Short- and Long-Term Effects of AFDC Receipt on Families
ABSTRACT Using a sample of women with children from the Panel Study of Income Dynamics, I look at the effects of receipt of AFDC assistance between 1968 and 1972 on attitudes and behaviors measured in 1972. I also look at the effect of AFDC receipt on the women’s hours worked and economic status up to 20 years after the measurement of AFDC receipt, and on the final educational attainments of the women’s children. While women receiving AFDC differ from non-recipients along the measures of some attitudes and behaviors, they do not differ in hours worked and economic status 20 years later. The children of recipients, however, do obtain fewer years of education than do those of non-recipients. These findings are less clear for long-term AFDC recipients.
Short- and Long-Term Effects of AFDC Receipt on Families
Using a sample of women with children from the Panel Study of Income Dynamics, I look at the effects
of receipt of AFDC assistance between 1968 and 1972 on attitudes and behaviors measured in 1972. I
also look at the effect of AFDC receipt on the women’s hours worked and economic status up to 20
years after the measurement of AFDC receipt, and on the final educational attainments of the women’s
children. While women receiving AFDC differ from non-recipients along the measures of some attitudes
and behaviors, they do not differ in hours worked and economic status 20 years later. The children of
recipients, however, do obtain fewer years of education than do those of non-recipients. These findings
are less clear for long-term AFDC recipients.
Short- and Long-Term Effects of AFDC Receipt on Families
Christopher Jencks (1993) proposes four ways of identifying individuals who comprise the
so-called “underclass” of American society. Two of his classification schemes, income level and
income sources, deal directly with monetary measures of poverty. The other two, cultural skills and
moral norms, speak to a larger issue brought forth when the lives of the poor are discussed and
researched—the extent to which low-income people do or do not “think, talk and act like those who
manage America’s major institutions” (p. 144). It is often the perceived cultural and moral poverty
among the poor that engenders the most concern, rather than the financial realities of their lives.
In particular, the attitudes, behaviors, and outcomes of recipients of Aid to Families with
Dependent Children (AFDC) receive a great deal of attention. Many people, from scholars to
politicians, suggest that women who receive AFDC demonstrate attitudes and behaviors that differ
from those of other women. These differences make women’s welfare receipt harmful to their
future, and, more precariously, the future of their children, as women acquire and promote actions
and beliefs that are at odds with middle-class values.
In this paper I will investigate the extent to which a sub-set of the “underclass,” women
receiving Aid to Families with Dependent Children (AFDC), differ from other women in their
attitudes and behaviors. I will also compare their own and their children’s long-term outcomes to
those of non-recipients. The use of longitudinal data allows for a unique look at the effects of
welfare receipt on families many years down the road. Through this work, I am able to shed some
light on the often contradictory discussions revolving around welfare recipients, their behaviors and
attitudes, and the outcomes of their children.
In this section, I highlight some areas in which welfare recipients are thought to differ from the rest
of society. First, I discuss the history of cultural models of poverty. Women and families receiving
welfare make up a significant part of what is often considered the culture of poverty. In the second
section, I perform a brief review of the literature on the link between welfare and single-motherhood.
Finally, I discuss recent work on potential attitudinal differences between AFDC recipients and non-
Culture of Poverty
Lewis (1966) was among the first academic researchers to identify and catalogue what he saw as the
“culture of poverty” among the poor. In a study of Puerto Rican slums, he identified a “strong
feeling of fatalism, helplessness, dependence, and inferiority” among the residents (23), concluding
that psychiatric treatment may be the best way of addressing the pathos that he observed.
More modern ethnographic work on the poor provides rich descriptions of lives that seem
shockingly foreign to the usual upper-middle class reader, implying vast cultural differences between
classes. Anderson’s Streetwise (1990) describes life in an urban neighborhood called Northton. In
the chapter entitled “Sex Codes and Family Life Among Northton’s Youth,” he describes the unique
cultural attitudes regarding male-female relationships found in this depressed community; attitudes
which appear at odds with larger, societal, views about relationships. In Northton, men play on
women’s desires for love and marriage in order to gain sexual favors, often leaving them alone and
pregnant in the end.
Anderson linked these seemingly dysfunctional sexual attitudes to the dire economic situation
of the area. He calls the lack of sexual and parental responsibility among Northton men, “a mean
adaptation to blocked opportunities and profound lack, a grotesque form of coping (113).
In The Moynihan Report and the Politics of Controversy (1967), Rainwater analyzes the
passions often stirred by the debate about potential cultural differences between the poor and the rest
of society. He discusses the heated controversy sparked by Daniel Patrick Moynihan’s report,
published in the late 1960’s, in which he examined trends of family deterioration among African-
Americans. In calling attention to such disturbing patterns as increasing out-of-wedlock births,
welfare use, and female-headed homes, Moynihan identified the “pathology” of urban black families
as a primary causal agent of these problems (75).
