 if only use spouse or self, 0 otherwise 1 if only use kin care, 0 otherwise 1 if only use center or informal (nonkin) care, 0 otherwise 1 if only use self-care 2 if only use parental care 3 if only use kin care 4 if only use market care

Table 2 shows that mothers using self-care are more likely to: live in the South, have an older first child, have a larger age gap between

Brandon

508

eldest and youngest children, and have more income. Their attachments to the labor force-measured in either weekly hours of work or years of work experience-equal or exceed those of mothers using market care. But they are less likely to: live in the North-East, have a mother nearby and have a college degree. TABLE 2 Selected Statistics of Mothers with School-Aged Children, by Primary Childcare Arrangements

Mother’s Characteristics Black (%) Work experience (years) Blue-collar worker (%) Education (years) Hours worked College degree (%) Unmarried (%) Family characteristics Grandmother close by (%) Coresident kin Number of children Price of childcare Family income (in thousands of dollars) Hours of care Chlsage Sibgap Regional characteristics South (%) West (%) Northeast (%)

Self-Cafe

Parental Care

Kin Care

Market Care

17.4 5.9 20.7 13.2 36.3 4.4 27.4

14.5 3.7 28.5 13.0 15.9 4.2 14.9

25.7 5.4 23.5 13.3 30.3 6.4 40.9

11.1 6.4 17.3 13.5 35.5 9.3 40.1

41.4 1.80 2.1 n.a.

45.3 2.01 2.1 n.a.

71.5 1.96

45.7

:k906*

1.7 $3.30*

26.69 9.52 12.22 2.50

30.30 n.a. 10.25 2.33

24.86 16.60 10.37 2.08

27.87 18.32 8.69 1.69

43.8 16.8 9.9 156

27.1 14.1 22.3 380

46.1 7.8 19.5 181

43.3 16.7 14.5 201

1.74

N * Predicted. n.a.: not applicable Source: National Longitudinal Study of the High School Class of 1972, fifth follow-up survey.

Table 2 also shows that time spent in self-care is smaller than time spent in either market or kin care; typically, children spend about two hours a day unsupervised. Table 3 presents the descriptive statistics for the variables used in multivariate models predicting use of self-care for the full sample of

Self-Care Arrangements

509

mothers (N=918), for mothers with only two children (N=7 lo), and for mothers with only one child (N=322). Table 3 shows that the incidence of self-care is lower in one-child families (14.2 percent) compared to two-child families (16.3 percent).” Effects of Kin and Children S Ages on Self-Care Use The models in Tables 4, 5. and 6 examine how the age gap between youngest and oldest siblings, the age of the eldest sibling, and accessibility of kin affect the likelihood that self-care is chosen. When other determinants are held constant, the results suggest that an important determinant of self-care is availability of kin for care and supervision of school-aged children. Grandmothers nearby and kin coresiding decrease the chance that self-care is chosen relative to kin provided care. Across all three tables, the estimated effect of nearby grandmothers is to strongly decrease self-care use in favor of kin care, possibly their care. If access to kin increased the likelihood of selfcare arrangements then that would suggest that only a weak condition for self-care was necessary: kin nearby-that does not happen according to these data. The estimated effects of coresident kin are weaker but still unambiguous. Live-in kin lower the likelihood that self-care is used in families with more than one child, presumably because live-in kin are part of the kin network established to care and supervise children. The other case, shown in Table 6, where coresident kin in one child families increase rather than decrease use of self-self is also a consistent result, even though it appears to contradict findings reported in Tables 4 and 5.

‘*For families with three or more children, the incidence of self-care rises to 19.2 percent. Estimates are lower and less biased than other estimates as they are not conditioned on mothers’ labor force participation. Prior estimates have been based on surveys of labor force participation among women; in computing them, it was assumed that if the mother worked full-time but did not use market child care, then her children were at home unsupervised. Hence, such estimates that cannot account for spouse- or kin-provided child care upwardly bias the incidence of self-care.

