The Use of Community Services by Rural Families in Wisconsin
by
Jim Cavaye, Ron Shaffer, & Sandra Wraith1
Staff Paper no. 97.4
October 1997
The Use of Community Services by Rural Families in Wisconsin
The changing United States economy impacts the welfare of rural families and alters policy choices [Deavers & Hoppe, 1991, 1992; Galston & Baehler, 1995; Lobao, 1990; Reich, 1988]. For example, the rising proportion of low wage job opportunities, general wage inequality, the shrinking middle class, and the increasing number of children in poverty affect family employment choices and have major implications for welfare reform [Nightingale & Haveman, 1995; Reich, 1988]. While these changes affect both urban and rural areas, rural families face unique problems and limited options making rural policy choices particularly important [Bokemeier & Garkovich, 1991; Deavers & Hoppe, 1991, 1992; Galston & Beahler, 1995; Lobao, 1990].
The shifting nature of rural communities, and their changing role in the national economy, increases the importance of examining the economic and social wellbeing of rural families. In the 1970's and 1980's, research on rural communities, and rural policy centered on farm families and their struggle with declining farm incomes. However, most rural residents are members of nonfarm families and in the late 1980's researchers recognized that the growing proportion of rural families depending on wage and salary incomes were also experiencing economic distress. In the past decade, nonfarm families have had to manage declining real earnings, rising unemployment and an increasing number of low paying jobs [Kassel & Gibbs, 1996; Kusmin & gibbs, 1996]
Many factors contribute to the declining economic security of rural people. Economists point to increased involuntary parttime employment, loss of real wages and fringe benefits, and shifts in occupations caused by changing technology and international competition. Tickamyer and Duncan [1991] and Bluestone and Harrison [1988] attribute much of the increase in low wage jobs to a rise in service employment. However, Gorham [1992] hypothesizes that it results from the decay of union power, the disintegration of the standard minimum wage and the internal restructuring of wage systems. Regardless, researchers agree that recent changes in rural employment have reduced living standards for the average rural resident.
Families cope with these economic and social changes by making decisions (consciously or not) on a broad spectrum of resource allocation issues, such as adjusting expenditures, selling assets, migration, family size, education, housing, informal economic activities, reliance on family support networks and the allocation of members' time to paid work and unpaid work.
Many families select a combination of coping behaviors to make up an overall economic "survival strategy" [Moen and Wethington, 1992]. In a general sense, the hypothesized elements of the household survival strategy are:
- alter level of involvement in formal economic activities by one or more household members(such as participation in the labor force);
- alter level of informal economic activity to increase income or reduce expenditures(such as the exchange of goods and services for cash or by barter); and
- alter the use of community services (such as local daycare provision, food pantries, jobtraining programs, transportation services, and financial support).
For example, rural families are resorting to a number of creative alternatives, or supplements, to formal wage labor as part of their overall survival strategy [Allen,1991; Duncan, 1992; Fitchen, 1992, lyson & Falk, 1993].
Hence, the formulation of appropriate rural policy, and development of relevant community services for rural families, depend on greater understanding of the choices available to families and how they build and modify a survival strategy.
Government policy has largely overlooked family decision-making. Policy makers have centered community development policy on job creation and education/training. Research has shown that these approaches have merit. However, reviews are mixed regarding their overall application to rural community economic development and their success in terms of raising rural living standards [Summers et al, 1976].
Family Decision Making
How families perceive and respond to change is by no means simple. Figure 1 describes a three stage process where family members incorporate information and resources from within and outside the family and prepare to adopt a single coping behavior. In Stages II and III, the family chooses from a set of options they perceive available which alters the family's wellbeing. In reality the process is less linear and more simultaneous or circular and the choice of one coping behavior may influence the choice of other options.
Moen and Wethington [1992] outline three models that also conceptualize the development of family survival strategies - structuralist, rational choice, and life course.
The structuralist model presents family behaviors as constrained, and to some extent determined, by external factors. For example, structuralists explain family fertility behaviors as responses to economic crises, such as the Great Depression, and economic boom times, such as the postWorld War II period. Recently, structuralists have examined family adaptive strategies as a response to massive economic restructuring and dislocation in the 1980's.2
The structural model assumes that families can partly control their immediate economic activities, and the complexity of family structure, household composition, and relationships among family members affect the process. Moen and Wethington [1992] argue that "social structural forces have impact not only on the adaptations that are possible, but also on which familiesand which individuals within familiesreceive the most benefit from a given strategy". They feel that four social structural systems influence economic opportunities - social status, educational stratification; gender relationships; and age/generational hierarchy.
The rational choice position forms the basis of the New Home Economics (NHE) model. This model assumes householders make a rational choice to allocate resources to maximize the joint utility (satisfaction) of the household subject to constraints on time, income, and the production of home goods [Becker, 1965}. This approach ignores individualism and lumps each family member's preferences into a collective decision unit. New Home Economists assume either perfect "altruism," where household members subordinate their individual preferences for the good of common household goals, or the existence of a benevolent dictator who acts unilaterally in everyone's best interests [Katz, 1992].
IntraHousehold Bargaining theorists improve on the NHE model by recognizing that each member of a household has individual preferences and that household decisions are made Through negotiation based on the relative bargaining power of each member [Bourguignon & Chiappori, 1992; Horney & McElroy, 1981; Manser & Brown, 1980; Thomas, 1990]. This bargaining power is determined by each individual's potential satisfaction if he or she were to leave the household.
"Life course" models combine aspects of structural and rational choice theories. They place family and individual strategies in a broad context of shifting opportunities and constraints over time. Social, institutional, and economic transformations change family resources and needs (or aspirations). They also prompt families to adopt patterns of behavior designed to reconcile needs and resources [Moen & Wethington, 1992]. These models realize the strength of the concept of family adaptive strategies by bridging the gap between social structures, social change, and individual lives. Family adaptive strategies are not just dependent variables to be explained by external forces and family interests, but independent and intervening variables in models of how family strategies facilitate or hinder changes.
Regardlesss of how these models conceptualize family decision-making processes, they all emphasize that survival strategies are formed through interactions of individuals, families and communities (Figure 2). Families and individuals are not atomistic units, but involved in a complex web of interactions. As Unger and Sussman [1990, p.1] argue;
"Understanding the life situations of families, identifying their problems, and developing new solutions require an ecological framework that recognizes that families are embedded in a matrix of relationships within community and larger social systems."
Households represents a resource pooling unit where the actions of individuals affect other member's decisions. Individuals work together to ensure the survival of the family unit [Tickamyer, et al., 1993].
Various approaches focus on the individual, family and community aspects shown in Figure 2. Traditional human capital models focus on the characteristics of the individual in determining the rate of labor force participation. An individual's willingness to participate in the labor force is largely determined by the market wage available to that particular individual. The individual's own personal characteristics and qualifications play a major role in determining the level of the wage offered. These personal characteristics and qualifications are most often measured by observable attributes such as education level, years of work experience, age, special skills and, more recently, sex and race [Bryant, 1990].