In his report, Moynihan linked self-defeating attitudes among blacks to the deprivations
imposed on them by whites through “three centuries of sometimes unimaginable mistreatment” (39).
Thus, in his view, racial inequalities led to pathological cultural attitudes, which then led to more
poverty and family dysfunction. Others, however, interpreted the causality of his argument
differently, accusing him of “blaming the victim” by linking poverty to attitudes of the poor (which
was, in fact, only half of his argument). Because of the ire faced by Moynihan after his report,
researchers sympathetic to the plight of the poor have been tentative in addressing the existence and
effects of potential attitudinal differences among them.
More conservative researchers, on the other hand, often deliberately make the connection
between attitudes and poverty to explain the economic position of the poor. Rector (1993) argues
that the real cause of poverty is “a breakdown in the values and conduct that lead to the formation of
healthy families and stable personalities, and promote self-sufficiency” among the poor (3). He uses
his belief in such “behavioral poverty” to call for large-scale reductions in social welfare spending.
While describing the underclass, Wilson (1987) uses the term “social isolation,” rather than
“culture of poverty.” He does this deliberately, to highlight the fact that the characteristics of the
poor are responses to social and economic situations, not self-sustaining cultural traits.
The key theoretical concept, therefore, is not culture of poverty but social
isolation. Culture of poverty implies that basic values and attitudes of the
ghetto subculture have been internalized and therefore influence
behavior…[social isolation] highlights the fact that culture is a response to
social structural constraints and opportunities (61).
In Wilson’s view, understanding the culture of the poor, which includes their attitudes and behaviors,
plays a vital role in understanding poverty itself.
One of the “mainstream” moral norms named by Jencks is the desire to wait until marriage
to have children. The extent to which the poor, and particularly welfare recipients, do or do not
follow this norm is frequently examined by researchers. Often, people suggest that teenage and non-
marital pregnancies occur among women who demonstrate a lack of confidence in or aspirations
toward the future and respond instead to short-term incentive such as welfare benefits.
For example, in an extensive empirical review, Moffitt (1992) found small but significant
effects of AFDC benefit levels on non-marital births. These effects, however, are not enough to
explain the large increase in female headed families that occurred in the 1960’s and early 70’s.
Olsen and Farkas (1990) set out to investigate Wilson’s (1987) theory which links decreased
employment opportunities to low marriage rates, leading to out-of-wedlock childbearing. Using a
hazards model, these authors look at the effect of monthly, local employment rates on entry into a
cohabitational union, testing whether low employment opportunities lead to fewer marriages/instances
of cohabitation among low-income youth. The authors then test for a link between employment
opportunity and non-marital births. Olsen and Farkas predict that high rates of local employment
raise women’s “opportunity cost” of having children, decreasing births independent of changes in
The authors found positive effects between employment opportunity and increased levels of
cohabitation/marriage. Even stronger, however, were the negative effects of employment
opportunity on non-marital births. A 10% increase in the number of youths working in a given
location led to a one-sixth decrease in the probability that a woman had a child before age 17. The
research suggests that the presence of high levels of employment leads youths to expect more from
their economic futures and, subsequently, alter their behaviors by entering into long-term
relationships and avoiding early and non-marital pregnancies.
Thornton and Camburn (1987) studied the effects of parental attitudes about sex on teenage
attitudes and behavior, using a longitudinal study of white, non-poor children born in Detroit in 1961.
They found strong, positive, effects of maternal permissiveness towards premarital sex on the
attitudes of teenagers, indicating a significant amount of parental influence in the attitude-formation
process. This influence was strongest between mothers and daughters. The study also found a link
between maternal behavior, in the form of premarital pregnancy, and the sexual behavior of their
children, which persisted even when controlling for maternal attitudes. This evidence lends support
to Bandura’s (1986) social learning theory, in which people form their beliefs based not only on their
own experiences, but also on the process of observing the behavior of role models around them. In
this case, teenagers are seen as modeling both the attitudes and behaviors of their parents in relation
to teenage sexuality and childbearing. Findings such as this cause researchers and policymakers to
worry about extensive intergenerational transmission of AFDC dependency.
AFDC and Attitudes
It is often suggested that the attitudes of welfare recipients differ from those of other
women. It is unclear whether, if such differences exist, they are the result of time spent on welfare,
or whether they existed prior to, and are a factor causing, welfare receipt.
Plotnick, Klawitter, and Edwards (1997) use the NLSY to study an extensive array of factors
predicting young women’s initial entry into welfare. In hazard models, they find no evidence for
effects of self-esteem or locus of control on entry, but significant effects for school-related attitudes,
family background characteristics and IQ scores.
Popkin (1990) studied the link between AFDC receipt and recipients’ sense of efficacy.
