The allocation of food expenditure in married- and single-parent families.
As of the year 2000, 15% of children and adolescents (ages 6--19) were overweight (National Center for Health Statistics 2002). Being overweight is due in part to a combination of genetic factors, poor nutritional habits, and inactivity. Overweight youth have a 70% chance of becoming an overweight or obese adult, which is significant given the links between being overweight and obesity in adulthood and health problems (Whitaker et al. 1997). Although individuals cannot control their genetic endowment, exercise and choices regarding nutritious food consumption play an important behavioral role in a healthy lifestyle.
The incidence of child overweight and obesity has garnered the attention of policy makers and has led to legislation stressing the importance of good nutrition, regular exercise, and preventive health screenings. (1) The Agriculture Department, also responding to the nation's obesity problem, revamped the food pyramid by taking age, sex, weight, and exercise into consideration (USDA 2003). The new food pyramid (www.mypyramid.gov) emphasizes the importance of a healthy diet consisting of fruits, vegetables, whole grains, lean meats (including poultry, fish, beans, eggs, and nuts), and fat-free or low-fat milk products. Similarly, a healthy diet is one that is low in fats, salt, and added sugars.
Eating habits are formed early in life, and preferences and choices that are determined during childhood can shape lifelong eating habits. Thus, it is a concern that many children have poor diets. Prior research finds that 81% of the diets of two- to nine-year-olds were poor or in need of improvement (Lino et al. 2002) and that parents' nutrition knowledge affects children's knowledge and dietary practices (e.g., Contento et al. 1993; Oliveria et al. 1992). Households with low socioeconomic status have been found to have especially poor diets (Ramezani and Roeder 1995; Variyam et al. 1998).
Estimates from nationally representative datasets have examined the prevalence of obesity and overweight by race and socioeconomic status, but very little attention has been paid to family structure. An inverse relationship between income and obesity has been established among adult women, but not among men and children (for a review of this literature, see Sobal and Stunkard 1989). Specific analyses among children find conflicting results. Results suggest that income has an inverse relationship with obesity for white children and adolescents but that this relationship does not hold for Mexican American or black youth (National Center for Health Statistics 1998; Troiano and Flegal 1998). However, obesity and overweight rates among youth are highest for Mexican American males aged 6-11, black females aged 6-19, and adolescents between 12 and 19 from low-income households (U.S. Department of Health and Human Services 2000). Because single-parent families are more likely to be both poor and headed by a nonwhite parent, one might expect the prevalence of obesity and overweight among children residing in these households to be greater than that among other structures.
While no national estimates are available for obesity and overweight rates among children residing in single-parent families, one might expect that single-parent families have characteristics that could influence differences in nutrition, food expenditure, and consumption. This paper is primarily interested in the role of family structure in food expenditure decisions. Children growing up in single-parent families have an elevated risk of health problems (Dawson 1991) partly because of the association between single parenthood and poverty. However, other studies suggest that even when controlling for economic, poverty, and maternal education measures, residing in a family with a persistently unmarried parent or living in a family with a single mother early in life is related to chronic health problems (Lipman and Offord 1997). It is unclear if these health conditions are related to overweight and obesity.
In addition to having fewer economic resources on average, single parents also face greater time demands than married parents (Sandberg and Hofferth 2001) and single mothers have been found to be more socially isolated than married mothers (McLoyd 1990). Increased time demands might lead single parents to trade off good nutrition for the convenience of fast food or prepared foods (Zick and McCullough 1996). Single parents' social isolation might be associated with a lack of information or social support that would help them make better nutritional choices (Ramezani and Roeder 1995). Thus, the first empirical question we address in this paper is whether food expenditures vary between married-parent and single-parent families. All else equal, we expect to see "better," or more nutritionally sound, food expenditure decisions by married parents compared to single parents.
As single parents, both mothers and fathers may face increased time pressures and social isolation relative to their married counterparts. However, there is reason to believe that single mothers and fathers "parent" differently in ways that might be reflected in their food expenditure decisions. This belief is based on evidence from empirical studies of married couples that indicate, for example, that mothers' control over family resources is associated with improvements in children's health and nutrition (Thomas 1990, 1994). Similarly, increases in a wife's income relative to her husband's income are associated with greater expenditure on food, child care, and children's and women's clothing, and reduced expenditure on transportation (Phipps and Burton 1998). Other evidence that husbands and wives may differ in their preferences for expenditures comes from Lundberg and Pollak's (1996) and Lundberg, Pollak, and Wales's (1997) finding that when wives control more of the family resources than husbands, expenditures on children increase. Lundberg, Startz, and Stillman (2003) note that wives prefer to save more than do their husbands and that relative control over household decisions is affected by control over market income. These studies support a model of marital bargaining and suggest that children might fare better when their mothers control a larger fraction of family resources. Thus, all else equal, we expect to see better or more nutritionally sound food expenditure decisions by single mothers compared to single fathers. Therefore, the second empirical question addressed in this paper is whether single mothers and single fathers differ in their food expenditure decisions, both compared to each other and compared to married parents.
Previous research has examined differences between single-parent and married-parent families in both their overall expenditure levels and the shares allocated to major expenditure categories including food, housing, clothing, and transportation (Abdel-Ghany and Schwenk 1993; Boyle 1989; Horton and Hafstrom 1985) and between single mothers and single fathers (Lino 1990, 1991; Paulin and Lee 2002). Much less attention has been paid to differences in the allocation of food expenditures across family types, although Paulin and Lee (2002) found that single fathers spend a greater share of their total budget on food away from home than do single mothers.
This study uses data from the Diary Survey component of the Consumer Expenditure Survey (CEX) from 1990 to 2003. The CEX is conducted by the Bureau of Labor Statistics and collects information on the purchasing habits of the nation's households and families. The CEX sample is representative of the total noninstitutionalized urban and rural population in the United States. The data are collected in independent quarterly Interview and weekly Diary surveys of approximately 7,500 sample households (5,000 prior to 1999). During the last six weeks of the year, the Diary Survey sample is supplemented to twice its normal size to increase the reporting of types of expenditures unique to the holidays. Each survey has its own independent sample, and each collects data on household income and socioeconomic characteristics.
The Interview and Diary components of the CEX are two separate surveys, with two separate representative samples. The Interview survey is designed to obtain data on the types of expenditures respondents can recall for a period of three months or longer. Each consumer unit (2) is interviewed once per quarter for five consecutive quarters. The Diary Survey, on the other hand, is designed to obtain data on frequently purchased smaller items, including food and beverages, housekeeping supplies, tobacco, nonprescription drugs, and personal care products and services. In the Diary Survey, respondents keep track of all their purchases made each day for two consecutive one-week periods. The Diary Survey is especially valuable for collecting data on frequently purchased items such as food and beverages as these purchases are less likely to be recalled over a long period of time. The Diary Survey also collects employment and income information for all members of the household, as well as demographic information including family structure.
The sample used for the empirical analysis consists of 29,376 households with children in which the household head is married, never married, or divorced who participated in the Diary Survey in any year between 1990 and 2003. In creating this sample, households with either incomplete income or expenditure data (799 households) or where the household head was separated or widowed (1,801 households) were dropped. Those who were separated are not included because definitions of what separation refers to will vary by person (i.e., whether they are legally separated and have filed for divorce or whether the spouse is living outside of the household). Moreover, the separation process has multiple outcomes including staying married or eventually divorcing. Because we have no way to determine the ultimate family structure decision in this dynamic process, we have no basis for assigning them to the married or single group. Furthermore, widowed heads are not considered single by choice, resulting in differences in unobservable characteristics associated with this type of family structure compared to other single parents. Prior research indicates that widowed-parent families look more like two-parent families than single-parent families (McLanahan and Sandefur 1994). (3) The final sample of 29,376 households with children contains 23,789 households in which the head is married and 5,587 in which the head is single. Of these single-headed families, 2,344 are never-married and 3,243 are divorced single-parent families. Also, 4,629 are single mothers and 958 are single fathers.