Brandon

TABLE 3 Means and Standard Deviations of Variables Used in Models Predicting Determinants of Self-Care Two-Children Families

One-Child Family

16.9 41.4 19.8 21.9

16.3 37.9 20.7 25.1

14.2 32.3 24.9 28.5

1.91 (0.65) 0.49 (0.50) 0.16 (0.37) 13.15 (1.60) 0.27

1.89 (0.69) 0.50 (0.50) 0.16 (0.37) 13.21 (1.65) 0.31 (0.46) 10.32 (2.17) 1.76 (1.98) n.a.

1.90 (0.75) 0.51 (0.50) 0.21 (0.41) 13.49 (1.78) 0.39 (0.49) 9.32 (2.10) n.a.

Full Sample Response Variable Care Choice Self-care Parental Kin Market Predictor Variables Coresident kin Grandmother

close by

Black Education Unmarried

(0.44) 10.79 (2.38) 2.30 (2.34) n.a.

Chlsage Sibgap Mattime Blue-collar

worker

Hours worked Family income Price of childcare South West Northeast N Source: National Longitudinal survey. na.: not applicable

58.8 (39.7) 0.24 0.23 0.21 (0.43) (0.42) (0.41) 25.70 27.30 29.70 (18.54) (17.74) (16.79) 28.78 27.52 26.63 (18.82) (16.88) (16.57) 3.76 3.63 4.14 (3.69) (3.15) (3.41) 0.36 0.38 0.38 (0.48) (0.49) (0.49) 0.14 0.15 0.15 (0.35) (0.36) (0.36) 0.18 0.17 0.20 (0.39) (0.38) (0.40) 918 710 322 Study of the High School Class of 1972, fifth follow-up

Self-Care Arrangements

511

TABLE 4 Multinomial Logit Analysis Predicting Use of Self-care: Full Sample (Parameter estimates and standard errors)

Intercept Coresident kin Grandmother close by Black Education Unmarried Chlsage Sibgap Blue-collar worker Hours worked Family income Price of care south West Northeast

N

Kin Care V/Self-Care 3.21** (1.33) 0.62* (0.35) 1.10*** (0.25) 0.38 (0.35) 0.08 (0.07) 0.39** (0.30) -0.39*** (0.06) 0.10* (0.05) 0.54* (0.29) -0.03*** (0.01) -0.01 (0.01) -0.01 (0.02) 0.31 (0.31) -0.75* (0.43) 0.42** (0.41)

Market Care V/Self-Care 5.62*** (1.30) 0.3 1 (0.37) -0.18 (0.25) 0.32 (0.44) 0.09 (0.07) 0.30 (0.30) -0.60*** (0.06) 0.09 (0.05) 0.23 (0.30) -0.01 (0.01) 0.01 (0.01) -0.11** (0.05) 0.17 (0.30) -0.21 (0.37) 0.12 (0.42)

Parental Care V/Self-Care 7.45*** (1.25) 0.10 (0.36) 0.11 (0.22) -0.23 (0.36) -0.03 (0.07) -0.66** (0.29) -O-36*** (0.06) 0.05 (0.05) 0.69** (0.26) -0.06*** (0.01) -4.43e-b (0.01) -0.0003 (0.02) -0.26 (0.27) -0.68** (0.35) -0.01 (0.38)

918

Log likelihood -964.98 Source: National Longitudinal Study of the High School Class of 1972. fifth follow-up survey. p < .lO; ** p < .05; *** p c .Ol