Through interviews with a sample of AFDC recipients from Chicago, she found length of time on
welfare to be the strongest, negative, predictor of efficacy (consisting of one measure of self-esteem,
and two measure of fate control). Low efficacy, in turn, affected respondents’ beliefs about the
difficulty of leaving welfare, with low efficacy respondents less likely to view work as a viable option
for leaving welfare.
Oritz and Bassoff (1987) performed a study of 53 teenage welfare recipients from California,
examining their views about education, careers, and optimism for the future. On their measure of
locus of control, the authors found teenage mothers to be less sure of the degree of control they had
over their lives than non-mothers. Additionally, a much higher proportion of teenage mothers, as
opposed to non-mothers, did not have specific career goals and thought that they would not graduate
from high school.
Alvarez et al. (1987) performed a study of low SES adolescent girls in Santiago, Chile,
comparing pregnant and non-pregnant teenagers along a host of sociocultural characteristics. Each
pregnant girl was matched with a non-pregnant teen of the same age and from the same
neighborhood to control for background characteristics. Girls were then interviewed on measures
such as expectations for the future, aspirations, peer group characteristics, and religious practices.
45.6% of the pregnant girls felt that the best way to improve their future was through a job,
compared to 75.8% of the non-pregnant teens. Additionally, 18.8% of the pregnant group expressed
aspirations of continuing their studies and attaining a professional title, while 55.3% of the non-
pregnant girls held these goals.
This study contains obvious problems of causality, as the pregnant girls were interviewed
after becoming pregnant. Their pregnancy could have affected their aspirations, creating the large
differences between the two groups. This study does, however, stand as one of the few that
explicitly works to establish a link between attitudes, specifically expectancies and aspiration, and the
behavioral outcome of teenage pregnancy.
Unlike the above research, the current study, because of its longitudinal nature, looks at the
effect of AFDC receipt on the attitudes and behaviors of welfare recipients, not just at a point in time,
but also up to 25 years later. This allows for the investigation of long-ranging effects of welfare
receipt on women’s work behavior, as well as children’s outcomes, to become evident.
Current dialogue, as well as some of the research discussed above suggests the existence of
sharp attitudinal and behavioral differences between welfare recipients and non-recipient women. In
this study, I test the hypothesis that welfare recipients will differ from non-recipients along several
behavioral and attitudinal measures. AFDC receipt is measured between 1968 and 1972, and the
outcomes, including various attitudes and behaviors, are measured in 1972 (and in one case in 1975).
Hypothesis 1: Women who received AFDC assistance between 1968 and 1972 will exhibit different
attitudes and behaviors measured in 1972 and 1975 than women who did not receive AFDC assistance.
Additionally, in this study I test for potential long-term effects of the hypothesized behavioral
and attitudinal differences among AFDC recipients on the future labor force participation and
economic outcomes of welfare recipients, as well as on the educational attainment of their children.
Thus, the behaviors and attitudes of welfare recipients takes on a mediating role between welfare
receipt and long-term outcomes.
Hypothesis 2: Women who received AFDC assistance between 1968 and 1972 will exhibit different long-
term labor market outcomes than non-recipients. Their children’s educational attainment will differ from
those of non-recipients as well.
Finally, I hypothesize that women receiving AFDC for an extended period of time will
demonstrate more extreme attitudinal and behavioral differences that occasional AFDC recipients.
Hypothesis 3: Length of time receiving AFDC will be associated with greater attitudinal differences
between recipients and non-recipients, and more extreme differences in long-term outcomes.
Data was obtained from the 1992 Panel Study of Income Dynamics (Hill, 1992). Since 1968
the PSID has followed and surveyed annually about 5000 families, calculating weights each year to
account for differential sampling and attrition. In each family, data is gathered from the household
head, who reports on him/herself, as well as gives information about other family members. Each
household head of the original PSID sample is then followed annually, regardless of changes in his or
her marital status and family composition. Additionally, children of the original PSID sample families
are followed when they split off and form their own households.
Unfortunately, under the original 1968 survey design of the PSID, now-antiquated
conventions dictated that, in any instance in which an adult male lives in a household, he is designated
the head. Data on the behavioral and attitudinal measures used in this study were only gathered for
heads of households from 1968-1972. Therefore, in order to be included in the sample used in this
study, a woman must have been a head of household (i.e. not married or cohabiting) at least once
between 1968 and 1972. Each such woman was then followed for all subsequent years of the
survey, despite the fact that a great number of them did marry by 1992. Because of the PSID’s
requirements for headship, however, these women are not a representative sample of the U.S. female
population, although they should be representative of female heads of households with children.
The sample used in this study includes all women who were heads of households at least
once between 1968 and 72, were aged 21-39 in 1972, and who were present in the survey at least
one year from 1968-72, when the attitude and behavioral measures were gathered. Additionally, all
women in this sample had at least one child in their household between 1968 and 1972. This controls
for the fact that all of the women receiving welfare, by nature of the program, have children, while
those not receiving welfare may not.
