is rural residency a risk factor for childhood poverty?

18

Click here to load reader

Upload: patricia-garrett

Post on 27-Sep-2016

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Is Rural Residency a Risk Factor for Childhood Poverty?

Rural Sonolo.<:l·59(1), 1994, pp. 66-83Copyright V 1994 b)' the Rural Sociological Society

Is Rural Residency a Risk Factor for Childhood Poverty?'

Patricia Garrell, Nicholas Ng'andu, anti Jolm FerronFrank Porter Graham ChillI Development Center;UlIit'ersil)' of North Carolina al Chapel Hill,ChapellIill, Nortli Carolina 27599-8040

AnSTRAGr The influence of rural variables on young children's povertystatus, adjusting for individual and family characteristics, is explored. Theliterature suggests that specific demographic variables exert an over­whelming influence on children's poverty status. This is confirmed withdata from the National Longitudinal Survey ofYouth. Results also suggestthat the residential histories of children have consequences for their pov­erty status, even after the influence of control variables has been takeninto account, The conclusion identifies the integration of survey andecological data as one promising direction for future research on child­hood poverty.

Introduction.

The mass media have presented a convincing case that it is danger­ous for children to live in inner-city ghettos. The implicit contrastis that the bucolic countryside provides an ideal child-rearing envi­ronment. Nevertheless, poverty is a problem in both types of com­munities because central cities and nonmetropolitan (nonmetro)areas alike have family poverty rates substantially above the nationalaverage (U.S. Bureau of the Census 1991).

Approximately one child in five in the United States was poor in1990. Minority children in all communities experience high levelsof poverty, and nonmetro minority children are among the mostdisadvantaged. As the Children's Defense Fund (Sherman 1992:4)observed: "The astronomical poverty rate for rural black children(53 percent) exceeds the rate for black children in cities (47 per­cent), as does the proportion of rural black children living in fam­ilies with incomes less than one-half the poverty line." Such highlevels of childhood poverty are virtually unknown in other industri­alized countries where proactive policies supportive of families andchildren mitigate the negative consequences of economic disadvan­tage (Garrett et aI. 1990; Korbin 1992; Smeeding et aI. 1988).

I Preparation of this manuscript was supported in part by grants from the W. KKcllogg Foundation of Battle Creek, Michigan, and the regional centers for ruraldevelopment. Additional funding was provided by the Pew Charitable Trusts, Theauthors, rather than the funding agencies, are responsible for all interpretations. "'ethank Peggy Ross Cook, Constance Hardesty, Naurine Lennox, Ann Tickamyer, andDeborah Tootle for helping clarity some important issues concerning rural childhoodpoverty, Wc also appreciate specific suggestions concerning an earlier draft of thismanuscript by janet Fitchen, Gene Summers, Michael Schulman, and reviewers forthis journal.

Page 2: Is Rural Residency a Risk Factor for Childhood Poverty?

Childhood Poverty - Garrett et al. 67

Recently, the Rural Sociological Society Task Force on PersistentRural Poverty (1993) explored persistent poverty. An underlyingtheme in the task force report is that there is a specificity aboutrural poverty that needs to be taken into account by researchers andpolicymakers. Historically, discussions of rural/urban differenceswere predicated on Durkheimian assumptions that the bases of so­cial solidarity differed substantially in societies with simple and com­plex divisions of labor. Concepts associated with Tonnies' (1887)distinction between gemeinschaft and gesellschaft, however, appearincreasingly anachronistic as national systems of transportation andcommunication penetrate rural areas, exerting strong homogeniz­ing influences on regional cultures, cuisines, and even accents. Ru­ral areas, nevertheless, retain their distinctiveness. Recent scholar­ship (Singelmann and Deseran 1993) focuses on the extent to whichthe structure of opportunity is qualitatively different in metropolitan(metro) and nonmetro areas. An underlying concern is how inter­actions between unequal opportunity structures and differential mi­gration produce both persistently poor places and people (RuralSociological Society Task Force 1993).

Social science theory has been insufficiently sensitive to the spatialcontext within which social relationships occur (Lobao 1993). Thisweakness is manifested in empirical research, with its paucity of mul­tilevel models (Tickamyer and Bokemeier 1993). With few excep­tions, the poverty literature fails to analyze the joint influences ofindividual, family, and community characteristics on childhood pov­erty. Analyses of metro/nonmetro poverty are largely descriptive,failing to control for sociodemographic variables known to be im­portant determinants of poverty status. By contrast, multivariateanalyses that control for individual and household characteristicsordinarily have an implicit urban bias and ignore community char­acteristics as potentially relevant variables. These weaknesses createliteratures on rural and urban childhood poverty that are difficultto reconcile.

The purpose of this paper is to explore the joint influences ofindividual, family, and residential characteristics on young children'spoverty experiences. Two aspects of rural residency-the child'sbirth into a rural community and the proportion of life lived in arural area-are considered, as are three aspects of children's povertyexperiences-birth into poverty, the persistence or chronicity ofpoverty, and its depth.

Childhood poverty

Relatively little research has focused explicitly on the dynamics ofpoverty among rural families and children (Garrett and Lennox1993). With the important exception of some qualitative studies of

Page 3: Is Rural Residency a Risk Factor for Childhood Poverty?

68 Rural Sociology, Vol. 59, No.1, Spring 1994

poor communities that necessarily consider how parents and chil­dren experience poverty (e.g., Fitchen 1981, 1991), most scholar­ship is informed by a demographic perspective. Research on child­hood poverty in rural areas is surprisingly recent. Studies documentrates of poverty, variation in rates by household structure and racial!ethnic group membership, and change in rates over time," Althoughthere is substantial evidence that rural residents face a dispropor­tionate risk of poverty, little research documents the effects of resi­dence after individual and family characteristics are taken into ac­count.

