Self-Esteem and Chronic Welfare Dependency

Leonard Schneiderman, Walter M. Furman, and Joseph Weber

Introduction

California State Assembly Bill 3659, the act that created the California Task Force to Promote Self-Esteem and Personal and Social Responsibility, identifies chronic welfare dependency, along with crime and violence, alcoholism and drug abuse, teenage pregnancy, child abuse, and educational failure, as "intractable social problems" of major concern to the citizens of California. It declares that the most cost-effective and humane solution to such social problems is to "discover, address, eradicate, and thereby prevent" their causes. This legislation calls specific attention to low self-esteem as a possible root of these social problems:

Low self-esteem may well have a wide-ranging, negative influence on individual human conduct, the costs of which both in human and societal terms are manifested in a number of ways, many of which convert into significant expenditure of state moneys. If so, these human costs and the costs to government could be reduced by raising the self-esteem level of our citizenry.

This chapter will review the scholarly literature linking self-esteem and chronic welfare dependency. At least three possibilities exist:

1. Variations in levels of self-esteem are related to variations in welfare dependency. That is, low self-esteem "causes" prolonged dependence on public assistance. Or, high self-esteem allows resistance to negative public attitudes and bureaucratic obstacles and "causes" indi-

viduals to use welfare benefits on a long-term basis when no other means of support is available.

2. Chronic dependence on welfare affects recipients' levels of self-esteem. Dependence on welfare "causes" feelings of helplessness and loss of control and lowers self-esteem. Or, dependence on welfare may enhance self-esteem through providing basic resources and improving living conditions (Nichols-Casebolt 1986).

3. No scholarly evidence links self-esteem to chronic welfare dependence.

All available research on the relationship between chronic welfare dependency and self-esteem necessarily rests on certain assumptions about the nature and causes of poverty and dependency. One set of assumptions holds that poverty and dependency are the consequences of external events over which affected persons can exercise little control, such as an economic system that does not provide opportunities for all to earn an adequate income. Another view assumes that poverty reflects a lack of ability or other personal deficiencies (perhaps low self-esteem) among the poor themselves and is the result of their failure to take advantage of existing opportunities (Gilder 1981; Murray 1984).

Differences also exist concerning the relative importance of public policies designed to reduce poverty and public policies designed to reduce dependency . To the extent that individuals do not have enough income to maintain an adequate standard of living, they are thought to be "poor." To the extent that individuals receive income from government transfers, they are thought to be "dependent." Some advocate public transfers (hence dependency) to reduce or eliminate the injurious effects of poverty, based on the conviction that an affluent society has a moral obligation to those in need. Others oppose using government transfers to reduce poverty, based on the conviction that such aid will undermine character and weaken self-sufficiency.

Reducing poverty and reducing dependency are compelling goals of social policy, but they are clearly not fully compatible. Different legislatures, in different times, with different leadership have seen these problems in different ways and have advocated different public policy responses (Morris and Williamson 1987).

Liberals and conservatives have tended to divide according to the explanations of poverty they find most convincing and the relative priority they assign to poverty reduction and dependency reduction, although recent signs indicate decreasing polarization and increasing consensus.

Both systemic and individual factors appear to play a role in reducing poverty, often interacting in complex ways. There is growing recognition that both liberals and conservatives share the goals of independence and self-sufficiency, although they may differ on public policies designed to promote the achievement of these goals. Liberals have generally held that adequate levels of income achieved through nonstigmatizing public income transfers and services (i.e., job creation and training, education, health and social services) may increase dependency in the short term but will lead to self-reliance in the long term. Conservatives hold that limiting transfer income may increase poverty in the short term but will be more likely, in the long term, to lead to adequate incomes through strengthened self-sufficiency; dependence on government is seen as the greater problem.

Across the political spectrum, concern that we have not been doing enough to aid the poor has been joined by a fear that we may be doing the wrong things. As a consequence, there has been a resurgence of interest in redesigning welfare programs (especially the program of Aid to Families with Dependent Children), with the aim of providing more adequate assistance to the poor without reinforcing or inducing values and habits that encourage prolonged dependency (Novak et al. 1987).

To what extent are persistent poverty and chronic welfare dependency associated with character deficiency and the failure of personal initiative, responsibility, and self-sufficiency? To what extent does relieving poverty through public income transfers have a detrimental effect on self-esteem and personal and social responsibility? To what extent can enforced self-reliance, accompanied by poverty, promote the long-term enhancement of independence, self-esteem, and personal and social responsibility? Can self-esteem prosper in an environment of material deprivation and limited opportunity? Do programs that improve material conditions and open new opportunities enhance self-esteem? What is the appropriate role of government in ensuring adequate incomes?

Questions so deeply enmeshed in personal values and ideology cannot be fully resolved through scientific inquiry, but there are some findings, drawn from available research, that can help us understand and approach the problem. In the pages that follow, we discuss "welfare" and attempt to delimit the social problem of primary concern here, that is, chronic welfare dependency. We look especially at the program of Aid to Families with Dependent Children and examine its role as the major public program designed to address the poverty of families and children in the United States and as the program most commonly thought of as "welfare." We then look at the dynamics of welfare utilization,

patterns of entering and leaving the welfare system, and characteristics of the chronically dependent. We discuss the concept of self-esteem and then review evidence linking self-esteem to patterns of welfare utilization. We end with suggestions for research regarding chronic welfare dependency and discuss public policy implications emerging from the review.

Chronic Welfare Dependency as a Social Problem

Assembly Bill 3659 identifies the long-term use of legislatively created entitlements (welfare benefits) as an "intractable social problem." Unlike the other social problems listed in the bill, however, welfare dependency could not exist without legislatively mandated programs and legislatively authorized expenditures. Welfare benefits exist to fulfill the legislature's intention to relieve the poverty of those in dire need. In the case of the program of Aid to Families with Dependent Children (AFDC), welfare benefits are targeted to meet the needs of children who are deprived of parental support.

Why, then, does the California legislature regard the use of entitlements it has itself created as a problem? What is "welfare dependency," and at what point does the use of welfare become excessive? To answer these questions, we begin with a brief review of the range of social welfare benefits offered under governmental auspices in the United States.

Total public spending (federal, state, and local) for social welfare in the United States in 1984 amounted to $ 672 billion, or 18.2 percent of the gross national product (Social Security Administration 1986). Social insurance programs designed to protect against the loss of income as a result of old age, disability, death, or unemployment accounted for approximately 51 percent of the total. Health and medical programs (excluding Medicare and Medicaid) amounted to 6 percent, veterans' programs to 4 percent, education 23 percent, and public aid to the poor only 13 percent (Social Security Administration 1987).

The percentage of government spending devoted to social welfare has been declining steadily at federal, state, and local levels since 1979 (Social Security Administration 1987, 31). Excluding trust-fund social insurance programs, federal social welfare expenditures for public aid (including AFDC) have fallen sharply as a percentage of all federal spending, from 13.7 percent in 1979 to 10.5 percent in 1984, a 24 percent reduction in spending.

Of the approximately $ 90 billion in public aid targeted for the poor

in 1984, $ 14.5 billion, or 16 percent, was expended as cash benefits through AFDC. The remaining 84 percent of the total went to the needy aged, blind, and disabled, to provide health benefits and food stamps. Put another way, of the $ 672 billion expended for social welfare purposes under public auspices in 1984, some $ 14.5 billion, or 2.2 percent of the total, went to AFDC, the one program usually identified as welfare (Social Security Administration 1986).

The enormous amount of attention given to so small a part of the nation's total social welfare effort reflects the symbolic importance of "welfare" and the deeply held public attitudes and values concerning it. It could be argued that all human beings are dependent to a greater or lesser extent on other people and on institutions. In a population of some 245 million Americans, for example, approximately 115 million are in the labor force. More than half of all Americans are therefore "dependent" on the productivity of others for the goods and services they need and want.

Such an example treats dependence as normative and basically benign. Ample evidence exists, however, that public attitudes toward dependence vary with its presumed degree of "legitimacy." If a particular group—such as children or the aged or disabled—is not expected to be self-supporting, or if a particular person's loss of independence is seen as involuntary, society views the dependence as legitimate. But if dependence occurs among people who are expected to be self-reliant, the public view is negative (Rainwater 1982).

Chronic welfare dependency is viewed as an intractable social problem to the extent that such dependence is not regarded as legitimate. Although dependence on the largest and most costly of the nation's welfare programs (Social Security) is largely seen as legitimate, this is not generally the case for the program of aid to poor children and their families. It is the question of legitimacy, and not just the dollars expended or the duration of benefits received, that defines welfare dependency as a problem and that led to our decision to confine this review of welfare dependence to the AFDC program.

Views about AFDC—whether dependence on the program is legitimate, for example, or whether such dependence lasts too long (that is, becomes chronic)—are powerfully influenced by how society feels about women as heads of households. If women who head families with children are expected to hold jobs and to be economically independent, even a short period of dependency may seem too long. If women who head families are not expected to participate in the labor force, then du-

ration and degree of economic dependence may not seem like an intractable social problem.

Maternal employment has until recently been viewed negatively, because it may reduce the mother's participation in the child's educational activities and emotional development, as well as her monitoring and supervision of the child's activities. According to this argument, single mothers who are employed disadvantage their children doubly: such children spend less time with absent fathers and less time with their mothers. Our current AFDC program was enacted at a time when the prevailing judgment was that mothers should stay at home and be fulltime caretakers of their children. Although this view is still strongly held, nearly everyone also agrees that heads of households should support their families. Thus, when a woman voluntarily assumes this role, she implicitly assumes the same obligations. This presents an unresolvable contradiction in values: single mothers should both hold jobs and stay at home (Garfinkel and McLanahan 1986).

Today, more than half of married mothers work outside the home at least part of the day. Given this change in the pattern of women's participation in the labor force, it is not surprising that welfare recipients are often viewed as lazy and undeserving and that welfare dependency is viewed as a problem. Despite the increasing expectation that women should hold jobs, the AFDC program continues to reflect ambivalent public attitudes by operating in a manner that discourages employment. Benefits are drastically reduced as earnings increase, and earnings replace rather than supplement welfare payments, often leaving those who are employed in a situation that is no better—and is sometimes worse (without medical care, for example)—than the situation of those who remain entirely dependent on AFDC. Sawhill (1976) found that women on welfare in the early 1970s had very low earning capacity; even if they had worked full time, more than half would still have earned less than they received in welfare grants. Schneiderman et al. (1987a) found that 90 percent of AFDC recipients in Los Angeles County in 1986 were totally dependent on AFDC benefits for their incomes.