Other conventional economic approaches have ignored family and community dimensions of an individual's decision to enter, or increase participation in the labor force. Studies have addressed the relationship of child care costs or marital separation and women's labor force participation [Johnson & Skinner, 1986; Mason & Kuhlthau, 1992]. However, most research of the labor force participation decisions of both men and women generally focuses on the effect of public assistance and various demographic characteristics such as gender, race, education, fertility, and previous work experience.
Community Resources and Family Strategies
Other work focuses on the interaction between family and community resources in determining survival strategies. In one of the few studies to address the community dimension of family decision making, Allen [1991] examines how family income generation decisions influence community structure and cohesion.
There is a weight of research that addresses the converse relationship - the impact of community resources on family decisions. In the late 1980's, social scientists began to study anew the relationships between economic distress from economic restructuring and family outcomes. The topic gained credence and was covered in the Journal of Marriage and Family decade review for the first time in 1990. In her review, Voydanoff [1990] discussed the relevance of studying families as economic units and the ways in which families experience economic distress.
Sociologists and anthropologists have long understood the importance of community in shaping the decision making processes of its residents. Duncan and Lamborghini [1994] point out that communities provide an environment for the shaping of aspirations and expectations of escaping or avoiding poverty. The authors assert that individuals' opportunities to overcome obstacles and change their lives depend greatly on the tool kits they develop and the resources offered by the community. They feel that "culture [is] a tool kit from which individuals draw to solve problems they encounter" [p. 439].
Studies have focused on the social aspects of community effects on family choices and prospects, such as positive role models and attitudes about the future. Duncan and Lamborghini [1994] argue that rural communities suffer from the 'resource side of isolation' entailing a lack of sufficient job opportunities, fewer contacts for obtaining jobs and less money and influence over public goods. They also emphasize the 'social side of isolation' involving the lack of sufficient numbers of positive role models, and the existence of destructive peer influences which may suppress attitudes of local residence toward the future.
A growing number of studies recognize the value of community social capital in shaping family economic choices. Duncan and Lamborghini [1994] define social capital as the "features of social organization, such as networks, norms, and trust that facilitate coordination and cooperation for mutual benefit" [p. 438]. This broad definition easily accommodates the myriad community institutions which provide services to local residents such as local service clubs, food pantries, church donations, and volunteer job counseling and training programs.
Community factors affect many of the choices individuals make or even consider. There is some research that discusses community and neighborhood influences on the perceived quality of life and on the levels of involvement in community development activities. O'Brien et al. [1989] demonstrate how social support systems affect individuals' perceptions of their life satisfaction derived from their urban neighborhood. Warren [1981, p. 61] notes that differences in social support services among neighborhoods are particularly crucial when differences among individuals' abilities to establish other supporting links among geographically dispersed sources are considered. This is a phenomenon particularly relevant to rural families. Sorter [1987] finds access to selected social services is affected not only by physical distance, but such behavioral dimensions as supportive relationships, peer group relationships, and social relationships. Abrahams [1992] also notes the importance of such community factors as social structure, economic structure, political structure and service capacity in judging the impact of a social development model of community development.
Finally, community resources are intimately connected to family and economic life. A good example of the interplay of informal activities, family networks, gender roles, and community resources can be found in Levitan and Feldman's [1991] study of the interhousehold informal economic exchanges in a rural community in New York. This research documents the types and prevalence of nonmonetary exchange and the relationship of these behaviors to household structure and the rurality of the community context.
Levitan and Feldman show that families engage in nonmonetary exchange for a variety of reasons, including social ones such as neighborliness. They note that rural areas are more conducive than urban areas to certain types of nonmonetary exchange activities, especially those that utilize natural resources. Spatial considerations also play a role; a sparsely distributed population can make the provision of community or market services difficult or unprofitable. In these cases, interhousehold exchanges from social networks may serve as a lifeline to ensure wellbeing. This has implications for rural development policy in that the cumulative and thus communitywide value of informal arrangements needs to be taken into account; those without access to these networks and their services are especially vulnerable [Levitan & Feldman, 1991, p. 168]. They conclude that:
"Nonmarket activities of the informal sector may be sufficient to decouple the ranking of the quality of life in a community from a positioning based upon formal sector income and employment data. Support generated within social networks may be the catalyst which provides the time, economic relief or social nourishment that enables the household to function."
Complexity
It's clear that the economic wellbeing of families results from a complex process of utilizing community and interpersonal resources, allocating family labor to formal and informal economic activities, and making key decisions about a variety of conditions from education, to living arrangements, to financial management. Individuals make decisions, such as labor force participation, within the context of family needs and capabilities and a complex web of community opportunities and constraints.
Policy approaches are shifting from interpreting economic choices as simple individual decisions to more complex considerations of family needs and considerations. Questions that we now realize are important include: How many children are in the family? What are the ages of these children? How many working individuals are in the family? What is the overall family income? In other words, students of the labor force behavior are now paying closer attention to the characteristics of the individual's other family members. What makes families choose their particular family work pattern, the mixture of formal and informal work among its members?
To help develop better policy approaches from this complexity, our study aimed to assess and quantify the relative impact of different family characteristics (such as income, employment status, number and age of children) on the use of community services by families (such as health care services, transportation services, child care, public assistance). The fundamental hypothesis is that particular household characteristics influence the level of use of particular community services to different degrees.
THE MODEL
The model can be represented in the followng fashion (see also Table 1):
Y = f (HH, ECON, EMPL, ECOPT, PUB)
Where
Y = use/nonuse of selected community services
HH = household characteristics includung age, size, children, housing;
ECON = income, financial needs, ...
EMPL = work status, experience, ...
ECOPT = perceived economic options if lost main income
PUB = use of other forms of public assistance
The following section describes the survey used to collect data to test this model.
THE SAMPLE
Between November 1995 and April 1996, the University of Wisconsin Letters & Sciences Survey Center (LSSC) conducted a 30 minute telephone survey.3 The survey was a random sample of households4 in telephone exchange areas characterized as nonmetropolitan. Our target was to contact rural (nonfarm and farm) families and collect information from them regarding what they were doing to support themselves, i.e., survival strategies.
There were 1611 completed and useable surveys representing a completion rate of 55.9% from the households contacted in 52 nonmetro counties. The LSSC contacted the same household repeatedly (up to 20 times) until a completed interview or refusal occurred. We solicited information from households that were in rural areas, were not adult sibling or room mate households, and were not male single head of household.5
The survey instrument elicited information from the respondent on household composition, formal wage work, selfemployment, informal economic activities, networks, and community social services used. Each respondent was asked the same questions (except for skip patterns) from a survey instrument. Questions required a yes/no response (e.g. employed or not), a choice from a series of options, or an open answer soliciting specific information (e.g. number of persons in a household).