Brandon

512

Live-in kin in the one-child families increase the probability of self-care because their impact on the use of self-care resembles the impact that an available eldest sibling has on use of self-care. Obviously, qualitative distinctions between childcare provided by livein kin in one child families and childcare provided by elder siblings in two child families become blurred. Other results support my theory that the eldest child’s age and the gap in ages between youngest and oldest sibling affect self-care use. The older the eldest child, or the older an only child, the more likely parents will choose self-care over alternatives. These results in Tables 4, 5,. and 6, support my prediction that the older an elder sibling becomes, presumably gaining maturity as they grow, the more alike they are to adult childcare providers. Or, for one child families, (Table 6) the older the only child becomes, the more likely they are to pursue their own activities and provide their own home care. The variable measuring the age gap between youngest and oldest siblings produces the predicted effect on use of self-care too. The estimated coefficient for Sibgap indicates that when there is too greater a disparity in ages between these two children, the chances that elder siblings double as childcare providers are lowered; parents, by necessity, must choose among the other alternatives to self-care. Clear confirmation of this effect is found in two-child families where estimated coefficients for Sibgap in columns 1 and 2 of Table 5 are large and significant. Those coefficients show that a large divergence in ages between the two siblings reduces the chances that self-care is used relative to kin or market care. Overall, the estimated effects for Ch lsage, Sibgap, Coresident Kin and Mother close by are consistent with the argument that kin networks and eldest siblings provide childcare services and are preferred alternatives to self-care arrangements. Analyses document that kin remain a source of low-cost informal childcare well into children’s formative school years, which parents will continue to use over other modes, like self-care. Efsects of Maternal

Characteristics

on Self-Care Use

I found no effect of maternal levels of education on use of selfcare. That says more about the data than a theory about mothers’ edu-

Self-Care Arrangements

513

cation levels and childcare choices, however. Because this sample is conditioned on high school graduation, an important source of variation, namely, high school dropouts, is removed from the analysis.

TABLE 5 Logit Analysis Predicting Use of Self-care: Two-Children Families (Parameter estimates and standard errors) Kin Care Market Care Parental Care V/Self-Care V/Self-Care V/Self-Care Intercept 3.88** 7.09*** 8.48*** (1.59) (1 S4) (1.51) Coresident kin 0.38 -0.10’ 0.13 (0.39) (0.41) (0.40) Grandmother close by 1. lo*** -0.36 0.16 (0.30) (0.29) (0.27) Black 1.47*** 1.23** 0.62 (0.58) (0.58) (0.58) Education 0.12 0.10 -0.003 (0.08) (0.08) (0.08) Unmarried 0.41* 0.51 -0.8 1** (0.35) (0.34) (0.34) Chlsage -0.48*** -0.74*** -0.47*** (0.08) (0.08) (0.08) Sibgap 0.16*’ 0.22*** 0.08 (0.07) (0.08) (0.07) Blue-collar worker 0.49 0.05 0.50 (0.33) (0.35) (0.31) Hours worked -0.03*** -0.01 -0.06*** (0.01) (0.01) (0.01) Family income -0.01 0.01 -0.01 (0.01) (0.01) (0.01) Price of care -0.14** -0.23*** -0.11* (0.07) (0.06) (0.06) South 0.49 0.47 -0.07 (0.35) (0.34) (0.32) West -0.74 -0.11 -0.47 (0.48) (0.41) (0.39) Northeast 0.47 0.40 -0.18 (0.48) (0.48) (0.44) N 710 Log likelihood -747.02 Source: National Longitudinal Study of the High School Class of 1972, fifth follow-up survey. * p c .lO; ** p < .05; *** p c .Ol Multinomial

Brandon

514

TABLE 6 Multinomial Logit Analysis Predicting Use of Self-care: One-Child Families (Parameter estimates and standard errors)

Intercept Coresident

kin

Grandmother

close by

Black Education Unmarried Chlsage Sibgap Blue-collar

worker

Hours worked Family income Price of care South West Northeast

N

Kin Care V/Self-Care 8.43** (2.68) -1.13* (0.57) 1.87*** (0.50) 2.36*** (0.87) 0.05 (0.12) 1.09** (0.55) -0.87*** (0.17) 0.24*** (0.01) 0.44 (0.56) -0.07*** (0.02) 0.01 (0.01) -o-13** (0.09) 0.10 (0.63) -1.75** (0.81) -1.02 (0.80)

Market Care V/Self-Care 11.66*** (2.63) -0.34 (0.59) -0.42 (0.48) 1.56* (0.84) 0.01 (0.12) 0.58 (0.53) -1.07*** (0.17) 0.026*** (0.01) -0.08 (0.56) -0.03 (0.01) 0.03 (0.01) -0.21** (0.09) 0.29 (0.60) -1.02 (0.7 1) -1.24 (0.81)