Persistently poor counties are disproportionately rural and familypoverty rates are higher in nonmetro than metro communities(Hoppe 1993). Indeed, poverty among families with children under18 is nearly as prevalent in nonmetro areas (19%) as in central cities(25%). For white and Hispanic families, poverty rates were higherin central cities than in nonmetro areas. For black families, however,poverty rates in nonmetro areas exceeded those in central cities(U.S. Bureau of the Census 1991). These data challenge stereo­typical notions about the nature of family poverty in the UnitedStates.

The literature on childhood poverty is substantially smaller thanthat on family poverty. Taking the child rather than the family asthe unit of analysis provides different estimates of the incidence anddepth of poverty (Smith 1989). At the national level, childhood pov­erty experienced a rapid drop from 26 percent in 1960 to 16 percentin 1970. The War on Poverty was effective and, when it ended, pov­erty rose. Childhood poverty increased to 20 percent by 1988 (Eg­gebeen and Lichter 1991). The incidence of childhood poverty ishigher in nonmetro than metro communities. The gap was mostdramatic in 1960 when the poverty rate of nonmetro children ex­ceeded that of metro children by 19 percent. By 1990, the nonmetrochildhood poverty rate stood at 21 percent compared with 18 per­cent for metro children (Lichter and Eggebeen 1992). Over threedecades, the poverty gap between nonmetro and metro childrendecreased, but official poverty levels among all children increased.Variables other than community of residence are relevant.

Poverty in the United States is associated with two major demo­graphic variables-ethnic minority parentage and female-headedhouseholds. Considering the United States as a whole, black chil­dren were approximately three times as likely as white children to

2 Poverty can be measured in many ways. In this paper, all commentaries aboutpoverty status refer to official U.S. government thresholds, which relate cash incometo family size (Orshansky 1965). The income-to-needs ratio (Duncan and Rodgers1988, 1991) captures the essence of poverty as measured by official statistics becauseit relates pre-tax earned income to family size and expresses cash income as a mul­tiple of the poverty threshold.

Page 4: Is Rural Residency a Risk Factor for Childhood Poverty?

Childhood Poverty - Garrett et at. 69

experience poverty (Eggebeen and Lichter 1991). Ethnic minoritiesresident in rural areas historically have high poverty rates (jensenand Tienda 1989; Kraly and Hirschman 1990). Nevertheless, howpoor, rural, minority children compare to each other and to theirwhite counterparts in terms of household composition and parentalcharacteristics is basically unknown (Snipp et al. 1993).

Growing up in a female-headed household is clearly a risk factorfor poverty. Since 1970 the poverty rate for children living in female­headed households has been approximately five times higher thanthat for children living in married couple families (Eggebeen andLichter 1991). Place of residence may have a differential effect. Lich­ter and Eggebeen (1992) compared observed child poverty rates inmetro and nonmetro areas to those standardized to reflect specificdemographic changes. They concluded that increased rates of childpoverty attributable to the higher prevalence of female-headedhouseholds more than offset the improvements attributable to in­creases in female employment, rising levels of parental education,and decreasing family size. The rise in female householders was re­sponsible for a larger share of the increase in childhood poverty innonmetro than metro areas during the 1980s. A reasonable hypoth­esis is that employment opportunities, and therefore income, aremore limited for women in rural than urban communities (Me­Laughlin and Sachs 1988; Wenk and Hardesty 1993).

Many rural families and children who are poor live in two-parent,married-couple households with one or more workers (O'Hare1988). Having two parents is no guarantee that children will enjoyan adequate standard of living. In 1990, poverty levels for married­couple families with children under 18 were 7 percent for metroresidents and 11 percent for nonmetro residents. National figuresmask important racial!ethnic group differences. Specifically, thepoverty level for married couple, nonmetro blacks was 28 percent,substantially above the 16 percent central-city level (U.S. Bureau ofthe Census 1991).

Considering the period 1979 to 1986, O'Hare (1988) demonstrat­ed that real median family income declined by 10 percent, whilethe poverty rate increased 55 percent among young adults in ruralareas. One-fourth of rural children lived in poor households by1986. In rural households headed by persons 18 to 29, the povertyrate rose from 19 to 32 percent between 1979 and 1986, with dis­proportionate increases among minorities. Nearly one-third ofyoung rural families with one wage earner were poor in 1986, andalmost one-tenth of those with two or more workers did not earnenough to bring family income above the poverty line.

Unemployment rates in rural areas are higher than in urban ar­eas. Facing labor markets with limited employment opportunities,many rural residents accept part-time jobs or stop looking for work.

Page 5: Is Rural Residency a Risk Factor for Childhood Poverty?

70 Ruml Sociology, Vol. 59, No.1, Spring 1994

When official unemployment rates are adjusted for these conditions,rural unemployment is almost one-third higher than urban unem­ployment. Unemployment, underemployment, and low wages are allimplicated in high levels of nonmetro poverty (O'Hare 1988).

Children born into poverty can realistically expect to spend muchof their pre-adult lives in poverty (Bane and Ellwood 1986). Never­theless, longitudinal data reveal that children make frequent tran­sitions in and out of poverty, reflecting changes in parents' livingarrangements and earnings capacity (Adams and Duncan 1990;Duncan and Rodgers 1988). The dynamics of persistent povertyamong nonmetro children have been studied only at the bivariatelevel. Ross and Morrissey (1989) found that nonmetro children ex­perienced higher rates of poverty than metro children. Whereaschildren in nonmetro male-headed families tended to be temporar­ily poor, those in female-headed families disproportionately experi­enced persistent poverty. Morrissey (1991) found that approximately12 percent of young adults had been persistently poor as both chil­dren and adults. Children raised in unmarried or female-headedhouseholds were more likely than others to experience second-gen­eration poverty. Intergenerational poverty was higher in nonmetroareas, among nonwhites, Southerners, and those who failed to mi­grate from rural to urban areas. Rogers (1991) found that thestrongest predictors of the adequacy of income to family needs wereparental education, number of siblings, and marital status. Metro/nonmetro residence had a statistically significant effect, after takinginto consideration other variables in the model. Both the incidenceand chronicity of family poverty increased during the 1980s (Devineet al. 1992).