Although the Diary Survey collects data for two consecutive weeks from respondents, not all respondents complete both weekly records. These analyses use all weeks that a consumer unit participates in the survey. Each diary is treated independently by the bureau, and this analysis does so as well (though all standard errors reflect the correlation of observations across weeks within families). Therefore, the data reflect 44,362 weeks for married-parent families, 4,194 weeks for never-married parents, 5,960 weeks for divorced parents, 8,370 weeks for single mothers, and 1,784 weeks for single fathers.
We developed measures for categories of food and beverage expenditure guided by the U.S. Department of Agriculture Food Pyramid (www. mypyramid.gov), defining 11 mutually exclusive expenditure categories that sum to a family's total weekly food expenditure. In our analyses, expenditures serve as proxies for consumption at the household level. (4) These expenditure categories include the six groups outlined within the pyramid, as well as categories for discretionary calories. Specifically, the expenditure categories in this study are the following: (groups) grains, vegetables, fruits, milk, meat and beans, and oils; and (discretionary calories) alcoholic beverages, nonalcoholic beverages, desserts and snacks, prepared foods and condiments, and food away from home. We constructed share measures for every category by taking the amount spent on each category as a fraction of the total family food and beverage expenditure. Therefore, all 11 of the dependent variables range from 0 to 100, where the value indicates the proportion of the total amount spent on food and beverages that category represents.
The independent variables of interest are several categorizations of family structure, taking into consideration both the marital status of the head of household and the sex of the head of household. The analyses also control for several important household and socioeconomic variables. These variables include head of household characteristics, household composition, economic characteristics, geographic characteristics, and year-fixed effects.
Family structure was created using the reported marital status of the head of household. The primary family structures that are compared in the analyses are married-parent families and single-parent families. Single-parent families are a combination of divorced and never-married parents, regardless of the sex of the head of household. Single-parent families are also examined separately by marital status (divorced and never married) as well as by sex (single mother and single father).
Family Head Characteristics
We also control for characteristics of the family head that may affect expenditure decisions, allowing us to more accurately identify the specific role of family structure. The controls specific to the family head include his or her age measured continuously in years, as well as his or her race. Race of the family head is coded as white or nonwhite, with nonwhite as the reference category in the regressions. For analyses that do not specify family structure according to the sex of the family head, sex of the head is included as a covariate (male head is omitted). Further, education level of the family head was divided into four categories: less than a high school diploma, high school diploma or equivalent, some college, and college graduate or more (high school graduate is omitted). Finally, a dichotomous variable whether the family head was employed at the time of the survey was included (not employed is omitted).
Three groups of variables account for family composition. First, eight continuous variables were created representing the number and sex of own children (biological or adoptive relationship with the head of family) residing in the family. Children in various age groups and of different sexes are likely to consume goods differently; parents may also make different expenditure decisions based on the composition of children in the family. The continuous measures describe the number of boys and girls between zero and five years of age, between 5 and 10 years of age, between 10 and 15 years of age, and between 15 and 17 years of age. These age groups were designed, broadly, to capture different developmental stages, which may have an impact on family purchasing decisions and resources. Second, the number of other children living in the family is included as a continuous measure. These other children refer to brothers or sisters of the head of family, nieces or nephews, or other unrelated children. Finally, the number of other adults residing in the household is included. This measure is a continuous measure capturing the economies of scale of family consumption. These other adults can include, where applicable, spouses, adult biological and adopted children, aunts, uncles, grandparents, and other relatives.
The economic resources of the family are also controlled for in the analyses to account for differences in the quality of food and beverage expenditures. Economic resources are measured in three different ways. First, annual income (in 2000 dollars) is measured with six dichotomous variables: negative income, less than or equal to 20,000, greater than 20,000 and less than or equal to 40,000, greater than 40,000 and less than or equal to 60,000, greater than 60,000 and less than or equal to 80,000, and greater than 80,000 (greater than 40,000 and less than or equal to 60,000 omitted). Economic resources are also measured with two dichotomous variables based on the whether the family received any public assistance (did not receive public assistance is omitted) and whether the family received food stamps (did not receive food stamps is omitted).
Geographic location was divided (by the CEX) into Northeast urban, Midwest urban, South urban, West urban, and all rural areas of the United States. An additional dichotomous variable was created to account for whether the family resided in a standard metropolitan statistical area (SMSA) (did not reside in SMSA omitted). By controlling for geographic location, we are attempting to account for differences in the prices of food and beverage items faced by families in different areas of the country.
Finally, the analyses control for the year the family participated in the Diary survey by including year-dummy variables for the year in which the survey was administered. These variables control for any time trends that may be associated with different purchasing decisions.
Multivariate linear regression analyses (OLS) for each of the 11 food and beverage expenditure shares are conducted. Although the errors across equations are likely to be correlated with one another, because the regressors in each of the 11 share regressions are identical, there are no efficiency gains over OLS in running these in a seemingly unrelated regression framework. (5) In each of these analyses, the coefficient on a dummy variable for family structure in each of the 11 linear regressions is reported. These coefficients can be interpreted as the percentage point difference between the family structure variable and the omitted family structure in the budget share allocated to a particular category. (6) Standard errors are corrected to reflect the fact that the regressions use up to two observations per household (Moulton 1990). Corrections are made to the standard errors because while the weekly observations are independent across households, they are not independent within households. The regressions are also weighted using the sampling weight provided by the CEX designed to represent the population and is used for estimations of total or mean expenditures. The weights for a consumer unit can differ for each week in which they participate in the survey as the household may represent a different number of consumer units with similar characteristics.
Recall all regressions control for the independent variables, including sensitivity tests with a continuous measure of income, and all findings were robust to either specification. The explanatory variables of interest in these models are the family structure variables discussed above. The first set of regressions assesses the differences in expenditure shares between single-parent families and married-parent families. The second set of regressions illustrates the differences between never-married parents and married parents, as well as divorced parents and married parents. The final set of analyses presents the differences between single fathers and married parents as well as single mothers and married parents.
Table 1 presents weighted descriptive statistics for the sample. Statistics for all families are presented in the first column followed by those for married-parent families and single-parent families. The last two columns present statistics for the sample of single-parent families divided into never-married and divorced single parents.
Predictable demographic differences exist across the family types. Married household heads are older and more likely to be white, male, better educated, and employed, and have higher incomes than single parents (t-tests not shown). They also, not surprisingly, have higher weekly total food and beverage expenditure than single-parent families ($134 versus $90). Divorced heads, while better off financially compared to never-married heads, are worse off economically than married-parent families (t-tests and [chi square] tests not shown). Divorced and never-married heads are less likely to be employed than are married heads and are more likely to receive food stamps and public assistance ([chi square] tests not shown). Never-married heads are more likely to receive public assistance than are divorced heads ([chi square] tests not shown).
Also presented in Table 1 are the average share of total weekly expenditures on food and beverages for married parents, all single parents, never-married parents (a subset of single parents), and divorced parents (also a subset of single parents). Married-parent families spend a smaller share (t-tests not shown) of their total expenditures on food and beverages (32% of their total weekly purchases go toward food and beverages on average) than single-parent families (36%). Among single-parent families, those with a never-married head spend, on average, a greater share on food and beverages overall (38%) than do those with a divorced head, who spend an average share of 35%. These differences are not surprising given the well-known empirical relationship known as Engel's law, in which richer households tend to spend a lower share of their overall budget on food; households with a never-married head are poorer than households with a divorced head, both of which are poorer than households with a married head.