The survey instrument was built from our review of the literature (conceptual and empirical) and from focus group interviews conducted in January to March of 1995 [Tigges, et al, 1995]. The survey instrument was pretested in interviews with 25 respondents chosen at random from the target population. Several questions were modified to improve phrasing, and to ensure that they required mutually exclusive responses.6
STATISTICAL ANALYSIS
The study examined only the use of selected community services. Figure 3 shows the hypothesized household characteristics expected to influence the use of different forms of assistance by households.7
The independent variables used to measure each household characteristic in Figure 3 were based on prior empirical and conceptual work reviewed earlier. The survey elicited data on a respondent's use of numerous community services, and their involvement in both formal and informal economic activities. The large array of proxy measures of the relationship among household characteristics and the use of community service in the service was reduced to a manageable number.
The 'full' empirical model for each dependent variable - the use of different forms of assistance is displayed in Table 1. This reflects an initial effort to uncover statistically significant relationships within this sample, although causal relationships were not identified. The full range of possibilities came from the literature, but response to many specific questions were insufficient to justify further analysis, and several of the questions gave insight to similar phenomena. Thus the 'winnowing' to improve testing efficiency.
The winnowing process was Cramer's V statistic, a modified chisquare statistic which measures the interaction between variables on a scale of 0 to 1 [Everritt, 1977; Feinberg,1977]. A threshold of significance (p < 0.05) was used to select those interactions between variables to be included.
HOUSEHOLD CHARACTERISTICS USE OF ASSISTANCE
Nature of Household
Employment
Education
Economic Options
Other Socio-Economic Characteristics
Assistance with Costs
Housing Assistance
Public Health Services
Supplemental Income
Child Care
Figure 3. Household Characteristics Used to Predict the Use of Assistance by Households.
| HOUSEHOLD CHARACTERISTICS |  |
USE OF ASSISTANCE |
| Nature of Household | Assistance with Costs | |
| Employment | Housing Assistance | |
| Education | Public Health Services | |
| Economic Options | Supplemental Income | |
| Other Socio-Economic Characteristics | Child Care |
Table 1
Empirical Model for the Households use of Community Assistance
| Hypothesized independent variables | Hypothesized use of assistance | Cramer's V interaction* |
|
|
||
| Nature of household | ||
| Household type | - or + | a b c d |
| Household size | + | c d |
| Children in household | + | a c d |
| Household income | - | a b c d |
| Shared housing with others | + | a b c d |
| Employment | ||
| At least one adult employed | - | a b c d |
| Type of employment (self vs other) | - or + | c |
| Years of employment | - | a b c d |
| Job shift (night, day) | - or + | |
| Job works same shift | - or + | a c d |
| Job location (home or other) | - | a b c d |
| Works for insurance | + | a |
| Education | ||
| Educational attainment | - | a b |
| Acted to improve education | - | |
| Economic options if lost main income (e.g. borrow money, live off savings) | - or + | a b |
| Public Assistance | ||
| Use supplemental income | - or + | a c d |
| Use housing assistance | - or + | b c d |
| Use public health services | - or + | a b c |
| Use assistance with cost | - or + | a b d |
| Other Socio-economic Characteristics | ||
| Age of respondent | - or + | a b c d |
| Feel financial needs are not met | + | a b c d |
| Home owned | - | a b c d |
| Use public transport | + | a b c d |
| Have insurance | - | a b c d |
| * indicates a Cramer's V of at least 0.1 (range 0-1) for internation with each of the variables - use of housing assistance (a), supplemental income (b), assistance with costs (c), and the use of public health services (d). | ||
Community Service Measures:8
Five types of community services public health services, child care, community assistance with household costs, supplemental income, and housing assistance were explored in greater detail. The use/nonuse of these become the dependent variables in the following regression analysis, and all were dichotomous, i.e., either used or did not use.
Whether through lack of availability of a specific community service or sample bias (see weighing) the frequency of use of the numerous specific community services proved insufficient for detailed statistical analysis. However, the detailed information requested represents components of more generalized community services. The composite community service does have sufficient observations for further analysis (Table 2). For each generalized community service the use of at least one of the component forms of community assistance was deemed to be use of that community service. For example, use of AFDC only, and use of both AFDC and food stamps, were coded as the use of the supplemental income community service. A type of community assistance was deemed to not be used if none of the component forms of community assistance were used. The use of child care was a single uncombined variable analyzed only for households with children under 13 years of age.
Hypothesized Influences on the Use of Community Assistance: Table 1 contains the hypothesized forces and the nature of the influence. These are the independent variables in the regression analysis.These influences were described by three forms of data. First, variables, such as, household income were measured directly. Second, variables measured by dichotomous yes/no data (e.g. respondent was employed or not) were coded as 0 or 1. Third, variables with more than two classes of nominal data, such as, options if the household lost its main income, were coded as categorical dummy variables.
Weighing. Based on income, the sample gained from the telephone survey was not representative of households in nonmetropolitan counties in Wisconsin [U.S. Bureau of the Census, 1990]. Income for nonmetropolitan households in the state was calculated by adjusting 1990 census data to 1996 dollars using a CPI increase of 13.9% since 1990 [U.S. Department of Labor, (19901996)]. The survey sample included a disproportionately high number of households earning greater than median income, and low income households were underrepresented (Figure 4).
Because use of public community assistance was likely to be strongly influenced by income, the sample was adjusted to approximate the distribution of household income for nonmetropolitan counties. This was done by weighing the data based on the proportion of households in the sample, and in nonmetropolitan counties, for each household income category.
For each income bracket in Figure 4, the percentage of households in nonmetropolitan counties was divided by the percentage of households in the sample. This gave a weight by which the number of households in each income bracket in the sample was multiplied (Table 3). This meant that data from a low income household in the sample was included in data analysis more frequently, and data from a high income household was included less frequently.
Table 2
Community Services
|
Combined Variable |
Households using at least one form of assistance |
Component Forms of Community Assistance |
Households Using each form of assistance |
|
|
|||
| Public health | 351 | Free immunization Low cost clinics In-home nursing Healthy Start Other public health service |
197 31 9 112 65 |
| Assistance with costs | 191 | Help with food Help with clothing Help with paying bills |
118 73 67 |
| Supplemental income | 570 | Social security or SSI Unemployment compensation Workers compensation AFDC Food Stamps or WIC General assistance |
197 255 78 78 186 3 |
| Housing assistance | 177 | Heating or cooling assistance Public housing |
116 68 |
Figure 4: Differences in Sample and Nonmetro Income Distributions
Table 3. The weighting factors to adjust the income distribution of the sample to that of households in non-metropolitan counties in Wisconsin.
Household Income Non Metro Sample Households Weighting factor
(1996 dollars) Households %
%
< $13,900 16.6 4.4 3.773
$13,900 - $22,779 22.1 11.5 1.922
$22,780 - $34,169 20.2 18.1 1.116
$34,170 - $45,559 16.4 27.6 0.594
$45,560 - $56,949 10.8 15.0 0.720
$56,950 - $68,339 6.0 10.2 0.588
$68,340 - $85,424 4.1 66 0.621
$85,425 - $113,899 2.2 3.0 0.733
$113,900 - $142,374 0.8 1.2 0.667
$142,375 - $170,849 0.3 0.6 0.500
> $170,849 0.6 1.7 0.353
Logistic Regression
Data were analyzed using logistic regression. Logistic regression was chosen because the dependent variables were dichotomous, i.e., people either did or did not use supplemental income. Furthermore, the aim was to use household characteristics to predict whether forms of community assistance would be used or not. This suits logistic regression where independent variables are used to estimate the probability of an event occurring (in this case, the use of community assistance) [Hosmer & Lemeshow, 1989].