Parental Care V/Self-Care

11.96.*** (2.60) -0.85 (0.57) 0.55 (0.45) I .22 (0.84) -0.07 (0.11) -0.41 (0.50) -0.70*** (0.16) 0.011* (0.01) 0.18 (0.52) -0.08*** (0.01) 0.01 (0.01) -0.09 (0.09) -0.08 (0.59) -0.95 (0.69) -0.97 (0.75)

322

Log likelihood -327.43 Source: National Longitudinal Study of the High School Class of 1972, fifth follow-up survey. * p < .lO; ** p < .05; *** p < .Ol

Self-Care Arrangements

515

The insignificant education effect reflects this lack of variability in the data. Self-care arrangements are found to less frequently occur among Black mothers. The effect of race is most apparent in families with two children. Blacks more than whites are predisposed to use kin rather than self-care, a result consistent with the existing childcare literature (Stack, 1974). Further, Tables 4 and 5 indicate that unmarried mothers are less likely than married mothers to use alternatives to self-care when those alternatives exist. If kin are available, then unmarried mothers are less likely to choose self-care. Again this finding concurs with previous findings showing that lowincome mothers’ prefer informal kin provided care for their preschoolaged children (Berger & Black, 1992; Connelly, 1991). My contention that childcare choices for school-age children are like decisions for pre-school children is supported. In contrast to my prediction, and results in Cain and Hofferth’s (1989) study, however, models show a positive correlation between work hours and use of self-care. Mothers’ hours of work increase use of self-care relative to kin care or parental care. This unexpected result, though hard to interpret, may mean that as work hours grow, mothers become less able to work all those hours while their children are in school. Inevitably, expanding work hours must increase the number of hours overlapping with children’s non-school hours. The variable measuring hours at work may be showing this dynamic.” Table 6 shows that exclusive use of maternal childcare in the past, when the child was pre-school aged, lowers the likelihood of self-care use when the child is school-aged. This effect was expected: a longer period out of the labor force during the child’s preschool-aged years diminish the likelihood that self-care is chosen during the child’s school-aged years. This is evidence, at least for one-child families, showing that mothers prefer close substitutes for their care and that self-care is not one of them.

‘*able 6 shows that when the choice is between self-care and parental care mothers in blue-collar jobs are more likely to choose self-care compared to mothers in other jobs. The result supports Presser and Cain’s (1983) study that found two-earner blue-collar couples stagger work schedules so that one parent can care for the children while the other works.

Brandon

516

The result also indicates that childcare choices early in a child’s life may predict future childcare choices. The particular case here showed that a mother’s past decision not to substitute for her own care while the child was pre-school aged strongly predicts use of adult care providers once the child is school-aged. Among the alternative adult providers, it seems reasonable to conclude that the closest possible substitutes for her own care once the child is school-aged is probably an adult relative. That conclusion is supported in these data. Table 6 shows that the magnitude of the estimated coefficient for Stime between use of kin care relative to use of self-care is much larger then the other two estimated Stime coefficients: for parental care and for market care. Effects of Economic Factors on Self-Care

Use

Tables 4, 5, and 6 reveal no effect of income on choice of selfcare. The expected direction of an income effect on use of self-care is observed, yet the coefficient is statistically insignificant. This puzzling result may suggest that either non-wage income or earnings from spouses are the sources of income affecting use of self-care. Higher earnings of the spouse may decrease the labor supply of mothers, thereby making mothers more likely to provide home care services. If mothers decrease work hours, then, most likely, the margin they cut on are those hours overlapping with children’s school hours. The price of market childcare is found to significantly affect use of self-care. Tables 4, 5, and 6 show that self-care is sensitive to a predicted market price of childcare. This implies that the demand of mothers for market childcare for their school-aged children is price sensitive, just like their demand for market childcare for younger children is price sensitive (Ribar, 1989; Heckman, 1974; Blau and Robins, 1988). The price effect confirms that self-care arrangements are a lowcost form of care, belonging to the set of childcare alternatives from which mothers choose. Also, the estimated price effect is consistent with the prediction that a rise in the price of market childcare should increase mothers’ demand for lower-cost substitute forms of childcare. Here, a rise in the price of market childcare induces mothers to switch to low-cost alternatives, of which self-care is one.