Table 2 provides weighted summary statistics on single fathers and single mothers. Single fathers are slightly older than single mothers, have slightly fewer coresident children, and have a greater number of coresident adults in their households. However, the most significant difference between single fathers and single mothers is that single fathers have a higher level of income ($31,360 versus $20,630), and single fathers spend more on food and beverages (t-tests not shown).
Table 3 presents the average expenditure share and standard deviation for each consumption category, the proportion of the sample that has zero expenditure in that category, and the average level (in 2000 dollars) of expenditures in that category for each family type (including those with zero expenditures). These statistics are presented for married-parent families and single-parent families, as well as the subsamples of single-parent families based on marital status (never married and divorced) and sex of the head (single fathers and single mothers).
Food purchased away from home is the single largest share of both married and single parents' household budgets, representing 33% of married parents' budgets and 31% of single parents' budgets. On average, married parents spend about $44 per week (in 2000 dollars) on food away from home and single parents spend $26. Meat and beans represent the largest share of the at-home food and beverage budget, corresponding to 16% of a married parent's budget (average of $22 per week) and 17% of a single parent's budget (average of $17 per week). The majority of married-parent families spent money in each of the created categories, but 51% and 69% did not purchase any oils or alcoholic beverages, respectively. Likewise, 59% and 78% of single-parent families did not purchase oils or alcohol, respectively.
Similarly, among single-parent families, single fathers spent 35% of their food and beverage budget on food away from home, while single mothers spent 30% of their budget on this category (not shown). Both single fathers and single mothers allocated the smallest share of their food and beverage budget toward oils. Single fathers spent over 13% of their total food and beverage expenditures on beverages (half on alcohol and the other half on nonalcoholic beverages) and spend about 8% on vegetables and fruits. Single mothers, in contrast, spent significantly less on alcohol (t-tests not shown) and significantly more on vegetables and fruits (t-tests not shown). Finally, whereas 20% of the sample of single mothers purchased any alcohol during the week, 35% of the sample of single fathers did so. Of the single fathers and single mothers who purchased any alcoholic beverages, single fathers on average spent 18% of their total food and beverage expenditures on alcohol and single mothers spent 14%.
Table 4 compares the food and beverage expenditure shares on the 11 expenditure categories for families with children with married heads and single heads (all families with single heads are also divided into families with never-married and divorced heads). Specifically, Table 4 presents two regressions: (1) the first column (Model 1) presents the coefficients for a regression consisting of married and single heads (married is omitted) and (2) the second and third columns (Model 2) present the coefficients of a second regression consisting of married, never-married, and divorced heads (married is omitted).
How are food and beverage expenditures allocated in single-parent families and married-parent families? Because single-parent families and married-parent families have different levels of income, levels of education, and numbers and ages of children, it is important to control for these differences when comparing how each family type allocates its food and beverage expenditure. Model 1 in Table 4 reports the coefficient on a dummy variable for whether the family is headed by a single parent from 11 separate linear regressions; recall, these coefficients are interpreted as the percentage point difference between single-parent families and married-parent families in the budget share allocated to a particular category, with shares ranging from 0% to 100%.
The findings suggest that, after controlling for the economic and demographic characteristics of the families, single-parent families devote a smaller share of their food and beverage budget to vegetables and fruits than do married-parent families. Furthermore, single-parent families spend a larger share of their food budget on alcohol and nonalcoholic beverages than do married-parent families. Specifically, single-parent families spend .44 and .34 percentage points less on vegetables and fruits, respectively, compared to married-parent families. Similarly, they spend .42 percentage points more on alcoholic beverages and .31 percentage points more on nonalcoholic beverages. These percentage point differences in vegetable and fruit purchases translate to a .08 standard deviation in vegetable expenditures (10% difference) and a .05 standard deviation in fruit expenditure (6% difference). Similarly, single-parent families spend .05 standard deviation more on alcohol, or 11% more.
In order to determine whether the overall composition of the food and beverage budget for single- and married-parent families differs (i.e., whether on average, the families' total food budget differs), a joint test was conducted on the 11 coefficients on the single-parent dummy variable from all food expenditure regressions. The joint test is a maximum-likelihood F test estimated after running the entire system of equations (all 11 regressions jointly), performed on the estimate of the covariance matrix. For our purposes, the joint test examines if all the coefficients on the family structure variable are jointly equal to zero. The hypothesis that all coefficients are zero can be rejected (F test = 6.39, p < .001). This suggests that the overall distribution of food and beverage choices differ for single-parent families compared to married-parent families.
Table 4 Model 2 (second and third columns) reports the regression coefficients on variables indicating whether the family head is never married or divorced. These coefficients reflect the percentage point difference between families with a never-married or divorced head, respectively, and married-parent families in the budget share allocated to a particular category. Both never-married and divorced single-parent families spend a smaller share of their food budgets on vegetables and fruits compared to married-parent families. As illustrated in the table, divorced and never-married families differ in the proportion of their food and beverage budget devoted to vegetables. When comparing divorced families to never-married families, the latter spent .31 percentage points less (.06 standard deviation or 7%) on vegetables compared to the former (analysis not shown). Additionally, families with a divorced head (but not families with a never-married head) also spend a greater share of their food budget on alcohol, compared to married-parent families (a difference of .05 standard deviation or 12%).
A joint test that the 11 coefficients on the never-married dummy variables (from the 11 food expenditure regressions) are zero can be rejected (F test = 5.79,p < .001). An identical test that the 11 divorced dummy variable coefficients are equal to zero can be rejected (F test = 4.11,p < .001). These tests demonstrate that the overall composition of the food budget differs for never-married families versus married families, and for divorced families versus married families.
Table 5 presents the coefficients on indicator variables for whether the family is headed by a single father or a single mother. These coefficients indicate percentage point differences in food budget shares between families headed by a single father and married-parent families and between families headed by a single mother and married-parent families, controlling for differences in economic and demographic characteristics. The findings suggest that families headed by a single father spend a smaller share of their food and beverage budgets on vegetables, fruits, meat and beans, desserts and snacks, and prepared foods compared to married-parent families. The findings also suggest that families headed by a single father, but not those headed by a single mother, spend a larger share of their food budgets on alcohol (2.62 percentage points more, which is equivalent to .30 standard deviation or 68% more than married parents) and on food away from home (1.90 percentage points more, which translates into .07 standard deviation or 6% more than married parents). Conversely, families headed by single mothers, but not those headed by single fathers, spend a larger share on grains and nonalcoholic beverages, and a lesser share on alcohol compared to married-parent families. Joint tests that the coefficients on single fathers and on single mothers are zero in each of the 11 food expenditure regression can be rejected (F test = 18.95, p < .001, and F test = 5.43, p < .001, respectively). This analysis suggests that overall expenditure patterns of both single fathers and single mothers differ from those of married families.
The last column of Table 5 presents statistics from F tests performed after each regression, testing specifically whether or not the coefficients on single father and single mother were significantly different from one another in each of the 11 share regressions. The findings illustrate that single mothers, compared to single fathers, spend a greater share of their food and beverage budget on grains, vegetables, fruit, milk, meat and beans, desserts and snacks, and prepared foods. Conversely, single mothers spend a smaller share of their budget on alcohol and food away from home compared to single fathers.
Thus far, the regression models have simply controlled for the employment status of the family head. That is, married dual-earner couples have not been distinguished from married single-earner couples. Time constraints could be a factor that contributes to the higher likelihood of single fathers' spending a greater share of his food and beverage budget on food away from home and a lesser share on vegetables, fruit, and meat and beans (although time constraints do not seem as plausible an explanation for single fathers' greater spending on alcohol). We might expect married dual-earner couples to spend more than their married single-earner counterparts on convenience foods, such as prepared food and food away from home, and less on fresh foods that need to be cooked, as well. In a series of additional analyses (presented in Table 6), we find some support for this conjecture. The regression coefficients presented are relative to the omitted category of single-earner married-parent families where fathers, but not mothers, are employed outside the home.