A study by Ellwood (1986b) reports that only one woman in five who left welfare was able to do so as a result of increased income from earnings and that earnings adequate for self-support required year-round, full-time work at the level of two thousand hours annually. Using data from the Panel Study of Income Dynamics (Survey Research Center, Institute for Social Research 1984) for the late 1970s, Ellwood shows that only 8 percent of all wives (poor and nonpoor) with preschool chil-

dren actually worked full time, with an additional 15 percent working between fifteen hundred and two thousand hours per year. If self-sufficiency based on earnings requires nearly full-time work, and if only one-quarter of all wives with young children work three-fourths of the time or more all year, complete self-sufficiency through earnings for single-parent households may not be a realistic goal for AFDC recipients with small children.

The strength of the legitimacy issue and its impact on the provision of benefits for women and children is evident when one compares public benefits available to widows and their children with benefits available to single mothers (divorced, separated, never married) and their children. Whereas 51 percent of all families headed by single mothers are poor, only 34 percent of families headed by widows are poor. Some of this difference is related to discrepancies between those social welfare benefits most commonly involving men and those in which women are predominantly involved—that is, differences between Social Security survivors' insurance, the program that aids widows and orphans, and AFDC, the program that aids primarily families headed by single mothers. Among whites, nearly 90 percent of all widows receive survivors' insurance, as do 70 percent of all widows among blacks. Only 23 percent of white divorced women and 34 percent of black divorced women receive AFDC, and the percentage of separated and never-married women, both white and black, who receive AFDC ranges from 40 to 59 percent. Yet it is difficult to argue that the financial needs of children living in these female-headed households are different from the needs of children in families headed by widows. In addition, the average Social Security survivors' insurance benefit is nearly twice the AFDC benefit among whites and more than double the AFDC benefit among blacks (Garfinkel and McLanahan 1986, 26).

The division between social insurance and public assistance has bifurcated social welfare across class lines. With a strong, articulate, middle-class constituency, social insurance, especially social security, carries no stigma, and its expanded benefits have reduced drastically the amount of poverty among the elderly. Public assistance, which has become synonymous with welfare, is, of course, restricted to the very poor. Its recipients carry the historic stigma of the unworthy poor, and, as a consequence, they are treated meanly. Their benefits, which do not lift them out of poverty, remain far below those paid by social security. (Katz 1986, ix)

Williamson (1974), in a study of 375 men and women in the Boston area in 1972, tried to determine how well informed people were on a

number of factual issues related to welfare. He asked respondents, "What percent of welfare recipients are able-bodied unemployed males?" Statistics compiled by the U.S. Department of Health, Education, and Welfare (HEW) for 1971 showed the official government estimate of this group to be less than 1 percent of the total number of recipients. But the mean estimate in Williamson's sample was 37 percent. He also asked, "What percent of welfare recipients lie about their financial situation?" HEW reported in 1971 that "suspected incidents of fraud or misrepresentation among welfare recipients occur in less than four-tenths of one percent (0.4 percent) of the total caseload in the nation." The mean response among the sample was 41 percent. Concerning the question of how many children under eighteen there were in the average AFDC family, official HEW figures showed 2.6 children per family. The mean response among the sample was 4.8. Williamson's data show evidence of negative misconceptions about welfare held by respondents at all levels of income and education. Surprisingly, even welfare recipients themselves shared these misconceptions.

Poverty and AFDC

Families headed by single women with children are the poorest of all major demographic groups in the United States, regardless of how poverty is measured (U.S. Bureau of the Census 1985). Their economic position relative to that of other groups, such as the aged or the disabled, has declined steadily during the past two decades.

Children being raised by their mothers alone are four times as likely to be poor as children with both parents at home. The poorest children are those with never-married mothers. (See Table 6.1.) More than 70 percent of children of never-married mothers—black, white, or Hispanic—lack enough money to reach the poverty level (U.S. Congress, Congressional Research Service and Congressional Budget Office 1985). Whether poverty is measured before or after government transfer payments and whether the income counted includes or excludes noncash benefits and money paid as taxes, poverty rates among children rose sharply from 1979 to 1983.

Roughly two-thirds of the nation's poorest families gained at least some of their income in 1985 from their own earnings. Welfare payments (principally AFDC) were the most frequent source of nonwage income for the poor; 37 percent received such income (U.S. Bureau of the Census 1985).

Inadequate income levels can have important consequences for the well-being of children and families. Health, nutrition, and access to prenatal and postnatal care are all associated with income level. Schneiderman and his colleagues (1987b) found in their study of 1,046 families receiving AFDC in 1986 that 18 percent of AFDC family cases and 29 percent of AFDC-UP (two-parent household) cases reported health-related conditions considered severe enough to interfere with employment.

AFDC is the only federal program explicitly designed to provide cash assistance to impoverished families with children. Maximum benefit levels for a single-parent, two-child family with no income vary in the forty-eight contiguous states, from 16 percent of a 1986 poverty-level income (Alabama) to 85 percent (California).^[*] (See Figure 6.1.) In fiscal year 1986, AFDC monthly enrollment averaged 11 million persons, 7.3 million children in 3.8 million families (U.S. Congress, Congressional Research Service and Congressional Budget Office 1985).

The AFDC program is targeted specifically for children who are deprived of parental support because of the absence or disability (and in some cases unemployment) of a supporting parent. The size of the targeted population can be appreciated when we recognize that half of all American children born today will spend part of their childhood in a

[*] * Only in Alaska and Hawaii do the combined AFDC and food stamp benefits exceed the poverty level. The national poverty level, however, is not corrected for the cost of living in a particular state, and the cost-of-living indexes in these two states tend to be quite high.

Figure 6.1.
Combined AFDC and Food Stamp Benefits for a Family of Three with No Income, January 1987
Source: "Welfare and Poverty Among Children" 1987, 6.

family headed by a mother who is divorced, separated, unwed, or widowed. About half of all such families are poor and dependent on welfare (Bumpass 1984). To some extent, these dependent families are in periods of transition (between marriages or jobs, for example), averaging five years for white families and seven years for black families. From the child's point of view, these are long periods of time. Furthermore, a large and growing minority of black children are being born to nevermarried women and can expect to spend their entire childhood in a mother-only family (Garfinkel and McLanahan 1986, 7). Whereas 45 percent of white children will spend part of their youth in a family headed by a woman, the estimate for black children is 86 percent (Bumpass 1984).

The number of families in the AFDC program has been relatively stable since the mid-1970s, rising slightly during the recession of 1980. The number of children per family has declined, however. Consequently, the actual number of children receiving AFDC benefits has dropped steadily, but children still account for two-thirds of all recipients (U.S. Congress, House Select Committee on Children, Youth, and Families 1987, 86).

In 1984, 38.2 percent of children receiving AFDC were deprived of parental support as a result of divorce or separation, 46.4 percent as a result of being born out of wedlock, 1.9 percent because of the death of the father, 8.6 percent because of the father's unemployment, and 3.6 percent because the father was incapacitated (U.S. Congress, House Select Committee on Children, Youth, and Families 1987, 29). Schneiderman et al. (1987b) found in their June 22, 1986, sample of Los Angeles County AFDC recipients that 9.7 percent of those AFDC cases involved families of unemployed workers. Seventy percent of all the AFDC recipients were children.

How effectively were poor children and their families assisted by the AFDC program? In 1985, the national child poverty rate was reduced 4 percent by cash welfare benefits. Even though Social Security imposes no means test for eligibility, it reduced the child poverty rate by 6 percent in 1985 ("Welfare and Poverty Among Children" 1987). Because of sharply differing public attitudes, however, children and their caretaker parents who depend on Social Security benefit programs are not defined as being among the "chronically welfare dependent," regardless of the degree or duration of their reliance on Social Security payments.

The number of AFDC-assisted children as a percentage of all poor children grew until 1974, when the number of AFDC children per 100 poor children started dropping (U.S. Congress, Congressional Research

Crtoon by Wasserman. © 1987, Boston Globe. Reprinted by permission of Los Angeles Times Syndicate.

Service and Congressional Budget Office 1985, 212). The average number of children on AFDC per 100 children in poverty increased from 13.4 in 1960 to 83.6 in 1973; the number then declined continually from 77.5 per 100 in 1977 to 53.3 per 100 in 1983.

These figures suggest that concern about chronic welfare dependency must be balanced with concern about the destructive impact of the chronically unrelieved poverty of a shockingly large proportion of the nation's children. Increasing the number of poor children aided and increasing the level and adequacy of the benefits will necessarily relieve poverty and increase dependency. Limiting dependency through restrictive eligibility and inadequate benefit levels will discourage reliance on public benefits and contain cost but will also leave poverty essentially unrelieved. Questions of value and choice are involved here, which research findings can amplify but not resolve.

Who Are the Welfare Dependent?

Understanding chronic welfare dependency requires a review of patterns of welfare use over time—that is, the dynamics of entering and

leaving the welfare system. The AFDC caseload is not static; daily, people enter and exit the program. Point in time, or cross-sectional, studies of welfare populations cannot adequately address factors that cause welfare dependency, cannot accurately measure past welfare use by a household or how long a current welfare episode will last. Cross-sectional studies do give an accurate picture of the characteristics of the caseload at a particular time, but that picture reflects the preponderance at any point of long-term welfare recipients and underestimates the amount of cycling on and off welfare that occurs among short-term recipients (Bane and Ellwood 1983; Wiseman 1985). For this reason, our analysis focuses on longitudinal studies of welfare and the lessons that have been learned from observing large samples over many years—decades, in some instances.

The two primary sources for longitudinal data on patterns of welfare dependency are the Panel Study of Income Dynamics (Survey Research Center, Institute for Social Research 1984), widely known as the PSID, and the National Longitudinal Survey (Center for Human Resource Research, Ohio State University 1980), known as the NLS. The first portion of the NLS was conducted in 1968 and sampled approximately five thousand women between the ages of fourteen and twenty-four. The PSID, conducted under the direction of the Survey Research Center at the University of Michigan, began in 1968 with a nationally representative sample of 2,930 households, plus another 1,872 households drawn from a previous census study (the Survey of Economic Well-Being) that concentrated on low-income households.

The analysis of information from the longitudinal surveys has involved different methods of characterizing the degree of welfare dependency. These methods include (1) simply counting years in which welfare was received at a criterion level (e.g., $ 250 per year) during the period of observation; (2) computing "spells" of welfare use; and (3) estimating "careers" of welfare use, including not only single spells but also probabilities of recidivism (Ellwood 1986a).

