at SciVerse ScienceDirect

Social Science & Medicine 74 (2012) 1754e1764

Contents lists available

Social Science & Medicine

journal homepage: www.elsevier.com/locate/socscimed

Feeding her children, but risking her health: The intersection of gender,household food insecurity and obesity

Molly A. Martin*, Adam M. LippertPennsylvania State University, 211 Oswald Tower, University Park, PA 16802, United States

a r t i c l e i n f o

Article history:Available online 20 December 2011

Keywords:OverweightObesityGenderFood insecurityParentingIncomeUSA

* Corresponding author. Tel.: þ1 814 863 5508.E-mail address: [email protected] (M.A. Mart

0277-9536/$ e see front matter � 2012 Elsevier Ltd.doi:10.1016/j.socscimed.2011.11.013

a b s t r a c t

This paper investigates one explanation for the consistent observation of a strong, negative correlation in theUnited States between income and obesity among women, but not men. We argue that a key factor is thegendered expectation that mothers are responsible for feeding their children. When income is limited andhouseholds face food shortages, we predict that an enactment of these gendered norms places mothers atgreater risk for obesity relative to child-free women and all men. We adopt an indirect approach to studythese complex dynamics using data on men and women of childrearing age and who are household heads orpartners in the 1999e2003 waves of the Panel Study of Income Dynamics (PSID). We find support for ourprediction: Food insecure mothers are more likely than child-free men and women and food insecure fathersto be overweight or obese and to gain more weight over four years. The risks are greater for single mothersrelative to mothers in married or cohabiting relationships. Supplemental models demonstrate that thispattern cannot be attributed topost-pregnancy biological changes thatpredispose mothers toweight gain oran evolutionary bias toward biological children. Further, results are unchanged with the inclusion of physicalactivity, smoking, drinking, receipt of food stamps, or Women, Infants and Children (WIC) nutritionalprogram participation. Obesity, thus, offers a physical expression of the vulnerabilities that arise from theintersection of gendered childcare expectations and poverty.

� 2012 Elsevier Ltd. All rights reserved.

Scholars argue that it takes money to maintain a healthy weightin America’s obesogenic environment (Poston & Foreyt, 1999)because healthy food is relatively expensive and calorie-dense,nutrient-poor food is cheap (Drewnowski & Specter, 2004).Although weight is a function of both caloric intake and expendi-ture, materialist arguments focus on the costs of food and predictgreater caloric intake and consequent body fat among low versushigh income people (Glass & McAtee, 2006). In the U.S., there isa strong, negative correlation between income and the likelihood ofbeing overweight or obese, but only among women; this is notobserved among men (for reviews, see McLaren, 2007; Sobal &Stunkard, 1989). This sex difference is puzzling, particularly toscholars who look beyond individual explanations to consider therole of shared environments for health because the majority of menand women live together (Casper & Bianchi, 2002) and sharesocioeconomic resources and weight-related behaviors (French,Story, & Jeffery, 2001; Mitchell et al., 2003). Given these common-alities, one would expect greater similarity between the sexes.

We hypothesize that the key distinction is not between allwomen and all men, but between mothers and non-mothers. We

in).

All rights reserved.

argue that the confluence of two factors e the experience of foodinsecurity and the gendered nature of childcare e intersect andcontribute to the observed sex differences in the association ofincome and body weight. Food insecurity is highly correlated withpoverty (Sarlio-Lähteenkorva & Lahelma, 2001) and occurs whena household faces budgetary constraints that limit the quantity orquality of food they can purchase (Wunderlich & Norwood, 2006).Yet food insecurity is a “managed process” (Radimer, 1990),meaning that families strategize and diligently work to avoidhunger. That responsibility, however, falls more heavily on womengiven traditional discourses about family life and “women’s work”that place greater expectations on women for feeding andnurturing their family, especially when children are present(DeVault, 1991). Given that food insecurity is correlated with poordietary behavior and obesity (for a review, see Institute of Medicine,2011), we assert that food insecurity mediates the associationbetween income and weight, but that the management of foodinsecurity intersects with gender to create differential risks forobesity between mothers and non-mothers.

To investigate these dynamics, we study men and women ofchildrearing ages (i.e., 18e55) who are heads or partners of U.S.households in the 1999, 2001 and 2003 waves of the Panel Study ofIncome Dynamics (PSID). We test whether the association between

M.A. Martin, A.M. Lippert / Social Science & Medicine 74 (2012) 1754e1764 1755

household food insecurity and the likelihood of being overweightor obese differs across groups defined by sex and parenthood incross-sectional models of weight status and longitudinal models ofweight change. We also examine how partner co-residence furthermoderates these processes due to the gendered norms aboutparental custody (Coltrane & Adams, 2003) and the greater prev-alence of food insecurity among single parents (Rose, Gundersen, &Oliveira, 1998).

Food insecurity and weight

Household food security exists along a continuum but can becategorized into a four-point ordered scale: food secure, foodinsufficiency, low food security, and very low food security (Bickel,Nord, Price, Hamilton, & Cook, 2000; Wunderlich & Norwood,2006). Most Americans are food secure, but some face food insuf-ficiency, meaning, they worry about having enough money to buyfood for the month, but actually make no or few changes to theirdiet (Wunderlich & Norwood, 2006). Food insecurity occurs whenthose fears become a reality. Low food security, or not having themeans to buy the kinds of food desired, reduces the quality andvariety of people’s diets (Wunderlich & Norwood, 2006). Very lowfood security occurs when people do not have the means to buy thequantity of food needed and leads people to skip meals and reducetheir food intake (Wunderlich & Norwood, 2006). Those with either“low food security” or “very low food security” are considered “foodinsecure” (Wunderlich & Norwood, 2006). In 2009, 14.7% of U.S.households were food insecure (Nord, Coleman-Jensen, Andrews, &Carlson, 2010), while in 2003, the year corresponding to our study,the prevalence was 11.2% (Nord, Andrews, & Carlson, 2004).

Because poverty predicts food insecurity (Sarlio-Lähteenkorva &Lahelma, 2001), there are several parallels found in research on theroleof food security for body weight. Keyamong them are consistentsex differences, such that low food security is linked to being over-weight (Adams, Grummer-Strawn, & Chavez, 2003; Dinour, Bergen,& Yeh, 2007; Lyons, Park, & Nelson, 2008; Townsend, Peerson, Love,Achterberg, & Murphy, 2001) and gaining 5 pounds or more in oneyear (Wilde & Peterman, 2006), but only among women. Very lowfood security is associated with being underweight, but again onlyfor women (Wilde & Peterman, 2006).

Several studies suggest that food insecurity is linked to over-weight and obesity due to management strategies people adopt inthe face of economic constraints. Food insecure individuals aremore likely to consume high-calorie but nutritionally-poor food toavoid feelings of hunger (Dixon, Winkleby, & Radimer, 2001;Drewnowski & Specter, 2004; Kirkpatrick & Tarasuk, 2008), eatirregular meals or skip breakfast (Kempson, Keenan, Sadani, Ridlen,& Rosato, 2002; Ma et al., 2003), and consume less milk, fruit andvegetables, especially later in the month (Tarasuk, McIntyre, & Li,2007). According to public health and nutrition research, thesedietary practices are associated with being overweight (Ledikweet al., 2006; Ma et al., 2003) and weight gain (Berkey, Rockett,Gillman, Field, & Colditz, 2003). In the next section, we detailhow the management of food insecurity is gendered.

Gender, childcare, and food insecurity management

Traditional discourses about “family” life and “women’s work”since the industrial revolution include expectations that womenare responsible for caring for their family members and managinghousehold tasks (Rothman, 1978; Sokoloff, 1980). When childrenare present in the home, those responsibilities multiply (Hays,1998) and the gendered division of household labor becomesmore unequal (Coltrane, 2000). For example, there is greatergender equity in the total number of hours spent on housework in

child-free cohabiting and married couples than among similarcouples with children (Sanchez & Thomson, 1997; South & Spitze,1994). Therefore, mothers are more likely to be subjected to,internalize, and reflect traditional gender expectations about theirroles and responsibilities than child-free women.