Married dual-earner families, compared to married families where only the father is employed, spend a smaller proportion of their total food and beverage expenditures on grains, vegetables, fruits, milk, meat and beans, oils, and desserts and snacks, and a larger proportion on alcoholic beverages and food away from home. Single fathers who are employed display a very similar pattern, differing significantly on these same dimensions from married families in which only fathers are employed, while also spending a smaller fraction on prepared foods. The patterns for married dual-earner families are also mirrored in most categories by single mothers who are employed.
We hypothesize that married dual-earner families should be under similar time constraints as employed single fathers and mothers: in all cases, all parents work outside the home. Thus, in additional analyses (not presented in Table 6), we test whether the coefficients on married-parent dual-earner, employed single-father, and employed single-mother families differ from one another. Findings suggest that married-parent dual earners do not significantly differ from employed single mothers in the share of food and beverage dollars allocated to vegetables and fruits. However, employed single fathers do significantly differ from both dual-earner and employed single-mother families by allocating a smaller share of their food and beverage dollars toward vegetables and fruits.
Further, married dual-earner families, employed single-father families, and employed single-mother families also differ significantly from married single-earner families in expenditures on food away from home. Married dual-earner families spend 4.18 percentage points more (. 16 standard deviation or 15% more), employed single-father families spend 5.30 percentage points more (.20 standard deviation or 19% more), and employed single-mother families spend 2.82 percentage points more (. 11 standard deviation or 10% more) than do married-parent families where only the father is employed. In postregression estimations (not shown), we find that married dual-earner families and employed single-father families do not differ in the share of food and beverage dollars spent on food away from home, but both spend larger shares compared to employed single mothers.
In contrast, single mothers who are not employed spend a significantly smaller share on food purchased away from home compared to married-parent families in which only the father is employed (which explains why the coefficient from the regressions that combined the two groups of single mothers was not different from zero in this expenditure category). Further, married parents where neither is employed also spend a smaller share on food away from home compared to married-parent families in which only the father is employed. Thus, the time constraint hypothesis seems especially apt for the case of expenditure on food away from home for all employed families compared to their counterparts with fewer employed individuals, irrespective of family structure.
This paper finds evidence that family structure and parental employment status, in combination, are associated with food and beverage expenditure patterns across families. Our results suggest, first, that families headed by a single parent allocate their food budgets differently than do married-parent families. Given prior findings that children in single-parent households do not have as good health outcomes as children in married-parent households, the impetus for this study was to assess whether nutritional priorities and eating habits--as reflected in food and beverage purchasing decisions--exist between these different types of families. These findings might help to explain some of these differences in children's health and development across family structure. Although we cannot, based on our data analysis, link either nutrition or the allocation of food expenditure to child development or health, we have shown that the allocation of family expenditures on different food items does differ depending on marital status and sex of the family head. Our measure of the allocation of food is notable in that we measure it as a proportion of total food expenditures rather than as a proportion of total family income. Thus, we compare families in terms of how they allocate their food budgets regardless of income. We cannot determine, with cross-sectional data, the extent to which individuals with different nutritional priorities and eating habits or tastes may sort into different family structures. That is, one should not interpret the findings as indicating a causal effect of family structure itself (e.g., in terms of the time constraints it imposes on parents) on the allocation of food budgets. Rather, these findings provide a descriptive analysis of differences seen in samples of families with varying structures and provide an important first step in examining nutrition differences across families, starting with what children have access to, specifically the food that is purchased.
Several findings from the multivariate regression analysis are noteworthy. First, compared to married parents, we find that families headed by single parents allocate a smaller proportion of their food budgets to vegetables and fruits. This association is seen for single-parent families, whether divorced or never married, and is also true for single-father and single-mother families. Children's consumption of fruits and vegetables is extremely important given the negative association between eating them and dietary fat intake (Birch and Fisher 1998). According to data from the Canadian Community Health Survey (Shields 2004), children and adolescents who reported eating fruits and vegetables five or more times a day were less likely to be overweight or obese than those who consumed them less frequently.
Because we control for income and geographic location, this effect is not likely due to the expense of produce or to access to quality grocers, recognizing that some other omitted variable effect might be present. Therefore, these choices might reflect nutritional priorities or eating habits that differ by family structure. They might also reflect differences in access to information about the importance of fruits and vegetables for a nutritious diet. It is also possible that these differences reflect a time or energy constraint in single-parent families to the extent that instilling habits of consuming fruits and vegetables (perhaps especially among children) is something that two parents can more easily do than one. It is also possible that children in single-parent families have more responsibility for preparing their own meals and are perhaps less likely to choose fruits and vegetables in doing so.
Second, we find that families headed by single fathers spend a greater share of their food budgets on alcohol compared to married-parent families and also single-mother families. Unfortunately, we cannot determine whether this relatively larger alcohol share reflects a higher quantity versus quality of alcohol. While there is research suggesting that alcohol can be good for health when drunk in moderation (Corrao et al. 2000), it is possible that the choices that single fathers make in how to allocate their food budgets in terms of alcohol expenditure may not be ideal for parent or child well-being in the home.
Finally, and perhaps most interestingly, we find important differences based on the employment status of parents in the family. These findings provide further insight into the differences between single- and married-parent families. Families where all parents are employed, irrespective of family structure, spend a greater share of their food budgets on food purchased away from home and a lesser share on vegetables, fruits, milk, and meat and beans compared with married-couple families in which the mother is not employed. All these differences could reflect time constraints faced by employed parents. When all parents in the family are employed, there may be a real or perceived lack of time for preparing balanced meals made from scratch; again, there may be less time or energy for encouraging children to do so. It is also possible that children in families where all parents are employed take on more responsibility for meal preparation and choose easier options less likely to include meals from scratch similar to findings in the single-parent family analysis presented above.
Food away from home in this analysis may include fast food, restaurants, and vending machine purchases. This finding raises concerns because restaurant portion sizes have increased, with the greatest increases in food consumed at fast food establishments (Nielsen and Popkin 2003) and often exceed Food and Drug Administration and U.S. Department of Agriculture standard portion sizes (Young and Nestle 2002). Further, food consumed outside the home has higher fat density and lower fiber and calcium density compared to foods prepared in the home (Lin, Guthrie, and Blaylock 1996). Prior research on nutrient intake suggests that children in father-headed single-parent families have higher-than-recommended intakes of saturated fat, sodium, and iron and that these intakes exceed those of children in single-mother and two-parent households (Lin, Guthrie, and Blaylock 1996). Our results suggest that these prior findings may be partially a function of a greater likelihood of eating food away from home among the children of single fathers. Moreover, recent work has suggested a link between maternal employment and child obesity that may be due to time constraints faced by employed mothers (Anderson, Butcher, and Levine 2003); our work suggests that this too could be due to a greater consumption of food away from home, coupled with a decrease in the intake of vegetables and fruits.
The results from our analysis are important in the current policy environment, particularly as current public policies increasingly focus on children's health and well-being. The differences in the allocation of food expenditures suggest that employed-parent families, perhaps because of the time constraints they face, or the role their children have in choosing their own foods, spend smaller shares of their food and beverage dollars on fresh food and more on foods purchased away from home. These choices come at a nutritional cost and may be a contributing factor to the rapidly rising rates of overweight and obesity in children. More research needs to be conducted in order to understand these linkages so that families and policy makers can best address the nutritional needs of children and families.
Abdel-Ghany, Mohamed and F.N. Schwenk. 1993. Differences in Consumption Patterns of Single Parent and Two-Parent Families in the United States. Journal of Family and Economic Issues, 14 (1): 299-315.