Persistent poverty figures prominently in the literature. Povertydramatically limits children's life chances and exposes them to bio­medical, environmental, and social-psychological risks that can re­sult in conditions difficult, even impossible, to remediate (Garrettand Lennox 1993). Young children face increasing developmentallevels in rapid succession. Consequently, deprivations associated witheven episodic poverty can have dire consequences.

Rural residence is likely to compound developmental risks asso­ciated with childhood poverty. Public and private social services aregenerally less available in rural communities (Martinez-Brawley andBlundall 1989; Sherman 1992). Service utilization is low, reflectingsubtle combinations of regulations that effectively discriminateagainst intact working families, ideological commitments to self-re­liance, the stigma associated with making personal concerns a mat­ter of public record, and the simple inadequacy of available services(Duncan and Tickamyer 1988;]ensen 1989; Rank and Hirschi 1988;Rogers 1991; Waltman 1986). Ignorance is also relevant. Where wel-

Page 6: Is Rural Residency a Risk Factor for Childhood Poverty?

Childhood Poverty - Garrett et at. 71

fare dependence is not endemic, people may not know how to ac­cess services (e.g., food stamps or medicare) that children need.

In summary, the literature suggests that individual, family, andcommunity characteristics are all relevant to children's poverty ex­periences. Available scholarship, however, provides little evidenceabout how multiple determinants are interrelated. This study usesmultivariate analysis techniques to explore the joint influence ofdemographic and rural residency variables on children's poverty sta­tus. Is rural residency a risk factor for childhood poverty, after theeffects of other individual and family factors are taken into account?Attention focuses on young children because there are good reasonsto believe that youth and rural residency both put poor children atrisk for negative developmental outcomes.

Methodology

Sam/lie

The data analyzed come from the National Longitudinal Survey ofYouth (NLSY). This panel study was begun in 1979 to document thelabor-market experiences of young adults. Minorities and poor peo­ple were oversampled. When appropriately weighted, the sample isrepresentative of people born in the United States between 1958and 1965. The sample retention rate for women was 92.7 percentbetween 1979 and 1986 (Baker and Mott 1989).

Child assessments were incorporated into the 1986 NLSY proto­col. Of the women interviewed, 3,322 were mothers of 5,876 chil­dren. This study is restricted to children who were under four-yearsold at the time of the 1986 assessment. These children were all bornduring the early 1980s, a period of rising poverty. Young childrenare particularly vulnerable to the biomedical, environmental, andsocial-psychological risks attendant upon poverty status.

NLSY mothers may have had more than one child under four.Including multiple children in statistical analyses violates the as­sumption of independence in sample selection, and selecting spe­cific children (e.g., first- or last-born) cannot be justified on theo­retical grounds for this particular study. Accordingly, randomsampling procedures were used to select one age-eligible child perhousehold. All analyses were conducted on weighted data. Totalsample size was 1,733.3

1 Older children in the NLSY sample were generally born to younger, more dis­advantaged mothers. The younger children, by contrast, reflect more normativechildbearing. Collectively, the NLSY children represent the first 40 percent of child­bearing to the 1958-1965 birth cohort (Baker and Mott 1989). Consequently, theyare not a random sample of their age mates in the United States. Rather, NLSYchildren are a large and heterogeneous group whose poverty experiences can beanalyzed to explore relationships among theoretically-relevant variables. Therefore,

Page 7: Is Rural Residency a Risk Factor for Childhood Poverty?

72 Rural Sociology, Vol. 59, No.1, Spring 1994

Measurement

Rural residency variables are of central concern in this analysis. Thehypothesis is that rural residency is positively associated with child­hood poverty. Birth into a rural community is operationalized asmother's residence during the year of the child's birth. Residentialpatterns are dynamic (Wenk and Hardesty 1993). Consequently, itis appropriate to calculate the proportion of the child's life spent ina rural community, based on mother's residence at the time of theannual interview. Rural-urban residence is operationalized usingstandard census criteria identifying 50,000 persons as the thresholdfor urbanized populations. Respondents resident in a county withless than 50 percent urbanized population are coded in the NLSYdata set as living in a rural county (Center for Human ResourceResearch 1992).

Child's age at assessment is conceptualized as a cohort variable.Economic conditions vary from year to year, and child's age is in­tended to capture secular changes likely to impact family well-being.Maternal characteristics are thought to exert important influenceson children's poverty experiences. Mother's race/ethnicity is an in­terviewer-assigned category collapsed to black, white, and other.' Inall analyses, the reference category is "other." Maternal age andeducational attainment pertain to the year of child assessment. Ac­ademic ability was measured in 1979 by the standard score on theacademic ability composite of the Armed Services Vocational Apti­tude Battery (U.S. Military Entrance Processing Command 1989).

Household characteristics can change dramatically, even duringthe lifetime of a very young child. Birth into a female-headed house­hold and proportion of life lived in a female-headed household areincluded in different models. The average adult-to-child ratio wascomputed from the annual household census. The presence ofmother, father/companion, and other female adult each contrib­uted one unit to the numerator; children under 18 each contributedone unit to the denominator. The resulting ratios were then aver­aged.

it is not appropriate to extrapolate from the sample of NLSY children to the popu­lation of all children in the United States of the same age. It is appropriate, however,to analyze the poverty experiences of NLSYchildren, focusing on demographic vari­ables that previous scholarship has identified as important and adding residentialvariables as determinants of children's poverty experiences.