A key feminine responsibility is “feeding the family,” whichrequires a series of tasks: meal planning, monitoring the supply ofhousehold provisions, shopping, cooking, and cleaning (DeVault,1991). Beyond the practical goals, “feeding the family” alsosustains children’s emotional needs for love, support and security(DeVault, 1991).

In food insecure homes, mothers work hard to prevent hungeramongst their children. In a qualitative study with frequently foodinsecure young mothers, all insisted that their children only expe-rienced food insufficiency because they adopted several strategiesto protect them (Stevens, 2010), including prioritizing their chil-dren’s needs over their own (McIntyre et al., 2003; Stevens, 2010). AsDeVault notes “[t]hese women seem to be expressing a heightenedsense of the more widespread notion that’s women’s own food isless important than that prepared for others” (1991, p.199). As onewoman in a cash-strapped household noted: “If it gets down to it, webuy to feed the kids” (DeVault, 1991, p.191).

To manage food insecurity, mothers adopt a variety of strategies.Some strategies focus on grocery shopping, like buying in bulk,shopping at different stores to get the best prices, or using coupons(DeVault, 1991; Wiig & Smith, 2008). Other strategies involvemothers’ food intake. Food insecure mothers skip meals, wait to eatuntil later in the day, or eat less to spare their children from hungerand nutritional deprivation (Badun, Evers, & Hooper,1995; DeVault,1991; McIntyre, Connor, & Warren, 2000; McIntyre et al., 2003). Asa result, women in food insecure households are at risk of nutrientdeficiencies in Vitamin A, folate, iron, and magnesium (Tarasuk &Beaton, 1999). We suspect that these behavioral patterns under-gird the unexplained sex differences in the association betweenfood insecurity and weight (Adams et al., 2003; Dinour et al., 2007;Lyons et al., 2008; Olson, 1999; Townsend et al., 2001; Wilde &Peterman, 2006) and why food insecurity is typically not corre-lated with children’s weight (Gundersen, Garasky, & Lohman, 2009;Martin & Ferris, 2007), but for an exception see Gundersen andKreider (2009). Unfortunately we do not have direct measures onpeople’s dietary behavior or food insecurity management practicesto fully explore this sequence, but we do have the requisite data totest our primary hypothesis:

H1. There is a statistically significant association between foodinsecurity and being overweight or obese for mothers, but notchild-free women or all men.

We know of only one paper about food insecurity and obesitythat emphasizes parenthood. With a sample of parents (65% ofwhom were single mothers), Martin and Ferris (2007) founda positive association between food insecurity and obesity, but theydid not explore whether there was a differential associationbetween mothers and fathers. Therefore, the current analysismakes a significant contribution by offering an initial test of thishypothesis.

The role of marriage and cohabitation

We predict that the living arrangements of heterosexual menand women further condition the differences between mothers andnon-mothers. Prior research demonstrates that caretaking dutiesamong separated parents are largely performed by the custodialparent, typically the mother (Furstenberg & Cherlin, 1994;Marsiglio, Amato, Day, & Lamb, 2000). Therefore, the risks ofoverweight due to food insecurity should be exacerbated among

M.A. Martin, A.M. Lippert / Social Science & Medicine 74 (2012) 1754e17641756

single mothers and relatively lower for mothers in co-residentialcouple households. Likewise, single fathers should be at greaterrisk of obesity when they are food insecure. Unfortunately, we havetoo few single fathers in our data to fully explore this possibilitybecause most single parents are single mothers (Casper & Bianchi,2002), reflecting a “community division of labor” (DeVault, 1991,p.193) whereby women routinely have custody after parentsseparate. Our second hypothesis is:

H2. The association between food insecurity and being over-weight or obese is stronger for single mothers versus married orcohabiting mothers.

It is important to note, however, that the causal relationshipsbetween overweight, family formation, union dissolution, andhousehold food security are complex. In fact, the causal processcould work in the opposite direction: Overweight women may beless likely to form unions and bear children given feminine beautyideals emphasizing thinness (Allon, 1982).

Alternative explanations

We predict that food insecurity and its management increasesthe risks of overweight and obesity for mothers given the genderedexpectations of childrearing. We recognize, however, that there arecompeting explanations and we do our best to address them.

First, one may agree with our prediction but disagree with ourinterpretation. One may consider any observed risks for mothers asreflecting, not childrearing, but biological risks of childbearing. Ifmetabolic changes related to pregnancy predispose birth mothersto gain weight, then food insecure biological mothers would be atgreater risk of overweight and obesity than “social” mothers. Suchdifferences could also arise if, due to evolutionary pressures,mothers are more protective of their biological children (Daly &Wilson, 1980). To test whether the experience of pregnancy orbiological kinship creates unique risks, we conduct two supple-mental analyses. First, we restrict our sample to only women livingwith children (50% of the sample) and compare whether the risk ofobesity for food insecure mothers is lower among women livingwith children they did not give birth to (i.e., they are adoptive, step,or foster mothers), controlling for the number of children present.Because most women live only with biological children, statisticalpower issues may limit our ability to detect a significant difference.Second, we restrict our sample to women who have ever given birthby 2003 and examine whether the risks of household food inse-curity increase as parity increases, regardless of whether theirchildren currently live with her and controlling for her age andother demographic characteristics. Because 89% of the women inour sample have given birth by 2003, power is less of a problem inthese analyses. If metabolic changes associated with pregnancyundergird our findings, then one would expect those risks toaccumulate with each birth and, thus, translate into a statisticallysignificant interaction between parity and food insecurity amongbiological mothers.

Second, one might argue that the statistical association betweenfood insecurity and overweight is a function of other sociodemo-graphic factors besides income. Thus, we control for status char-acteristics, like age, education and race/ethnicity in all models.

Third, one might expect that other mediating factors explainthese patterns, especially given that we do not have self-reportedmeasures of energy intake or, even better, data from doubly-labeled water tests to measure their energy intake (Schoeller,1990). We test several alternative mechanisms. Because foodinsecure mothers may have fewer opportunities for recreationalphysical activity, we test whether differences in self-reportedphysical activity reduce the association between food insecurity

and weight among mothers. We also test whether the consumptionof alcohol or smoking cigarettes explains the observed patterns.Because of the stresses associated with poverty and food insecurity(Huddleston-Casas, Charnigo, & Simmons, 2009), which wouldlikely feel more threatening to parents, food insecure parents couldbe more likely to self-soothe themselves with alcohol and nicotine.Yet these behaviors are associated with being overweight (Mokdadet al., 2003; Slattery et al., 1992). Lastly, given the longstandingdebate about whether receiving food stamps (now officially theSupplemental Nutrition Assistance Program) increases the risks foroverweight and obesity (Borjas, 2004; Gibson, 2003; Institute ofMedicine, 2011), we test whether our results change with theinclusion of food stamps receipt. We also include a measure ofparticipation in the Women, Infants and Children (WIC) nutritionalprogram.

In sum, we bridge several empirical literatures to develop a newtheoretical model about how gendered patterns of childcareintersect with household economics to increase the risk of over-weight among poor, food insecure mothers. We recognize thatthere are several alternative explanations and, thus, do our best totest them with the available data. Our aim is to provide an initialexamination of whether overweight and obesity are physicalexpressions of the vulnerabilities that arise from the intersection ofgender, parenthood, and poverty.

Data and methods

Data

We use data from the Panel Study of Income Dynamics (PSID)because it is the only study that collects data on individuals’ weight,income, household food insecurity, and household composition.Unfortunately, PSID does not have information about individual’senergy intake and food insecurity management.

PSID is a longitudinal household-based study that began col-lecting data in 1968 for a nationally representative sample and anoversample of low-income, Southern households (Hill, 1992). ThePSID contains longitudinal data for all individuals who were ever ina PSID household, even if they move out (Hill, 1992). Interviewssince 1997 are conducted biennially. Given that the PSID has beenfielded for almost 50 years, sample attrition could pose a problem,but several studies have found that attrition has not affected PSID’srepresentativeness (Becketti, Gould, Lillard, & Welch, 1988;Fitzgerald, Gottschalk, & Moffitt, 1998). PSID is not representative,however, of immigrant groups arriving in the U.S. after 1968.