Anderson, Patricia M., Kristin F. Butcher, and Phillip B. Levine. 2003. Maternal Employment, Child Care and Childhood Obesity. Journal of Health Economics, 22 (3): 477-505.
Birch, Leann L. and Jennifer O. Fisher. 1998. Development of Eating Behaviors among Children and Adolescents. Pediatrics, 101 (3): 539-549.
Boyle, Maureen. 1989. Spending Patterns and Income of Single and Married Parents. Monthly Labor Review, 112 (3): 37-41.
Contento, Isobel R., Charles Basch, Steven Shea, Bernard Gutin, Patricia Zybert, John L. Michela, and Jill Rips. 1993. Relationship of Mothers' Food Choice Criteria to Food Intake of Preschool Children: Identification of Family Subgroups. Health Education Quarterly, 20 (2): 243-259.
Corrao, Giovanni, Luca Rubbiati, Vincenzo Bagnardi, Antonella Zambon, and Kari Poikolainen. 2000. Alcohol and Coronary Heart Disease: A Meta-Analysis. Addiction, 95 (10): 1505-1523.
Dawson, Deborah A. 1991. Family Structure and Children's Health and Well-Being: Data from the 1988 National Health Interview Survey on Child Health. Journal of Marriage and the Family, 53 (3): 573-584.
Horton, Sally E. and Jeanne L. Hafstrom. 1985. Income Elasticities for Selected Consumption Categories: Comparison of Single Female-Headed and Two-Parent Families. Home Economics Research Journal, 13 (March): 292-303.
Lin, Biing-Hwan, Joanne Guthrie, and James R. Blaylock. 1996. The Diets of America's Children: Influence of Dining Out, Household Characteristics, and Nutrition Knowledge. Agricultural Economic Report No. 746, U.S. Department of Agriculture, Economic Research Service, Washington, DC.
Lino, Mark. 1990. Factors Affecting Expenditures of Single Parent Households. Home Economics Research Journal, 18 (March): 191-201.
--. 1991. Expenditures on a Child by Single Parent Families. Family Economics Review, 4 (1): 2-7.
Lino, Mark, Peter P. Basiotis, Shirley A. Gerrior, and Andrea Carlson. 2002. The Quality of Young Children's Diets. Family Economics and Nutrition Review, 14 (1): 52-60.
Lipman, Ellen L. and David R. Offord. 1997. Psychosocial Morbidity among Poor Children in Ontario. In Consequences of Growing Up Poor, edited by Greg J. Duncan and Jeanne Brooks-Gunn (239-287). New York: Russell Sage.
Lundberg, Shelly and Robert A. Pollak. 1996. Bargaining and Distribution in Marriage. Journal of Economic Perspectives, 10 (4): 139-158.
Lundberg, Shelly J., Robert A. Pollak, and Terence J. Wales. 1997. Do Husbands and Wives Pool Resources?: Evidence from the UK Child Benefit. Journal of Human Resources, 32 (3): 463-480.
Lundberg, Shelly, Richard Startz, and Steven Stillman. 2003. The Retirement Consumption Puzzle: A Marital Bargaining Approach. Journal of Public Economics, 87 (5-6): 1199-1218.
McLanahan, Sara and Gary Sandefur, eds. 1994. Growing Up with a Single Parent: What Hurts, What Helps. Cambridge, MA: Harvard University Press.
McLoyd, Vonnie C. 1990. The Impact of Economic Hardship on Black Families and Children: Psychological Distress, Parenting, and Socioemotional Development. Child Development, 61 (2): 311-346.
Moulton, Brent R. 1990. An Illustration of a Pitfall in Estimating the Effects of Aggregate Variables on Micro Units. The Review of Economics and Statistics, 72 (2): 334-338.
National Center for Health Statistics. 1998. Health, United States, 1998 with Socioeconomic Status and Health Chartbook. Hyattsville, MD: National Center for Health Statistics.
--. 2002. Health, United States, 2002 with Chartbook on Trends in Health of Americans. Hyattsville, MD: National Center for Health Statistics.
Nielsen, Samara Joy and Barry M. Popkin. 2003. Patterns and Trends in Food Portion Sizes, 1977-1998. Journal of the American Medical Association, 289 (4): 450-453.
Oliveria, Susan A., R. Curtis Ellison, Lynn L. Moore, Matthew W. Gillman, Ellen J. Garrahie, and Martha R. Singer. 1992. Parent-Child Relationships in Nutrient Intake: The Framingham Children's Study. American Journal of Clinical Nutrition, 56 (September): 593-598.
Paulin, Geoffrey D. and Yoon G. Lee. 2002. Expenditures of Single Parents: How Does Gender Figure In? Monthly Labor Review, 125 (7): 16-37.
Phipps, Shelley A. and Peter S. Burton. 1998. What's Mine Is Yours? The Influence of Male and Female Incomes on Patterns of Household Expenditure. Economica, 65 (260): 599-613.
Ramezani, Cyrus A. and Claudia Roeder. 1995. Health Knowledge and Nutritional Adequacy of Female Heads of Households in the United States. Journal of Consumer Affairs, 29 (2): 381-402.
Sandberg, John F. and Sandra L. Hofferth. 2001. Changes in Children's Time with Parents: United States, 1981-1997. Demography, 38 (3): 423-436.
Shields, Margot. 2004. Measured Obesity: Overweight Canadian Children and Adolescents. http:// www.statcan.ca/english/research/82-620-MIE/2005001/pdf/cobesity.pdf.
Sobal, Jeffrey and Albert Stunkard. 1989. Socioeconomic Status and Obesity: A Review of the Literature. Psychological Bulletin, 105 (2): 260-275.
Thomas, Duncan. 1990. Intra-Household Resource Allocation: An Inferential Approach. Journal of Human Resources, 25 (4): 635-664.
--. 1994. Like Father, Like Son; Like Mother, Like Daughter: Parental Resources and Child Height. Journal of Human Resources, 29 (4): 950-988.
Troiano, Richard P. and Katherine M. Flegal. 1998. Overweight Children and Adolescents: Description, Epidemiology, and Demographics. Pediatrics, 101 (3): 497-504.
U.S. Department of Agriculture (USDA). 2003. Backgrounder for Revision of Food Guide Pyramid. http://www.usda.gov/news/releases/2003/09/bg0308.htm.
U.S. Department of Health and Human Services. 2000. Healthy People 2010: Understanding and Improving Health. 2nd edition. Washington, DC: U.S. Government Printing Office.
Variyam, Jayachandran N., James R. Blaylock, David Smallwood, and P. Peter Basiotis. 1998. USDA's Healthy Eating Index and Nutrition Information. Technical Bulletin No.: 1866, Economic Research Service and Center for Nutrition Policy and Promotion, Washington, DC.
Whitaker, Robert C., Jeffrey A. Wright, Margaret S. Pepe, Kristy D. Seidel, and William H. Dietz. 1997. Predicting Obesity in Young Adulthood from Childhood and Parental Obesity. New England Journal of Medicine, 337 (13): 869-873.
White House. 2002. Healthier US: The President's Health and Fitness Initiative. http://www. whitehouse.gov/infocus/fitness/execsummary.html.
Young, Lisa R. and Marion Nestle. 2002. The Contribution of Expanding Portion Sizes to the US Obesity Epidemic. American Journal of Public Health, 92 (2): 246-249.
Zick, Cathleen D. and Jane McCullough. 1996. Trade-offs between Purchased Services and Time in Single Parent and Two-Parent Families. Journal of Consumer Affairs, 30 (1): 1-23.