Each of these methods yields a somewhat different picture of the characteristics of individuals, assistance programs, and labor market conditions that affect entry into the welfare system, exits from welfare, and becoming chronically welfare-dependent. There is, however, no agreed-upon criterion in the empirical literature to delimit when a case is termed chronic or, as some analysts prefer, "persistently dependent" (Ellwood 1986a). In addition to the purely temporal measure of duration (length of time during which assistance is received), the extent of

the family's reliance on AFDC for its household income (AFDC as a percentage of total income) is also often considered as a measure of dependency. Several analysts (Rein and Rainwater 1978; Coe 1981) note from the PSID data that income from AFDC for a family on aid generally constituted less than half of the entire household income reported in the multiyear period. (Rein and Rainwater found that two-thirds of the households receiving aid in at least four of the seven years observed received less than half of their entire income in the period from welfare.) In contrast, as mentioned earlier, recent research on AFDC recipients in Los Angeles County revealed that 90 percent of recipients had no income other than AFDC (Schneiderman et al. 1987a), a finding supported by Garfinkel and McLanahan's observation that about 85 percent of women with children do not work during any month in which they receive aid (1986, 38).

How then shall we define chronic welfare dependency? Depending on the method used to analyze total time on welfare, various authors (e.g., Coe 1981; O'Neill et al. 1984; Ellwood 1986a) have characterized persistent dependence as starting from two to ten years after the initial receipt of welfare aid. Most analysts view receiving welfare for five years or more as chronic or persistent dependency, and we will accept this rough guideline for delimiting the characteristics of the population chronically dependent on AFDC. (But we will apply it with some flexibility in reviewing work concerning the relationship between self-esteem and chronic welfare dependency.)

Going on Welfare

We have already noted that AFDC families become eligible for aid as a result of divorce or separation, out of wedlock births, or the father's unemployment, incapacity, or death. In our recent cross-sectional survey of the Los Angeles County caseload, the officially coded reasons for eligibility included 43.9 percent listed as separation, divorce, or desertion; 37.2 percent as an "absent parent" (i.e., children needing support in a household in which the parents never were married or lived together); 8.8 percent as loss of work (most of these cases were in the AFDC-UP program); and the remainder in various categories such as death or disability of a parent (Schneiderman et al. 1987b, 79–80).

These data point to the overwhelming primacy of changes in family or household composition, as opposed to changes in income levels, as the factors that precipitate welfare episodes. Bane and Ellwood ana-

lyzed the PSID sample data for women who had ever received AFDC (n = 676) to investigate reasons associated with going on welfare; their findings from the 554 observations of new welfare episodes are shown in Table 6.2. The authors conclude that "three-fourths of all spells of AFDC begin with a relationship change whereby a female-headed family with children was created. Only 12 percent of beginnings can be traced to earnings decreases" (1983, 19).

Short-Term AFDC Use

Although the subject of this analysis is chronic welfare dependency and its relationship to self-esteem, an aside about short-term welfare use provides a valuable context for the discussion. The NLS and PSID analyses focusing on welfare dynamics have promoted a new understanding that most welfare assistance is temporary and that a fairly large proportion of the general population (an estimated one-quarter) has at some point received welfare assistance (here defined to include AFDC, General Assistance, SSI, or food stamps received by the household head or spouse [Duncan and Hoffman 1987, 4]). Coe (1981, 140) shows that in the PSID sample only 25.8 percent of the individuals who ever received welfare between 1969 and 1978 received aid for more than five years during that period. And although it has been estimated that one-quarter of the general population has received welfare at some time, only an estimated 6.5 percent received it in six or more years during the ten-year

period of observation. A recent summary of the body of evidence on welfare dynamics concludes: "Occasional welfare receipt is common, persistent welfare receipt is not. Movement on and off welfare rolls is widespread" (Duncan and Hoffman 1987, 3). Clearly, then, most welfare assistance does not involve chronic welfare dependence, and any policy or programmatic efforts to alter and reform the AFDC program must consider the large volume of short-term reliance on AFDC.

Leaving AFDC

Most analysis directly relating to welfare dependency concerns factors that differentiate short-term from long-term AFDC recipients. Ellwood (1986a) presents data from the PSID on "the primary reason for spell completion" for all observations of completed spells in the sample (Table 6.3). These data reveal that "becoming a wife is the primary reason that women leave welfare" and that "earnings gains account for only 21 percent of all exits" (1986a, 55–56). Other common reasons for terminating an AFDC episode are children reaching the age of eighteen (11.2 percent) and increases in other transfer income (14.2 percent—representing alimony, child support, Social Security [such as survivors' or disability benefits], and other transfers).

O'Neill and her colleagues (1984) analyzed the PS1D and the NLS data, as well as national AFDC caseload statistics between 1965 and 1982, in order to estimate durations of AFDC spells and to identify variables associated with long-term dependency. Table 6.4 shows the probability of remaining on welfare after a given number of years (post-case opening), as found in the three data sources. This table also presents results from Bane and Ellwood (1983), using a slightly different criterion for receipt of aid during the year.

The probability of remaining on welfare beyond five years was .34 in Bane and Ellwood's 1983 analysis, .18 for all races in the NLS data, and .22 for the cohort starting AFDC in 1965, but it was .16 and .12, respectively, for the 1970 and 1975 AFDC cohorts, indicating a reduction in long-term dependence over the time period. These findings are for single-spell durations, however, and ignore the effect of recidivism on the total welfare career of a recipient. Given this limitation, the authors note that they were able to piece together a distinct profile of recipients who "are likely to have long or short spells, based on their characteristics at the start of a welfare spell" (O'Neill et al. 1984, 10–11). The following independent variables were associated with this distinctive profile and thus with duration of welfare dependency:

1. Market productivity . Histories of work and high earnings led to quick exits from AFDC.

2. Children . A larger family size led to a slower exit.

3. Education . More years of schooling were associated with earlier exits, particularly exits resulting from marriage.

4. Being a teenager when first child is born . Although this variable in fact did not directly affect exit rates, we will later show that it seems indirectly associated with greater dependency.

5. Race . Longer durations were found for black women, even when other relevant factors were controlled.

6. Attitudinal factors . Attitudinal factors (in particular, the Rotter locus of control measure, which we will discuss below) were weakly related to spell duration.

7. Welfare grants . The authors noted mixed evidence tying increased grant levels to lower probabilities of exit.

Ellwood's (1986a) analysis of the PSID data appears to be the most comprehensive identification of recipient characteristics associated with

careers of long-term AFDC dependence. Using the multiple-spells approach, Ellwood found that earlier analyses had underestimated the extent of persistent welfare dependence and that more than 40 percent of welfare recipients had multiple spells of welfare receipt. When multiple spells are considered,

fifty percent of those who ever receive AFDC will receive it for no more than four years. Yet nearly 25 percent will eventually use AFDC in ten or more years. The estimated average total period of AFDC receipt is nearly seven years. . . . The majority of those who receive welfare do not receive it for very long. Nonetheless, a sizable minority appear to have very long periods of receipt. Consequently, of those who are on AFDC at any point in time, almost 60 percent are in the midst of periods of welfare use that will include ten or more years of receipt. (Ellwood 1986a, 25)

Table 6.5 presents Ellwood's marginal and aggregate findings concerning factors associated with length of time on welfare. Marginal findings show the influence of a factor by itself, with all other factors remaining equal. The aggregate analysis allows factors to naturally covary, so that, for example, changing marital status at spell beginning may also then change mean age.

Ellwood summarizes the findings on the marginal effects of recipient characteristics: "Education, marital status, number of children, work experience, and disability status are all closely correlated with first-spell durations. These same variables generally have substantial influence on recidivism and later-spell durations as well" (1986a, 31). As to the aggregate duration estimates, marital status seems to be associated with the most difference in the expected length of time during which women receive assistance. Never-married women averaged 9.3 years of AFDC receipt, whereas divorced women averaged less than 5 years. Ellwood also notes that race and work experience distinguished different sub-groups of AFDC recipients and that education also seemed to affect total duration on aid. Although AFDC benefit levels are not individual characteristics, Ellwood did note that "persons in high-benefit states averaged longer spells and were likely to return to welfare" (1986a, 32) when all else was held constant.

Intergenerational Welfare Dependency

The transmission of welfare dependency between generations is a topic of both theoretical and policy interest. Views of poverty and wel-

fare dependence that hold individual failings responsible for the problems often describe the transmission of a parent's dysfunctional values and attitudes to a child through early socialization as the operative mechanism that induces "doomed to fail" attitudes.

The dynamics of this aspect of chronic welfare dependency are less well understood than are the various factors causing individuals to enter and exit AFDC. Only recently have reliable findings on the association between the welfare dependence of parents and that of their adult children become available. Table 6.6, compiled from the PSID data base over a nineteen-year period of observation, shows the relationship between the welfare dependence of parents (when the female child was thirteen to fifteen years old) and the child's subsequent welfare dependence between the ages of twenty-one and twenty-three.

The table reveals a clear relationship between parents' dependence and that of children, but it also reveals that 64 percent of daughters from highly dependent families (defined as receiving welfare in all three study years) were not at all dependent during the follow-up period. Duncan, Hill, and Hoffman (1988, 469–470) point out that although women from welfare backgrounds have a higher incidence of welfare dependence than do women whose parents did not receive welfare (20 percent of the former group were classified as highly dependent, as opposed to 3 percent of the latter group), the analysis did not control for other factors related to receiving welfare, such as the lack of material resources available to children in impoverished mother-only families. The authors conclude that "the welfare dependence of these daughters is affected by factors other than the welfare dependence of their parents, but as yet more elaborate attempts to estimate the extent to which welfare dependence is transmitted between generations, controlling for other factors, have been inconclusive" (1988, 470).

Patterns of Welfare Use and Social-Psychological Variables

We possess a considerable body of evidence regarding factors—both personal and programmatic—that affect entering and staying in the welfare system. It is apparent that psychological or social-psychological variables such as self-esteem are not highlighted in these analyses. Goodwin argues that these studies are conducted by economists and that the data sources say "virtually nothing about the social-psychological experiences of respondents" (1983, 8). O'Neill and her colleagues did find that locus of control exercised some influence on exit probabilities, with those individuals who had a greater internal locus of control being more likely to exit quickly from welfare, but they note that this association is weak (1984, 14). Before reviewing the evidence implicating self-esteem in chronic welfare dependency, we must first discuss the theoretical and measurement issues pertaining to self-esteem (and related constructs) as used in this literature.

The Concept of Self-Esteem

Self-esteem is currently an ambiguous and poorly defined construct in the literature. Numerous conceptualizations, operational definitions, and measurement methodologies exist and often vary depending on the theoretical orientation of the researcher, the context of measurement, and the research goals. Crandall (1973) argues that self-esteem has been found to relate to a wide variety of other variables at one time or another and has no agreed-upon theoretical or measurement definition. Wells and Marwell state that "self-esteem is a deceptively slippery concept about which there is a good deal of confusion and disagreement" (1976, 5). In a compilation of thirty major self-esteem inventories (Robinson and Shaver 1973), little common ground and few consensus findings are noted among the tests reviewed.

