The 1996 welfare-to-work reforms were supposed to end the era of "welfare queens" and put single moms to work.
Dramatic declines in cash assistance caseloads got loud political applause, but academics wondered whether the incentives intended to encourage self-sufficiency were, in fact, working.
It was clear these mothers were no longer on the rolls, but did they have jobs? If they did, had their incomes risen or fallen?
In 2004, Gregory Acs and Pamela Loprest did one of the most-cited studies in the wave of research that followed.
They found, using data from 1997 through 1999, that not quite half of welfare leavers were working.
However, state data showed as many as 20 percent had not worked in the past year, had no working partner and were not receiving cash assistance or disability payments - in short, had little to no income.
Many others moved in and out of work or relied on disability payments.
Similar trends were found in 2002 data.
Acs and Loprest also analyzed the reasons mothers weren't working.
Three in 10 said it was because they were taking care of their family. One in 10 was in school or training. And another 3 in 10 had poor health.
Other reasons for not working included inability to find a job, lack of child care and lack of transportation.
Later research zeroed in on "disconnected mothers" - women who have at least one child under 18 living with them; income less than 200 percent of the poverty line; no government cash assistance or disability payments; and are not in school. To count, they had to be in that situation for at least four months.
A 2011 study by Loprest and Austin Nichols for the Urban Institute found that the proportion of low-income single mothers who are "disconnected" has increased over the past 15 years.
About 1 in 8 fit that definition in 1996. By 2008, the ratio had jumped to 1 in 5.
About one-third of those mothers lived alone. They were much more likely than other low-income moms to live with a partner or other adults.
They were also much more likely to have an infant or struggle with a health problem but less likely to be a U.S. citizen.
About half received food stamps, and a similar proportion received public health insurance.
The dominant reason for becoming disconnected was losing a job.
Contact reporter email@example.com or 573-4197. On Twitter: @carlibrosseau