We make several restrictions to arrive at our analytic sample.First, we must rely on data collected in 1999, 2001, and 2003, theyears in which PSID collected data on both weight and food inse-curity. Second, we restrict our analysis to those who were either thehead of a PSID household or their marital or cohabiting partner in1999, 2001, and 2003 (n ¼ 9935) because PSID only collects data onbody weight for those individuals. While this provides fora consistent sample across the various models, it makes the samplemore selective with regard to family structure stability. Oursubstantive findings are unchanged, however, in analyses wherethe data are multiply imputed to include anyone who meets therestrictions listed below and was ever in the PSID between 1999and 2003, regardless of their relationship to the household head.Third, we restrict the analysis to heads and partners between theages of 18 and 55 in 1999 (n ¼ 8151) to focus on adults most at riskfor living with minor children and, thus, the hypothesized patterns.The next two restrictions eliminate outlier cases that would chal-lenge the statistical homogeneity of our analysis. Fourth, we dropthose who report being foreign born (n ¼ 82) or who can bereasonably assumed to be foreign born because they have five or

M.A. Martin, A.M. Lippert / Social Science & Medicine 74 (2012) 1754e1764 1757

fewer years of completed schooling (and the minimum age ofcompulsory schooling in the United States is 16) (n ¼ 51). Theseindividuals are unique in both unobserved and observed ways (i.e.,their means and correlations for food insecurity, number of chil-dren, marital status and weight differ significantly) because thePSID is not representative of immigrants. The absence of immi-grants reduces the prevalence of food insecurity in the study(Borjas, 2004). Fifth, we omit women who are pregnant at the timeof the 2003 interview (n ¼ 85). Specifically, we omit womenreporting a live birth in the PSID’s Childbirth and Adoption HistoryFile within 9 months following their 2003 interview date. Afterthese restrictions, our sample is 7931 adults.

Missingdata dueto item non-response is relatively minor in thesedata. There are actually no missing data for people’s sex, age, part-nership status, the number of co-residential children, urbanicity, andhousehold income (because the PSID has imputed it). There is minoritem non-response on food insecurity (n1999 ¼ 9, n2001 ¼ 11,n2003 ¼ 20), self-rated health (n2003 ¼ 86), race (n ¼ 129), andwomen’s fertility histories (n ¼ 36). The items with the most missingdata are body mass index (n1999 ¼ 379 [4% of the original 9935sample], n2001 ¼ 257 [3%], n2003 ¼ 305 [3%]) and education(n2003 ¼ 596 [6%]). We utilize multiple imputation handle item non-response, which replaces missing values with predictions frominformation observed in the sample (Rubin, 1987). We use thesupplemental program “ice” within STATA 11.0 (Royston, 2005a,b) tocreate five imputed data sets. The imputation models include all ofthe variables and their interactions that are used in the empiricalmodels, as well as the respondent’s work status, occupation, andregion (all in 2003), the number of adults in the household (in 1999,2001, 2003), whethertheylivewithayoungchild(ages0e5;in 2003)and whether PSID imputed their income. We estimate the empiricalmodels for each imputed data set and then combine the results,accounting for the variance within and between the imputedsamples to calculate the coefficients’ standard errors (Rubin, 1987).

Measures

Body weightWe determine people’s weight classification in three steps. First,

because PSID only has self-reported weight and because self-reported weight is generally biased downward among women andupward among men (Cawley & Burkhauser, 2006), we use theCawley (2004; Cawley & Burkhauser, 2006) adjustments to improvethe accuracy of our dependent variable. Specifically, we multiplyrespondents’ self-reported weight by race- and sex-specific coeffi-cients from Cawley’s regressions of measured weight on self-reported weight. Second, we calculate their body mass index(BMI) [weight (kg)/height2 (m2)] from their self-reported height andtheir Cawley-adjusted self-reported weight. Third, we follow WorldHealth Organization (2000) guidelines to classify BMI into thefollowing weight categories: underweight (BMI < 18.5), normalweight (18.5 � BMI < 25), overweight (25 � BMI < 30) and obese(BMI � 30). In the cross-sectional models, we predict whethera person is (1) normal weight or underweight, (2) overweight, or (3)obese in 2003. Because less than 2% of the sample is underweight,we cannot model underweight as a separate category. For thelongitudinal models, we predict their weight change (in pounds)between 1999 and 2003, simply calculated as their Cawley-adjusted2003 weight minus their Cawley-adjusted 1999 weight.

Household food insecurityWe use the U.S. Department of Agriculture’s Food Security Scale

(Bickel et al., 2000). Respondents were asked a sequential series of18 questions if they live with children and 10 questions if they donot. The different series are made equivalent (and thus orthogonal

to the presence of children) in the final 10-point scale and cate-gorical measure of food security. Following the USDA’s guidelines,households are classified as food insecure (¼1) if they score a 2.2 orhigher on the Food Security Scale (Bickel et al., 2000). We measuretheir household food insecurity in 2003 and create a longitudinalmeasure that counts the survey years with reported householdfood insecurity between 1999 and 2003 (values: 0, 1, 2, or 3).

SexSex is a dichotomous indicator for whether the person is female

(1 ¼ yes) or male.

ChildrenPSID participants report the number of children between the

ages of zero and 17 years currently in the household, regardless oftheir biological relationship to the household head or their partner.We create a dichotomous measure indicating children are present(¼1) and a count of children present.

We use the PSID’s Childbirth and Adoption History (1985e2007)data to create two variables. First, among those living with childrenin 2003, we determine whether the woman gave birth to everychild present and create a dichotomous variable equal to one if shedid not. Because very few women live with a mix of biological andnon-biological children (N ¼ 14), the results primarily reflectwhether women who did not give birth to any of the childrenpresent (N ¼ 307) are different. Second, we calculate the totalnumber of children a woman has ever borne.

In the longitudinal models, we use a variable that equals thedifferencebetween thenumberofchildrenpresentin2003and 1999.

Partner co-residenceTo compare adults in different residential relationships, we

estimate models separately for those who are single and those whoare who are living with a romantic partner, whether married orcohabiting.

Alternative mediating variablesSupplemental models include the following variables, reported

in 2003: being a “current smoker” (¼1), the number of alcoholicdrinks consumed per day (0 ¼ none,1 ¼ less than one a day, 2 ¼ 1 to2 per day, 3 ¼ 3 to 4 a day, and 4 ¼ 5 or more a day), bouts of“heavy” physical activity during the last month (PSID-providedexamples include aerobics, running, swimming, strenuous house-work), bouts of “light” physical activity during the last month(PSID-provided examples include walking, golfing, gardening,bowling), receipt of food stamps in 2001 (¼1), and receipt of WIC in2002 (¼1).

Control variablesTo control for confounding variables, we include age (in years),

education (in years of completed schooling), poor self-rated health(0 ¼ “good,” “very good,” or “excellent,” 1 ¼ “poor” or “fair”),disability status (1 ¼ at least one limitation in the Activities of DailyLiving Scale, 0 ¼ none), and metropolitan residence (0 ¼ non-metropolitan area, 1 ¼ metropolitan area). Race is measured withthree dichotomous variables to compare (1) non-Hispanic AfricanAmericans, (2) Hispanics, and (3) non-Hispanic other racial groupsto non-Hispanic Whites (the reference category).

Analysis

For the cross-sectional analysis, we estimate several ordinallogistic regression models in STATA (v. 11) to predict 2003 weightcategories. The results are substantively similar to those frommultinomial logistic regression models. (Results available upon

M.A. Martin, A.M. Lippert / Social Science & Medicine 74 (2012) 1754e17641758

request.) For the longitudinal models, we make an additional datarestriction. We omit people who report gaining (n ¼ 73 [averagedacross imputations]) or losing (n ¼ 50 [consistent across imputa-tions]) at least 75 pounds in between 1999 and 2003 because suchdramatic changes likely reflect a reporting error in either year orvery unique weight-related experiences. We then use an OLSregression to predict their change in weight (in pounds) between1999 and 2003.