The relatively young age range of the women was selected to minimize the chances that the
women’s attitudinal measures would reflect their concurrent economic situations. However, the
small number of female heads in the PSID forced the inclusion women aged up to 39 years in order
to obtain a large enough sample for analysis. In order to correct for this, all analyses control for age.
Additionally, all analyses are weighted by the most recent weight available for the respondent in a
given set of regressions.
Welfare Receipt. I created a dummy variable indicating whether the respondent had received AFDC
(Aid to Families of Dependent Children) at any time between 1968 and 1972, using PSID questions
regarding the amount, if any, of such aid received by each household.
For each year that a woman was the head of her household and received at least one dollar
of AFDC, she was counted as having received welfare in that year. The PSID questions used to
create this dummy variable ask specifically about money from AFDC (called ADC in 1968), and do
not pertain to any other social assistance programs. From this information, I used three dummy
variables to divide the sample into three groups. The first category contains women who received no
AFDC payments at all in any year between 1968 and 1972. I call this group no welfare receipt. The
next two categories identify women who received AFDC during some, but not all, of the years in the
five-year span, (some welfare receipt) and, finally, those who received AFDC during all five years
(welfare receipt in all years).
Years of completed schooling. Reports of years of completed schooling were taken, when possible,
from self-reports in the 1985 interviewing wave, when an extensive battery of schooling-related
questions were asked. In cases where the women did not survive in the PSID sample until 1985, I
took the most recent report of completed schooling, gathered between 1968 and 1972. All analyses
reported in this paper control for the respondent’s level of completed schooling.
Cognitive skills. As a measure of cognitive skills I took the PSID's sentence-completion test score
from the 1972 interview. As documented in Veroff, McClelland, and Marquis (1971) this test was
adapted from the Lorge-Thorndike intelligence test.1 Sample items are: "The ragged _____ may prove
a good horse", with alternatives: "Puppy", "Child", "Calf", "Lamb" and "Colt"; and "The coward
threatens only when he is _____", with the alternatives: "Afraid", "Surrounded", "Safe", "Conquered"
and "Happy", with "Colt" and "Safe" taken to be the correct responses. The sample distribution of
test scores indicate a much greater degree of differentiation at the low than at the high end of the
distribution. For our sample of 21-39 year-old women with children the mean on the 13-item test was
9.08 with a standard deviation of 1.86.2 The sentence completion test measure was re-standardized
with division by the standard deviation to have a standard deviation of one. All analyses control for
this measure of cognitive skills.
Motivational measures. The motivational measures used as dependent variables in these analyses
come from questions posed in the PSID in 1972 and, for one measure, again in 1975.
In 1972 a series of forced-choice questions comparing challenge vs. affiliation, challenge vs. power
and power vs. affiliation were posed in the PSID. Only the challenge vs. affiliation is used in this
analysis. It was created by averaging responses to the following two questions: "Would you like to
have more friends, or would you like to do better at what you try?" and "Would you prefer a job
where you had to think for yourself, or one where you work with a nice group of people?"
(correlation = .19). A score of 1 on each of these measures indicates a preference for challenge over
affiliation, while respondents choosing to have more friends, or to work with nice people, received
scores of 0.
The measure of personal control followed Hill et al. (1985) in averaging respondents' mean
scores for each of the following questions: "Have you usually felt pretty sure that your life will work
out the way you want it to, or have there been more times when you haven't been sure about it?",
"When you make plans ahead, do you usually get to carry out things the way you expected, or do
things usually come up to make you change your plans?" and "Would you say that you nearly always
finish things once you start them, or do you sometimes have to give up before they are finished?"
(alpha = .79). In this study, I use responses to these questions gathered in 1972, and again in 1975,
averaging responses from these two years for each of the three components.
Responses to the three components of the personal control measure were scored as 1 for a
positive response, (i.e. pretty sure life will work out, carrying out plans, and finishing things), .5 for
an equivocal response, and zero for a negative response.3 In creating this measure of personal
control, we averaged each respondent's score over as many of the five years between 1968 and 1972
for which data exists on all three of the components. The alphas for the three components of our
measure of personal control range from .73 to .76, indicating reasonable levels of reliability.
Additionally, the alpha for our index of personal control, in which these three components are
combined, has an alpha of .85, again pointing to a highly reliable measure.
The measure of fear of failure was constructed by averaging scores for the following
questions, all of which were asked only in 1972: "When taking tests would you say that you get very
upset, somewhat upset, or not upset at all?", " When working on tests, does your heart beat very
fast, faster than normal, or about normal?", "During tests, would you worry a lot about what it would
mean to fail, worry some, or not worry at all?", and "When taking a test, do you perspire a great deal,
more than usual, or not at all?" (alpha = .54). Respondents whose answers to the above questions
indicated any type of averse reaction to test-taking were given a score of 1 on fear of failure, (those
who reported getting very upset and those who reported getting somewhat upset were both given
scores of 1), those who reported no reactions to test-taking were scored as -1, and respondents who
did not or could not answer were coded as zero.