Other statistical techniques did not suit the data or the objectives of the study as well as logistic regression. Discriminant analysis allows prediction of group membership with a dichotomous dependent variable. Predicting the classification of the population into groups that did or did not use forms of assistance would be a way of inferring the influence of household characteristics on the use of community assistance. However, for the predicted classification to be optimal, discriminant analysis assumes that independent variables are distributed normally, and that the variancecovariance matrices of the two groups are equal [Klecka, 1980; Lachenbruch, 1975]. Neither of these represented the current data set. Logistic regression requires far fewer assumptions [Hosmer & Lemeshow, 1989; Rao, 1973].
Multiple least squares regression is not appropriate because a dichotomous dependent variable violates the assumptions for hypothesis testing. For example, errors cannot be normally distributed.
Factor analysis would have allowed the large number of independent variables to be condensed into several factors that would explain most of the variation in the dependent variable. However, at least interval data is required for factor analysis to explain accurately variation [Kim & Mueller, 1978; Rummel, 1970]. The dichotomous nature of the dependent variables (and many of the independent variables) limited the variation of the data for any analytical technique chosen, but factor analysis suffered most from this limited variation.
Interpreting Logistic Regression. In multiple least squares regression the coefficient of each independent variable measures the extent of change in the dependent variable for every unit change in that independent variable, holding all others constant. Logistic regression is similar but uses a logarithmic function.
Odds = Probability (event)/Probability (no event) = e B0 + B1X1 + .+ BnXn
The coefficient Bi is the change in the log odds of an event occurring that results from a one unit increase in the ith independent variable, holding all others constant. Hence, e raised to the exponent Bi, (described as (exp)B) is the factor by which the odds change when the ith independent variable increases by one unit, holding all others variables constant.
If the coefficient Bi is positive, the probability of the dependent event occurring increases. In this case a one unit increase in the ith independent variable increases the odds of households using community assistance. If the coefficient Bi is negative, the probability of the dependent event occurring decreases with an increase in the respective independent variable.
For example, in Table 4 with regard to the use of child care, a categorical ("0 or 1") independent variable describing whether households are covered by insurance or not has a coefficient of 1.613 and an exp (B) of 5.020. This means that a household with insurance has 5.02 times the odds of using child care than a household without insurance. In the same table, the variable describing the age of respondent has a coefficient of 0.191 and an exp (B) of 0.826. Hence, for every additional year of age, a respondent's odds of using child care are multiplied by 0.826 thus is decreased.
Goodness of Fit. Comparing the predicted use (or non use) of forms of assistance by households, with their actual use (or non use), provides a measure of the goodness of fit of the logistic regression model. Classification tables were used to determine the percentage of households that were correctly predicted to use, and not to use, different forms of community assistance. The accuracy of predicting both the use, and non use, of community assistance provided a dual measure of the goodness of fit of the model. A robust model should be able to predict both accurately. The prediction of the use of community assistance was always less accurate than the prediction of nonuse, despite income weighing, because there were far fewer cases of use than nonuse. The overall prediction accuracy of the equations varied from 74.0% (supplemental income) to 90.5% (housing assistance). The prediction of the use of community assistance varied from 32.0% (use of housing assistance) to 51.8% (use of supplemental income). The prediction of nonuse varied from 87.3% (supplemental income) to 98.0% (housing assistance).
Child Care
The use of child care by households with a child und 13 was influenced by a combination of household structure and economic factors (Table 4). Household structure affected household arrangements for caring for children. As household size increased, households were less likely to use child care presumably because older children could care for younger siblings or one of the parents remained home. Households with older respondents (probably with older children) were less likely to use child care than households where respondents were younger.
Economically secure households tended to use child care. Better educated, higher income people with longer employment histories, and who would borrow money in a crisis were considerably more likely to use child care. These proxy situations where both partners were employed, in better than minimum wage jobs. Hence, apart from household structure, the economic ability to pay for child care, was a second major determinant of its use by households. Supporting evidence was single parenthood. Single mothers were 4.2 times more likely than couples with children to use child care, holding others variables constant. The economic factors associated with the use of child care suggest that many single mothers chose to remain employed in relatively high quality jobs. The results challenge the image of a low income, poorly skilled single mother using child care to attend a lowpaying job.
The use of child care was not significantly (p<0.10) associated with use of community assistance. This is not surprising since community assistance, particularly supplemental income, allows unemployed adults to remain with children.
One unusual result was the influence of insurance. Households covered by any form of insurance (e.g. health, home, life) were over five times more likely to use child care than households without insurance. This may reflect the use of child care by more affluent households since lower income households are more likely to be unprotected or selfinsured.
Table 4
Use of Child Care by Households with Children <13. ( n = 651)
| Variable | Coefficient | S.E. | exp (B) | |||
|
|
||||||
| STATISTICALLY SIGNIFICANT (p<0.10) | ||||||
| Covered by any form of insurance | 1.613 | 0.377 | 5.020 | |||
| Single mothers with children < 18 compared to couples with children < 18 | 1.453 | 0.554 | 4.275 | |||
| If lost main income: Borrow money compared to living off savings | 0.381 | 0.167 | 1.464 | |||
| College degree compared to high school diploma or less | 0.361 | 0.179 | 1.435 | |||
| Household income ($/10000) | 0.100 | 0.045 | 1.106 | |||
| Respondent years of employment | 0.084 | 0.029 | 1.088 | |||
| Age of respondent (years) | -0.191 | 0.032 | 0.826 | |||
| Household size (persons) | -0.323 | 0.106 | 0.724 | |||
| Constant |
4.062 |
1.121 |
||||
| NOT STATISTICALLY SIGNIFICANT | ||||||
| Household Characteristics | ||||||
| Other household types compared to couples living alone | -1.071 | 1.052 | 0.343 | |||
| Shared housing with others | 0.171 | 0.252 | 1.186 | |||
| Employment | ||||||
| At least one adult in the household is employed | -0.210 | 0.590 | 0.811 | |||
| At least one adult in the household is self employed | -0.208 | 0.386 | 0.812 | |||
| At least one adult in the household works at home | -0.736 | 0.395 | 0.479 | |||
| Works for insurance | 0.148 | 0.265 | 1.159 | |||
| Education | ||||||
| Vocat.degree/college experience compared to high school diploma or less | -0.042 | 0.152 | 0.959 | |||
| At least one adult in the household continued education in last ten years | -0.266 | 0.207 | 0.766 | |||
| Economic Options | ||||||
| If lost main income: Sell assets compared to living off savings | -0.099 | 0.180 | 0.906 | |||
| If lost main income: Use govt. assistance compared to living off savings | -0.114 | 0.222 | 0.893 | |||
| Public Assistance | ||||||
| Use at least one form of supplemental income | 0.071 | 0.236 | 1.074 | |||
| Use at least one form of assistance with costs | -0.286 | 0.315 | 0.751 | |||
| Use at least one form of public health service | 0.379 | 0.224 | 1.461 | |||
| Use at least one form of housing assistance | -0.230 | 0.336 | 0.795 | |||
| Other Socio-economic Characteristics | ||||||
| Used public transport | 0.126 | 0.249 | 1.134 | |||
| Home owned | 0.358 | 0.261 | 1.430 | |||
| Feel household financial needs are not met |
0.348 |
0.268 |
1.416 |
|||
| Fit of Model | n | % of cases | ||||
| Correctly predicted non-use of child care | 429 | 88.2 | ||||
| Correctly predicted use of child care | 228 | 57.6 | ||||
| Overall prediction accuracy | 77.6 | |||||
|
|
||||||
Supplemental Income
The use of supplemental income was determined almost solely by economic status (Table 5). If a household was already using other types of community assistance, or felt that their financial needs were not met, or considered government assistance as the first option if they lost their main income, they were considerably more likely to use at least one form of supplemental income assistance. There was also the greatest interaction among community services and the use of supplemental income.