Despite the ambiguity in the research literature, there are certain common elements normally understood as "self-esteem." These core elements include (1) a notion of reflexivity in the self, with the self necessarily divided into an observing component and an object of that observation; (2) a notion of evaluation as the self is viewed as an object; and (3) a notion of affective response to the evaluation, whereby the individual has feelings and cares about the evaluation.

Despite these common elements, the research literature is characterized more by diversity than by convergence. The lack of consistent definition is evident in the diversity of names used for self-esteem and its cognates. Coopersmith (1967) employs the term self-attitudes; Wylie (1968, 1974) uses self-regard and self-worth . Other terms include self-confidence, self-efficacy, ego-acceptance, self-acceptance, self-concept, self-affection, self-appreciation, self-evaluation, self-description, and so on. The sparsity of empirical tests of association makes it difficult to determine whether these various constructs converge into what is commonly called self-esteem. As we will see later, the evidence in support of empirical convergence is weak. Thus, in the absence of good reliability and validity tests, we cannot confidently assume that studies examining self-efficacy, for example, are indeed measuring a component or covariate of self-esteem (Crandall 1973). Nevertheless, many of the above constructs are often used interchangeably with self-esteem in the literature, even though tests for convergent construct validity are lacking.

Regardless of the diverse measurement methodologies and the weak evidence for measurement convergence, there is considerable conceptual overlap among the various theoretical perspectives on self-esteem. In general, theories stress the importance of the individual's immediate social context—particularly the family—in determining self-esteem. Rosenberg's reference group theory (Rosenberg 1965, 1967; Rosenberg and Pearlin 1978) emphasizes the immediate reference groups that individuals employ in making self-evaluative social comparisons. Coopersmith (1959, 1967) assumes that similar groups largely determine the parameters of individual notions of personal success and aspiration. The emphasis is generally on the affect received from significant others within the individual's most immediate social context, with the attitudes and values of those external to this context—the society at large—being less important for self-esteem. Minority children and adults in objectively impoverished conditions, for example, can demonstrate a relatively high level of self-esteem, even higher than that demonstrated by individuals in far better economic conditions (Rosenberg and Pearlin 1978).

Almost all approaches assume a compelling drive to maintain one's level of self-value, or self-worth, even if this requires perceptual distortion, denial, rationalization, or other defense mechanisms. Rosenberg (1965) suggests that the motive to achieve and maintain self-esteem is possibly the most powerful in the entire human repertoire.

A third area of general consensus concerns the affective, evaluative

nature of self-esteem. Self-esteem is seen as a function of the individual's feelings about the self, feelings that may be independent of what could be called self-knowledge. It is possible to be aware of negative objective aspects of the self without actually feeling poorly about oneself. To the extent that self-esteem is determined by the feelings of significant others, negative self-knowledge may indeed be peripheral to self-esteem.

The widely held assumption that low self-esteem has predictable behavioral consequences that are necessarily associated with low motivation or lack of initiative or social responsibility is not supported by the empirical literature. Indeed, evidence suggests that some individuals with low and moderate levels of self-esteem may be more driven, competitive, and motivated than individuals with high self-esteem (Wells and Marwell 1976; Graf and Hearne 1970). Cohen (1959), for example, suggests that those with high self-esteem employ more defensive strategies and repressive tendencies, which prevent challenging or personally threatening information from entering into the self-concept. Consequently, Cohen argues, people with high levels of self-esteem are less open to change, because of their rigid thought patterns and excessive use of psychological defenses. Similarly, Taylor and Brown (1988) concur that individuals with low self-esteem are most likely to maintain more accurate and reality-based impressions of their abilities, potential, and self-image. Other research suggests that persons with low self-esteem have greater social competence (Achenbach and Zigler 1963) and higher interpersonal stability (Neuringer and Wanake 1966). In sum, this research concludes that common assumptions regarding the relationship between self-esteem and behavior are not unequivocally correct. Neither, however, is there unequivocal support for a positive relationship between low self-esteem and pro-social behavior. On balance, the relationship as studied to date seems indeterminate.

Related Constructs

Numerous psychological and social-psychological concepts that refer to the self have been shown to relate to self-esteem in both theoretical and, to a lesser extent, empirical domains. Again, the reported empirical associations vary depending on the particular definition of self-esteem employed.

Coopersmith (1967) suggests that one's sense of efficacy and ability is highly related to self-esteem, in terms of the influence of self-esteem on the individual's perceptions of personal success. Utilizing data from

both face-to-face interviews and a series of objective personality tests, Coopersmith operationalizes personal efficacy as individual competence in dealing with social, academic, and personal matters and reports that self-esteem was consistently positively related to the various indices of personal efficacy. Other research supports this finding, though less convincingly. For example, Woodworth (1958) suggests that efficacy and self-esteem are conceptually interrelated, but provides little supporting evidence.

A sense of personal effectiveness in dealing with life's challenges and obstacles undoubtedly affects how individuals evaluate themselves. But, as Korman (1970, 1976) suggests, there may be various types of self-esteem that are localized around specific behaviors or tasks confronted by the individual. He suggests that there are at least three independent kinds of self-esteem, which are differentially derived from social interaction, task-specific performance, and more general evaluations of the self. From this perspective, it is likely that an individual who is effective in one domain will have high self-esteem specific to that area of performance, whereas other domain-specific areas of self-evaluation will be unaffected.

Another construct related to self-esteem involves Rotter's (1966) notion of internal or external locus of control, which suggests that individuals differ in ascribing causality for personally relevant events. Some believe they are personally responsible for their successes and failures and are labeled "internals." Others tend to externalize the causes of events and are labeled "externals." One might expect that "internals" would feel more competent than would those individuals who believe that life events are controlled by outside forces, a belief that could lead to feelings of helplessness and inadequacy. Hesketh (1984), reviewing a number of studies measuring the association between self-esteem and locus of control, reports a range of correlations between. 24 and. 30 and suggests that the two variables are only moderately related. In general, it has been found that those with high self-esteem tend to have an internal locus of control, whereas those with low self-esteem are more external (Bhagat and Chassie 1978; Kishor 1981). It would appear that a moderate but statistically significant relationship does exist between the sense of personal control and efficacy and measures of self-esteem. Wells and Marwell (1976), however, suggest that many researchers use the locus of control and sense of efficacy measures as expedient "proxy measures" of self-esteem, although the substitution is not entirely justified, given the relatively low levels of association between the two constructs. Never-

theless, numerous studies do present conclusions regarding self-esteem that are derived from items designed to measure either internal/external locus of control or personal efficacy.

The reported low levels of empirical association between self-esteem and certain related constructs does not necessarily imply that personal efficacy and locus of control are not components of self-esteem. Given the relatively loose association between the various indicators of selfesteem, it is not surprising to find that various conceptually related constructs bear a weak empirical relationship to commonly employed measures of self-esteem. It may be, once again, that these related constructs are highly associated with specific components or types of self-esteem (i.e., social self-esteem, or task-specific self-esteem), as Korman (1976) suggests, but the relationships may be obscured in empirical tests employing global self-esteem indices.

Relating Self-Esteem to Chronic Welfare Dependency

Only a small body of research exists that is directly relevant to the relationship between self-esteem and the related psychological and social-psychological constructs and chronic welfare dependency. Some of these studies are strictly cross-sectional (e.g., Carson 1967; Druga 1986; Goodban 1985; Jayaratne et at. 1980; Bishop 1974); others are based on longitudinal data (Handler and Hollingsworth 1969; O'Neill et al. 1984; Hill et al. 1985; Nichols-Casebolt 1986; Goodwin 1983). Duncan, Hill, and Hoffman point out the difficulty in making causal associations between welfare dependence and "values and attitudes" (1988, 239). They note that cross-sectional association between attitudes and welfare receipt is inadequate for gauging a link between welfare dependency and psychological characteristics. They suggest that three criteria are required to find a consistent link between receiving welfare and psychological variables:

1. Recipients should have values and attitudes that are measurably "worse" than those of nonrecipients;

2. Attitudes and values should be adversely affected by welfare receipt; and

3. Initially worse values and attitudes should increase the likelihood of future dependence.

In this formulation, the authors ignore possible positive relationships between receiving welfare and psychological variables such as self-esteem, although as a whole their review of welfare dependence clearly identifies the positive consequences of welfare benefits for recipients. Duncan and Hoffman (1987) note that most analyses of welfare dependence and its effects have only crudely estimated the characteristics of the welfare system itself that potentially affect the recipient's behavior and that most research has not adequately addressed the nonwelfare alternatives available to the recipient.

Cross-sectional evidence for a relationship between self-esteem and welfare dependence is inconclusive. Carson (1967) tested relationships between welfare receipt (yes/no) and self-esteem, using the Social Vocabulary Index (SVI) with 81 welfare and 116 nonwelfare clients applying for services at the Utah Division of Vocational Rehabilitation. According to Carson, the SVI yields scores relating to self-concept, self-acceptance, ideal self-concept, and concept of others. This early study found no differences between welfare recipients and nonrecipients on the SVI measures. To further test the relationship, length of time on welfare was subdivided into short, intermittent, and long (more than three years) and tested as an independent variable in relation to SVI scores. Again, expected variations in self-concept among subgroups were not found. The author offers a post hoc explanation that the excessive defensiveness of welfare clients may keep them from admitting to a lowered self-concept, a view that Carson finds corroborated by the fact that chronic welfare clients scored highest on scores of self-concept and self-acceptance.

Bishop (1974), in a study of 231 black women who had been on welfare at least once, did find expected differences—with current welfare recipients evidencing lower self-esteem. Druga (1986) also looked at length of time on aid (but measured at a particular point) as it related to the social-psychological makeup of a sample of 75 AFDC recipients. The long-term recipients (on AFDC for six consecutive years) were not found to differ significantly from the shorter-term recipients on eleven items chosen from Rotter's (1966) locus of control scale. Druga concludes from these data that welfare use may best be understood by examining systemic rather than personal variables.

Goodban (1985) analyzed certain psychological effects of receiving welfare assistance among a group of one hundred single, black AFDC recipients in New Haven, Connecticut. (Fifty-five had been on aid more than five years.) Self-esteem was defined with a single item asking a

woman "whether welfare made her feel differently about herself" (1985, 415). Fifty-five percent of the respondents stated that welfare did indeed make them feel differently about themselves. Goodban reports a relationship between this self-attitude and (1) lower subjective social class; (2) a belief in not having control over going on welfare; and (3) a belief that opportunities are generally available in America. Goodban summarized her data as attesting to the extreme psychological (and economic) hardship of the welfare role, presumably also referring to the fact that a majority of respondents reported being stigmatized (embarrassed or ashamed) by being on welfare and receiving food stamps. Although more than half of the respondents had been on aid for six years or more, no data are reported tying length of time on aid to the socialpsychological variables. Furthermore, the self-esteem measure used in the study is suspect—we are unclear as to how the item references the self-esteem concept, to say nothing of the reliability or validity of the measure. Nevertheless, based on her finding that recipients who believe in individualistic explanations for going on welfare and who report feeling a high degree of control over going on welfare are more affected in terms of their self-esteem, Goodban concludes that this constellation of attitudes is counterproductive in that it discourages legitimate activism by the recipients (i.e., seeking all legitimate entitlements and possibly joining a welfare rights organization).