The measure of whether a respondent wants her child to be a leader was comprised of
responses to two questions, both of which were asked in 1972: "Would you rather have your child be
popular with his or her classmates or be a leader?" and "Would you rather have your child be a leader
or do the work the teacher expects?" (correlation = .22). For each measure, those respondents
preferring leadership roles for their children were given scores of 1, those preferring popularity or
obedience were given scores of -1, and all missing data was coded as zero.
Finally, I used data from a measure of parental expectations gathered in 1972. PSID
respondents with children in school were asked to estimate the amount of education that they expect
their children to complete. These questions were recoded to form a 2 to -2 scale, in which higher
scores indicate high parental expectations for their children’s educational achievement. A recoded
score of 2 indicates that the parents expect all of their children to go to college, a score of 1 means
that parents expect some will go to college, zero indicates that the parents think that all will finish
high school, -1 means that the parents expect some to Finnish high school, and -2 indicates parents
expect at least one or more of their children will not finish high school.
Behavioral measures Several behavioral measures, all taken in 1972, are also used as dependent
variables in these analyses. Indicators of social connections include: i) how often one attends church
and iii) and how often the family eats together. Each of these variables was scaled as number of
times per week. Single-person families were given missing values for the measure of eating together.
Frequency of bar attendance is measured as estimated times per year that the respondent
goes to a bar or tavern, and annual expenditures on alcohol is measured in 1994 dollars.
Finally, I included several miscellaneous attitudes and behaviors my analyses. Included in
this category was an assessment of the cleanliness of the respondent's home, taken by the interviewer
at the time of the annual interview. For this measure, respondents assessed as having a "very clean"
home were coded as a 1, those with a "clean" home were coded as zero, and those with "so-so", "not
very clean", and "dirty" homes were given scores of -1. In cases in which the interview took place
outside of the respondent's home, a missing score was given.
The two other measures included in the miscellaneous category are behavioral indications of
time spent watching television, and time spent reading newspapers. The newspaper measure
indicates how many times the respondent read a paper, while the television indicator measures the
hours per week a respondent reported watching T.V. during weekdays. Both measures are scaled,
respectively, as number of times or number of hours per week.
Finally, housework hours is the average hours that the wife or unmarried female head of the
family spent on housework in 1972.
Children’s outcomes. As a measure of outcomes for the children of the women in this sample, I
used a measure of average years of completed education for each respondent's child. Data on this
measure was gathered each year up until 1992.
Background variables. All analyses controlled for the following set of demographic and background
characteristics of the respondents: age in 1972; race (black =1, other =0); number of children; age of
youngest child in 1972; and dummy variables for: i) whether the respondent's father was a laborer,
farmer, clerical worker, craftsman, operative, self-employed, or (the omitted group) a professional or
manager; ii) whether the respondent grew up in a city, a town, or farm/other location (the omitted
group); iii) whether the respondent grew up in the South; iv) whether the respondent reported her
family of origin as being poor, average (the omitted group) or well-off; and iv) whether or not the
respondent had a health condition in 1972 that limited the amount or type of work she could do. The
background variables also include the respondent's number of siblings; and years of completed
education for the respondent's father. In order to minimize cases lost to missing data, I created
dummy variables to account for missing data on father's occupation and the respondent's health
limitations, and included these in our series of background controls. Finally, I assigned a full-sample-
mean value of 9.27 on years of father's education to the missing cases on that variable. Additionally,
some analyses, as indicated in Table 3, control for the percent of time that each woman was married.
Table 1 presents the total weighted percentages of women in the sample who 1) never
received welfare, 2) received some welfare, or 3) received welfare all 5 years, between 1968 and
1972. Of my entire sample of 587 female heads with children, 67.7% did not receive any money
from AFDC at all between 1968 and 1972. 27.3% received AFDC in at least one, but not all, of these
five years, and 5.0% received AFDC in all five years.
Table 2 presents weighted means, standard deviations, and minimum and maximum values
for the variables used in these analyses, shown for non-welfare-receiving and for welfare-receiving
women separately. (Women receiving welfare during all or some of the five-year period are pooled
together here in one category of women receiving welfare). Weighted t-tests were used to determine
instances in which the difference in means between these two groups was significantly greater than
zero (at p<.05).
Results from the t-tests point to several measures on which women receiving welfare differ
from those who did not. For example, women receiving AFDC have a significantly lower income-to-
needs ratio, both in the initial, 1968-1972 time period, and in the final, 1988-92 period than non-
recipients. Additionally, recipients demonstrate significantly lower measures of hourly wage than
non-recipients. Finally, both recipients and their children have significantly lower measures of years
of completed schooling than non-recipients.