All models include PSID 2003 sampling weights to account forthe PSID’s attrition and oversampling of low-income Southernhouseholds and, thereby, make the findings generalizable to the2003 U.S.-born population. For ease of presentation, we presentresults stratified by sex, but we estimate supplemental modelsusing a pooled sample of men and women to directly test whetherthe interaction between food insecurity and the presence of chil-dren is significantly different by sex.

Results

Table 1 presents weighted descriptive statistics for our fullanalytic sample and for men and women separately. Key amongthese is that over 60% of the sample is overweight or obese in both1999 and 2003. On average, women are more likely to be over-weight or obese in both years and women gain more weightbetween 1999 and 2003 (p < .01). In 1999, 6.8% of the sample wasfood insecure, while only 4.8% of the sample was food insecure in2003. These estimates are lower than the national averages forthese years, reflecting our restriction to U.S.-born individuals. Assuch, our tests rely on the comparison of small subpopulations.There are 174 food insecure men (101 are fathers) and 293 foodinsecure women (196 are mothers). In 2003, the average samplemember was 39 years old, which partially accounts for theobserved decline in the proportion living with children between

Table 1Sample means and percentages, weighted and adjusted for sampling design.

Full sample (N ¼ 7931)1999 2003

Weight, Cawley-adjusted self-reportBody mass index 28.3 29.2Weight classificationUnderweight 1.1% 1.0%Normal weight (reference) 34.7% 30.1%Overweight 33.6% 33.6%Obese 30.6% 35.3%

Weight change (in pounds), 1999e2003 5.3Household food insecurity 6.8% 4.8%Household income (in $1000s) e 79,892Female (¼1) e 53.1%ChildrenCo-reside with children (¼1) 55.0% 49.9%Number of children present 1.0 0.9Relationship to co-residential children (among those living with children)All borne by her e eSome or all not borne by her e e

Number of children ever borne e eAge (range: 18e55 in 1999) e 39.4Education in years (range: 6e17) e 13.5Married or cohabiting (¼1) e 73.3%Poor or fair self-rated health (¼1) e 11.1%Disabled (¼1) e 6.2%Race/ethnicitynon-Hispanic White (reference) e 80.1%non-Hispanic African American e 9.9%Hispanic e 6.1%non-Hispanic other racial group e 3.9%

Metropolitan residence (¼1) e 76.0%Note: Two-tailed tests of a significant difference between men and women are noted as

1999 and 2003. Half the sample lives with children in 2003 andwomen are slightly more likely than men to live with children(p < .05).

Prior research has consistently found a linear, negative rela-tionship between income and obesity among women, but not men.Given that this sex difference motivates our study, we first testwhether we find similar patterns in these data. To do so, weexamine coefficients from an OLS regression of standardized BMI(i.e., mean ¼ 0, standard deviation ¼ 1) in 2003 on standardized2003 household income separately for men and women in modelsthat include PSID sampling weights. We test for significant sexdifferences in this association in a supplemental model thatincludes men and women together and an interaction betweenhousehold income and sex. For men, the standardized coefficientfor household income is �0.01 and not statistically significant(p ¼ .29), while for women it is �0.20 and statistically significant(p < .001). This sex difference is statistically significant (p < .000).We next explore the relationship between household income andthe likelihood of being overweight or obese in 2003 in a similarmanner, but using a logistic regression model and measuringincome in its original metric in ten-thousand dollar units. House-hold income does not predict whether a man will be overweight orobese (p ¼ .34), but the odds that a woman will be overweight orobese declines by 0.01 with every ten-thousand dollar increase inincome (p < .0001). This sex difference is also statistically signifi-cant (p < .0001). Thus, we replicate prior research findings withthese data.

Table 2 presents the results for our cross-sectional ordinallogistic regression models predicting individuals’ weight classifi-cations in 2003. Model 1 presents the additive model and finds that,for both men and women, neither household food insecurity northe presence of children predict being overweight. Model 2provides the test of our hypothesis that mothers are at a higher risk

Women (N ¼ 4337) Men (N ¼ 3594)1999 2003 1999 2003

31.1 32.2 25.3z 25.8z

0.2% 0.1% 2.0%z 2.0%z17.9% 13.9% 53.7%z 48.4%z34.7% 32.5% 32.3% 34.8%47.1% 53.5% 11.9%z 14.7%z

6.7 3.7z7.9% 5.6% 5.5%z 3.9%z

e 76,460 e 83,783ze e e e

56.2% 51.2% 53.6% 48.4%y1.1 1.0 1.0 0.9

e 57.5% e ee 42.5% e ee 2.03 e ee 39.3 e 39.5e 13.4 e 13.6ye 70.7% e 76.3%ze 12.4% e 9.7%ze 7.2% e 5.1%z

e 78.8% e 81.8%ze 11.4% e 8.1%ze 6.1% e 6.1%e 3.7% e 4.0%e 76.3% e 75.7%

follows: y: p < .05, z: p < .01.

Table 2Coefficients from cross-sectional ordinal logistic regression models predicting a heavier weight classification in 2003 (N ¼ 7,931).

Women Men

Model 1 Model 2 Model 1 Model 2

Child present �0.034 (0.09) �0.085 (0.09) �0.127 (0.08) �0.122 (0.09)Household food insecurity 0.085 (0.21) �0.487 (0.31) �0.062 (0.22) 0.027 (0.34)Child present * Household food insecurity e 1.106 (0.40) ** e �0.150 (0.42) zAge 0.013 (0.01) ** 0.012 (0.01) ** 0.003 (0.01) 0.003 (0.01)Married or cohabiting �0.298 (0.01) ** �0.303 (0.10) ** 0.363 (0.12) ** z 0.363 (0.12) ** zEducation �0.132 (0.02) *** �0.130 (0.02) *** �0.062 (0.02) *** z �0.062 (0.02) *** yPoor or fair health 0.064 (0.17) 0.076 (0.16) 0.586 (0.16) *** y 0.589 (0.16) *** yDisabled 0.309 (0.18) 0.331 (0.18) �0.057 (0.23) �0.062 (0.23)Race/ethnicity (reference: non-Hispanic White)African American 1.295 (0.12) *** 1.287 (0.12) *** �0.440 (0.12) *** z �0.441 (0.12) *** zHispanic 1.537 (0.24) *** 1.503 (0.24) *** �0.350 (0.18) z �0.345 (0.18) zOther race �0.042 (0.20) �0.047 (0.20) �0.248 (0.19) �0.249 (0.19)Metropolitan area �0.158 (0.09) �0.162 (0.09) �0.025 (0.09) z �0.025 (0.09)

Standard errors in parentheses * p < .05, ** p < .01, *** p < .001.Note: The ordered weight classifications are (1) Normal or Underweight, (2) Overweight, and (3) Obese. Models are weighted and adjusted for sampling design. Two-tailedtests of a significant difference between men and women are noted as follows: y: p < .05, z: p < .01.

M.A. Martin, A.M. Lippert / Social Science & Medicine 74 (2012) 1754e1764 1759

for overweight and obesity when challenged with food insecurityrelative to child-free women and all men. The results support ourhypothesis. The coefficient for the interaction between householdfood insecurity and child co-residence among women is large inmagnitude (e.g., comparable in size to the coefficients for beingAfrican American or Hispanic) and highly significant (p < .01). Infact, the likelihood that a woman is in a heavier weight classifica-tion is 202% (¼[e1.106 � 1]*100%) higher if she is a food insecuremother relative to child-free food secure woman. Moreover, bychanging the reference category, we can see that the likelihood ofbeing in a heavier weight classification is 177% (¼[e1.02 � 1]*100%)higher for food insecure mothers versus food insecure child-freewomen (p < .05). Among men, the interaction between house-hold food insecurity and the presence of children is not statisticallysignificant. This sex difference is statistically significant (p < .01).