• The interviewer-assigned category is actually black, Hispanic, and non-black/non­Hispanic. Accordingly, the category labeled "white" in this analysis includes smallnumbers of nonblack/nonHispanic minorities. There are multiple measures of eth­nicity in the NLSYdata set, but substantial amounts ofmissing data make it impossibleto combine variables and create an improved measure. The operationalization ofrace/ethnicity in the NLSY is discussed in detail in the NLS Users' Guide (Centerfor Human Resource Research 1992).

Page 8: Is Rural Residency a Risk Factor for Childhood Poverty?

Childhood Poverty - Garrett et al. 73

Economic characteristics reflect household patterns of labor-forceattachment. There are two measures-the number of workers at thetime of birth and the average number of workers during the child'slifetime. These measures of labor-market involvement were designedto capture the differential reliance on income from employmentrather than transfer payments in rural and urban areas (O'Hare1988).

Poverty experiences are the outcome measures of interest. Thethree dependent variables are each anchored in the child's personalhistory. Initial disadvantage is captured by the child's birth into pov­erty. This is measured by whether the family fell below the officialpoverty threshold during the year of the child's birth. The persis­tence of poverty is reflected in the proportion of the child's lifespent below the official poverty threshold. Finally, the adequacy offamily income or, alternatively, the depth of poverty is measured bythe income-to-needs ratio (Duncan and Rodgers 1988, 1991). Familyincome was operationalized as the sum of mother's and spouse's(when present) cash income from wages and self-employment. Thissum was then expressed as a proportion of the official povertythreshold for a family of the same size (U.S. Bureau of the Census1991). Following these procedures, yearly income-to-needs ratioswere calculated and then averaged to reflect the child's general ex­perience with poverty. An income-to-needs ratio of unity (1) reflectsthe official poverty line.

Data analysis

To determine whether rural residence had an effect on children'spoverty experiences, after the influence of demographic variablesknown to be important had been taken into account, the study fo­cused on the contrast between the restricted model, including onlycontrol variables, and the inclusive model, adding rural residencyvariables. Hierarchical modeling was the appropriate data analyticstrategy. When the dependent variable was continuous (e.g., income­to-needs ratio), ordinary least squares regression was used. Whenthe dependent variable was categorical (e.g., born or not into pov­erty, spending more or less than half one's life in poverty), logisticregression was used. Logistic regression describes the relationshipbetween a binary or dichotomous outcome variable and a set ofexplanatory variables (Hosmer and Lemeshaw 1989). The methodsemployed in logistic regression follow the same general principlesas linear regression. The coefficients were estimated using maxi­mum likelihood techniques. Interpretation of the estimated param­eters is similar to that of parameters obtained using ordinary linearregression methods. Interpretation is facilitated if parameters areexpressed as an odds ratio by exponentiating the estimated param-

Page 9: Is Rural Residency a Risk Factor for Childhood Poverty?

74 Rural Sociology, Vol. 59, No.1, Spring 1994

Table 1. Means of variables by place of birth of the child-

Variables

Child characteristicsAge at assessment (in months)

Poverty characteristicsBorn into povertyPercent of life in povertyIncome-to-needs ratio

Maternal characteristicsPercent whitePercent blackPercent other (Latinos)Age (years)Educational level (years)Academic ability (score)

Household characteristicsBorn into a female-headed householdPercent of life in a female-headed householdAverage adult-to-child ratio

Economic characteristicsNumber of workers at year of birthAverage number of workers during child's life

Rural variablePercent of life lived in rural area

Born in Born inurban area rural area(N = 1,342) (N = 391)

22.27 23.39

0.18 0.2221.53 25.97

2.05 1.64

48.58 73.1529.51 23.0121.91 3.8425.89 25.7112.26 11.83

-0.19 -0.28

0.23 0.2021.48 19.26

1.86 1.80

1.87 1.651.74 1.54

2.26 93.11

• Weighted means calculated using SUDAAN (Shah et al. 1991).

eter [exp(b)] as described by Aldrich and Nelson (1984), DeMaris(1990), and Hosmer and Lemeshaw (1989).

Results

Children in the NLSY sample who were and were not born into ruralareas have very similar characteristics (Table 1). Nevertheless, thepoverty variables all suggest a higher incidence of economic disad­vantage among those born in rural areas. The single largest differ­ence is the proportion of life lived in rural areas, which, as expected,is much higher for children born into rural communities.

All child, maternal, household, and economic variables are sig­nificantly related to poverty outcomes (Table 2). This indicates thatit is appropriate to introduce them as control variables into modelsexploring the influence of rural residency on poverty status. At thebivariate level, rural variables are also associated with the povertyvariables. The correlation between being born in a rural area andthe proportion of life lived in a rural area is very high (r = 0.95).

Page 10: Is Rural Residency a Risk Factor for Childhood Poverty?

Childhood Poverty - Garrett et al. 75

Table 2. Pearson correlations of predictor variables by poverty out­come variables

Predictor variables

Child characteristicsAge at assessment

Maternal characteristicsPercent whitePercent blackAgeEducational levelAcademic ability

Household characteristicsBorn into a female-headed house­

holdPercent of life in a female­

headed householdAverage adult-to-child ratio

Economic characteristicsNumber of workers in year of birthAverage number of workers during

child's life

Rural variablesBorn into rural area (1 = yes, 0 = no)Percent of life lived in rural area

Poverty outcome variables

Percentof life Income-

Born into lived in to-needspoverty poverty ratio

0.06* 0.08** -0.14**

0.07** 0.11** -0.09**0.28** 0.31** -0.30**

-0.21** -0.18* 0.26**-0.28** -0.32** 0.41**-0.37** -0.41** 0.44**

0.42** 0.42** -0.36**

0.44** 0.50** -0.45**-0.33** -0.41** 0.45**

-0.37** -0.34** 0.18**

-0.37** -0.44** 0.27**

0.05* 0.05* -0.11**0.04 0.06* -0.11**

* fl < 0.05; **fl < 0.01.