Non-recipients score higher than AFDC recipients on the measure of preference for
challenge, and the measure of personal control. They also score lower than recipients on a measure
of a fear of failure (indicating that AFDC recipients demonstrate a higher degree of fear of failure
In terms of the behavioral measures, non-recipients go to bars a significantly greater number
of times per year than recipients. They also spend less time performing housework activities,
although they score significantly higher on the measure of whether their home is clean. Finally,
recipients spend a greater amount of time watching TV than non-recipients.
Additionally, there are several demographic characteristics on which recipients and non-
recipients differ. Recipients are much more likely to be black (53% black) than non-recipients (of
whom 22% are black). Finally, the fathers of recipients are significantly more likely to have been
laborers than those of non-recipients, while recipients were more likely to grow up in a town than
The next series of analyses, presented in Table 3, uses the two categories of welfare receipt,
some and all, (with no receipt as the omitted category) along with education, Sentence Completion
Test, and the entire set of background controls, to predict women’s attitudes and behaviors measured
in 1972, after the period of welfare receipt used in this study (1968-72). My interest here is
determining the instances in which, with all of the above-named controls, different levels of welfare
receipt demonstrate a significant relationship with women’s attitudes and behaviors.
Of the four motivational variables investigated in these analyses, some welfare receipt
demonstrates a significant relationship with three. In Table 3, we see that, compared to a women
who never received welfare, one who did experience some welfare receipt has .09 point lower score
on the personal control measure (where mean and standard deviation for non-welfare receivers are
.53 and .31). Additionally, women who received some welfare demonstrated a .18 point lower score
on preference for challenge (where the scale is 0 to 1 and the mean and standard deviation for non-
welfare receivers are .69 and .30).
Finally, we see that women who had received some welfare during the 1968-72 time period
scored .18 points higher on the measure of a fear of failure. This suggests a significantly greater fear
of failure among welfare recipients than among their non-welfare-receiving counterparts. Compared
to women who never received welfare, those who received AFDC either some or all of the time did
not demonstrate any significant differences in the motivational measure of whether the mother wants
her child to be a leader or on the measure of parental expectations for children’s success. These are
the only motivational measures on which welfare-receiving and non-welfare receiving women did not
differ significantly. Interestingly, the dummy variable indicating welfare receipt in all years did not
demonstrate a significant relationship with any of the motivational variables.
In addition to the motivational measures, there are several behavioral measures on which
welfare and non-welfare receivers demonstrated significant differences. Women receiving welfare in
all five years spent about 505 more hours per year doing housework than women never receiving
As indicated in the t-test comparisons, women receiving welfare are less likely to go to bars,
shown by the negative and significant relationship between welfare receipt and bar attendance. A
woman receiving some welfare goes to bars or taverns over 10 times per year less frequently than a
non-receiver, and a woman receiving welfare all 5 years goes to a bar almost 20 times less frequently
than someone who never received welfare.
However, women who receive welfare in all five years spend significantly more money on
alcohol than do those who never received AFDC during this time period. We see that women who
received AFDC during all five years spent about $181 per year more on alcohol than did non-
There are also several behavioral measures in which welfare-receivers and women never
receiving welfare in this time period did not differ significantly. No differences between the two
groups of women in weekly newspaper reading, daily television watching, church attendance,
cleanliness of the home, or the number of times per week that a family eats together emerged from
these mulitivariate analyses.
Finally, also in Table 3, I used welfare receipt measured between 1968 and 1972 to predict
future events in the lives of the women and children in my sample. First, I predicted each woman’s
annual work hours, averaged and measured (in thousands) between 1973 and 1992. Here, a
significant and negative relationship between AFDC receipt and hours worked emerged, indicating
that women who had received some welfare between 1968 and 1972 worked, on average, about 270
hours less per year than non-welfare receivers between 1973 and 1992. Those receiving welfare in
all five years worked about 440 hours per year less, on average.
However, when predicting average annual hours worked in the more distant future, the
significance of welfare receipt decreased. When predicting hours worked five years after the
measurement of welfare receipt (1973-77), some welfare receipt is associated with 430 fewer hours
per year of work, and welfare receipt all five years is associated with 730 fewer hours of work.
When predicting hours worked in 1978-82, ten years after the period of measured receipt, the
coefficient on some receipt drops to -.250, remaining significant, and that on welfare receipt all years
drops to -240, losing significance.
Finally, there was no significant relationship between some welfare receipt and hours worked
between 1983 and 1987 (coefficient of -.11), or hours worked between 1988 and 1992 (coefficient
of -.02). Welfare receipt in all years did demonstrate a significant negative relationship with hours
worked between 1983 and 1987, but not with hours worked in the last five-year period (1988-1992).