(1.) In 2002, U.S. Senators Bill Frist, Jeff Bingaman, and Christopher Dodd introduced bipartisan, comprehensive legislation aimed at reducing obesity, particularly among children and adolescents. The Improved Nutrition and Physical Activity Act, or IMPACT, stressed the importance of good nutrition and regular exercise. In June 2002, President Bush introduced his HealthierUS initiative, which calls for regular exercise, portion control of nutritious foods, and preventive health screenings (White House 2002).
(2.) A consumer unit is either all members of a particular household who are related by blood, marriage, adoption, or other legal arrangements; a person living alone or sharing a household with others, as a roomer in a private home or lodging house, or in permanent living quarters in a hotel or motel, but who is financially independent; or two or more persons living together who use their incomes to make joint expenditure decisions. The Bureau of Labor Statistics uses the terms consumer unit, family, and household interchangeably for convenience. However, the proper technical term for purposes of the CEX is consumer unit.
(3.) All analyses presented were also run including separated and widowed families, and all findings are robust to inclusion.
(4.) Expenditures as proxies for consumption are imperfect as they are measured at the household level, and we have no knowledge of who is consuming the goods purchased. All we can say, then, is that these food and beverage expenditures are available for consumption of those who reside in the home.
(5.) Because the sum of the shares adds up to 100, the unconstrained OLS estimation provides predicted shares that add up to 100 across all 11 equations, as well as the sum of the coefficients on individual variables summing to 0. Estimating the equations separately does not lead to bias in the coefficients or the standard errors, and running them as a system leads to identical coefficients and standard errors. This joint equation estimation procedure, however, is used when testing the joint significance of family structure variables across all 11 equations.
(6.) Coefficients on other variables in the estimation are available upon request.
Kathleen M. Ziol-Guest is a Robert Wood Johnson Foundation Health and Society Scholar in the Department of Society, Human Development, and Health at the Harvard School of Public Health, Harvard University, Boston, MA (email@example.com). Thomas DeLeire is a senior analyst at the Congressional Budget Office, United States Congress, Washington, DC (firstname.lastname@example.org). Ariel Kalil is an associate professor at the Harris Graduate School of Public Policy Studies, University of Chicago, Chicago, IL (email@example.com).
This research was partially supported by a grant from the National Institute of Child Health and Human Development through the Population Research Center at the National Opinion Research Center, University of Chicago. The authors thank Association for Public Policy Analysis and Management conference participants for their helpful suggestions.
DeLeire is on leave from the Department of Economics at Michigan State University.
The views expressed in this paper are those of the authors and should not be interpreted as those of the Congressional Budget Office.
TABLE 1 Weighted Summary Statistics for Different Family Structures All Families Married Variable Mean SD Mean SD Family head characteristics Age 37.69 8.33 38.24 8.15 White 0.81 -- 0.85 -- Male 0.62 -- 0.73 -- Less than high school 0.14 -- 0.12 -- High school diploma 0.31 -- 0.29 -- Some college 0.27 -- 0.27 -- College graduate or more 0.28 -- 0.31 -- Employed 0.89 -- 0.90 -- Family composition Number of children Number of boys <5 0.32 0.56 0.33 0.57 Number of girls <5 0.30 0.55 0.32 0.56 Number of boys >5 0.28 0.53 0.29 0.53 and [less than or equal to] 10 Number of girls >5 0.26 0.51 0.27 0.52 and [less than or equal to] 10 Number of boys >10 0.27 0.52 0.28 0.52 and [less than or equal to] 15 Number of girls >10 0.26 0.50 0.26 0.51 and <15 Number of boys >15 0.10 0.31 0.10 0.31 and [less than or equal to] 17 Number of girls > 15 0.09 0.30 0.09 0.30 and < 17 Number of other children 0.05 0.30 0.03 0.24 Number of other adults 1.06 0.67 1.22 0.57 Economic characteristics Income (year 2000 dollars) 40.15 41.19 44.32 42.85 <0 0.01 -- 0.02 -- [greater than or equal to] 0.34 -- 0.28 -- 0 and < 20,000 [greater than or equal to] 0.23 -- 0.22 -- 20,000 and <40,000 [greater than or equal to] 0.19 -- 0.21 -- 40,000 and <60,000 [greater than or equal to] 0.11 -- 0.13 -- 60,000 and <80,000 [greater than or equal to] 0.12 -- 0.14 -- 80,000 Receives cash assistance 0.05 -- 0.02 -- Receives food stamps 0.12 -- 0.08 -- Geographic characteristics Region of the country Northeast urban 0.17 -- 0.17 -- Midwest urban 0.20 -- 0.20 -- South urban 0.29 -- 0.28 -- West urban 0.20 -- 0.21 -- All rural 0.14 -- 0.14 -- SMSA 0.79 -- 0.79 -- Weekly expenditure characteristics Food and beverage 126.10 93.50 134.62 95.07 (year 2000 dollars) Per capita food and beverage 32.89 24.55 33.49 24.14 All expenses 0.33 0.24 0.32 0.24 (year 2000 dollars) Number of household weeks 54,516 44,362 Single Never Married Variable Mean SD Mean SD Family head characteristics Age 35.33 8.64 30.44 7.87 White 0.64 -- 0.47 -- Male 0.17 -- 0.14 -- Less than high school 0.20 -- 0.27 -- High school diploma 0.37 -- 0.39 -- Some college 0.30 -- 0.27 -- College graduate or more 0.14 -- 0.07 -- Employed 0.82 -- 0.75 -- Family composition Number of children Number of boys <5 0.25 0.52 0.41 0.62 Number of girls <5 0.24 0.50 0.39 0.60 Number of boys >5 0.25 0.50 0.24 0.48 and [less than or equal to] 10 Number of girls >5 0.23 0.48 0.23 0.48 and [less than or equal to] 10 Number of boys >10 0.26 0.51 0.18 0.46 and [less than or equal to] 15 Number of girls >10 0.25 0.49 0.17 0.43 and <15 Number of boys >15 0.11 0.31 0.05 0.23 and [less than or equal to] 17 Number of girls > 15 0.10 0.31 0.05 0.23 and < 17 Number of other children 0.10 0.45 0.13 0.52 Number of other adults 0.42 0.68 0.47 0.73 Economic characteristics Income (year 2000 dollars) 22.42 26.81 16.94 19.82 <0 0.01 -- 0.00 -- [greater than or equal to] 0.56 -- 0.69 -- 0 and < 20,000 [greater than or equal