Welfare and Work in Camden in the 1970s

by Ted Goertzel

Although welfare reform has become politically prominent since the last congressional elections, the issues which are being raised are not new. There have been several previous attempts to reform the welfare system. Before the FDP (Family Development Plan) there was REACH (Realizing Economic Achievement), and before REACH there was WIN (Work Incentives Now). Each of these reforms attempted to help welfare mothers become independent. If they failed, it would be good to know why so we do not make the same mistakes again.

There were evaluation studies of each of these programs, although the results were often ignored because the programs had changed before the research was complete. In 1969, 1970, and 1973, for example, a research team led by Samuel Klausner of the University of Pennsylvania, with a grant from the U.S. Department of Labor, carried out a major longitudinal study of Camden welfare mothers. Klausner and his associates wrote several reports using the data from the each of the surveys, and we summarized some of their findings in an earlier issue of FPR News. But, so far as we have been able to determine, Klausner's team did not complete the analysis by combining all three data sets to look for trends over the four year period.

This data provides a historical record which cannot be replaced. The Forum for Policy Research contacted Professor Klausner and got his help in locating copies of the data on old computer tapes. With some difficulty, computer consultant Meg Koppel converted it to a format which could be read by modern computers. Alan Harrod merged the three files into one file for data analysis. Students in the fall, 1994, survey research class then took the massive data sets and searched for interesting findings.

Sample Characteristics Klausner's

MONTHS WORKED IN 1973

Did not work

51.2%

Less than one month

3.5%

1 to 3 months

8.7%

4 to 6 months

8.3%

7 to 9 months

4.5%

10 to 11 months

4.5%

12 months (inc. vacations)

19.4%

team interviewed 444 women in 1969, 447 in 1970 and 352 in 1973. They got the names and addresses from the Welfare Department, and the sample was drawn from women receiving welfare in Camden County as a whole. Alan Harrod was able to match up 289 cases in all three years for the analysis in this report. Most of the women (76%) were black, 17% were white, 5.5% were Puerto Rican and 1.4% "other." Today, of course, the percentage of Hispanics would be higher, but Klausner's sample was representative of welfare recipients in the County at the time.

All of these women were receiving at least some money from AFDC in 1969, 1970 and in 1973. This means that the women who were successful in getting off welfare completely were not included. Since we were able to find 289 of the 444 women from 1969 in the 1973 sample, we conclude that 65% of the women who were on welfare in 1969 continued to be on welfare in 1973. Clearly, welfare dependency is not a new phenomenon in Camden County. Because of the limitations of the data set, this paper is a study of these persistent welfare recipients, not of those who left welfare.

Correlates of Working While On Welfare. The file includes a tremendous amount of data including detailed financial information, family histories, attitudinal measures and psychological tests. A full analysis would make a fine dissertation topic for a D.S.W. student (if anyone is interested, please call the Forum). For this report, we will focus on the question of work behavior. We will ask, which variables, measured in 1969 or 1970, help to explain which women would be working in 1973? As Table One shows, although all the women were still receiving at least some welfare, almost half of them had worked at least part of the year in 1974. Nineteen percent of them worked all year.

Table Two lists a number of variables, measured in 1969 or 1970, which were significantly correlated with the number of months worked in 1973.

----------------------------------------------------------------------

Table Two

Correlations with Months Worked in 1973

Variables Measured Correlation Number of Significance

in 1969 or 1970 Coefficient Cases Level

Respondent wanted to work .19 288 .001

rather than stay home (1969)

Years of Education .23 287 .000

Belief in fatalism scale (1969) -.13 289 .013

Incentive to work (1969) -.03 289 .283

Maternalist Values Scale (1969) -.08 287 .080

Belief that people look down .15 288 .007

on welfare mothers. (1969)

Ideal Family Size (1969) -.16 227 .009

Age of youngest child (1973) .06 287 .175

Amount of money used to pay .17 289 .002

off debts in 1969

Resp. had work experience (1969) .10 287 .049

Earnings on the last job (1969) .06 260 .181

Intelligence in 1969 .05 286 .214

Intelligence in 1970 .16 275 .005

Intelligence in 1973 .17 267 .002

Respondent was married in 1969 -.10 289 .038

Respondents' father supported .15 289 .006

family when she was growing up

Months worked in 1970 .31 276 .000

----------------------------------------------------------------------

Respondents who said they preferred working rather than staying home were, in fact, more likely to be working four years later. However, most respondents chose this answer, and many were not working, so a preference for working is clearly not enough. Of course, women might prefer to work if they could find a well paying job and good child care, but not at the kind of job that they can realistically get. Only 17% of the women had graduated high school, only 2.8% had attended a business or vocational school, and only 1.4% had attended college. Half of them had completed only the 9th grade. Our correlational analysis showed that educational attainment was strongly correlated with getting off welfare and going to work.

Klausner included several attitude scales, measuring beliefs which were thought to be part of the "culture of poverty," a theory which was more popular in the 1960s than it is today. Women who scored high on a scale of "fatalism orientation" in 1969 were somewhat less likely to be working in 1973. A scale of the "incentive to work" as expressed in 1969, however, was not significantly related to working in 1973.

Another possibility is that a strong belief in primacy the maternal role keeps women for working. One of Klausner's conclusions was that the welfare mothers' strong commitment to traditional maternal roles kept them from working. Using his scale of maternal values in 1969, however, we found only a weak and marginally significant correlation with months worked in 1973. We got a higher correlation, however, when we measured maternalism by asking how large women thought the ideal family would be. Those who thought a larger family to be ideal were less likely to work.