Also in Table 3, I used the two welfare receipt categories to predict the average years of
educational attainment of the children of the women in my sample. Here, a significant and negative
relationship between some welfare receipt and children’s education emerged, with children of women
receiving welfare in some years attaining about one-half a year less of schooling (-.51) than children
of non-receivers. This negative relationship diminished slightly when controlling for women’s
income during the time of their welfare receipt, but remained significant (coefficient of -.47, analyses
not shown here). Interestingly, welfare receipt in all five years did not emerge as a significant
predictor of children’s education in my analyses.
Finally, in the last two columns of Table 3, I used the two AFDC-receipt categories to
predict women’s income-to-needs ratios in 1968-1972 and in 1988-1992, the first and last time
periods in this study. I used income-to-needs because it is a more reliable indicator of women’s
economic conditions than hourly wages, which often are low or non-existent.
These last regressions indicate that women who received AFDC during some years and
women receiving AFDC during all years demonstrate significantly lower income-to-needs ratios that
non-recipients in 1968-1972. However, neither category of AFDC receipt was significantly
correlated with income-to-needs in 1988-1992, contradicting the means-based t-test results presented
Finally, in analyses not shown here, I tested whether the attitudinal measures of the AFDC
recipients, (lower feelings of control, lower preference for challenge, and a greater fear of failure),
themselves affect future outcomes in the lives of the recipients. I used personal control, fear of
failure, and preference for challenge as independent variables to predict women’s future work hours
and the educational attainment of their children. None of the motivational measures demonstrated a
significant relationship with children’s education or with women’s work hours.
Women who received AFDC during some of the years between 1968 and 1972 demonstrate
significantly lower scores on the motivational measures of personal control and preference for
challenge, and higher scores on an index of fear of failure. These findings conform with some of the
assumptions of cultural theories of poverty, which hold that those living in poverty (including welfare
recipients) acquire self-defeating attitudes that then serve to promote their further poverty and
dependence. However, many cultural models also assume that the differences between the poor and
non-poor are exacerbated with time—the longer you are poor, the more “different” you become.
According to my results, women who received AFDC for the entire five-year period covered by my
data were more similar to non-receivers than those receiving AFDC only part of the time.
Additionally, another important tenet of cultural models of poverty suggests that the negative
attitudes held by the poor are harmful to their futures. My results contradict this theory. Despite
holding more negative attitudes than non-recipients, women receiving some welfare do not differ
from their non-recipient counterparts in work hours more than ten years after the measurement of
their receipt. Additionally, after 20 years, there are no significant differences between women
receiving welfare all five years and non-recipients. Finally, after 20 years, there are no significant
differences between the income-to-needs ratios of recipients and non-recipients.
However, the educational attainment of the children of occasional welfare recipients does
differ from that of non-recipients, as their children attain, on average, a half-year less schooling than
those of non-recipients. There are no significant differences in average years of schooling between
the children of women who received AFDC all five years and those who did not receive it at all.
Popkin (1990), through in-depth interviews with Chicago welfare recipients, illustrates how
participation in the AFDC program can lead to low feelings of control and self-esteem. My analyses
suggest that these feelings, while real, may not have a consistently negative effect on women and
their children many years down the road. This suggests that, while welfare receipt is correlated with
negative motivational attitudes, these attitudes themselves do not play a significant role in predicting
the long-term outcomes of women and their children.
In a cross-sectional survey, Sisco and Pearson (1994) found high rates of alcoholism and
drug abuse among AFDC recipients in Maryland. According to my results, both categories welfare
recipients demonstrated significantly less frequent bar attendance than non-recipients. However,
women receiving AFDC all five years demonstrated significantly higher levels of annual spending on
alcohol than non-receivers, while women receiving welfare in only some years demonstrated no
The results of this study should not be seen as an indication of the effects of AFDC receipt
itself on the people who receive it. In my analyses, AFDC receipt is only a proxy for economic and
other conditions that lead some women to require governmental assistance in maintaining their
households. Therefore, it must be stressed that it is not the AFDC receipt itself that is associated
with the outcomes presented here, but a whole host of economic and perhaps psychological
characteristics of the women in my sample, into which my measure of AFDC receipt is tapping.
A great deal more work needs to be done in order to fully understand the relationship
between AFDC receipt and the attitudes, behaviors, and outcomes of women. Most strange among
my findings is the suggestion that, while my “ever received some welfare” dummy variable indicated
a significant negative relationship with attitudinal measures, as well as children’s outcomes, the more
extreme measure of whether a woman received welfare all five years did not. Popkin’s (1990)
results indicated that length of time on AFDC was a predictor of women’s attitudes. My results
contradict this seemingly logical finding.