Conversely, greater economic status reduced the likelihood of a household using supplemental income. If at least one adult in the household was employed, a respondent had a college education vs high school or less, or was self employed (as opposed to working for someone else), the household was considerably less likely to receive supplemental income (the odds were multiplied by 0.061, 0.659, 0.579 respectively). Every $10,000 increase in income and each additional year of employment also reduced the odds of a households use of supplemental income (exp (B) = 0.855 and 0.976 respectively).
Public Health Services
Both household structure and economic status (Table 6) influenced the use of public health services. The presence of children made households most likely to use at least one public health service (exp (B) 3.064). Couples with children under 18 years were over 1.5 times as likely to use public health services than couples without children at home.
The use of other types of community assistance community assistance with costs and supplemental income increased the odds of the use of public health services by 2.345 and 1.768 times compared to households that did not use these forms of community assistance.
Conversely, households with higher economic status made less use of public health services. Having at least one adult employed, or having a vocational degree or college experience compared to a high school diploma or less reduced the odds most (exp (B) = 0.353 and 0.723 respectively). Each $10,000 increase in income reduced the odds, but to a lesser extent (exp (B) = 0.893).
The specific reasons why households with at least one adult working at home have almost double the likelihood of using public health services is unclear. It may be due to home work allowing greater flexibility to attend public health clinics, or visiting services may be utilized more if household members are at home during the day, or public health services represent the households implementation of their selfinsurance.
Table 5
Use of Supplemental Income ( n = 1201)
| Variable | Coefficient | S.E. | exp (B) | |||
|
|
||||||
| STATISTICALLY SIGNIFICANT (p<0.10) | ||||||
| Use at least one form of housing assistance | 0.942 | 0.251 | 2.566 | |||
| Feel household financial needs are not met | 0.731 | 0.172 | 2.077 | |||
| Use at least one form of public health service | 0.575 | 0.180 | 1.777 | |||
| Use at least one form of community assistance worth costs | 0.515 | 0.241 | 1.673 | |||
| If lost main income: Use govt. assistance compared to living off savings | 0.414 | 0.142 | 1.514 | |||
| Age of respondent (years) | 0.029 | 0.013 | 1.030 | |||
| Respondent years of employment | -0.024 | 0.012 | 0.976 | |||
| Household income ($/1000) | -0.157 | 0.043 | 0.855 | |||
| College degree compared to high school diploma or less | -0.417 | 0.145 | 0.659 | |||
| At least one adult in the household is self employed | -0.546 | 0.217 | 0.579 | |||
| At least one adult in the household is employed |
-2.794 |
0.764 |
0.061 |
|||
| NOT STATISTICALLY SIGNIFICANT | ||||||
| Household Characteristics | ||||||
| Couples with children < 18 compared to couples living alone | 0.001 | 0.161 | 1.001 | |||
| Single mothers with children < 18 compared to couples living alone | -0.326 | 0.197 | 0.722 | |||
| Other household types compared to couples living alone | 0.340 | 0.237 | 1.406 | |||
| Household size (persons) | 0.114 | 0.077 | 1.012 | |||
| Children < 13 years in household | -0.113 | 0.231 | 0.894 | |||
| Shared housing with others | 0.105 | 0.197 | 1.111 | |||
| Education | ||||||
| Vocat. degree/ college experience compared to high school diploma or less | 0.074 | 0.116 | 1.077 | |||
| At least one adult in the household continued education in last ten years | -0.126 | 0.146 | 0.881 | |||
| Economic Options | ||||||
| If lost main income: Sell assets compared to living off savings | -0.084 | 0.123 | 0.919 | |||
| If lost main income: Borrow money compared to living off savings | -0.064 | 0.125 | 0.938 | |||
| Other Socio-economic Characteristics | ||||||
| Home owned | -0.079 | 0.187 | 0.924 | |||
| Used public transport | 0.176 | 0.190 | 0.890 | |||
| Covered by any form of insurance | 0.019 | 0.258 | 1.019 | |||
| Works for insurance | 0.160 | 0.184 | 1.174 | |||
| Constant |
1.201 |
0.929 |
||||
| Fit of Model | n | % of cases | ||||
| Correctly predicted non-use of supplemental income | 725 | 87.3 | ||||
| Correctly predicted use of supplemental income | 431 | 51.8 | ||||
| Overall prediction accuracy | 74.0 | |||||
|
|
||||||
Table 6
Use of Public Health Services ( n = 1196)
| Variable | Coefficient | S.E. | exp (B) | |||
|
|
||||||
| STATISTICALLY SIGNIFICANT (p<0.10) | ||||||
| Children < 13 years in household | 1.120 | 0.297 | 3.064 | |||
| Use at least one form of assistance with costs | 0.852 | 0.237 | 2.345 | |||
| At least one adult in the household works at home | 0.692 | 0.278 | 1.998 | |||
| Use at least one form of supplemental income | 0.570 | 0.182 | 1.768 | |||
| Couples with children < 18 compared to couples living alone | 0.428 | 0.205 | 1.534 | |||
| Household income ($/10000) | -0.114 | 0.046 | 0.893 | |||
| Vocat. degree/ college experience compared to high school diploma or less | -0.325 | 0.133 | 0.723 | |||
| At least one adult in the household is employed |
-0.157 |
0.043 |
0.855 |
|||
| NOT STATISTICALLY SIGNIFICANT | ||||||
| Household Characteristics | ||||||
| Single mothers with children < 18 compared to couples living alone | 0.141 | 0.237 | 1.152 | |||
| Other household types compared to couples living alone | 0.273 | 0.359 | 1.314 | |||
| Household size (persons) | -0.011 | 0.077 | 0.989 | |||
| Shared housing with others | 0.376 | 0.206 | 1.456 | |||
| Age of respondent (years) | -0.007 | 0.017 | 0.994 | |||
| Employment | ||||||
| At least one adult in the household is self employed | 0.106 | 0.288 | 1.111 | |||
| Respondent years of employment | -0.020 | 0.015 | 0.981 | |||
| Works for insurance | 0.115 | 0.220 | 1.122 | |||
| Education | ||||||
| College degree compared to high school diploma or less | 0.296 | 0.153 | 1.344 | |||
| At least one adult in the household continued education in last ten years | -0.041 | 0.169 | 0.960 | |||
| Economic Options | ||||||
| If lost main income: Sell assets compared to living off savings | 0.237 | 0.144 | 1.268 | |||
| If lost main income: Use govt. assistance compared to living off savings | 0.115 | 0.163 | 1.122 | |||
| Public Assistance | ||||||
| Use at least one form of housing assistance | 0.384 | 0.245 | 1.468 | |||
| Other Socio-economic Characteristics | ||||||
| Used public transport | 0.292 | 0.200 | 1.339 | |||
| Feel household financial needs are not met | -0.027 | 0.208 | 0.973 | |||
| Covered by any form of insurance | 0.152 | 0.264 | 1.165 | |||
| Home owned | -0.221 | 0.203 | 0.802 | |||
| Constant |
-0.809 |
0.293 |
||||
| Fit of Model | n | % of cases | ||||
| Correctly predicted non-use of public health | 877 | 94.9 | ||||
| Correctly predicted use of public health | 272 | 39.4 | ||||
| Overall prediction accuracy | 81.7 | |||||
|
|
||||||
Housing Assistance
Economic need was important to the use of housing assistance (Table 7). The use of assistance with costs (exp (B) 6.303) and use of supplemental income (exp (B) 2.636) were the factors that most increased the likelihood of the use of housing assistance. Single mother households were also over twice as likely to use housing assistance than couples living alone. Every $10,000 dollar increase in income (exp (B) 0.599), or at least one adult being employed (exp (B) 0.264), or the household owning their home (exp (B) 0.325) were increases in economic status that reduced the likelihood of the use of housing assistance.