We also explore whether we find similar results using thenumber of children present in the home instead of a simple cate-gorical indicator for any children. Although the results are in theanticipated direction, the interaction is not statistically significantfor women (p ¼ .35) or men (p ¼ .62). We also explore whether we

Table 3Coefficients from cross-sectional ordered logistic regression models predicting a heavier

Women

Model 1

Panel A. Separate Models by Relationship StatusMarried or Cohabiting in 2003 (N ¼ 5,863)Child present �0.071 (0.10)Household food insecurity 0.620 (0.30)Child present * HH food insecuritya eSingle in 2003 (N ¼ 2,068)Child present 0.080 (0.17)Household food insecurity �0.271 (0.28)Child present * HH food insecuritya ePanel B. Women Living with Children in 2003 (N [ 2,592)Co-resides with children not borne by her 0.263 (0.16)Household food insecurity 0.749 (0.27)Co-resides with children not borne by her * HH food insecuritya eNumber of children present 0.066 (0.11)Panel C. Women Who Have Ever Had a Birth by 2003 (N [ 3,844)Number of children ever born 0.104 (0.04)Household food insecurity 0.332 (0.22)Number of children ever born * HH food insecuritya e

Standard errors in parentheses * p < .05, ** p < .01, *** p < .001.Note: The ordered weight classifications are (1) Normal or Underweight, (2) Overweighttests of a significant difference between men and women are noted as follows: y: p < .0

a “HH” is an abbreviation for “Household”.

find similar results if we substitute the number of years of house-hold food insecurity between 1999 and 2003 for the categoricalindicator of household food insecurity in 2003. Similar to our initialresults, we find that the risk of being overweight or obese increasesas the years of household food insecurity increases among mothers(b ¼ 0.338, p < .05), but not non-mothers. (Results for these twotests available upon request.) Thus, with these initial tests, we findsupport for our first hypothesis: household food insecurity isassociated with overweight and obesity among mothers, but notamong child-free women or all men.

Table 3 tests this basic finding across various specifications tobetter ascertain the factors that undergird this differential risk formothers. Panel A provides the test for Hypothesis 2 and examineswhether the risks of food insecurity are greater among singlemothers relative to married or cohabiting mothers. We stratify oursample not only by sex, but also by whether the person lives witha partner. Among married and cohabiting individuals, we do notfind that food insecure mothers are more likely to be overweight orobese than food insecure, child-free women or food secure women.We do see this among single food insecure mothers. Therefore, as

weight classification in 2003.

Men

Model 2 Model 1 Model 2

�0.073 (0.10) �0.050 (0.09) �0.064 (0.09)0.578 (0.48) �0.324 (0.26) y �0.738 (0.48) y0.064 (0.60) e 0.550 (0.55)

�0.074 (0.18) �0.842 (0.24) *** z �0.795 (0.25) ** z�0.890 (0.37) * 0.519 (0.40) y 0.600 (0.47) y1.471 (0.54) ** e �0.333 (0.86)

0.207 (0.16) e e** 0.588 (0.28) * e e

2.357 (1.07) * e e0.068 (0.11) e e

* 0.109 (0.04) * e e0.576 (0.54) e e

�0.095 (0.18) e e

, and (3) Obese. Models are weighted and adjusted for sampling design. Two-tailed5, z: p < .01. Models include all control variables measured in 2003.

M.A. Martin, A.M. Lippert / Social Science & Medicine 74 (2012) 1754e17641760

predicted, the risks of food insecurity are greater among singlemothers. We do not find similar and statistically significant risks forfood insecure, single fathers and the difference between singlefathers and single mothers is not statistically significant. But ourlimited sample sizes for food insecure single fathers (n ¼ 18) andfood insecure single mothers (n ¼ 104) limit the reliability of thesesex comparisons.

Panels B and C in Table 3 explore whether the greater risk ofobesity among food insecure mothers is related to pregnancy-specific metabolic changes that predispose biological mothers tobe heavier. Panel B focuses on women living with children in 2003;all other individuals are omitted from these models. The modelstest whether mothers who did not give birth to the children theylive with are less likely to be overweight relative to those motherswho did, while controlling for the number of children present. InModel 1, we confirm that all food insecure mothers are more likelyto be overweight or obese, but their biological relationship to thosechildren is not, on average, predictive of being overweight or obese.In Model 2, we test whether the risks of food insecurity for beingoverweight or obese are greater for those living with only biologicalchildren. If either pregnancy-specific metabolic changes orevolutionary-based preferences for biological children were drivingour results, then we would expect the coefficient for the interactionbetween household food insecurity and living with non-biologicalchildren to be statistically significant and negative. Although theinteraction is statistically significant, it has the opposite sign. Therisks of food insecurity for being in a heavier weight category areactually greater if the mother is not biologically related to all of thechildren present.

To further test whether pregnancy-related metabolic changesexplain the associations observed in Table 2, Panel C in Table 3examines these patterns among women who have ever givenbirth, regardless of whether those children currently live with themor not and controlling for the mothers’ social and demographic

Table 4Coefficients from cross-sectional ordered logistic regression models predicting a heavier wpathways.

Model 3

WomenChild present �0.087 (0.09)Household food insecurity �0.493 (0.30)Child present * HH food insecuritya 1.119 (0.40) **Physical ActivityFrequency of light activity �0.003 (0.00)Frequency of heavy activity �0.012 (0.01) **

Substance useAlcohol consumption eSmoking e

Means-Tested Nutrition Program ParticipationFood stamps eWIC e

MenChild present �0.135 (0.09)Household food insecurity 0.028 (0.35)Child present * HH food insecuritya �0.169 (0.42) zPhysical ActivityFrequency of light activity �0.001 (0.00)Frequency of heavy activity �0.009 (0.00) *

Substance useAlcohol consumption eSmoking e

Means-Tested Nutrition Program ParticipationFood stamps eWIC e

Standard errors in parentheses * p < .05, ** p < .01, *** p < .001.Note: The ordered weight classifications are (1) Normal or Underweight, (2) Overweighttests of a significant difference between men and women are noted as follows: y: p < .0

a “HH” is an abbreviation for “Household”.

characteristics. Model 1 demonstrates that the likelihood a motheris overweight or obese increases with parity. Interestingly, currenthousehold food insecurity among all women who have ever givenbirth is not predictive of overweight. Model 2 reveals, however, thatthe experience of household food insecurity does not interact witha woman’s parity to generate additional risks for being overweightor obese. Therefore, based on the results in Panels B and C ofTable 3, we conclude that metabolic changes associated withpregnancy do not explain why food insecure mothers are at greaterrisk of being overweight or obese.

Although the evidence is consistent with our theoreticalexplanation that food insecure mothers adopt strategies that striveto protect their children, but that create risks for being overweightor obese, we do not have direct measures of these behaviors to testthis. Instead, we test alternative mediating pathways and see if,with the inclusion of other indicators, the magnitude or statisticalsignificance of our key finding changes. Table 4 presents the resultsfrom a series of models that test whether the risk of being over-weight or obese for food insecure mothers differs with the inclu-sion of physical activity (Model 3), smoking and alcoholconsumption (Model 4), and food stamps and WIC participation(Model 5). Model 6 includes all six hypothesized mediating path-ways. Across all models, the interaction of food insecurity and thepresence of children is statistically significant and essentially thesame magnitude as reported in Table 2. The interaction for men isnever statistically significant.

We next estimate our longitudinal models. Because of theselection process into childbirth, adoption and forming a blendedfamily with non-biological children, we not only estimate ourlongitudinal models for the full sample, but also on a sub-sample ofindividuals already living with children in 1999. If selection intochild co-residence were a key factor, then we should find largerestimates for the interaction between household food insecurityand changes in the number of children in the full sample.

eight classification in 2003 with the inclusion of indicators for alternative mediating

Model 4 Model 5 Model 6

�0.130 (0.09) �0.106 (0.09) �0.152 (0.09)�0.454 (0.31) �0.502 (0.30) �0.476 (0.31)1.117 (0.41) ** 1.012 (0.40) * 1.041 (0.41) *

e e �0.003 (0.00)e e �0.012 (0.01) **

�0.301 (0.06) *** e �0.289 (0.06) ***�0.447 (0.11) *** e �0.455 (0.11) ***

e 0.314 (0.21) 0.317 (0.21)e 0.469 (0.24) 0.443 (0.25)

�0.15 (0.09) �0.117 (0.09) �0.16 (0.09)0.119 (0.34) 0.029 (0.34) 0.119 (0.35)

�0.159 (0.41) z �0.130 (0.42) z �0.174 (0.41) z

e e �0.001 (0.00)e e �0.010 (0.00) *

�0.072 (0.05) z e �0.070 (0.05) z�0.744 (0.10) *** z e �0.757 (0.11) *** z

e �0.215 (0.24) z �0.085 (0.23) ze 0.061 (0.23) y 0.076 (0.23) y

, and (3) Obese. Models are weighted and adjusted for sampling design. Two-tailed5, z: p < .01. Models include all control variables measured in 2003.