Consequently, these variables are not introduced into the same mod­el in order to avoid problems of multicolinearity.

Models predicting the poverty experiences of young children areincluded in Tables 3, 4, and 5. Model 1 in each table includes onlythe control variables (i.e., child, maternal, household, and economiccharacteristics). Across the tables, most controls are significantly re­lated to poverty status outcomes. The sole exception is that beingwhite is not significantly related to the income-to-needs ratio.

In Table 3 the dependent variable measures initial disadvantage(whether or not the child was born into poverty). The independentvariable of interest is whether the child was born in a rural com­munity, which implies a contrast between Models 1 and 2. Ruralbirth is significantly associated with poverty status at birth. Specifi­cally, the odds of being born into poverty for children born in arural area are estimated to be 1.12 times higher than those for chil-

Page 11: Is Rural Residency a Risk Factor for Childhood Poverty?

76 Rural Sociology, Vol. 59, No.1, Spring 1994

Table 3. Logistic regression results of the relationship between con­trol and rural variables and being born into poverty (N = 1,631)

Outcome variable:Born into poverty

VariablesModel 1 Model 2

B (SE of B) B (SE of B)

Control variablesAge at assessmentWhiteBlackMother's ageMother's educational levelMother's academic abilityBorn in female-headed household

(1 = yes, 0 = no)Adult-to-child ratio at birthNumber of workers in household

at year of birth

Rural variableBorn in a rural area

(I = Yes, 0 = No)

Constant-2 log likelihood

** 11 < 0.01.

0.003 (0.0001)**-0.049 (0.0006)**-0.115 (0.0005)**-0.254 (0.0001)**-0.114 (0.0001)**-0.614 (0.0003)**

1.413 (0.0003)**-0.874 (0.0003)**

-0.958 (0.0002)**

8.58P< 0.0001

0.004 (0.0001)**-0.055 (0.0006)**-0.074 (0.0005)**-0.256 (0.0001)**-0.103 (0.0001)**-0.601 (0.0003)**

1.467 (0.0004)**-0.924 (0.0003)**

-0.966 (0.0001)**

0.117 (0.0004)**

8.52P< 0.0001

dren born in urban areas. There are no statistically significant in­teractions between the control and rural variables (p> 0.01).

Models predicting the chronicity of poverty, specifically the pro­portion of life lived in poverty, appear in Table 4. Most childrenspent either none or all of their lives in poverty, which is consistentwith the restricted age range of the sample. Cross-tabulations dem­onstrated that birth into poverty was related to, but different from,proportion of life lived in poverty because children experiencedtransitions both into and out of poverty. Given these considerations,it seemed appropriate to recode the original variable, distinguishingbetween children who spent more or less than 50 percent of theirlives in poverty. Logistic regression was used to analyze the relation­ship between the chronicity of poverty and rural residency, adjustingfor control variables.

The child, maternal, household, and economic variables are allstatistically significant. When rural residency variables are added tothe model, they also attain statistical significance. Generally, thismeans that the chronicity of poverty is related to the proportion oflife lived in rural areas. For children born in rural areas, the oddsof spending the majority of their lives in poverty are 1.27 timeshigher than those for children not born into rural areas. Similarly,for children who have lived half or more of their lives in rural areas,

Page 12: Is Rural Residency a Risk Factor for Childhood Poverty?

Tab

le4.

Log

isti

cre

gre

ssio

nre

sult

so

fth

ere

lati

on

ship

bet

wee

nco

ntr

ol

and

rura

lva

riab

les

and

per

sist

ence

of

po

ver

ty(N

=1,

631) O

utc

om

eva

riab

le:

Liv

ing

hal

fo

rm

ore

of

life

inpo

vert

y

Var

iabl

es

Co

ntr

ol

vari

able

sA

geat

asse

ssm

ent

Whi

teB

lack

Mo

ther

'sag

eM

oth

er's

educ

atio

nal

leve

lM

oth

er's

acad

emic

abil

ity

Per

cen

to

flif

ein

afe

mal

e-he

aded

ho

use

ho

ldA

dult

-to-

chil

dra

tio

Nu

mb

ero

fw

orke

rsin

ho

use

ho

ldat

year

of

bir

th

Rur

alva

riab

les

Chi

ldb

orn

inru

ral

area

Per

cen

to

flif

ein

rura

lar

ea

Co

nst

ant

-2lo

gli

keli

hood

*fJ

<0.

05;

**fJ

<0.

01.

Mo

dell

B(S

Eo

fB

)

-0.0

16

(0.0

001)

**0.

391

(0.0

005)

**0.

218

(0.0

005)

**-0

.17

7(0

.000

1)**

-0.0

89

(0.0

001)

*-0

.62

1(0

.000

2)**

0.01

9(0

.000

1)**

-0.9

16

(0.0

002)

**

-1.3

38

(0.0

002)

**

7.64

4fJ

<0.

0001

Mod

el2

B(S

Eo

fB

)

-0.0

15

(0.0

001)

**0.

454

(0.0

005)

**0.

330

(0.0

005)

**-0

.19

1(0

.000

1)**

-0.1

03

(0.0

001)

**-0

.55

7(0

.000

2)**

0.02

0(0

.000

1)**

-0.9

17

(0.0

002)

**

-1.3

83

(0.0

002)

**

0.17

2(0

.000

3)**

8.13

5fJ

<0.