Longitudinal or experimental studies are more appropriate than strictly cross-sectional studies for analyzing questions about relationships between self-esteem and chronic welfare dependency. Analyses of both PSID and NLS data, as well as two studies that collected data at two points in time only (Handler and Hollingsworth 1969; Goodwin 1983), are discussed below.

Handler and Hollingsworth (1969) investigated feelings of stigma, other attitudes, AFDC program use, and the probability of leaving welfare among a sample of 767 AFDC recipients in Wisconsin over a period of approximately two years, beginning in June 1967. Stigma was conceptualized as (1) feeling embarrassed or uncomfortable when interacting with people who do not receive AFDC; and (2) a generalized feeling held by the recipient about the community's degree of understanding or hostility toward AFDC recipients. More than half of the subjects reported feeling stigmatized in both respects. Although this is not a direct measure of self-esteem, the idea of stigma (Goffman 1963) incorporates both the devaluation of individual attributes by society and the internalization of this general opinion, reflected in a devalued self-

estimation. Handler and Hollingsworth note that recipients who reported feeling stigmatized differed in several respects from the nonstigmatized group, feeling less satisfied with the AFDC program and less accepting of unannounced home visits by the caseworker. But the authors report apparently paradoxical findings, for stigmatized recipients were more aggressive in asserting their rights to a variety of program benefits, were likely to leave the welfare system sooner than those recipients who did not report feeling stigmatized, and were more likely to exit AFDC as a result of their own efforts (including finding work and getting married). The authors conclude that "feelings of stigma . . . seem to reflect an independent cast of mind. Recipients who have these feelings are upset about being on welfare. . . . Moreover, they do something about their situation" (1969, 17–18).

Goodwin (1983) studied AFDC recipients enrolled in the Work Incentive Program (WIN) in 1978, as well as a comparison group of individuals receiving unemployment insurance in New York and Chicago. Goodwin expressly attempted to uncover the attitudinal and social-psychological factors associated with both economic and attitudinal outcome measures from a one-year follow-up. Regrettably, he was able to achieve only a 57 percent reinterview rate. Among the attitudinal measures included were (1) "expecting economic independence"—intending or expecting to work and not receive aid at a later time; (2) feeling that it was proper to support one's family by working; and (3) a measure of general self-confidence incorporating items seemingly related to locus of control and to self-esteem. Two items from Goodwin's self-confidence scale provide the flavor of this measure: "When I make plans, I am almost certain that I can make them work"; "I am able to do most things as well as other people" (1983, 23).

Goodwin found achieving economic independence at time-2 to be related to several factors at time-1: the expectation of achieving independence, the length of time spent on aid, and previous job status. The self-confidence measure was not reported as significant in the regression model. Further, Goodwin noted a spiral of causation, in which expectations for independence at follow-up were influenced both by the earlier level of those expectations and by the independent influence of having achieved economic independence in the interim. Again, the self-confidence variable was found to be irrelevant.

Several analyses of the large-scale longitudinal data bases—the NLS and the PSID—have attempted to uncover the relationships between psychological factors and economic mobility. One study (Hill et al.

1985), while not limiting its research to welfare cases, reviewed the PSID findings on motivation and economic success. The PSID employed Atkinson's theory of motivation (Atkinson 1964, cited in Hill et al. 1985), differentiating between more basic motives (including need for achievement, need for affiliation, and need for power, as well as the negative motivation, fear of failure) and expectancies (defined as the "individual's assessment of the chances that his or her own performance will, in fact, lead to a desired outcome" [Hill et al. 1985, 4]). This motivation theory holds that action is precipitated by the stable basic motives plus an expectation that action can affect an outcome in a particular set of circumstances. To measure expectancy in the PSID, a personal efficacy index was constructed, consisting of two items asking respondents whether they "usually felt pretty sure their life would work out the way they wanted" and whether, "when they made plans ahead, they usually carried them out" (PSID items cited in Hill et al. 1985, 28). These items seem to connote some certainty about controlling outcome, making this efficacy index seem more akin to a locus of control concept than to the generalized self-evaluation suggested by self-esteem.

The findings on motivation and economic well-being show expected correlations (rather weak) between the efficacy index and the other motivational indices and current measures of economic well-being, with the exception that welfare income correlated negatively with the motivational measures. These data support an inference that receiving welfare is related to motivational and efficacy states. As the authors note, however, causality was indeterminate in this analysis. To measure the direction of causation between attitudes and economic status, the authors constructed a model testing the effects of prior motivational levels on subsequent economic well-being. With this procedure, "no more than modest and usually insignificant effects of basic motives on economic outcomes [were found] and . . . no consistent effect of expectancies on outcomes" (Hill et al. 1985, 7). Rather, changes in the efficacy index seem to be predicted by "various economic and non-economic events" (77). Thus, when the cross-sectional correlations are subject to longitudinal causal analysis, it appears (in the PSID) that the attitudes (expectancies) are more reflective than predictive of economic status.

O'Neill et al. (1984) analyzed the contribution of psychological factors to welfare dependency in both the PSID and the NLS samples. As previously noted, the PSID contained an efficacy index, and the NLS contained items from Rotter's locus of control inventory that the O'Neill study equates with the PSID efficacy index, with internal control denot-

ing greater efficacy, that is, certainty that plans will turn out as intended. For the NLS (Rotter scale) data, a more external attitude was associated with a greater degree of welfare dependency, although the effect was not significant. Furthermore, locus of control affected welfare dependency in the case of welfare exit as a result of earnings only (with external outlook decreasing the probability of exit as a result of earnings), but it did not affect exits that occurred because of marriage. Replicating that analysis in the PSID sample produced equivalent results: the efficacy index (in the weighted regression) had the right sign in relation to welfare dependency, but the result was not statistically significant.

In these studies, the factors delineated earlier in our discussion of welfare dynamics (individual factors such as education, marital status at first birth, labor market experience, and disability, as well as program factors) are the variables that seem to control long-term dependence on aid, with the efficacy index as measured not contributing significantly to the observed outcomes over time. The O'Neill et al. study, however, despite controlling for a variety of relevant variables, was unable to account for the effect of duration itself on subsequent duration. In an attempt to discover whether an individual might "lose motivation and self-confidence as a result of being on welfare and [whether] these attitudes . . . perpetuate welfare dependency" (1984, 85), an analysis was conducted of the effects of enrollment in AFDC on attitudes over time. The effect of AFDC was, as expected, negative on efficacy, but again the effect was not statistically significant. Analyses with both the PSID and the NLS data yielded similar results on this question.

Nichols-Casebolt (1986) conducted an analysis similar to that reported by O'Neill et al. (1984) with the PSID, focusing her entire inquiry on the psychological effects of income-tested programs (those which provide assistance to low-income clients only), in order to evaluate the "stigma" argument—that recipients "incorporate the general negative views of themselves held by others, thus reducing the recipient's feelings of self-worth" (1986, 288). The psychological concept employed was "psychological well-being" (PWB), measured by four PSID items covering satisfaction with self and sense of control over intended outcomes. This conceptualization seems to blend a notion of self-esteem with the concept of locus of control. Nichols-Casebolt, however, argues that these dimensions of self-satisfaction and personal competence "incorporate essential elements in overall PWB" and relate to the major arguments about the effects of stigma as they focus "on the loss of self-esteem and the increased dependency of recipients" (1986, 289).

As did O'Neill et al., Nichols-Casebolt presents both cross-sectional

and longitudinal findings. Table 6.7 reproduces her findings for cross-sectional data by race for a subgroup of PSID subjects (female heads of household less than fifty-four years old, with children under the age of eighteen, from data collected between the years of 1968 and 1972) for different categories of income support. As before, cross-sectional data support the implication that income-tested support is associated with a decrease in PWB in a statistically significant fashion. To test the causal relationships in the data, a panel model was used that incorporated data on all welfare cases, short-term as well as long-term,^[*] and evaluated change in PWB as a function of (1) change in AFDC status in the prior year, and (2) possible changes in other personal characteristics during that time (income, number of children, age of youngest child, and work status). In this model, PWB was differentiated into its two measured components, self-satisfaction and personal competence, and the analysis was conducted separately for blacks and whites. Results reveal that entering AFDC had a statistically significant impact in lowering feelings of self-satisfaction and personal competence for blacks, but not for

[*] * Nichols-Casebolt proposes that a test of the model using changes in PWB during an AFDC spell would be advisable; in fact she ran such an analysis and reported that the "results were in the expected direction—increasing years on AFDC was related negatively to PWB—but the coefficient was not significant" because of sample sizes (1986, 302n. 20).

whites. Six of eight coefficients for blacks and whites were, however, in the expected direction.

Nichols-Casebolt concludes that these results, taken together with the cross-sectional data, show that for blacks the lower psychological well-being of AFDC recipients was a result of being on AFDC. She also argues that her data may underestimate the effects of AFDC on PWB, because of the method of analyzing each year of aid as a distinct entity, and that initial AFDC receipt may capture a "relief" phenomenon, presumably associated with the lessening of stress the recipient experiences when benefits are first received. Nichols-Casebolt speculates that "once the financial crisis is alleviated, however, the negative psychological consequences of AFDC begin to be felt" (297). We should point out that no data are presented in the analysis to either confirm or reject this speculation and that the data concerning effects on PWB subsequent to exiting AFDC, in which one would predict improved PWB as the negative stimulus ends, do not support that interpretation.

Has the Relationship Between Self-Esteem and Chronic Welfare Dependency Been Tested?

From the available evidence, we conclude that there has been only a partial test of a possible relationship between self-esteem and chronic welfare dependency in the empirical literature. Adequate testing would include (1) adequate measurement of self-esteem in the mainstream tradition of that literature; (2) a focus that allows us to differentiate between chronic and nonchronic welfare use, as these populations differ considerably; and (3) relatively clear causal paths that help to resolve the problem of what comes first, attitudes or dependency. We might also wish for a measure of welfare dependence that takes into account nonwelfare income and for a data set that allows us to understand how variations in AFDC programs influence observed differences in self-esteem and chronic use.

Given these criteria, we can see the limitations in the evidence available to judge self-esteem's contribution to this social problem. Perhaps the most severe drawback is in the measurement of the self-esteem variable. Jayaratne and colleagues note that although the concept of self-esteem is widely studied in social psychology, "very little information is available on adults in general, and virtually none on welfare recipients" (1980, 285; emphasis added). The cross-sectional studies all used different measures of self-esteem, and, as we have noted, studies measuring

locus of control and efficacy are not equivalent to self-esteem studies, for associations between these constructs are weak. The longitudinal studies use explicit theories of motivation (needs and expectancies) to guide their instrumentation. Although these theories try to identify the most likely attitudinal or motivational contributors to economic outcomes, they do not provide direct measures of self-esteem.