Self-Care

Arrangements

517

In sum, many predictions receive empirical support suggesting that self-care is one home-care option that parents consider when making childcare arrangements for school-aged children, Results confirm that many of the variables affecting parents’ past childcare decisions also affect later childcare decisions in the same way, even when the hours of childcare demanded have been reduced. Moreover, empirical tests validate my contention that the unique nature of selfcare arrangements requires a special set of family conditions.

Conclusions Self-care arrangements evoke apprehension among many people. Knowledge about its determinants is needed to allay public disquiet and to paint a more accurate picture of why parents choose it. Until the public is better informed, many will continue to perceive school-age children who care for themselves as living in poor families and left isolated. This study’s results cast doubt on this commonly held perception, however. Results from this study suggest that self-care arrangements do not necessarily imply that employed mothers leave their children unsupervised. Rather, my findings suggest that a complex set of familial and economic factors lead mothers to make age-appropriate childcare decisions, of which self-care is one. The concern over this childcare practice is part of the childcare policy debate. That debate deserves as much new data as possible because without understanding which factors lead mothers to use selfcare, it seems hard to know how, and where, government should aim its intervention. The study’s findings at least suggest that policies aimed subsidizing childcare for school-aged children may diverge from policies needed to address the latch-key child syndrome, Results strongly imply that children in latch-key conditions are perhaps a very distinct group differing from the population of children in self-care arrangements. Second, as federal financing of childcare services has fallen (Robins, 1989), gauging effects of economic constraints on mothers’ childcare choices for school-aged children is important. Unless the effects of prices and income on mothers’ childcare decisions for all

Brandon

518

children are determined, the childcare policy debate for school-aged children could remain under-informed, and it will be impossible to fully ascertain the effects of specific childcare proposals, like subsidizing the supply of after-school care. Evidence suggests that subsidizing the price of childcare for school-aged children has the same impact on childcare demand as subsidizing the price of childcare for preschool-aged children. Hitherto, few studies have been able to show this connection between different sets of childcare choices for the same set of children. Policy makers need to also debate whether self-care arrangements for schoolaged children are a good thing as compared to other forms of childcare. The results here cannot speak to this issue. Nor can it address the possibility the factors like children’s maturity and children’s own preferences affect mothers’ decisions on self-care.

References Aldrich, J. H., & Nelson, F.D. (1986). Linear probability, logit, and probit models Sage University Papers, No 45. Beverly Hills: Sage University Press. Berger, D.M., & Black, D.A. (1992). Childcare subsidies, quality of care, and the labor supply of low-income, single mothers, Review of Economics and Statistics, 74.635-642. Blau, D.M. & Robins, PK. (1988). Child-care costs and family labor supply. Review ofEconomics and Statistics, 70(4):374-g 1. Blau, D.M. (Ed.), (1991). The Economics of childcare. Russell Sage Foundation. Bruno, R.R. (1987). After-school care of school-age children: December 1984. Current Population Reports, Series p-23, no. 149. Washington, DC: U.S. Government Printing Office. Children’s Defense Fund. (1982). The childcare handbook: needs, programs, and possibilities. Washington, DC: Children’s Defense Fund. Cain, V. & Hofferth, S. (1989). Parental choice of self-care for school-age children, Journal of Marriage and the Family 5 1 (February): 6577. Connelly, R., (1990). The effect of childcare costs on the labor force participation and AFDC recipiency of single mothers, Discussion Paper 92090 (Institute for Research on Poverty. Connelly, R., (1991). The importance of childcare costs to womens decision marking, in David Blau, ed., The economics of childcare. (pp. 8390), New York: Russell Sage Foundation.