However, some contradictory results suggest that housing assistance may not depend simply on economic adversity. Households with insurance, and households where at least one adult had continued their education were 2.183 and 1.702 times as likely to use housing assistance than households without insurance, and where no adult have improved their education, respectively. Also households where one member works just for insurance had less than half the chance (exp (B) 0.429) of using housing assistance than households where members do not work for insurance.
Assistance with Costs
A household's use of at least one form of community assistance with costs was associated with a contradictory mix of factors suggesting both low and reasonable economic status (Table 8). Low economic status factors considerably increased the likelihood of the use of community assistance with costs. These factors included the use of housing assistance (exp (B) 6.231) and public health services (exp (B) 2.186), feeling that financial needs are not met (exp (B) 2.725) and being a single mother compared to couples living alone (exp (B) 2.445).
However, criteria that pointed to reasonable economic status were also associated with use of at least one form of community assistance with costs. These criteria included at least one adult in the household being employed (exp (B) 3.412), households owning their home (exp (B) 2.075), and borrowing money as the first option if the household lost its main income (exp (B) 1.471).
This suggests that, more than any other type of community assistance, assistance with costs may be used most by "the working poor" and even households with some assets, as opposed to only households in serious economic need. This may reflect the true economic strategies of households, or it may mean that in qualifying for assistance with costs, households must met an employment stipulation or less stringent economic criteria. The results may also be inaccurate due to the relatively low number of households using assistance with costs in the sample (n = 140). Note the relatively low probability of correctly predicting use of assistance with costs.
Table 7
Use of Housing Assistance (n = 1196)
| Variable | Coefficient | S.E. | exp (B) | |||
|
|
||||||
| STATISTICALLY SIGNIFICANT (p<0.10) | ||||||
| Use at least one form of assistance with costs | 1.841 | 0.301 | 6.303 | |||
| Use at least one form of supplemental income | 0.969 | 0.262 | 2.636 | |||
| Covered by any form of insurance | 0.781 | 0.343 | 2.183 | |||
| Single mothers with children < 18 compared to couples living alone | 0.736 | 0.307 | 2.087 | |||
| At least one adult in the household continued education in last ten years | 0.532 | 0.253 | 1.702 | |||
| Household income ($/10000) | -0.513 | 0.110 | 0.599 | |||
| Works for insurance | -0.846 | 0.383 | 0.429 | |||
| Home owned |
-1.125 |
0.271 |
0.325 |
|||
| At least one adult in the household is employed |
-1.330 |
0.486 |
0.264 |
|||
| NOT STATISTICALLY SIGNIFICANT | ||||||
| Household Characteristics | ||||||
| Couples with children < 18 compared to couples living alone | 0.220 | 0.295 | 1.246 | |||
| Other household types compared to couples living alone | -0.712 | 0.601 | 0.491 | |||
| Household size (persons) | 0.137 | 0.108 | 1.147 | |||
| Children < 13 years in household | 0.052 | 0.399 | 1.053 | |||
| Shared housing with others | -0.195 | 0.300 | 0.823 | |||
| Age of respondent (years) | -0.022 | 0.026 | 0.979 | |||
| Employment | ||||||
| At least one adult in the household is self employed | 0.562 | 0.432 | 1.754 | |||
| At least one adult in the household works at home | 0.532 | 0.426 | 1.703 | |||
| Respondent years of employment | 0.043 | 0.024 | 1.044 | |||
| Education | ||||||
| Vocat.degree/college experience compared to high school diploma or less | 0.096 | 0.212 | 1.101 | |||
| College degree compared to high school diploma or less | -0.288 | 0.289 | 0.750 | |||
| Economic Options | ||||||
| If lost main income: Sell assets compared to living off savings | -0.081 | 0.230 | 0.922 | |||
| If lost main income: Borrow money compared to living off savings | 0.122 | 0.198 | 1.130 | |||
| If lost main income: Use govt. assistance compared to living off savings | 0.141 | 0.203 | 1.151 | |||
| Public Assistance | ||||||
| Use at least one form of public health service | 0.379 | 0.260 | 1.461 | |||
| Other Socio-economic Characteristics | ||||||
| Used public transport | -0.298 | 0.292 | 0.743 | |||
| Feel household financial needs are not met | -0.459 | 0.284 | 0.632 | |||
| Constant |
-1.454 |
1.014 |
||||
| Fit of Model | n | % of cases | ||||
| Correctly predicted non-use of housing assistance | 1020 | 98.0 | ||||
| Correctly predicted use of housing assistance | 130 | 32.0 | ||||
| Overall prediction accuracy | 90.5 | |||||
|
|
||||||
Table 8
Use of Assistance with Costs (n = 1191)
| Variable | Coefficient | S.E. | exp (B) | |||
|
|
||||||
| STATISTICALLY SIGNIFICANT (p<0.10) | ||||||
| Use at least one form of housing assistance | 1.830 | 0.302 | 6.231 | |||
| At least one adult in the household is employed | 1.227 | 0.537 | 3.412 | |||
| Feel household financial needs are not met | 1.003 | 0.274 | 2.725 | |||
| Shared housing with others | 0.986 | 0.259 | 2.681 | |||
| Single mothers with children < 18 compared to couples living alone | 0.894 | 0.411 | 2.445 | |||
| Use at least one form of public health service | 0.782 | 0.251 | 2.186 | |||
| Home owned | 0.730 | 0.290 | 2.075 | |||
| If lost main income: Borrow money compared to living off savings | 0.386 | 0.187 | 1.471 | |||
| Respondent years of employment | -0.045 | 0.021 | 0.956 | |||
| Covered by any form of insurance | -1.062 | 0.322 | 0.346 | |||
| At least one adult in the household works at home | -2.062 | 0.637 | 0.127 | |||
| Constant |
-5.334 |
1.069 |
||||
| NOT STATISTICALLY SIGNIFICANT | ||||||
| Household Characteristics | ||||||
| Couples with children < 18 compared to couples living alone | -0.003 | 0.406 | 0.997 | |||
| Other household types compared to couples living alone | -1.533 | 1.017 | 0.216 | |||
| Household size (persons) | 0.132 | 0.106 | 1.142 | |||
| Children < 13 years in household | 0.530 | 0.407 | 1.699 | |||
| Age of respondent (years) | 0.028 | 0.022 | 1.029 | |||
| Employment | ||||||
| At least one adult in the household is self employed | -0.381 | 0.504 | 0.683 | |||
| At least one adult in the household works other than a daytime shift | 0.152 | 0.247 | 1.165 | |||
| Works for insurance | 0.004 | 0.322 | 1.004 | |||
| Education | ||||||
| Vocat.degree/college experience compared to high school diploma or less | -0.117 | 0.176 | 0.890 | |||
| College degree compared to high school diploma or less | 0.388 | 0.211 | 1.473 | |||
| At least one adult in the household continued education in last ten years | -0.431 | 0.239 | 0.650 | |||