M.A. Martin, A.M. Lippert / Social Science & Medicine 74 (2012) 1754e1764 1761

It is important to note that these models are vulnerable to issuesof power. First and foremost, the interactions in these longitudinalmodels create more combinations of rare events. As such, weconsider p-values of less than .10 as statistically significant. Also, itis unknown whether a four-year window is a sufficient time frameto observe shifts in weight as a function of the combination of foodinsecurity and childcare among women. Finally, there is thepotential for differential ceiling effects. In 1999, 30% of women inthe sample are already obese, whereas only 12% of men are obese in1999. Similarly food insecure women are already heavier in 1999than food secure women.

Table 5 displays the results for OLS regression models of weightchange between 1999 and 2003 after omitting those who gain orlose 75 pounds or moreover this four-year period. In Model 1, wesee that increases in the number of children are not, on average,predictive of weight gain for either women or men in the fullsample. Similar patterns are observed for women in the sub-sample restricted to those living with children in 1999. Yet menliving with children in 1999 are predicted to lose weight as thenumber of children increases. Years of food insecurity is associatedwith weight loss, but this association is only statistically signifi-cant among women in the full sample and men living with chil-dren in 1999.

Model 2 includes the interaction between changes in thenumber of children present and the count of years of food insecu-rity. We find a significant positive association for both women andmen in the full sample and a significant positive association forwomen in the sub-sample of people living with children in 1999. Itis important to note that the magnitude of the interaction is largerfor women already living with children in 1999 than for all women.Thus, selection into biological or social motherhood is not the keyfactor. For men, however, the interaction is larger in the full sampleand, thus, selection might underlie these results.

Given the unexpected statistically significant interaction formen in the full sample, we graph the predicted probabilities fromthis model in Fig. 1. The predicted probabilities are calculated byletting the number of children, years of household food insecurity,and their interaction to freely vary, but we set all other variables totheir means or modal categories. For ease of presentation, Panel Ashows the patterns when individuals live with fewer children in2003, Panel B shows the patterns when the number of children isunchanged, and Panel C shows the patterns when individuals livewith more children in 2003. In each panel, the y-axis is the numberof pounds gained or lost between 1999 and 2003 and the x-axis isthe years of food insecurity. The dark bars are for men and thelighter bars are for women.

Panels A and B reveal similar patterns for both men and women.Whether the number of co-residential children declines (Panel A)or remains the same (Panel B), men and women gain the mostweight (i.e., 5 pounds for men and 7 pounds for women) if they are

Table 5Coefficients from OLS regression models predicting weight change (in pounds; 2003e199years of food insecurity (1999, 2001, 2003).

Full Sample (N ¼ 7,808)a

Women Men

Model 1 Model 2 Model 1 Mode

Change in count of children 0.184 (0.47) �0.274 (0.49) �0.397 (0.38) �0.61Years of food insecurity �1.690 (0.85) * �1.298 (0.87) �1.062 (0.80) �0.92Change in child count *

Years of food insec.e 1.911 (0.95) * e 1.15

Standard errors in parentheses þ p <.10, * p < .05, ** p < .01, *** p < .001.Note: Models are weighted and adjusted for sampling design. Two-tailed tests of a signifiz: p < .01. Models include all control variables measured in 2003.

a Individuals who lost or gained 75 or more pounds between 1999 and 2003 are omit

never food insecure. The amount of weight they gain, however,declines as the years of food insecurity increases. In fact, in Panel A,men and women who live with fewer children in 2003 and expe-rience 3 years of food insecurity actually lose weight (i.e., 4 poundsfor men and 3 pounds for women). Panel C, which shows thepatterns for individuals who live with more children in 2003, isrevealing. For men living with more children in 2003, their weightgain over this period (approximately 4 pounds) is insensitive toincreases in the number of years of food insecurity. Among womenwho live with more children in 2003, however, the amount ofweight they gain increases as their years of food insecurityincreases. Those who are never food insecure gain 6 pounds andthose who are food insecure for 3 years gain 10 pounds. Althoughthis is only a four-pound difference, we see that as childcareresponsibilities increase in conjunction with greater exposure tohousehold food insecurity, women gain more weight. For men, thissignificant interaction translates into them being less likely to loseweight as they gain children and increase their years of householdfood insecurity. Therefore, the longitudinal results buttress ourcross-sectional findings.

Discussion

This manuscript explores whether the sex differences inoverweight and obesity related to food insecurity, and incomemore broadly, can be better characterized as differences betweenmothers and non-mothers. In support of Hypothesis 1, we findthat food insecure mothers are more likely to be overweight andobese than their food insecure, but child-free female counter-parts. In contrast, food insecure fathers are not at greater risk ofbeing overweight or obese. These risks of motherhood do notappear to be the result of metabolic changes associated withpregnancy per se, nor do they diminish with the inclusion of self-reported physical activity, smoking, drinking, food stampsreceipt, and WIC participation. Furthermore, the longitudinalfindings suggest that women are at risk of gaining weight as theygain childcare responsibilities and additional years of householdfood insecurity.

The combined risks of childcare and household food insecurityare particularly problematic for single mothers, as we predicted inHypothesis 2. Single mothers are not only more at risk of experi-encing food insecurity than their married or cohabiting counter-parts, but the consequences of household food insecurity for theirweight are greater. At the individual-level, this likely reflects thechallenges of being both the sole provider and caretaker withina household. But there is an important cultural dimension as well.Most single parent families are headed by women because oftraditional, gendered views of childcare. Together, these individualand cultural factors place single mothers at greater risk of poverty,food insecurity, and obesity.

9) based on the change in the number of children present (2003e1999) and count of

Living with Children in 1999 (N ¼ 4,904)a

Women Men

l 2 Model 1 Model 2 Model 1 Model 2

9 (0.41) �0.145 (0.57) �0.873 (0.61) �0.887 (0.52) þ �1.012 (0.55) þ2 (0.80) �1.645 (1.04) �0.789 (1.05) �1.649 (0.92) þ �1.495 (0.95) !3 (0.57) * e 2.549 (0.95) ** e 0.520 (0.63) !

cant difference between men and women are noted as follows: !: p <.10, y: p < .05,

ted from these analysis.

B Individuals who Live with the Same Number of Children in 2003 and 1999

A Individuals who Live with Fewer Children in 2003

-10

-50

510

15W

eigh

t gai

ned

or lo

st, 2

003

– 199

9, in

pou

nds

0 years 1 year 2 years 3 yearsYears of Household Food Insecurity, 1999, 2001, 2003

Men Women

-10

-50

510

15W

eigh

t gai

ned

or lo

st, 2

003

– 199

9, in

pou

nds

0 years 1 year 2 years 3 yearsYears of Household Food Insecurity, 1999, 2001, 2003

Men Women

Individuals who Live with More Children in 2003

-10

-50

510

15W

eigh

t gai

ned

or lo

st, 2

003

– 199

9, in

pou

nds

0 years 1 year 2 years 3 yearsYears of Household Food Insecurity, 1999, 2001, 2003

Men Women

C

Fig. 1. Predicted change in weight 1999 and 2003 depending on changes in thenumber of co-residential children and years of food insecurity. Panel A. Individualswho live with fewer children in 2003. Panel B. Individuals who live with the samenumber of children in 2003 and 1999. Panel C. Individuals who live with more childrenin 2003.