Self-Care Arrangements

519

Duncan, G.J. & Hill, R.C. (1974). The childcare mode choice of working mothers. In G. Duncan & Morgan, J.N. (eds.), Five thousand american familiespatterns of economic progress (4): 301-324 Ann Arbor: Institute for Social Research. Fine, M., (1986). Why urban adolescents drop into and out of public school. Teachers College Record, 87:393-409. Garbarino, J. (198 1). Latchkey children: how much of a problem? Education Digest, 46(6): 14- 16. Hayes C., Palmer, J. & Zaslow, M. (1990). Who cures for America’s children? Washington D.C.: National Academy Press. Heckman, J.J. (1974). Effects of child-care programs on women’s work effort. Journal Of Political Economy 82(2, Pt2):S 136-63. He&man, J.J. (1979). Sample selection bias as a specification error. Econometrica 47:93 l-960. Hogan D., Hao, L. & Parish, W. (1990). Race, kin networks, and assistance of mother-headed families. Social Forces 68(3):797-g 12. Hour, V.J. & Kilburn, M.R. (1991). Childcare decisions: are there differences by women’s working status? Chicago: The University of Chicago. mimeo. Jaynes, G. D. & Williams, R.M., (1989). A common destiny: blacks and American society. Washington, D.C.: National Academy Press. Judge, G., Griffiths, W.E., Hill, R.C., Lutkepohl, H., Lee, T. (1985). The theory and practice of econometrics Second Edition Wiley Lehrer, E. (1983). Determinants of childcare mode choice: an economic perspective. Social Science Research 12:69-80. Leibowitz, A., Waite, L., & Witsberger, C. (1988). Childcare for preschoolers: differences by child’s age. Demography 25: 205219. Long, T. J. & Long, L. (1982). Latchkey children: The child’s view of self cure. Washington, DC: Catholic University of America. (ERIC Document Reproduction Service No. ED 211 229). Presser, H.P. (1986). Shift work among American women and childcare. Journal of Marriage and the Family 48: 55 l-563. Ribar, D. C. (1989). Childcare and the labor supply of married women: Structural evidence. Providence: Brown University. Mimeo. Riccobono, J., Henderson, L.B., Burkheimer, G.J., Place, C., & Levinsohn, J.R. (1981). National longitudinal study: base year (1972) through forth followup (1979) data file users manual. Washington D.C.: National Center for Educational Statistics. Robins, P & Spiegelman, R. (1978). An Econometric Model Of The Demand For Childcare. Economic Inquiry 16:83-94. Robins, P. (1989). Federal financing of childcare: Alternative approaches and economic implications. The University of Wisconsin. IRP Discussion Paper Series, 890-89.

520

Brandon

Robinson, B. E., Rowland, B. H. & Coleman, M (1986). Latchkey kids. Lexington, MA: Lexington Books,. Rodman, H. & Pratto, P.J. (1987). Child’s age and mother’s employment in Journal of relation to greater use of self-care arrangements for children. Marriage and the Family, 49,573-578. Rodman, H., Pratto, P.J. & Smith Nelson, R. (1987). Childcare arrangements and children’s functioning: A comparison of self-care and adult-care children. Developmental Psychology, 2 I,4 13-4 18. Stack, C. (1974). All our kin: Strategies for survival in a Black community. New York: Harper and Row. Steinberg, L. (1986). Latchkey children and susceptibility to peer pressure: An ecological analysis. Developmental Psychological, 22:1-7. Stolzenberg, R. & Waite, L. (1984). Local labor markets, children and labor force participation of wives. Demography 21, 157-170. Spencer, B., P. Sebring & Campbell, B. (1987). NLS-72 Fifth followup sample design report. Washington D.C.: National Center for Education Statistics. U.S. Bureau of the Census. (1982). Trends in Childcare Arrangements of Working Mothers, Current Population Reports, Special Studies, Series P-23, No. 117. Zigler, E. F. & Long, T.J. (1983). Testimony before the U.S. Senate Children’s Caucus: Policy Forum to Examine Problems Confronting Latchkey Children (9 June). Zucker, H. L (1944). Working parents and latchkey children. Annals of the American Academy, 236,43-50.