| Economic Options | ||||||
| If lost main income: Sell assets compared to living off savings | -0.223 | 0.232 | 0.800 | |||
| If lost main income: Use govt. assistance compared to living off savings | -0.052 | 0.214 | 0.949 | |||
| Public Assistance | ||||||
| Use at least one form of supplemental income | 0.481 | 0.257 | 1.617 | |||
| Other Socio-economic Characteristics | ||||||
| Used public transport | 0.460 | 0.269 | 1.585 | |||
| Household income ($/10000) |
-0.051 |
0.073 |
0.950 |
|||
| Fit of Model | n | % of cases | ||||
| Correctly predicted non-use of assistance with costs | 1004 | 97.8 | ||||
| Correctly predicted use of assistance with costs | 140 | 37.2 | ||||
| Overall prediction accuracy | 90.4 | |||||
|
|
||||||
SUMMARY AND CONCLUSIONS
The key findings were that low income and less economically secure households are using multiple activities to support themselves. This initial analysis only hints at some of the activities and their linkage. However, our data has two important implications. First, our results confirm earlier work and observations indicating that rural Wisconsin households choices do parallel other areas and contexts. Second, the analysis uncovers some community assistance progrms that rural households are linking together in imaginative ways that previously have not been confirmed.
Four composite forms of community services were examined. Public health was composed of use of free immunization, low cost clinics, inhome nursing care, and healthy start. Supplemental income was composed of using social security (remember these were preretirement age households) and/or SSI, unemployment comp, workers comp, AFDC, food stamps and/or WIC, and general assistance. Housing assistance was composed of using heating and/or cooling assistance and living in public housing. Assistance with costs was composed of help with food, help with clothing, and help with paying bills. Use of child care services both public and private was the fifth service examined.
Younger respondents and increased household income increased the use of child care services. The implications for the current debate on child care and welfare reform suggests that younger and income, single parent families are more vulnerable to availability of child care services.
The analysis indicates that people were using supplemental income when their economic circumstances worsened and use it in conjunction with several other forms of community support.
Public health services use increased with the loss of economic status and were also used with several other forms of community services.
Single moms with lower income were households with increased use of housing assistance. Unexpectedly, however, households with insurance and continued education were also frequent users of housing assistance.
The analysis suggests households experiencing financial difficulties did use community services that help reduce costs. There was an unconfirmed hint that even those just exceeding a threshold found this form of support useful.
Community services played a substantial role in the choices made by households. The dominance of economic status/security implies low income households ability to acquire economic security has great implications for the demand for community services. Community services are of great importance to low income households. It also means that the need for community services will be particularly susceptible to economic status (real and perceived) of its residents.
The analysis reported suggests that the tendency to treat community development, social development, and economic development as separate policy spheres of influence increases the risk of adverse or unintended consequences. As Wisconsin moves forward on its welfare reform experiment, this data confirm the need to link these policy initiatives for some family types, in particular single moms. It is particularly crucial to remember welfare reform and poverty elimination are not the same policy issue.
The results provide valuable insight into the linkages between community and family economic survival. Illuminating this important connection will draw attention to the need for rural development policies which target those community institutions and interpersonal resources directly affecting rural family welfare.
FOOTNOTES
1. Respectively graduate research assistant, Development Studies, Professor of Agricultural and Applied Economics, former graduate research assistant Agricultural & Applied Economics. The study was supported by CREES-USDA National Research Initiative Project #144-EL-77 and College of Agriculture and Life Sciences, UW-Madison Hatch #142-N660. Leann Tigges and Ann Ziebarth co-principle investigators on NRI #144-EL-77.2. See Clay and Schwarzweller (1991) for studies on rural areas, and Voydanoff (1990) for a review of research primarily with an urban focus.
3. The Letters and Science Survey Center, University of Wisconsin-Madison conducts research projects for university administration, faculty, staff, and service departments. The Center has conducted a wide variety of survey research projects mostly utilizing Computer Assisted Telephone Interviewing (CATI) applications.
4. Households rather than families were sampled. While the initial contact to a residential phone number was random, households were screened to determine if they fit our sample criteria.
5. It was anticipated that this category would be so small that serious analysis would not be meaningful. So attention was directed to household types more prevalent.
6. A copy of the survey instrument is available, on request, from the NRI principle investigators.
7. Assistance, public assistance, community services, and services are used interchangeably through out this report.
8. The terms community services, community assistance, public assistance, assistance will be used interchangeably.
REFERENCES
Abrahams, C. 1992. "A Social Development Practice Model for Community Development," Journal of the Community Development Society, 23, 2: 103-115.
Allen, J. C. 1991. "Household Income Generation in Rural Communities: A Comparison of Downeast Maine and Eastern Washington State." Paper presented at the Annual Rural Sociological Meeting, Columbus, Ohio, August.
Becker, G. S. 1965. "A Theory of the Allocation of Time," Economic Journal, 75 (Sept): 493-517.
Bluestone, B. & Harrison, B., 1988, "The Growth of LowWage Employment: 1983-86," American Economic Review, 78, 2 (May): 124-128.
Bokemeier, J.L. & Garkovich, L.E., 1991,. "Meeting Rural Family Needs," in Rural Policies for the 1990s, Cornelia B. Flora & James A. Christenson (eds.) Boulder, CO: Westview Press: 114-127.
Bourguignon, F. & Chiappori, P.A. 1992. "Collective Models of Household Behavior," European Economic Review, 36, 2/3 (April): 355-364.