0001

Mod

el3

B(S

Eo

fB

)

-0.0

17

(0.0

001)

**0.

471

(0.0

005)

**0.

301

(0.0

004)

**-0

.17

9(0

.000

1)**

-0.1

01

(0.0

001)

**-0

.57

5(0

.000

2)**

0.02

0(0

.000

1)**

-0.9

01

(0.0

002)

**

-1.3

19

(0.0

002)

**

0.00

3(0

.000

1)**

7.65

7fJ

<0.

0001

Q ~ ;::. c c ~ ~ ~ ~ I ~ ~ :::: ~ I;l l'"'< '" '"

Page 13: Is Rural Residency a Risk Factor for Childhood Poverty?

Tab

le5.

Mu

ltip

leli

nea

rre

gre

ssio

nre

sult

so

fth

ere

lati

on

ship

bet

wee

nco

ntr

ol

and

rura

lva

riab

les

and

inco

me­

to-n

eed

sra

tio

(N=

1,63

9)

~ ? l:l ...

Var

iabl

es

Co

ntr

ol

vari

able

sA

geat

asse

ssm

ent

Whi

teB

lack

Mo

ther

'sag

eM

oth

er's

edu

cati

on

alle

vel

Mo

ther

'sac

adem

icab

ilit

yP

erce

nt

of

life

Ina

fem

ale-

head

edh

ou

seh

old

Adu

lt-t

o-ch

ild

rati

oN

um

ber

of

wor

kers

inh

ou

seh

old

atye

aro

fb

irth

Rur

alva

riab

les

Chi

ldb

orn

inru

ral

area

Per

cen

to

flif

ein

rura

lar

ea

Co

nst

ant

Adj

uste

dR

"

*jl<

0.05

;**

jl<

0.01

.

Mo

dell

B(S

Eo

fB

)

-0.0

07

(0.0

02)*

*-0

.05

7(0

.127

)-0

.24

3(0

.108

)*0.

130

(0.0

15)*

*0.

160

(0.0

22)*

*0.

168

(0.0

49)*

*

-0.0

10

(0.0

01)*

*0.

540

(0.0

46)*

*

0.25

4(0

.047

)**

-4.2

99

0.44

Ou

tco

me

vari

able

:In

com

e-to

-nee

dsra

tio

Mod

el2

B(S

Eo

fB)

-0.0

07

(0.0

02)*

-0.1

47

(0.1

28)

-0.2

97

(0.1

08)*

0.12

7(0

.015

)**

0.15

7(0

.021

)**

0.15

8(0

.049

)**

-0.0

10

(0.0

01)*

*0.

535

(0.0

46)*

*

0.23

0(0

.047

)**

-0.3

03

(0.0

74)*

*

-4.0

69

0.44

Mod

el3

B(S

Eo

fB

)

-0.0

07

(0.0

02)*

*-0

.15

4(0

.128

)-0

.30

2(0

.108

)*0.

128

(0.0

15)*

*0.

159

(0.0

21)*

*0.

154

(0.0

49)*

*

-0.0

10

(0.0

01)*

*0.

533

(0.0

46)*

*

0.22

8(0

.047

)**

-0.0

03

(0.0

01)*

*

-4.0

82

0.44

~ ~.

Q g ~- ~ :- I..J

l~'O ~ ~.......

~ ;:1.

~ .......

'0 ~

Page 14: Is Rural Residency a Risk Factor for Childhood Poverty?

Childhood Poverty - Garrett et al. 79

the odds of spending the majority of their lives in poverty are esti­mated to be 1.32 times higher than those for children who lived themajority of their lives in urban communities. Again, there are nostatistically significant interactions between the control and ruralvariables (1) > 0.01).

The income-to-needs ratio is the dependent variable predicted inTable 5. Model 1 provides the contrast. Rural residency variables,introduced in Models 2 and 3, are significantly and negatively relat­ed to the income-to-needs ratio, after adjusting for the control vari­ables. Children born in rural areas were significantly more likely toexperience inadequate family incomes. Birth into a rural ratherthan an urban area is associated with a 0.3 decrease in the income­to-needs ratio. A similar relationship exists for children who spenta large proportion of their lives in rural areas. Every 10 percentincrease in the proportion of life lived in a rural area is associatedwith a 0.03 decrease in the income-to-needs ratio. No statisticallysignificant interactions occur between the control and rural vari­ables (1) > 0.01).

Discussion

The literature on childhood poverty emphasizes the demographiccharacteristics of the family as critical determinants of childhoodpoverty status. The results of these analyses suggest that this empha­sis is entirely appropriate. The maternal, household, and economiccharacteristics included in the models are consistent and strong pre­dictors of young children's poverty status.

Rural residency variables also attain statistical significance, afterthe influence of the control variables has been taken into account.These results are consistent with Rogers (1991), who found statisti­cally significant effects of residence on the adequacy of income tofamily needs. Nevertheless, in separate metro and nonmetro models,the variables behaved similarly, suggesting that the same fundamen­tal processes were at work.

In this study, the effects of the rural residency variables are statis­tically significant but small. The characteristics introduced as controlvariables in the models are known to be distributed differently inmetro and nonmetro areas. This is especially relevant for two vari­ables-female-headed households and labor-market involvement.These are included precisely because the literature suggests thatthey are related to the specificity of rural poverty. Poor rural chil­dren are more likely than their urban counterparts to live in mar­ried-couple families with one or more workers. Under these cir­cumstances, it is noteworthy that rural residency variables attain sta­tistical significance, after controlling for these and other variablesin the models.

Page 15: Is Rural Residency a Risk Factor for Childhood Poverty?