Somewhat contradictory to the belief that a large family is desirable is the belief that people in general look down on welfare mothers. This belief, in 1969, was correlated with the number of weeks worked in 1973.

On a more practical level, it could be argued that women work if they need the money to supplement their grant. The 1969 survey included a measure of how much money the women were spending each month to pay off debts. Women who were spending a lot on paying off debts were significantly more likely to be working in 1973. The amount of money earned on the last job, however, was not correlated with the likelihood of working.

An obvious hypothesis is that women are more likely to work outside the home when their children are older. This seems so obvious as to be self-evident, yet it did not work out with our sample. There was no significant correlation with age of the youngest child in any of the years and the number of months worked outside the home in 1973. Apparently, welfare mothers who need or want to work find a way to do so regardless of the age of their youngest child.

Eight-six percent of the respondents reported, in 1969, that they had had some work experience. This variable correlated only weakly with the likelihood of working in 1973, however.

The controversial book The Bell Curve has recently argued that welfare dependence is due to low intelligence. Klausner and his colleagues thought that perhaps 20% of the respondents were unable to work because of low intelligence. They included a measure of intelligence (the digit-symbol items from a standard IQ test) in their survey in all three years. The measure was reliable in that the scores were consistent from year to year (alpha reliability = .86). We found that there was no correlation between IQ in 1969 and the likelihood of working in 1973. There was, however, a correlation between intelligence in 1970 and in 1973 and working in 1973. This data suggests that staying home with the kids for years tends to depress a woman's IQ score, while working outside the home increases it. We excluded the 16 Puerto Rican women from these analyses because their scores were exceptionally low, probably due to cultural differences.

Only 43 of the respondents were married in 1969, and these women were somewhat less likely to be working in 1973 than the unmarried women. There was no difference in this respect between the women who had never married and those who were separated, divorced or widowed. Almost none of the women in the sample married during the four years of the study, probably because women who marry generally leave welfare and were lost to the study.

Some people believe that welfare dependency is a cultural factor which is passed along from generation to generation. Klausner included a question asking who supported the respondent's family when she was growing up. The traditional pattern is for the father to support the family, and 69.2% of the respondents had grown up in families of this sort. Seventeen percent had been supported by the mother and 8% by other relatives. Only 3% had lived in families supported primarily by welfare, and 2.4% in institutions or other arrangements. Seventy percent reported that their mothers had never been on welfare, and those whose mothers had been on welfare were no less likely to have worked in 1973. Thus, welfare dependency was not an intergenerational family tradition. We did find, however, that those women whose fathers had supported their families in their childhood were somewhat more likely to be among those who were working in 1973.

Finally, and unsurprisingly, the number of months worked in 1970 is positively correlated with the number of months worked in 1973.

A Causal Model. One of the advantages of a longitudinal data set such as this one is that we can look at changes over time. This gives us a firmer basis for drawing causal inferences than we would have from survey data collected at one point in time. The particular challenge with this data set was to develop an understanding of what factors cause women who are on welfare to work outside the home.

For a causal analysis, we need to go beyond the bivariate correlations in Table One and look at the complex ways in which the variables interact with each other. The results of this analysis can be seen in the path diagram in Figure 1. In the Figure, the causal paths move mostly from left to right as indicated by the arrows. The numbers on the arrows (beta weights) indicate the strength of the causal links. [Note: Figure 1 is missing from this file, but is in the printed report, email me if you want a copy at goertzel@camden.rutgers.edu]

The strongest link is between months worked in 1970 and months worked in 1973. This confirms the common observation that the most important thing is to get a mother out of the home and into the workforce as quickly as possible. This apparently establishes a pattern and changes her expectations, as well developing skills.

Years of education is also important, both in the short term and in the long term. Women with more education, as measured in 1969, are more likely to be working in both 1970 and 1973. This is true despite the fact that the average level of education in this sample was quite low. If a larger percentage had graduated high school or gone to college, the correlation might be higher.

Two other 1969 variables are related to months worked in 1973, even when other factors are controlled. One is the amount of money the woman was paying each month to service her debts. This may be an indirect measure of a desire for consumer goods, or simply difficulty in living within a very limited income. It may reflect a lack of income support "under the table" from family or boyfriends. The other is ideal family size, how many children a woman thought desirable. This may reflect a basic value question, is it more important to stay home and have a lot of children or go out into the world and be able to buy and do things?

It is interesting to note the variables which do not show up in the figure. The age of the youngest child would seem to be an obvious factor, but it did not work out statistically. Fatalism and other value orientation questions seem to be byproducts of education and work, not causes. This is also true with intelligence. A woman's intelligence as measured in 1969 did not product whether she would be working in 1973, but there was a bivariate correlation between months worked and IQ in 1973. This leads us to conclude that working has a positive impact on the intelligence test score.

The finding about debt recalls an anecdote about the first capitalists to open factories on the Hawaiian islands. At first, they had no trouble recruiting local women to work in the factories. After a few months, however, most of them quit as they had already purchased everything they needed and wanted to stay home with their families. The capitalists were unhappy with this outcome, however, so they distributed Sears catalogs to all the women. In our society, of course, there is plenty of advertising promoting consumer goods. Perhaps what is needed is to give welfare mothers credit cards so they will run up debts and have to work to pay them off.

Of course, this suggestion is made tongue-in-cheek, not as a serious suggestion for public policy. It does show, however, that looking at data may lead to conclusions different from what we had expected.