Bryant, K. 1990. The Economic Organization of the Household, Cambridge University Press.
Clay, D. C. & Schwarzweller, H.K., 1991. Research in Rural Sociology and Development, A Research Annual: Household Strategies, 5, JAI Press Inc.:Greenwich, CT.
Deavers, K. L. & Hoppe, R.A., 1992, "Overview of the Rural Poor in the 1980s," in Rural Poverty in America, C. Duncan (ed.) Westport, CT: Auburn House: 320.
Deavers, K. L. & Hoppe, R.A., 1991, "The Rural Poor: The Past As Prologue," in Rural Policies for the 1990s, C. B. Flora & J. A. Christenson (eds.) Boulder, CO: Westview Press: 85-101.
Duncan, C. (ed.), 1992. Rural Poverty in America, Westport, CT: Auburn House.
Duncan C. M. and N. Lamborgini, 1994, "Poverty and the Social Context in Remote Rural Communmities," Rural Sociolgy, 59, 3: 437-461.
Everitt, B.S., 1977, The Analysis of Contingency Tables,. New York, Halsted Press.
Feinberg, S.E., 1977, The Analysis of Crossclassified Categorical Data,. Cambridge, MIT Press.
Fitchen, J. 1992. Poverty in Rural America: A Case Study, Westview Special Studies in Contemporary Social Issues, Westview Press.
Galston, W. A. & K. J. Baehler, 1995. Rural Development in the United States: Connecting Theory, Practice, and Possibilities, Washington DC: Island Press.
Gorham, L., 1992. "The Growing Problem of Low Earnings in Rural Areas," in Rural Poverty in America, Duncan, C. (ed.): 21-40.
Horney, M.J. McElroy, M., 1981. "Nash Bargained Household Decisions: Toward a Generalization of the Theory of Demand," International Economic Review, 22, 2 ( June)
Hosmer, D.W., & Lemeshow, S. 1989. Applied Logistic Regression. New York, John Wiley and Sons.
Johnson, W. & Skinner, J. 1986. "Labor Supply and Marital Separation," American Economic Review, 76, 3: 45569.
Kassel, K. & Gibbs, R.M., 1996, "Rural Nonfarm Earnings Edge Up," Rural Conditions and Trends, 7 #3: 22-24.
Katz, E. 1992, "Intra-Househhold Resource Allocation in the Guatemalan Central Highlands: The Impact of Non-Traditional Agriculatural Exports," Dissertation, Department of Agricultural Economics, University of WisconsinMadison.
Kim, J. & Mueller, C.W., 1978, Factor Analysis: Statistical Methods and Practical Issues. Sage University Paper Series, Sage Publications, Beverly Hills, CA.
Klecka, W.R., 1980, Discriminant Analysis: Quantitative Applictions in the Social Sciences. Sage University Paper Series, Sage Publications, Beverly Hills, CA.
Kusmin, L. & Gibbs, R.M., 1996, "Rural Employment Gains Continue, But Pace Slows," Rural Conditions and Trends, 7 #3: 18-21.
Lachenbruch, P.A. 1975. Discriminant Analysis. Hafner, New York, NY.
Levitan, L. & Feldman, S. 1991. "For Love or Money: Nonmonetary Economic Arrangements Among Rural Households in Central New York," in Research in Rural Sociology, A Research Annual: Household Strategies, D. Clay and H. Schwarzweller (eds.), 5, JAI Press Inc.: Greenwich, CT: 149-172.
Lobao, L. M., 1990. Locality and Inequality: Farm and Industry Structure and Socioeconomic Conditions, State University of New York Press: Albany NY.
Lyson, T. A. & W.A. Falk (eds.), 1993. Forgotten Places: Uneven Development in Rural America, Lawerence KS: University of Kansas Press.
Manser, M. & Brown, M. 1980. "Marriage and Household DecisionMaking: A Bargaining Analysis," International Economic Review, 21, 1 ( February):
Mason, K. & Kuhlthau, K. 1992. "The Perceived Impact of Child Care Costs on Women's Labor Supply and Fertility," Demography, 29, 4 (November) 523-43.
Moen, P. & Wethington, E. 1992. "The Concept of Family Adaptive Strategies," in Annual Review of Sociology, 18: 233-251.
Nightingale, D. S. & R. Haveman (eds) 1995, The Work Alternative: Welfare Reform and the Realities of the Job Market, Washington DC: The Urban Institute Press.
O'Brien, D., McClendon, M. & Ahmed, A. 1989. "Neighborhood Community and Quality of Life," Journal of the Community Development Society, 20, 2: 59-71.
Rao, C.R., 1973, Linear Statistical Inference and its Application. 2nd Edition, Wiley, New York, NY.
Reich, R. B., 1988. "The Rural Crisis and What to Do About It," Economic Development Quarterly, 2, 2: 38.
Rosenzweig, M., 1990, " Programme Interventions, IntraHousehold Allocation, and the Welfare of Individuals: Economic Models of the Household". in Intrahousehold Resource Allocation: Isuues and Methods for Development Policy and Planning, Beatrice Lorge Rogers & Nina P. Schlossman (eds.), Tokyo: United Nations, University Press,
Rummel, R.J., 1970, Applied Factor Analysis. Northwestern University Press, Evanston, IL.
Sorter, B. 1987. "Community Development Analysis Through Behavioral Patterns," Journal of the Community Development Society, 18, 2: 42-53.
Summers, G. F., S. D. Evans, F. Clemente, E. M. Beck, & J. Minkoff, 1976, Industrial Invasion of Nonmetropolitan America: A Quarter of Century fo Experience, New York: Praeger.
Thomas, D., 1990. "IntraHousehold Resource Allocation: An Inferential Approach," Journal of Human Resources, 25, 4, Fall:. 634-664.
Tickamyer, A.R. & C. M. Duncan, 1991. "Work and Poverty in Rural America," in Rural Policies for the 1990s, C. B. Flora & J. A. Christenson (eds.) Boulder, CO: Westview Press: 102-113.
Tickamyer, A., J. Bokemeier, S. Feklman, J.P. Jones, D-A. Wenk, R. Harris, 1993. "Women and Persistent Rural Poverty." in Persistent Poverty in Rural America, g. F. summetrs (ed), boulder, CO: Westview Press: 200229 .
Tigges, L., A. Ziebarth, & R. Shaffer, 1995, "Wisconsin Community Decussion Group Summary Report," UW-Madison, Dpet of Rural Sociology.
Unger, D. G. & M. B. Sussman,(ed) 1990, Families in Community Settings: Interdisciplinary Perpsectives, Binghamton, NY: The Howorth Press.
U.S. Bureau of the Census 1990. Census of Population . US Bureau of the Census, Washington D.C..
U.S. Department of Labor 1990 through 1996. Detailed CPI Report. Bureau of Labor Statistics, Washington D.C..
Warren, D.I. 1981. Helping Networks, West Bend IN: Notre Dame University Press.
Voydanoff, P. 1990. "Economic Distress and Family Relations: A Review of the Eighties," Journal of Marriage and Family, 52 (November): 1099-1115.