M.A. Martin, A.M. Lippert / Social Science & Medicine 74 (2012) 1754e17641762

This study is not without limitations. Because of the PSID’sdesign, our findings are representative of US-born individuals andcannot speak to the risks for immigrant mothers, an important andgrowing segment of the US population. Because immigrants havehigher rates of food insecurity (Borjas, 2004), our sample hasa lower rates of food insecurity than is found in official reports. This,in turn, limits our power for some analyses given that we interacthousehold food insecurity (5% of the sample), sex (50% is female),and co-residence with children (50% of the sample) in all models. Inlight of these power limitations it is somewhat surprising that wecan detect statistically significant differences for single mothers,women living with some non-biological children, and for allmothers in our longitudinal analyses. Our sample is also restricted toindividuals who head their own household (either alone or witha partner) over these four years. Our findings may not be applicableto those who live in subfamilies or with their parents, though wefind substantively similar results when we relax this requirement.We prefer the results presented here that use this sample restrictionbecause the decision to move in with someone else is likelyendogenous to the processes we are interested in (i.e., themanagement of poverty and food insecurity) and thus deservesgreater attention and direct investigation. A key limitation is that weonly have indirect evidence that mothers adopt strategies to protecttheir children, but place themselves at greater risk of obesity. Finally,we cannot claim to have uncovered a causal relationship, even in thelongitudinal models, because we have observational data. Fora stronger test, we also ran fixed effects models to examine changesover time. The interaction between household food insecurity andthe presence of children was not statistically significant for men orwomen in these models (see Appendix Table 1). Thus, unobservedfactors may be contributing to the patterns we observe. However,fixed effects models place even greater demands on power and thismay partially explain the absence of a significant effect. Weencourage scholars to conduct additional research and collect datawith larger samples and direct measures of individual’s dietarybehavior to better adjudicate the relationship between householdfood insecurity, motherhood, and weight.

These caveats aside, our research makes a significant contribu-tion by demonstrating that that the documented sex difference inthe relationship between food insecurity and weight is bettercharacterized as a difference between mothers and non-mothers.The results are consistent with our arguments, derived from priorresearch on food insecurity, that mothers adopt numerous, butunhealthy strategies to protect their children when the family facesthreats to their food supply. These findings, thus, provide a new lensby which to consider our commonly held biases against overweightwomen. Obesity is highly stigmatized (Dejong, 1980; Puhl &Brownell, 2003) and people frequently characterize those who areoverweight as lazy and stupid (Crandall & Schiffhauer, 1998). To theextent that poor, food insecure mothers are at greater risk of obesity,the prior literature suggests that it is certainly not attributable tosloth or absent-mindedness. Instead, we suspect that their activemanagement and protection of their children creates risks forobesity and weight gain. Beyond drawing our attention to thegenerally hidden work of poor mothers, these results demonstratethat our socially constructed roles and responsibilities can generaterisks for individual health and well-being.

Acknowledgments

Funding for this research was provided by NICHD grant R01-HD050144 (PI: G.D. Sandefur). We thank Michelle Frisco and JasonHoule, the Robert Wood Johnson Foundation Health & SocietyScholars Program Working Group on Gender and Health at ColumbiaUniversity, and anonymous reviewers for their valuable comments.

M.A. Martin, A.M. Lippert / Social Science & Medicine 74 (2012) 1754e1764 1763

Appendix

Table 1Coefficients from OLS fixed effects regression models predicting weight, 1999e2003

Women Men

Model 1 Model 2 Model 1 Model 2

Number of childrenpresent

�0.150 (0.32) �0.177 (0.33) 0.173 (0.24) 0.167 (0.24)

Household foodinsecurity

�0.619 (0.81) �0.966 (1.29) �1.284 (0.76) �1.365 (1.04)

Number of children* Household foodinsecurity

e 0.213 (0.58) e 0.054 (0.47)

Standard errors in parentheses.Note: This model has 3 explanatory variables, 7,808 individuals (N), and 3 timepoints (T). The total number of observations equals N * T ¼ 23,424, while the degreesof freedom lost equals NþK�1 ¼ 7,810.

References

Adams, E. J., Grummer-Strawn, L., & Chavez, G. (2003). Food insecurity is associatedwith increased risk of obesity in California women. Journal of Nutrition, 133,1070e1074.

Allon, N. (1982). The stigma of overweight in everyday life. In B. Wolman (Ed.),Psychological aspects of obesity: A handbook (pp. 130e174). New York: VanNostrand.

Badun, C., Evers, S., & Hooper, M. (1995). Food security and nutritional concerns ofparents in an economically disadvantaged community. Journal of the CanadianDietetic Association, 56(2), 75e80.

Becketti, S., Gould, W., Lillard, L., & Welch, F. (1988). The PSID after fourteen years:an evaluation. Journal of Labor Economics, 6(4), 472e492.

Berkey, C. S., Rockett, H. R., Gillman, M. W., Field, A. E., & Colditz, G. A. (2003). Longitu-dinal study of skipping breakfast and weight change in adolescents. InternationalJournal of Obesity and Related Metabolic Disorders, 27(10), 1258e1266.

Bickel, G., Nord, M., Price, C., Hamilton, W., & Cook, J. (2000). Guide to measuringhousehold food security. Revised 2000. Alexandria, VA: U.S. Department ofAgriculture, Agriculture, Food and Nutrition Service.

Borjas, G. (2004). Food insecurity and public assistance. Journal of Public Economics,88(7e8), 1421e1443.

Casper, L. M., & Bianchi, S. M. (2002). Continuity and change in the American family.Thousand Oaks, CA: Sage Publications.

Cawley, J. (2004). The impact of obesity on wages. Journal of Human Resources, 39(2),451e474.

Cawley, J., & Burkhauser, R. V. (2006). Beyond BMI: The value of more accuratemeasures of fatness and obesity in social science research. Cambridge, MA: NBERWorking Paper 12291.

Coltrane, S. (2000). Research on household labor: modeling and measuring thesocial embeddedness of routine family work. Journal of Marriage and Family,62(4), 1208e1233.

Coltrane, S., & Adams, M. (2003). The social construction of the divorce ‘problem’:morality, child victims, and the politics of gender. Family Relations, 52(4),363e372.

Crandall, C., & Schiffhauer, K. (1998). Anti-fat prejudice: beliefs, values, and Amer-ican culture. Obesity Research, 6, 458e460.

Daly, M., & Wilson, M. (1980). Discriminative parental solicitude: a biologicalperspective. Journal of Marriage and Family, 42(2), 277e288.

Dejong, W. (1980). The stigma of obesity: the consequences of naïve assumptionsconcerning the causes of physical deviance. Journal of Health and Social Behavior,21, 75e87.

DeVault, M. L. (1991). Feeding the family. Chicago: University of Chicago Press.Dinour, L. M., Bergen, D., & Yeh, M.-C. (2007). The food insecurity-obesity paradox:

a review of the literature and the role food stamps may play. Journal of theAmerican Dietetic Association, 107(11), 1952e1961.

Dixon, L. B., Winkleby, M. A., & Radimer, K. L. (2001). Dietary intakes and serumnutrients differ between adults from food-insufficient and food-sufficientfamilies: Third National Health and Nutrition Examination Survey,1988e1994. Journal of Nutrition, 131(4), 1232e1246.

Drewnowski, A., & Specter, S. E. (2004). Poverty and obesity: the role of energydensity and energy costs. American Journal of Clinical Nutrition, 79(1), 6e16.

Fitzgerald, J., Gottschalk, P., & Moffitt, R. (1998). An analysis of sample attrition inpanel data: the Michigan panel study of income dynamics. Journal of HumanResources, 33(2), 251e299.

French, S. A., Story, M., & Jeffery, R. W. (2001). Environmental influences on eatingand physical activity. Annual Review of Public Health, 22, 309e335.

Furstenberg, F. F., & Cherlin, A. J. (1994). Divided families: What happens to childrenwhen parents part. Cambridge, MA: Harvard University Press.

Gibson, D. (2003). Food Stamp Program participation is positively related to obesityin low income women. Journal of Nutrition, 133, 2225e2231.