Taken as a whole, the studies have included a roster of approximations for measured self-esteem, ranging from single items about whether individuals feel "different" about themselves or are satisfied with themselves to multiple-item clusters representing efficacy, psychological wellbeing, stigmatization, evaluative self-concept, and self-esteem. All told, the record of inquiry includes a dozen or so studies of evaluative self-attitudes and how these relate to receiving welfare. A number of the studies, especially those which by design delve into the causal paths, may contain not self-esteem measures but efficacy measures. Some authors (e.g., Nichols-Casebolt) maintain that these measures directly pertain to self-esteem. Our review of self-esteem measurement leads to the conclusion that these measures are not synonymous but appear to be conceptually related. Although it appears that not a single study adequately measured the self-esteem variable, this is probably true precisely because no consensus or rationale for adequate measurement of the ambiguous concept has emerged.

Does Level of Self-Esteem Cause Welfare Dependency?

We have reviewed the major direct evidence on this question and find that although cross-sectional studies might lean toward positing a relationship between low self-esteem and dependency, the longitudinal data generally negate the observed finding. It is true that the longitudinal data bases provide less adequate measures of self-esteem than do certain of the cross-sectional studies. Goodwin (1983), however, did include a self-concept scale in his follow-up research and did not conclude that self-attitudes were important for later status and expectations of independence.

As to the possibility that high self-esteem causes welfare dependency, the evidence to evaluate this relationship seems inadequate. Two theoretical avenues have been suggested. The first argues that individuals must have considerable persistence and resilience to overcome the bureaucratic barriers associated with the welfare system and that these

traits should accompany high self-esteem. The second suggestion, more theoretical, is based on certain research findings that more aggressive and "driven" individuals may have lower self-esteem. But the data are inadequate to test these assertions, although the study by Handler and Hollingsworth (1969) does in fact suggest that lower self-esteem (measured as part of a "stigma" concept) is associated with earlier exit from welfare. We find no evidence relevant to the idea that high self-esteem allows an individual to cope with the public welfare bureaucracy.

Does Welfare Dependency Affect Self-Esteem?

The evidence implicating chronic welfare dependency in the lowering of self-esteem, though essentially still inconclusive, seems stronger than the evidence linking prior self-esteem to the likelihood of chronic welfare dependency. Some investigators note that welfare provides material benefits that, at least in the short run, reduce uncertainty and may resolve an economic crisis for the individual initially qualifying for aid. Thus any decrease in self-esteem as a result of long-term welfare use would appear over a longer period and may at first be masked by this short-term relief. As Nichols-Casebolt (1986) notes, however, no data set exists with which to test this hypothesis. Duncan, Hill, and Hoffman (1988) point out that research has generally avoided analyzing the positive contribution welfare makes to individual and family circumstances and that the overall social judgment of the program should include the benefits in education, nutrition, health care, household stability, assistance with child-rearing, and so forth that are received especially by children.

Several analysts summarize the evidence in this area as mixed but generally indicating some deterioration in self-esteem as a result of welfare dependency. Rainwater (1982) argues that the stigmatizing effects of receiving welfare may be evident not only in the lowering of measured self-esteem but also in depression, denial, and defining oneself as sick or disabled, areas not subject to clear-cut observation by written selfesteem tests. Nichols-Casebolt (1986) points out the surprising lack of evidence addressing this deeply held belief that welfare stigma undermines psychological well-being (including self-satisfaction). She finds in her PSID research that, for blacks, receiving AFDC does result in lowered psychological well-being. In general, Nichols-Casebolt believes that many research designs tend to be biased against finding any relationship at all between the variables (principally because of measure-

ment and time-lag problems); thus she views the emergence of even weak findings as suggestive of a trend implicating welfare dependence in lowered psychological well-being. O'Neill and colleagues (1984), however, while noting some findings tying overall exposure to AFDC with lowered efficacy, conclude that the relationship is not strong enough to support the idea of a causal relationship between AFDC and efficacy.

Although the evidence linking chronic welfare dependency to a loss of self-esteem remains inconclusive, certain trends in the data, combined with the methodological drawbacks of the existing research, lead us to conclude that some weak effect of welfare dependence on the variables of sense of control and personal efficacy deserves further exploration. As yet, no finding ties persistent welfare dependence to a generalized loss of self-esteem.

Theoretical and Research Directions

Further research on self-esteem and welfare dependency depends on two features: improved measurement of psychological variables and adequate longitudinal data. Experience to date reveals that when either feature is lacking, few findings of value emerge. Based on the concept of self-esteem as an overall evaluative self-attitude, which is strongly affected by social context and which individuals possess a powerful drive to sustain at a functional level, there is little reason to believe that self-esteem is a strong influence on chronic welfare dependency. If parenthood (single or otherwise), possible adherence to deviant attitudes, or membership in social groups that do not stigmatize welfare dependence can sustain self-esteem—and theoretically there is no reason to suspect that they cannot—then self-esteem per se is unlikely to emerge as a strong causative factor in predicting which individuals will become chronically welfare dependent.

Learned Helplessness . There is weak evidence to suggest that certain psychological factors do play a role in persistent welfare dependency. Research on caseload dynamics has found that duration of welfare itself influences the probability of remaining on aid. This is true even when the research controls for all of the other factors related to duration, such as family size, educational level, labor market involvement, and so forth. It is certainly reasonable to assume that chronic material deprivation below the poverty level (AFDC standards are consistently below the poverty level), coupled with the external controls associated with welfare

dependency, may have psychological consequences. According to Kane (1987), victimization and the loss of control associated with chronic poverty may breed dysfunctional psychological coping mechanisms, which are best explained by the theory of reactance/learned helplessness.

The reactance/learned helplessness model (Wortman and Brehm 1975; Taylor and Reitz 1968), which has substantial empirical support in the social-psychological literature, describes how individuals differentially react to threats to their personal control. The theory suggests that individuals, when faced with uncontrollable situations, initially react with anger and aggression, symptoms of "reactance," in efforts to reassert personal control. If the loss of control persists despite individual initiatives, the individual will experience helplessness, which is eventually internalized after prolonged repetition and can become a generalized chronic state. The helpless state is characterized by motivational deficits, apathy, resignation, and an inability to recognize and respond to new opportunity (Seligman 1975).

Pettigrew (1980) and Kane (1987) argue that the reactance/helplessness cycle helps to explain certain behaviors often characteristic of welfare recipients. These authors suggest that welfare clients who are most active, motivated, and assertive are representative of "reactive types," whereas others (possibly those chronically dependent on welfare) suffer from the motivational deficits induced by a conditioned sense of helplessness, which prevents them from taking advantage of opportunities when they do occur. Kane argues that the poor in general, and many welfare recipients in particular, experience nonresponsive bureaucratic environments over which they have little control. Such prolonged lack of control eventually generalizes even to situations where control can be asserted but is not. Learned helplessness may thus prevent welfare recipients from recognizing new opportunities. By implication, then, this view predicts that structural changes alone may not be sufficient to move chronically welfare-dependent individuals to a state of independence.

Further research using the reactance/learned helplessness paradigm may help to explain why chronic welfare dependency extends for a duration unexplained by any other measured factors.

Conclusion

Is welfare dependency caused by the attitudes and values of welfare recipients? Are welfare dependency and the attitudes and values of wel-

fare recipients the products of the persistent experience of economic disadvantage and powerlessness, including the experience of welfare itself? The data needed to answer these questions with confidence do not exist. We do know that the major immediate cause of chronic welfare dependency is chronic income deficiency. Those who are chronically welfare dependent constitute a diverse group, but they all share this basic overriding characteristic of too little income.

Only four lawful sources of income are available to Americans: (1) employment; (2) membership in a family with income to share; (3) capital ownership; and (4) transfers. Overcoming income deficiency means increasing income from one or more of these sources.

Nearly 50 percent of all Americans are in the labor force. The vast majority are able to exchange their labor for an income above the poverty level. About 6 percent of the labor force is involuntarily unemployed. Others are underemployed or employed at wages so low they cannot support their families at or above the poverty level. Still others are "discouraged workers," that is, they have ceased looking for work because of repeated failures in finding employment. Although employment is a significant source of income for most Americans, it is an income source now not fully available to all. There is little evidence that low self-esteem is a major barrier to participation in the labor force.

The largest group of Americans not in the labor force is composed of the nation's children. Although most children are members of families with income to share, changes in patterns of marriage, divorce, out of wedlock childbirth, and the structure of the (increasingly contingent) labor force have created growing numbers of families (particularly female-headed families) with income deficits. Clearly, not all Americans hold membership in families with income to share.

Capital ownership, in the form of dividends, rent, and interest, is a source of income for many Americans. But a family of four would need working capital in excess of $ 150,000 to generate an income at the poverty level from this source, and only a very small fraction of all American families have such capital holdings.

Transfers (Social Security and related public benefit programs) are an important income source for many Americans. Many elderly persons and dependent children are taken out of poverty by these programs. But others may not be entitled to assistance from any existing program or may receive benefits that are below the poverty level and thus do little to relieve economic need or promote independence.

When income is pooled from these several sources, some 13.6 per-

cent of all Americans are still left with income insufficient to meet their own and their families' needs. In 1983, 58 percent of all impoverished families with children in the United States received benefits from the AFDC program.

Social policy related to such poor children and their families is dominated by two conflicting objectives—to reduce their poverty by providing material benefits above a defined standard of health and decency, and to reduce or prevent their dependence on government. Efforts to reconcile these conflicting objectives within a single "welfare" program have left the nation with a very high level of unrelieved child poverty and a very substantial level of long-term dependence among welfare recipients. The program of Aid to Families with Dependent Children reduced the national child poverty rate by only 4 percent in 1985. More than half of all recipients depended on this aid for more than four years. The average length of dependence on AFDC benefits was almost seven years; slightly less than 25 percent of recipients depended on cash payments for ten or more years. The AFDC program has essentially functioned as a cash payment program, doing little to assist recipients in overcoming barriers to self-sufficiency (deficient language skills, lack of job skills, child care needs, poor health) and also doing little to relieve child poverty. Furthermore, some research plausibly suggests that long-term welfare dependence may undermine an individual's perceived capacity to exercise control over events in his or her personal life.

Public policies that fall so far short of their intended goals—to relieve poverty and to reduce dependency—are clearly in need of change. An effective public policy response should include both the provision of income adequate to bring poor children and their families above the poverty level and efforts to promote self-sufficiency. Much of the current debate on welfare reform has focused on public policy initiatives designed to reduce poverty and dependency by changing the mix of income sources available to poor families from employment, family support, and income transfers.