It may be the case that women who receive welfare for many consecutive years can count
on a more stable flow of money than those who only receive welfare sporadically. Perhaps it is this
stability that makes these women look more like non-recipients than do those who receive welfare
occasionally. It may also be the case that some women serve their children best by staying home
with them and continuing to receive AFDC, rather than entering the workforce. This may be
particularly true for families living in unsafe neighborhoods that require a great high level of parental
monitoring of their children (Puntenney, 1997). Because the PSID does not include indicators of
neighborhood characteristics, I am unable to test this hypothesis, however.
Finally, the number of women in my sample who received AFDC in all five years is
extremely low (unweighted N = 45), casting doubt on the reliability of my findings about this sub-
group. More work in this area needs to be done in order to fully investigate my initial results. There
are numerous ways to measure long-term welfare receipt, all of which might demonstrate different
relationships with the outcomes in this study.
This paper has important implications for current policy research and practice, especially in
light of the 1996 changes in welfare laws. These changes target long-term AFDC recipients, with the
presumption that length of time on welfare is associated with more negative outcomes for mothers
and children. However, my results suggest otherwise. Researchers and policy-makers need to do
more work investigating the varied effects of short- and long-term use of AFDC in order to better
inform the current policy debate.
In this paper, I offer evidence that contradicts some of conventional wisdom about women
and children on welfare. While women receiving some welfare do appear to differ from non-
recipients in their motivational measures immediately after the time of AFDC receipt, I find no
significant correlation between some welfare receipt and the hours of paid work by the mother ten
years after receipt. Nor do I find a significant relationship between either category of AFDC receipt
and income-to-needs measured during the final time period of this study. However, I do find a
significant relationship between occasional welfare receipt and the educational attainment of the
women’s children. These relationships become less clear when looking at women who experienced
long-term welfare receipt, opening the door for further research.
Alvarez, M., Burrows, R. Zvaighat, A. & Muzzo, S. (1987). “Sociocultural Characteristics of
Pregnant and Nonpregnant Adolescents of Low Socioeconomic Status: A Comparative Study.”
Adolescence, 22(85), 149-156.
Anderson, E. (1990). Streetwise. Chicago: University of Chicago Press.
Hill, M., Augustyniak, S., Duncan, G., Gurin, G., Gurin, P., Liker, J., Morgan, J., and Ponza, M.
(1985) Motivation and Economic Mobility, Ann Arbor: Institute for Social Research.
Jencks, C. (1992). Rethinking Social Policy. New York: Harper Perennial.
Lewis, O. (1966). “The Culture of Poverty,” Scientific American, 215, 19-25.
Moffitt, R. (1992). “Incentive Effects of the U.S. Welfare System: A Review,” Journal of
Economic Literature, 30, 1-61.
Olsen, R. & Farkas, G. (1990). “The Effect of Economic Opportunity and Family Background on
Adolescent Cohabitation and Childbearing among Low-Income Blacks.”
Journal of Labor Economics, 8(3), 341-363.
Oritz, E. & Bassoff, B. (1987). “Adolescent Welfare Mothers: Lost Optimism and Lowered
Expectations,” Social Casework: The Journal of Contemporary Social Work, 400-405.
Plotnick, R., Klawitter, M. & Edwards, M. (1997). “Do Psychosocial Characteristics Affect
Socioecnomic Outcomes? The Case of Welfare Use by Young Women,” unpublished manuscript.
Popkin, S. (1990). “Welfare: Views from the Bottom,” Social Problems, 37, 64-79.
Puntenney, D.L. (1997). Poor mothers at work: Balancing family and labor market responsibilities
in an inner-city community. Doctoral dissertation 378 NU 1997. Evanston, IL: Northwestern
Rainwater, L. (1967). The Moynihan Report and the Politics of Controversy, Cambridge: MIT
Rector, R. (1993). “Why Expanding Welfare Will not Help the Poor,” Heritage Foundation Lecture
Series, Washington, DC.
Sisco, C. & Pearson, C. (1994). “Prevalence of Alcoholism and Drug Abuse among Female AFDC
Recipients,” Health and Social Work, 19, 75-77.
Wilson, W. (1987). The Truly Disadvantaged. Chicago: University of Chicago Press.
Veroff, J., McClelland, L., and Marquis, K. (1991) Measuring Intelligence and Achievement
Motivation in Surveys. Final Report to HEW, OEO, Contract No. OCO-4180.
Table 1: Weighted Percentages of Women in AFDC receipt categories Between 1968 and 1972
NO AFDC RECEIPT IN ANY YEAR
AFDC RECEIPT IN AT LEAST ONE YEAR
NO AFDC RECEIPT IN SOME, NOT ALL YEARS
AFDC RECEIPT IN SOME, NOT ALL YEARS
NO AFDC RECEIPT IN ALL YEARS
AFDC RECEIPT IN ALL YEARS
Total Number of Women