80 Rural Sociology, Vol. 59, No.1, Spring 1994

It is easy to incorporate rural residency into a multivariate modelpredicting childhood poverty status. It is more difficult to theorizethe link between demographic and residency variables that makessuch statistical analyses appropriate. Whereas alternative theoreticalparadigms inspire research on rural poverty (Rural Sociological So­ciety Task Force on Persistent Rural Poverty 1993), analyses of child­hood poverty are more descriptive than theoretical. Indeed, it is notclear whether the determinants of family and childhood povertyshould be conceptualized as an artifact of demographic variables ora proxy for structural contexts that present differential economicopportunities (Lichter et al. 1993b). The results reported in thispaper support both interpretations, and the literature on regionallabor markets provides a potential framework for their integration.

The demographic characteristics of adults position them to takedifferential advantage of economic opportunities, which vary geo­graphically. Moreover, the structure of economic opportunities en­courages people to reposition themselves in space. Selective hiringand regional migration are two sides of the same coin. Person andplace intersect in complex ways, so demographic characteristics andresidency status jointly influence family income and, consequently,childhood poverty status.

People whose education and experience make them competitiveon the job market are not equitably distributed across space (Lichteret al. 1993a). For this reason, a model with only demographic char­acteristics provided the standard against which rural residency vari­ables were evaluated for all childhood poverty outcomes. The logicof the rural labor-market literature suggests that demographic andsocioeconomic characteristics are interrelated. Consequently, onewould not expect a substantial increase in variance explained whenrural residency variables were added to statistical models containingdemographic variables predicting childhood poverty.

In this paper, rural residency status is essentially used as a proxyfor local opportunity structures. The underlying assumption is thatregional economic organization generates opportunities that con­strain individuals' choices and income potentials, thereby structur­ing socioeconomic well-being in ways that cannot be reduced toindividual characteristics (Lobao 1993). Multilevel models are ap­propriate (Tickamyer and Bokemeier 1993).

Analyses of poverty or, more generally family income, necessarilydirect attention to regional labor markets. If labor markets are reallythe mediating variable of interest, they can be measured directly byincorporating more specific and sensitive ecological or context vari­ables. The analyses presented here suggest that this is a potentiallyfruitful direction for further inquiry. Several national data sets, in­cluding the National Longitudinal Survey of Youth and the PanelStudy of Income Dynamics, contain geocode files with data at dif-

Page 16: Is Rural Residency a Risk Factor for Childhood Poverty?

Childhood Poverty - Garrett et al. 81

ferent levels of geographic disaggregation. Other national data sets,including the Current Population Survey and the Public Use Micro­data Samples from the U.S. Bureau of the Census can be aggregatedto characterize regions. Some national data sets may lend themselvesto the study of childhood poverty in a refined ecological context.

This study is limited by the data set analyzed. The NLSY childrenare not representative of their age mates in the United States. Con­sequently, one cannot generalize from this sample to the populationof U.S. children. Nor can one anticipate that research on othersamples would replicate the magnitude of the effects reported here.Nevertheless, the results suggest that the children's residency pat­terns are worthy of consideration in the analysis of their povertyexperiences.

References

Adams, Terry K., and Greg]. Duncan1990 Long-Term Poverty in Nonmetropolitan Areas. Ann Arbor, Ml: University

of Michigan, Survey Research Center.Aldrich, John H., and Forrest D. Nelson

1984 Linear Probability, Logit, and Probit Models. Newbury Park, CA: Sage.Baker, Paula C., and Frank L. Mott

1989 NLS Child Handbook, 1989. Columbus, OH: Ohio State University, Centerfor Human Resource Research.

Bane, Mary Jo, and David T. Ellwood1986 "Slipping into and out of poverty: the dynamics of spells." Journal of Hu­

man Resources 21:1-23.Center for Human Resource Research

1992 NLS Users' Guide 1992. Columbus, OH: Ohio State University.DeMaris, Alfred

1990 "Interpreting logistic regression results: a critical commentary." Journal ofMarriage and the Family 52:171-77.

Devine,Joel A., Mark Plunkett, and James D. Wright1992 "The chronicity of poverty: evidence from the PSID, 1968-1987." Social

Forces 70:787-812.Duncan, Cynthia M., and Ann R. Tickamyer

1988 "Poverty research and policy for rural America." American Sociologist 19:243-59.

Duncan, Greg J., and Willard Rodgers1988 "Longitudinal aspects of childhood poverty." Journal of Marriage and the

Family 50:1007-21.1991 "Has children's poverty become more persistent?" American Sociological

Review 56:538-50.Eggebeen, David j., and Daniel T. Lichter

1991 "Race, family structure, and changing poverty among American children."American Sociological Review 56:801-17.

Fitchen,Janet M.1981 Poverty in Rural America: A Case Study. Boulder, CO: Westview Press.1991 Endangered Spaces, Enduring Places: Change, Identity, and Survival in Ru­

ral America. Boulder, CO: Westview Press.

Page 17: Is Rural Residency a Risk Factor for Childhood Poverty?

82 Rural Sociology, Vol. 59, No.1, Spring 1994

Garrett, Patricia, and Naurine Lennox1993 "Rural families and children in poverty," Pp. 230-58 in Rural Sociological

Society Task Force on Persistent Rural Poverty (eds.), Persistent Poverty inRural America. Boulder, CO: Westview Press.

Garrett, Patricia, DeeAnn Wenk, and Sally Lubeck1990 "Working around childbirth: comparative and empirical perspectives on pa­

rental-leave policy." Child Welfare 69:401-13.Hoppe, Robert

1993 "Poverty in rural America: trends and demographic characteristics." Pp. 20­38 in Rural Sociological Society Task Force on Persistent Rural Poverty(eds.), Persistent Poverty in Rural America. Boulder, CO: 'WestviewPress.