Employment

The current interest in "workfare" has as its focus a desire to remove long-term welfare recipients from the welfare rolls and to replace their welfare benefits with earnings. It seems clear that, with assistance (child care, education, training, job placement, counseling, and so on), potential exists to increase income from employment for some welfare recipients, though that increase will, at best, be modest (Gueron 1988). Such

programs often include motivational components designed to alter the attitudes (including self-attitudes) of participants. Efforts to change motivation are logical components of a work training program for welfare recipients, but no good research is available that tests their effectiveness. Achieving the dual goals of reducing dependency and poverty requires balanced attention to ensure that earnings—or some combination of earnings and public benefits—are at or above the poverty level.

Family Support

The AFDC program was conceived as a program to protect the children of dead or disabled fathers. Today, it also serves the children of absent fathers, who often contribute little to the support of their children. Child support programs are now being advocated to reduce both poverty and dependency through increased parental support. Uniform levels of child support awards, financed by an income tax on the absent father, have been identified and are being tested in pilot and demonstration programs (Garfinkel, McLanahan, and Wong 1988). When fully implemented, such programs could have a modest impact on the size of the AFDC caseload.

Income Transfers

The Tax Reform Act of 1986 made an important step forward by removing most poor families from the income tax rolls and reducing the extent to which marginal families are taxed into poverty. Additional tax reform proposals that might use our tax structure to help alleviate poverty are now under discussion. Among these proposals, the replacement of the $ 2,000 personal exemption (a regressive tax feature) with a per capita refundable tax credit is the most far-reaching.

Workfare programs could, of course, have their greatest immediate impact on income from salaries and wages. Education, training, improved health care, and social services offered as part of such programs could also have a long-term beneficial impact on children and parents who care for them that is not work-specific. Added income from employment and from child support programs could relieve the poverty and dependency of many families. Better-targeted support to poor families through the income tax system would add substantially to their incomes and avoid the dangers of stigmatization, bureaucratic control, the disincentive effects of high marginal tax rates on earnings, and the high administrative costs of welfare programs.

These tax-based reforms, together with programs designed to assist employable welfare recipients, could have a considerable impact on poverty and dependency. Nevertheless, many families would still suffer income deficits that leave them below the poverty level. The profound problems of education, training, health, housing, and child care that welfare recipients face suggest that solving the problem of chronic welfare dependency will require a comprehensive, urgent effort by every one of our human service institutions to address the extraordinary needs of these families. It will also require conscious efforts to open new job opportunities for those excluded from the labor force or only marginally employed.

Even with improved access to income from employment, family support, and transfers, a residual income-tested welfare program will be needed to meet the income deficits of some poor families. The new opportunities envisioned in these policy directions presuppose a target population with the motivation and ability to take advantage of opportunities. As suggested earlier, chronic welfare dependency and its associated chronic poverty may leave some recipients with neither the hope nor the confidence that different behaviors will be differently rewarded. Some will need help in overcoming learned behaviors that are no longer functional in an environment of new opportunity. Psychological services directed to enhance client motivation and ability may be useful under such circumstances. But such services are not likely to succeed as substitutes for genuinely expanded opportunities.

Future Research

This review of the research literature indicates that substantially more work needs to be done in conceptualizing and investigating the major variables identified. We have only a beginning understanding of the dynamics of poverty and welfare dependency and the psychosocial implications of those states for the individuals and families involved. More research is needed to inform policy responses to questions that are of increasing public concern and interest. Several promising research directions have been suggested.

Unfortunately, there is no quick pathway through the process of study and analysis. There is a desperate need for longitudinal data that can help us understand the movement of individuals and families across the life span and the factors that influence participation in the civic, social, and economic life of the community. Much more will have to be

invested in studies and in pilot and demonstration projects to test new ideas and approaches before moving to full implementation of new policies.

In Assembly Bill 3659, the California legislature called public attention to the problems of chronic welfare dependency, crime and violence, alcoholism and drug abuse, teenage pregnancy, child abuse, and academic failure. Although we have some idea of the number of people involved in each problem area, we have little knowledge about whether these are discrete or overlapping populations. Recently, increasing attention has been paid to the possible existence of an "underclass" in American life—a class of people who are outside the nation's economic and social mainstream. There is no rigorous or accepted definition of the term underclass, but Reischauer (1987) has summarized the most widely agreed-upon characteristics of the group as (1) persistently low income; (2) a weak attachment to the labor force; (3) a lack of education and skills exemplified by functional illiteracy and dropping out of school; (4) a sense of alienation from mainstream society, possibly accompanied by substance abuse, criminal activity, participation in the underground economy, fathering or bearing a child out of wedlock, or long-term welfare dependency; (5) membership in a minority group that is discriminated against; and (6) residence in a neighborhood with a high concentration of disadvantaged or dysfunctional behavior.

According to Reischauer (1987), if the underclass is defined as the nonaged, nondisabled population that has persistently low income, lacks skills or an education, and has a limited attachment to the labor force, it could represent as much as 2.4 percent of the nation's population and 15.8 percent of its poor. If one subtracts from this group those who are not black or Hispanic and who do not live in a large city, the size of the underclass shrinks to 0.8 percent of the total population and 5.6 percent of the nation's poor. Although these numbers may underrepresent single young black males, the homeless, and those engaged in illegal activities, the size of the problem population may be relatively small. But the profile of deprivation and disadvantage and its implications for public policy are very great.

It is important that problem-specific analyses of the "intractable problems" identified in Assembly Bill 3659 not lead us to overlook their overlapping existence in a single population group or the search for broader concepts of causation. Obviously, we know very little and need to know much more. Legislative support of research and testing relevant to public policies addressing these problems is needed.

Bibliography

Achenbach, T., and E. Zigler, eds. 1963. Projective Psychology: Clinical Approaches to the Total Personality . New York: Knopf.

Anastasi, A. 1961. Psychological Testing . New York: Macmillan.

Atkinson, J. W. 1964. An Introduction to Motivation . Princeton, N.J.: Van Nostrand.

Bane, M. J., and D. T. Ellwood. 1983. The Dynamics of Dependence: The Routes to Self-Sufficiency . Cambridge, Mass.: Urban Systems Research and Engineering.

———. 1986. "Slipping Into and Out of Poverty: The Dynamics of Spells." Journal of Human Resources 21(1): 1–23.

Bhagat, R. S., and M. B. Chassie. 1978. "The Role of Self-Esteem and Locus of Control in the Differential Prediction of Performance, Program Satisfaction, and Life Satisfaction in an Educational Organization." Journal of Vocational Behavior 13: 317–326.

Bishop, E. S. 1974. "The Self-Concept of the Welfare Mother: Some Sociological Correlates." Wisconsin Sociologist 2(1): 10–16.

Branch, A., J. Riccio, and J. Quint. 1984. Building Self-Sufficiency in Pregnant and Parenting Teens . New York: Manpower Demonstration Research Corp.

Brehm, J. W. 1966. A Theory of Psychological Reactance . New York: Academic Press.

Bumpass, D. L. 1984. "Children and Marital Disruption: A Replication and Update." Demography 21 (February): 71–82.

California. State Assembly Bill 3659. 1986. An Act to Create the California Task Force to Promote Self-Esteem and Personal and Social Responsibility . September.

Carson, G. L. 1967. "The Self-Concept of Welfare Recipients." Personnel and Guidance Journal 45(5): 424–428.

Center for Human Resource Research, Ohio State University. 1980. Findings of the National Longitudinal Survey of Young Americans, 1979 . Youth Knowledge Development Report 2.7, Research on Youth Employment and Employability Development. U.S. Department of Labor. Washington, D.C.: Government Printing Office.

Coe, R. D. 1981. "A Preliminary Examination of the Dynamics of Welfare Use." In Five Thousand American Families: Patterns of Economic Progress, vol. 9: Analyses of the First Twelve Years of the Panel Study of Income Dynamics, edited by M. S. Hill, D. H. Hill, and J. H. Morgan, 121–168. Ann Arbor, Mich.: Institute for Social Research.

Cohen, A. R. 1959. "Some Implications of Self-Esteem for Social Influence." In Personality and Persuasibility, by I. L. Janis et al., 102–120. New Haven,

Conn.: Yale University Press.

Coopersmith, S. 1959. "A Method of Determining Types of Self-Esteem." Journal of Abnormal and Social Psychology 59: 78–88.

———. 1967. The Antecedents of Self-Esteem . San Francisco: Freeman.

Crandall, F. 1973. "The Measurement of Self-Esteem and Related Constructs." In Measures of Social Psychological Attitudes, edited by J. P. Robinson and P. R. Shaver, 45–159. Ann Arbor, Mich.: Institute for Social Research.

Danziger, S. 1988. "Antipoverty Policy and Welfare Reform." Paper prepared for the Rockefeller Foundation Conference on Welfare Reform, Williamsburg, Va., February 16–18.

Druga, G. S. 1986. "Locus of Control and Patterns of Welfare Use." Ph.D. dissertation, Case Western Reserve University.

Duncan, G. J., M. S. Hill, and S. D. Hoffman. 1988. "Welfare Dependence Within and Across Generations." Science 29 (January): 467–471.

Duncan, G. J., and S. D. Hoffman. 1987. "The Use and Effects of Welfare: A Survey of Recent Evidence." Institute for Social Research, University of Michigan, Ann Arbor. Typescript.

Ellwood, D. T. 1986a. Targeting "Would-Be," Long-Term Recipients of AFDC . Princeton, N.J.: Mathematica Policy Research.

———. 1986b. "Working Off Welfare: Prospects and Policies for Self-Sufficiency of Female Family Heads." John F. Kennedy School of Government, Harvard University. Typescript.

Fish, B., and S. Karabenick. 1971. "Relationships Between Self-Esteem and Loss of Control." Psychological Reports 29: 784.

Fitch, G. 1970. "Effects of Self-Esteem, Perceived Performance, and Choice on Causal Attributions." Journal of Personality and Social Psychology 16(2): 311–315.

Fitts, W. 1964. Manual for the Tennessee Self-Concept Scale: Nashville, Tenn.: Counselor Recordings and Tests.

———. 1970. Interpersonal Competence: The Wheel Model . Nashville, Tenn.: Counselor Recordings and Tests.

Garfinkel, I., ed. 1982. Income-Tested Transfer Programs: The Case For and Against . New York: Academic Press.

Garfinkel, I., and S. McLanahan. 1986. Single Mothers and Their Children: A New American Dilemma . Washington, D.C.: Urban Institute Press.

Garfinkel, I., S. McLanahan, and P. Wong. 1988. "Child Support and Dependency." In Beyond Welfare: New Approaches to the Problem of Poverty in America, edited by H. R. Rodgers, Jr., 66–85. Armonk, N.Y.: M. E. Sharpe.