to] 0.28 -- 0.24 -- 20,000 and <40,000 [greater than or equal to] 0.10 -- 0.05 -- 40,000 and <60,000 [greater than or equal to] 0.03 -- 0.01 -- 60,000 and <80,000 [greater than or equal to] 0.02 -- 0.01 -- 80,000 Receives cash assistance 0.16 -- 0.25 -- Receives food stamps 0.30 -- 0.44 -- Geographic characteristics Region of the country Northeast urban 0.16 -- 0.18 -- Midwest urban 0.22 -- 0.23 -- South urban 0.32 -- 0.33 -- West urban 0.19 -- 0.17 -- All rural 0.11 -- 0.09 -- SMSA 0.80 -- 0.85 -- Weekly expenditure characteristics Food and beverage 89.88 76.56 81.26 71.59 (year 2000 dollars) Per capita food and beverage 30.33 26.10 26.22 23.24 All expenses 0.36 0.27 0.38 0.29 (year 2000 dollars) Number of household weeks 10,154 4,194 Divorced Variable Mean SD Family head characteristics Age 38.82 7.38 White 0.76 -- Male 0.19 -- Less than high school 0.14 -- High school diploma 0.35 -- Some college 0.32 -- College graduate or more 0.19 -- Employed 0.87 -- Family composition Number of children Number of boys <5 0.14 0.39 Number of girls <5 0.13 0.38 Number of boys >5 0.26 0.52 and [less than or equal to] 10 Number of girls >5 0.23 0.49 and [less than or equal to] 10 Number of boys >10 0.31 0.53 and [less than or equal to] 15 Number of girls >10 0.31 0.52 and <15 Number of boys >15 0.14 0.36 and [less than or equal to] 17 Number of girls > 15 0.13 0.35 and < 17 Number of other children 0.08 0.39 Number of other adults 0.37 0.65 Economic characteristics Income (year 2000 dollars) 26.32 30.24 <0 0.01 -- [greater than or equal to] 0.47 -- 0 and < 20,000 [greater than or equal to] 0.31 -- 20,000 and <40,000 [greater than or equal to] 0.14 -- 40,000 and <60,000 [greater than or equal to] 0.04 -- 60,000 and <80,000 [greater than or equal to] 0.03 -- 80,000 Receives cash assistance 0.10 -- Receives food stamps 0.21 -- Geographic characteristics Region of the country Northeast urban 0.14 -- Midwest urban 0.21 -- South urban 0.32 -- West urban 0.21 -- All rural 0.12 -- SMSA 0.77 -- Weekly expenditure characteristics Food and beverage 96.02 79.35 (year 2000 dollars) Per capita food and beverage 33.25 27.60 All expenses 0.35 0.26 (year 2000 dollars) Number of household weeks 5,960 Note: Source is CEX, Diary Component, 1990-2003. TABLE 2 Weighted Summary Statistics for Single Fathers and Single Mothers Single Father Variable Mean SD Family head characteristics Age 37.16 9.07 White 0.79 -- Male 100 -- Less than high school 0.19 -- High school diploma 0.38 -- Some college 0.28 -- College graduate or more 0.15 -- Employed 0.94 -- Family composition Number of children Number of boys [less than or equal to] 5 0.24 0.51 Number of girls [less than or equal to] 5 0.23 0.46 Number of boys >5 and [less than 0.20 0.47 or equal to] 10 Number of girls >5 and [less than 0.16 0.43 or equal to] 10 Number of boys >I 0 and [less than 0.27 0.50 or equal to] 15 Number of girls >10 and [less than 0.20 0.43 or equal to] 15 Number of boys > 15 and [less than 0.12 0.34 or equal to] 17 Number of girls >15 and [less than 0.08 0.28 or equal to] 17 Number of other children 0.13 0.46 Number of other adults 0.64 0.73 Economic characteristics Income (year 2000 dollars) 31.36 34.42 <0 0.01 -- [greater than or equal to] 0 and <20,000 0.38 -- [greater than or equal to] 20,000 and <40,000 0.34 -- [greater than or equal to] 40,000 and <60,000 0.17 -- [greater than or equal to] 60,000 and <80,000 0.06 -- [greater than or equal to] 80,000 0.05 -- Receives cash assistance 0.06 -- Receives food stamps 0.15 -- Geographic characteristics Region of the country Northeast urban 0.11 -- Midwest urban 0.22 -- South urban 0.26 -- West urban 0.26 -- All rural 0.15 -- SMSA 0.76 -- Expenditure characteristics Food and beverage (year 2000 dollars) 107.61 88.45 Per capita food and beverage 38.81 30.33 All expenses (year 2000 dollars) 0.36 0.26 Number of household weeks 1,784 Single Mother Variable Mean SD Family head characteristics Age 34.97 8.5 White 0.61 -- Male 0.00 -- Less than high school 0.20 -- High school diploma 0.36 -- Some college 0.30 -- College graduate or more 0.14 -- Employed 0.79 -- Family composition Number of children Number of boys [less than or equal to] 5 0.25 0.52 Number of girls [less than or equal to] 5 0.24 0.51 Number of boys >5 and [less than 0.26 0.51 or equal to] 10 Number of girls >5 and [less than 0.24 0.49 or equal to] 10 Number of boys >I 0 and [less than 0.25 0.51 or equal to] 15 Number of girls >10 and [less than 0.26 0.50 or equal to] 15 Number of boys > 15 and [less than 0.10 0.31 or equal to] 17 Number of girls >15 and [less than 0.10 0.31 or equal to] 17 Number of other children 0.10 0.45 Number of other adults 0.37 0.67 Economic characteristics Income (year 2000 dollars) 20.63 24.63 <0 0.01 -- [greater than or equal to] 0 and <20,000 0.59 -- [greater than or equal to] 20,000 and <40,000 0.27 -- [greater than or equal to] 40,000 and <60,000 0.09 -- [greater than or equal to] 60,000 and <80,000 0.02 -- [greater than or equal to] 80,000 0.02 -- Receives cash assistance 0.18 -- Receives food stamps 0.33 -- Geographic characteristics Region of the country Northeast urban 0.17 -- Midwest urban 0.22 -- South urban 0.34 -- West urban 0.18 -- All rural 0.09 -- SMSA 0.81 -- Expenditure characteristics Food and beverage (year 2000 dollars) 86.34 73.46 Per capita food and beverage 29.24 25.03 All expenses (year 2000 dollars) 0.36 0.28 Number of household weeks 8,370 Note: Source is CEX, Diary Component, 1990-2003. TABLE 3 Weighted Descriptive Statistics of Food Category Shares and Levels for Different Family Structures Married Single Average Average Level Level Variable Mean SD p ($) Mean SD p ($) Grains 7.62 6.99 0.13 9.72 8.02 8.73 0.20 6.72 Vegetables 4.66 5.34 0.24 6.25 4.37 5.67 0.35 4.25 Fruits 5.31 6.20 0.24 7.07 4.95 6.70 0.35 4.53 Milk 8.41 9.00 0.11 9.54 8.85 11.05 0.19 6.43 Meat and 15.85 14.11 0.18 22.26 17.41 16.49 0.25 17.48 Beans Oils 1.75 2.86 0.51 2.47 1.80 3.34 0.59 1.87 Alcoholic 3.86 8.83 0.69 6.15 3.48 9.65 0.78 3.79 beverages Nonalcoholic 5.81 7.01 0.24 7.43 6.60 9.77 0.31 5.39 beverages Desserts and 6.22 7.05 0.22 8.45 6.18 8.15 0.32 5.60 snacks Prepared 7.74 7.62 0.22 10.91 7.53 8.47 0.31 7.42 food and condiments Food away 32.76 26.88 0.16 44.37 30.80 30.14 