Glass, T. A., & McAtee, M. J. (2006). Behavioral science at the crossroads in publichealth: extending horizons, envisioning the future. Social Science & Medicine,62(7), 1650e1671.

Gundersen, C., Garasky, S., & Lohman, B. J. (2009). Food insecurity is not associatedwith childhood obesity as assessed using multiple measures of obesity. Journalof Nutrition, 139(6), 1173e1178.

Gundersen, C., & Kreider, B. (2009). Bounding the effects of food insecurity onchildren’s health outcomes. Journal of Health Economics, 28(5), 971e983.

Hays, S. (1998). The cultural contradictions of motherhood. New Haven, CT: YaleUniversity Press.

Hill, M. S. (1992). The panel study of income dynamics: A user’s guide. Newbury Park,CA: Sage.

Huddleston-Casas, C., Charnigo, R., & Simmons, L. A. (2009). Food insecurity andmaternal depression in rural, low-income families: a longitudinal investigation.Public Health Nutrition, 12, 1133e1140.

Institute of Medicine. (2011). Hunger and obesity: Understanding a food insecurityparadigm: Workshop summary. Washington, D.C.: The National Academies Press.

Kempson, K. M., Keenan, D. P., Sadani, P. S., Ridlen, S., & Rosato, N. S. (2002). Foodmanagement practices used by people with limited resources to maintain foodsufficiency as reported by nutrition educators. Journal of the American DieteticAssociation, 102(12), 1795e1799.

Kirkpatrick, S. I., & Tarasuk, V. (2008). Food insecurity is associated with nutrientinadequacies among Canadian adults and adolescents. Journal of Nutrition,138(3), 604e612.

Ledikwe, J. H., Blanck, H. M., Khan, L. K., Serdula, M. K., Seymour, J. D., Tohill, B. C.,et al. (2006). Dietary energy density is associated with energy intake andweight status in US adults. American Journal of Clinical Nutrition, 83(6),1362e1368.

Lyons, A.-A., Park, J., & Nelson, C. H. (2008). Food insecurity and obesity:a comparison of self-reported and measured height and weight. AmericanJournal of Public Health, 98(4), 751e757.

Ma, Y., Bertone, E. R., Stanek, E. J., III, Reed, G. W., Hebert, J. R., Cohen, N. L., et al.(2003). Association between eating patterns and obesity in a free-living USadult population. American Journal of Epidemiology, 158(1), 85e92.

Marsiglio, W., Amato, P., Day, R. D., & Lamb, M. E. (2000). Scholarship on fatherhoodin the 1990s and beyond. Journal of Marriage and Family, 62(4), 1173e1191.

Martin, K. S., & Ferris, A. M. (2007). Food insecurity and gender are risk factors forobesity. Journal of Nutrition Education and Behavior, 39(1), 31e36.

McIntyre, L., Connor, S. K., & Warren, J. (2000). Child hunger in Canada: results ofthe 1994 national longitudinal survey of children and youth. Canadian MedicalAssociation Journal, 163(8), 961e965.

McIntyre, L., Glanville, N. T., Raine, K. D., Dayle, J. B., Anderson, B., & Battaglia, N.(2003). Do low-income lone mothers compromise their nutrition to feed theirchildren? Canadian Medical Association Journal, 168(6), 686e691.

McLaren, L. (2007). Socioeconomic status and obesity. Epidemiological Reviews,(Advanced Access).

Mitchell, B. D., Rainwater, D. L., Hsueh, W.-C., Kennedy, A. J., Stern, M. P., &MacCluer, J. W. (2003). Familial aggregation of nutrient intake and physicalactivity: results from the San Antonio family heart study. Annals of Epidemi-ology, 13(2), 128e135.

Mokdad, A. H., Ford, E. S., Bowman, B. A., Dietz, W. H., Vinicor, F., Bales, V. S., et al.(2003). Prevalence of obesity, diabetes, and obesity-related health risk factors,2001. JAMA, 289, 76e79.

Nord, M., Andrews, M., & Carlson, S. (2004). Household food security in the UnitedStates, 2003. Washington, D.C.: U.S. Department of Agriculture, EconomicResearch Service.

Nord, M., Coleman-Jensen, A., Andrews, M., & Carlson, S. (2010). Household foodsecurity in the United States, 2009. Washington, D.C.: US Department of Agri-culture, Economic Research Service.

Olson, C. M. (1999). Nutrition and health outcomes associated with food insecurityand hunger. Journal of Nutrition, 129, 521e524.

Poston, W. S., 2nd, & Foreyt, J. P. (1999). Obesity is an environmental issue.Atherosclerosis, 146(2), 201e209.

Puhl, R. M., & Brownell, K. D. (2003). Psychosocial origins of obesity stigma: towardchanging a powerful and pervasive bias. Obesity Reviews, 4, 213e227.

Radimer, K. L. (1990). Understanding hunger and developing indicators to assess it.Ithaca, NY: Cornell University.

Rose, D., Gundersen, C., & Oliveira, V. (1998). Socio-economic determinants of foodinsecurity in the United States: Evidence from the SIPP and CSFII datasets. Tech-nical Bulletin #1869. Washington, DC: U.S. Department of Agriculture,Economic Research Service.

Rothman, S. M. (1978). Woman’s proper place: A history of changing ideas andpractices, 1870 to the present. New York: Basic.

Royston, P. (2005a). Multiple imputation of missing values: update. Stata Journal,5(2), 1e14.

Royston, P. (2005b). Multiple imputation of missing values: update of ice. StataJournal, 5(4), 527e536.

Rubin, D. (1987). Multiple imputation for nonresponse in surveys. New York: Wiley.Sanchez, L., & Thomson, E. (1997). Becoming mothers and fathers: parenthood,

gender, and the division of labor. Gender and Society, 11(6), 747e772.Sarlio-Lähteenkorva, S., & Lahelma, E. (2001). Food insecurity is associated with

past and present economic disadvantage and body mass index. Journal ofNutrition, 141, 2880e2884.

Schoeller, D. A. (1990). How accurate is self-reported dietary energy intake?Nutrition Reviews, 48(10), 373e379.

M.A. Martin, A.M. Lippert / Social Science & Medicine 74 (2012) 1754e17641764

Slattery, M. L., McDonald, A., Bild, D. E., Caan, B. J., Hilner, J. E., Jacobs, D. R., jr., et al.(1992). Associations of body fat and its distribution with dietary intake, physicalactivity, alcohol, and smoking in blacks and whites. American Journal of ClinicalNutrition, 55, 943e949.

Sobal, J., & Stunkard, A. J. (1989). Socioeconomic status and obesity: a review of theliterature. Psychological Bulletin, 105, 260e275.

Sokoloff, N. J. (1980). Between money and love: The dialectics of home and marketwork. New York: Praeger.

South, S. J., & Spitze, G. (1994). Housework in marital and nonmarital households.American Sociological Review, 59(3), 327e347.

Stevens, C. A. (2010). Exploring food insecurity among young mothers (15e24years). Journal for Specialists in Pediatric Nursing, 15(2), 163e171.

Tarasuk, V., & Beaton, G. H. (1999). Women’s dietary intakes in the context ofhousehold food insecurity. Journal of Nutrition, 129, 672e679.

Tarasuk, V., McIntyre, L., & Li, J. (2007). Low-income women’s dietary intakes aresensitive to the depletion of household resources in one month. Journal ofNutrition, 137, 1980e1987.

Townsend, M. S., Peerson, J., Love, B., Achterberg, C., & Murphy, S. P. (2001). Foodinsecurity is positively related to overweight in women. Journal of Nutrition,131(6), 1738e1745.

Wiig, K., & Smith, C. (2008). The art of grocery shopping on a food stamp budget:factors influencing the food choices of low-income women as they try to makeends meet. Public Health Nutrition, 12, 1726e1734.

Wilde, P. E., & Peterman, J. N. (2006). Individual weight change is associated withhousehold food security status. Journal of Nutrition, 136, 1395e1400.

Wunderlich, G. S., & Norwood, J. L. (Eds.). (2006). Food insecurity and hunger in theUnited States: An assessment of the measure. Washington, DC: National Acade-mies Press.