Gilder, G. 1981. Wealth and Poverty . New York: Basic Books.

Goffman, E. 1963. Stigma: Notes on the Management of Spoiled Identity . Englewood Cliffs, N.J.: Prentice-Hall.

Goodban, N. 1985. "The Psychological Impact of Being on Welfare." Social Service Review, September, 403–422.

Goodwin, L. 1983. Causes and Cures of Welfare . Lexington, Mass.: Lexington Books.

Graf, R., and L. Hearne. 1970. "Behavior in a Mixed Motive Game as a Function of Induced Self-Esteem." Perceptual and Motor Skills 31: 511–517.

Gueron, J. M. 1988. Reforming Welfare with Work . Occasional Paper 2. Ford Foundation Project on Social Welfare and the American Future. New York: Ford Foundation.

Handler, J., and J. Hollingsworth. 1969. "Stigma, Privacy, and Other Attitudes of Welfare Recipients." Stanford Law Review 22: 1–19.

Hesketh, B. 1984. "Attribution Theory and Unemployment: Kelly's Covariation Model, Self-Esteem, and Locus of Control." Journal of Vocational Behavior 24: 94–109.

Hill, M., S. Augustyniak, G. Duncan, G. Gurin, P. Gurin, J. Liker, J. Morgan, and M. Ponza. 1985. Motivation and Economic Mobility . Ann Arbor, Mich.: Institute for Social Research.

Jayaratne, S., W. A. Chess, J. Norlin, and J. Bryan. 1980. "Demographic Correlates of Self-Esteem Among Black and White AFDC Recipients." Journal of Sociology and Social Welfare 7(2): 285–297.

Kane, T. J. 1987. "Giving Back Control: Long-Term Poverty and Motivation." Social Service Review, September, 405–419.

Katz, M. B. 1986. In the Shadow of the Poorhouse: A Social History of Welfare in America . New York: Basic Books.

Kishor, N. 1981. "The Effect of Self-Esteem and Locus of Control in Career Decision-Making of Adolescents in Fiji." Journal of Vocational Behavior 19: 227–232.

Korman, A. K. 1966. "Self-Esteem Variable in Vocational Choice." Journal of Applied Psychology 50: 479–486.

———. 1970. "Toward an Hypothesis of Work Behavior." Journal of Applied Psychology 54: 31–41.

———. 1976. "Hypothesis of Work Behavior Revisited, and an Extension." Academy of Management Review 1: 50–63.

Maslow, A. 1968. Toward a Psychology of Being . New York: Van Nostrand Reinhold.

Morris, M., and J. B. Williamson. 1987. "Workfare: The Poverty/Dependence Trade-Off." Social Policy 18(1): 13–17.

Murray, C. 1984. Losing Ground: American Social Policy, 1950–1980 . New York: Basic Books.

———. 1987. "In Search of the Working Poor." Public Interest 89 (Fall): 3–19.

Neuringer, C., and L. Wanake. 1966. "Interpersonal Conflicts in Persons of High Self-Concept and Low Self-Concept." Journal of Social Psychology 68: 313–322.

Nichols-Casebolt, A. 1986. "The Psychological Effect of Income Testing Income-Support Benefits." Social Service Review, June, 287–302.

Novak, M., et al. 1987. The New Consensus on Family and Welfare: A Community of Self-Reliance . Washington, D.C.: American Enterprise Institute for Public Policy Research.

O'Neill, J., D. Wolf, L. Bassi, and M. Hannan. 1984. An Analysis of Time on Welfare . Washington, D.C.: Urban Institute Press.

Pettigrew, T. F. 1980. "Social Psychology's Potential Contributions to an Understanding of Poverty." In Poverty and Public Policy, edited by V. T. Covello. Cambridge, Mass.: Schenkman.

Platt, J., R. Eiseman, and A. Darbes. 1970. "Self-Esteem and Internal-External Control: A Validation Study." Psychological Reports 26: 162.

Quarantelli, E. L., and J. Cooper. 1966. "Self-Conceptions and Others: A Further Test of Median Hypotheses." Sociological Quarterly 7: 281–297.

Rainwater, L. 1982. "Stigma in Income-Tested Programs." In Income-Tested Transfer Programs: The Case For and Against, edited by I. Garfinkel, 1946. New York: Academic Press.

Rein, M., and L. Rainwater. 1978. "Patterns of Welfare Use." Social Service Review, December, 511–534.

Reischauer, R. D. 1987. "America's Underclass." Public Welfare 45(1): 7–31.

Ricketts, E. R., and I. V. Sawhill. 1986. "Defining and Measuring the Underclass." Paper presented to the annual meeting of the American Economic Association, New Orleans, December 28.

Robinson, J. P., and P. R. Shaver, eds. 1973. Measures of Social Psychological Attitudes . Ann Arbor, Mich.: Institute for Social Research.

Rosenberg, M. 1963. "Parental Interest and Children's Self-Conceptions." Sociometry 26 (1): 35–49.

———. 1965. Society and the Adolescent Self-image . Princeton, N.J.: Princeton University Press.

———. 1967. "Psychological Selectivity in Self-Esteem Formation." In Atti tude, Ego-Involvement, and Change, edited by C. Sherif and M. Sherif, 26–50. New York: Wiley.

Rosenberg, M., and L. I. Pearlin. 1978. "Social Class and Self-Esteem Among Children and Adults." American Journal of Sociology 84: 54–77.

Rosenberg, M., and R. G. Simmons. 1972. Black and White Self-Esteem: The Urban School Child . Washington, D.C.: American Sociological Association.

Rotter, J. B. 1966. "Generalized Expectancies for Internal vs. External Control of Reinforcement." Psychological Monographs 80: 1–28.

St. John, N. 1975. School Desegregation: Outcomes for Children . New York: Wiley.

Sawhill, I. V. 1976. "Discrimination and Poverty Among Women Who Head Families." Signs: A Journal of Women in Culture and Society 2 (Spring): 201–211.

———. 1986. "Anti-Poverty Strategies for the 1980s." Discussion Paper, Changing Domestic Priorities Project. Washington, D.C.: Urban Institute Press.

Schneiderman, L., W. Furman, P. Lachenbruch, M. B. Welch, and I. Chow. 1987a. "GAIN Participant Needs Assessment: A Profile of Recipients of Aid to Families with Dependent Children in Los Angeles County." Center for Child and Family Policy Studies, School of Social Welfare, University of California, Los Angeles.

Schneiderman, L., W. Furman, P. Lachenbruch, E. Fielder, M. B. Welch, and I. Chow. 1987b. "GAIN Participant Needs Assessment: Report Two." Center for Child and Family Policy Studies, School of Social Welfare, University of California, Los Angeles.

Schorr, A. 1986. Common Decency: Domestic Policies After Reagan . New Haven, Conn.: Yale University Press.

Seligman, M. E. P. 1975. Helplessness: On Depression, Development, and Death . San Francisco: Freeman.

Sherwood, J. J. 1965. "Self-Identity and Referent Others." Sociometry 28(1): 66–81.

Silber, E., and J. S. Tippett. 1965. "Self-Esteem: Clinical Assessment and Measurement Validation." Psychological Reports 16: 1017–1071.

Simmons, R. G., L. Brown, D. M. Bush, and D. A. Blyth. 1978. "Self-Esteem and Achievement of Black and White Adolescents." Social Problems 26: 6–96.

Social Security Administration. 1986. Social Security Bulletin, Annual Statistical Supplement. Washington, D.C.: U.S. Department of Health and Human Services.

Social Security Administration. 1987. Social Security Bulletin, Annual Statistical Supplement. Washington, D.C.: U.S. Department of Health and Human Services.

Sullivan, H. S. The Interpersonal Theory of Psychiatry . New York: Norton.

Survey Research Center, Institute for Social Research. 1984. User Guide to the Panel Study of Income Dynamics . Ann Arbor, Mich.: Inter-University Consortium for Political and Social Research.

Taylor, J., and W. Reitz. 1968. "The Three Faces of Self-Esteem." Research Bulletin no. 80. Department of Psychology, University of Western Ontario.

Taylor, S. E., and J. D. Brown. 1988. "Illusion and Well-Being: A Social Psychological Perspective on Mental Health." Psychological Bulletin 103(2): 193–210.

U.S. Bureau of the Census. 1985. "Money Income and Poverty Status of Families and Persons in the United States: 1984." Current Population Reports, Series P-60, no. 149. Washington, D.C.: Government Printing Office.

———. 1987. "Consumer Income." Current Population Reports, Series P-60, no. 158. Washington, D.C.: Government Printing Office.

U.S. Congress. Congressional Budget Office. 1985. Reducing Child Poverty . Washington, D.C.: Congressional Budget Office.

———. Congressional Research Service and Congressional Budget Office. 1985. Children in Poverty . Report prepared for the House Committee on Ways and Means. Washington, D.C.: Government Printing Office.

———. House. Select Committee on Children, Youth, and Families. 1987. U.S. Children and Their Families: Current Conditions and Recent Trends, 1987 . Washington, D.C.: Government Printing Office.

Warren, P., and D. Maxwell-Jolly, n.d. "How California's Welfare Dynamic Affects Work Programs Such as GAIN." Office of the Legislative Analyst, Sacramento, Calif. Typescript.

"Welfare and Poverty Among Children." 1987. Congressional Research Service Review 8, no. 7 (July).

Wells, L. E., and G. Marwell. 1976. Self-Esteem: Its Conceptualization and Measurement . Beverly Hills, Calif.: Sage.

Williamson, J. B. 1974. "Beliefs About the Welfare Poor." Sociology and Social Research 58: 163–175.

Wiseman, M. 1985. "The Welfare System." In California Policy Choices, vol. 2, edited by J. Kirlin and D. R. Winkler, 133–202. Los Angeles: University of Southern California, School of Public Administration.

Woodworth, R. 1958. Dynamics of Behavior . New York: Holt, Rinehart & Winston.

Wortman, C. B., and J. W. Brehm. 1975. "Responses to Uncontrollable Outcomes: An Integration of Reactance Theory and the Learned Helplessness Model." In Advances in Experimental Social Psychology, vol. 8, edited by L. Berkowitz, 277–336. New York: Academic Press.

Wylie, R. C. 1968. "The Present Status of Self-Theory." In Handbook of Personality Theory and Research, edited by E. F. Borgatta and W. W. Lambert, 728–787. Chicago: Rand McNally.

———. 1974. The Self-Concept . Vol. 1, A Review of Methodological Considerations and Measuring Instruments . Rev. ed. Lincoln: University of Nebraska Press.

Ziller, R., J. Hagey, M. Smith, and B. Long. 1969. "Self-Esteem: A Self-Social Construct." Journal of Consulting and Clinical Psychology 33: 84–95.