0.25 26.38 Number of 44,362 10,154 weeks Never Married Divorced Average Average Level Level Variable Mean SD p ($) Mean SD p ($) Grains 8.37 9.07 0.21 6.33 7.78 8.47 0.20 7.01 Vegetables 4.36 5.92 0.39 3.93 4.38 5.49 0.33 4.47 Fruits 5.19 7.24 0.37 4.27 4.78 6.28 0.33 4.72 Milk 9.27 12.02 0.22 6.06 8.56 10.30 0.17 6.70 Meat and 19.21 17.80 0.26 18.32 16.12 15.36 0.24 16.89 Beans Oils 1.80 3.42 0.61 1.78 1.79 3.29 0.58 1.93 Alcoholic 3.27 9.57 0.80 3.25 3.63 9.71 0.76 4.17 beverages Nonalcoholic 6.67 10.42 0.33 4.84 6.55 9.28 0.30 5.79 beverages Desserts and 3.05 8.49 0.34 5.01 6.27 7.91 0.30 6.03 snacks Prepared 7.20 8.68 0.34 6.62 7.77 8.31 0.28 7.99 food and condiments Food away 28.61 30.83 0.29 20.86 32.37 29.55 0.22 30.31 Number of 4,194 5,960 weeks Single Father Single Mother Average Average Level Level Variable Mean SD p ($) Mean SD p ($) Grains 7.01 8.01 0.22 6.94 8.22 8.85 0.20 6.68 Vegetables 3.92 4.77 0.36 4.40 4.46 5.83 0.35 4.22 Fruits 4.17 5.34 0.36 4.72 5.11 6.93 0.34 4.50 Milk 7.96 9.21 0.18 7.04 9.03 11.38 0.19 6.31 Meat and 15.09 14.53 0.25 18.12 17.87 16.81 0.25 17.36 Beans Oils 1.65 2.78 0.59 2.03 1.82 3.44 0.59 1.83 Alcoholic 6.43 13.03 0.65 7.74 2.89 8.70 0.80 3.00 beverages Nonalcoholic 6.58 10.10 0.30 6.14 6.60 9.70 0.31 5.25 beverages Desserts and 5.11 6.92 0.33 5.48 6.39 8.36 0.32 5.63 snacks Prepared 7.07 8.09 0.31 8.06 7.62 8.55 0.31 7.29 food and condiments Food away 35.00 30.28 0.19 36.95 29.97 30.05 0.26 24.27 Number of 1,784 1,784 weeks Note: Source is CEX, Diary Component, 1990-2003. p is the proportion of the sample with zero expenditure in the category. Average level is the average dollar amount, in year 2000 dollars, of expenditure in the category including those with zero expenditures. TABLE 4 Results of Regression Analysis for the Total Sample: Comparisons to Married Parents (n = 44,362 weeks) Model 1 All Single Heads Variable B SE [R.sup.2] Grains 0.16 0.14 0.03 Vegetables -0.44 *** 0.10 0.02 Fruits -0.34 ** 0.11 0.04 Milk -0.04 0.18 0.04 Meat and beans -0.23 0.27 0.06 Oils 0.01 0.05 0.02 Alcoholic beverages 0.42 * 0.19 0.02 Nonalcoholic 0.31 * 0.16 0.02 beverages Desserts and snacks -0.18 0.13 0.01 Prepared food -0.15 0.14 0.01 and condiments Food away 0.46 0.52 0.06 Number of weeks 10,154 Model 2 Never Married Divorced Variable B SE B SE F Test Grains 0.27 0.20 0.10 0.16 0.56 Vegetables -0.63 *** 0.14 -0.32 ** 0.14 4.34 Fruits -0.31 * 0.16 -0.35 ** 0.12 8.78 Milk -0.01 0.27 -0.06 0.20 0.02 Meat and beans 0.18 0.38 -0.48 0.31 2.32 Oils 0.02 0.08 0.01 0.06 0.00 Alcoholic beverages 0.35 0.25 0.47 * 0.21 0.20 Nonalcoholic 0.32 0.23 0.30 0.18 0.00 beverages Desserts and snacks -0.13 0.18 -0.21 0.15 0.16 Prepared food -0.24 0.20 -0.10 0.16 0.44 and condiments Food away 0.20 0.72 0.63 0.60 0.29 Number of weeks 4,194 5,960 Note: F test is a postregression estimation for the equality of the coefficients on never married and divorced in Model 2. * p < .05; ** p < .01; *** p < .001. TABLE 5 Results of Regression Analysis for Sex of Single Parent: Comparison to Married Parents (n = 44,362 weeks) Single Father Single Mother Variable B SE B SE F Test Grains -0.30 0.26 0.41 ** 0.15 6.87 ** Vegetables -0.68 *** 0.14 -0.23 * 0.11 8.18 ** Fruits -0.84 *** 0.16 -0.04 0.12 20.78 *** Milk -0.34 0.29 0.32 0.19 4.73 * Meat and beans -1.06 * 0.43 0.20 0.28 7.29 ** Oils -0.11 0.09 0.06 0.06 3.30 Alcoholic beverages 2.62 *** 0.42 -0.51 ** 0.17 53.03 *** Nonalcoholic 0.41 0.30 0.33 * 0.16 0.05 beverages Desserts and -0.98 *** 0.20 0.25 0.14 30.55 *** snacks Prepared food and -0.61 ** 0.23 0.16 0.15 9.73 ** condiments Food away 1.90 * 0.95 -0.96 0.54 7.95 ** Number of weeks 1,784 8,370 Note: F test is a postregression estimation for the equality of the coefficients on single father and single mother. * p < .05; ** p < .01; *** p < .001. TABLE 6 Results of Regression Analysis for Working Status of Parents: Comparisons to Married, Only Father Employed Families (n = 10,702 weeks) Married Married Mother Dual Earner Only Employed Variable B SE B SE Grains -0.45 *** 0.10 0.04 0.29 Vegetables -0.63 *** 0.08 -0.17 0.26 Fruits -0.68 *** 0.09 -0.64 ** 0.21 Milk -0.84 *** 0.14 0.30 0.42 Meat and beans -1.18 *** 0.20 0.74 0.53 Oils -0.11 ** 0.04 0.02 0.09 Alcoholic beverages 0.35 ** 0.13 0.15 0.34 Nonalcoholic beverages -0.17 0.10 0.17 0.27 Desserts and snacks -0.34 ** 0.10 -0.05 0.27 Prepared food and -0.14 0.11 0.26 0.28 condiments Food away 4.18 *** 0.38 0.65 1.04 Number of weeks 31,556 1,479 Married Parents Single Father Neither Employed Employed Variable B SE B SE Grains -0.03 0.42 -0.75 ** 0.26 Vegetables 0.01 0.32 -1.18 *** 0.15 Fruits 0.15 0.45 -1.36 *** 0.18 Milk 1.09 0.78 -1.07 *** 0.30 Meat and beans 2.10 * 0.95 -2.16 *** 0.46 Oils 0.15 0.20 -O.18 * 0.09 Alcoholic beverages -0.31 0.46 3.02 *** 0.45 Nonalcoholic beverages 1.00 * 0.47 0.34 0.32 Desserts and snacks 0.52 0.41 -1.28 *** 0.23 Prepared food and 0.40 0.45 -0.69 ** 0.25 condiments Food away -5.07 *** 1.33 5.30 *** 1.02 Number of weeks 625 1,674 Single Father Single Mother Not Employed Employed Variable B SE B SE Grains 1.88 1.62 -0.09 0.17 Vegetables -0.25 0.78 -0.68 *** 0.12 Fruits -1.14 0.68 -0.58 *** 0.14 Milk 2.68 1.91 -0.53 * 0.22 Meat and beans 3.23 2.25 -0.86 * 0.34 Oils -0.17 0.32 -0.02 0.06 Alcoholic beverages 0.26 1.00 -0.22 0.20 Nonalcoholic beverages 0.24 0.87 0.17 0.18 Desserts and snacks 0.12 0.69 -0.10 0.16 Prepared food and -0.73 0.90 0.08 0.17 condiments Food away -6.11 3.40 2.82 *** 0.64 Number of weeks 110 6,734 Single Mother Not Employed Variable B SE Grains 1.28 *** 0.35 Vegetables -0.57 * 0.29 Fruits -0.27 0.31 Milk 1.83 ** 0.53 Meat and beans 1.18 0.66 Oils 0.03 0.12 Alcoholic beverages -0.64 0.34 Nonalcoholic beverages 0.84 * 0.38 Desserts and snacks 0.84 * 0.36 Prepared food and 0.21 0.35 condiments Food away -4.73 *** 1.04 Number of weeks 1,636 * p < .05: ** p < .01: *** p < .001.
|Printer friendly Cite/link Email Feedback|
|Author:||Ziol-Guest, Kathleen M.; DeLeire, Thomas; Kalil, Ariel|
|Publication:||Journal of Consumer Affairs|
|Date:||Dec 22, 2006|
|Previous Article:||Gender differences in debt repayment problems after divorce.|
|Next Article:||The effect of demographic, economic, and nutrition factors on the frequency of food away from home.|