Hosmer, David W., and Stanley Lemeshow1989 Applied Logistic Regression. New York:John Wiley.

Jensen, Leif1989 "Rural-urban differences in the utilization and ameliorative effects of wel­

fare programs." Pp. 25-39 in H. Rodgers and G. Weiher (eds.), Rural Pov­erty: Special Causes and Policy Reforms. New York: Greenwood Press.

Jensen, Leif, and Marta Tienda1989 "Nonmetropolitan minority families in the United States: trends in racial

and ethnic economic stratification, 1959-1986." Rural Sociology 54:509-32.Korbin, Jill E.

1992 "Child poverty in the United States." American Behavioral Scientist 36:213­19.

Kraly, Ellen P., and Charles Hirschman1990 "Racial and ethnic inequality among children in the United States: 1940

and 1950." Social Forces 69:33-51.Lichter, Daniel T., and David]. Eggebeen

1992 "Child poverty and the changing rural family." Rural Sociology 57:151-72.Lichter, Daniel T., Lionel J. Beaulieu, Jill L. Findeis, and Ruy Teixeira

1993a "Human capital, labor supply and poverty in rural America." Pp. 39-67 inRural Sociological Society Task Force on Persistent Rural Poverty (eds.),Persistent Poverty in Rural America. Boulder, CO: 'WestviewPress.

Lichte,', Daniel, Gretchen T. Cornwell, and David]. Eggebeen1993b "Harvesting human capital: family structure and education among rural

youth." Rural Sociology 58:53-75.Lobao, Linda

1993 "Renewed significance of space in social research: implications for labormarket studies." Pp. 11-32 in]. Singelmann and F. Deseran (eds.), Inequal­ities in Labor Market Areas. Boulder, CO: 'WestviewPress.

Martinez-Brawley, Emilia E., and Joan Blundall1989 "Farm families' preferences toward the personal social services." Social

Work 34:513-22.Mcl.aughlin, Diane K., and Carolyn Sachs

1988 "Poverty in female-headed households: residential differences." Rural So­ciology 53:287-306.

Morrissey, Elizabeth S.1991 International poverty in metro and nonmetro areas. Paper presented at the

National Institute on Social Work and Human Services in Rural Areas, Nac­ogdoches, TX.

O'Hare, William P.1988 The Rise of Poverty in Rural America. Washington, DC: Population Refer­

ence Bureau, Report 15.Orshansky, Mollie

1965 "Counting the poor: another look at the poverty profile." Social SecurityBulletin 28:3-29.

Page 18: Is Rural Residency a Risk Factor for Childhood Poverty?

Childhood Poverty - Garrett et al. 83

Rank, Mark R., and Thomas A. Hirschi1988 "A rural-urban comparison of welfare exits: the importance of population

density." Rural Sociology 53:190-206.Rogers, Carolyn C.

1991 The Economic Well-being of Nonmetro Children. Economic Research Ser­vice, Rural Development Research Report 82. Washington, DC: U.S. De­partment of Agriculture.

Ross, Peggy J., and Elizabeth S. Morrissey1989 "Rural people in poverty: persistent versus temporary poverty." Pp. 59-74

in Proceedings of the National Rural Studies Committee. Eugene, OR: West­ern Rural Development Center;

Rural Sociological Society Task Force on Persistent Rural Poverty (eds.)1993 Persistent Poverty in Rural America. Boulder, CO: Westview Press.

Shah, B. V., B. G. Barnwell, P. N. Hunt, and L. M. LaVange1991 SUDAAN User's Manual. Release 5.50. Research Triangle Park, NC: Re­

search Triangle Institute.Sherman, Arloc

1992 Falling by the Wayside: Children in Rural America. Washington, DC: Chil­dren's Defense Fund.

Singelmann,joachim, and Forrest A. Deseran (eds.)1993 Inequalities in Labor Market Areas. Boulder, CO: Westview Press.

Smccding, Timothy, Barbara Boyle Torrey, and Martin Rein1988 "Patterns of income and poverty: tile economic status of children and the

elderly in eight countries." Pp. 89-119 in J. Palmer, T. Smeeding, and B.Torrey (cds.), The Vulnerable. Washington, DC: Urban Institute Press.

Smith, james P.1989 "Children among the poor." Demography 26:235-48.

Snipp, Matthew, Hayward D. Horton, Leifjensen,joane Nagel, and Refugio Rochin1993 "Persistent rural poverty and racial and ethnic minorities." Pp. 173-99 in

Rural Sociological Society Task Force on Persistent Rural Poverty (eds.),Persistent Poverty in Rural America. Boulder, CO: Westview Press.

Tickamyer, Ann R., and janet Bokemeier1993 "Alternative strategies for labor market analyses: multi-level models oflabor

market inequality," Pp. 49-68 in J. Singelmann and F. Deseran (eds.), In­equalities in Labor Market Areas. Boulder, CO: Westview Press.

Tonnies, Ferdinand1887 Gcmcinschaft und Gesellschaft, Leipzig: Fues's R. Reisland.

U.S. Bureau of the Census1991 Poverty in the United States: 1990. Current Population Reports. Consumer

Income, Series P-60, No. 175. Washington, DC: U.S. Government PrintingOfficc.

U.S. Military Entrance Processing Command1989 Counselor's Manual for tile Armed Services Vocational Aptitude Battery,

Form 14. North Chicago, IL: U.S. Military Entrance Processing Command.Waltman, Gretchen H.

1986 "Main street revisited: social work practice in rural areas." Social Casework67:466-74.

Wcnk, DccAnn, and Constance Hardesty1993 "The effects of rural-to-urban migration on tile poverty status of youth in

the 1980s." Rural Sociology 58:76-92.