Jay Hein, president of the Sagamore Institute for Public Policy Research in Indianapolis, served as director of the White House Office of Faith-Based and Community Initiatives from 2006 through 2008. During the 1990s he was instrumental in reforming the Wisconsin welfare system.

What was the Wisconsin welfare system like when you arrived? It needed to be transformed. We thought the only route out of poverty would be work, but the welfare "job" was to sit behind a computer and prove how poor you were to become eligible for benefits.

So your goal was to ... To say to this person full of promise, "You can realize your potential in the workforce." We had to transform the service delivery system.

Did you? We had 97,000 heads of household on welfare in 1993 in Wisconsin. All the sophisticated social science literature at the time said a third of them were gaming the system and a third could never be working. Within four years our 97,000 head-of-household case load-300,000 or so people-went to a 7,000 head-of-household welfare caseload.

What happened to those 90,000 heads of households? How many of them actually went to work? The vast majority. We proved the point that labor force attachment is not only statistically the only road out of poverty, but sustainably so. We experienced the same thing nationally after 1996 when the federal law was passed: For single mothers employment went to record highs, and poverty went to record lows.

How do you sustain that progress? It's hard work. People often need not only a job but a progression of jobs: What does career progression look like to someone who grew up in a welfare household that knew only inter-generational poverty? The answer is not getting a job: It's getting attachment to the workforce.

Did those who went to work stay attached? Has anyone tracked them? In the late 1990s and early 2000 we stopped talking about welfare reform. Each governor produced the press release declaring victory because of caseload reduction. We took it off the nation's agenda. The research atrophied. Now we are running on the fumes from those days.

As you were reforming welfare in Wisconsin, I visited the Kenosha County Job Center in that state and saw women being taught to forget about the possibilities of marriage-even though the life of a single mom is so tough. The officials were also decidedly opposed to religious influences. Right. You've diagnosed the difference between what government can do and what government shouldn't do. I would never trust government with the formation of character and the building of the person. That is the role of civil society and faith-based institutions.

Congress did learn enough from Wisconsin to change the name of one big welfare program from AFDC, Aid to Families with Dependent Children, to TANF, Temporary Aid for Needy Families. Wasn't the emphasis on "temporary" an improvement? Correct. I was thrilled to be a part of that experiment, depressed as you were to realize the job wasn't done, and hopeful when I discovered who was doing the job and had been doing it all along: people of faith and good will who were serving their neighbors in need.

That gets to the question of how government should interact with religion-based institutions. We want the best providers to help people recover from drugs or attach to the workforce. At the same time we need to respect the dynamic balance of the First Amendment. It protects the free expression of our religious faith and at the same time protects against the establishment of religion.

How does that work out in practice? You know the metaphor about two different types of organizations: salad groups (lots of lettuce, with "religion" mixed in like pieces of tomato or cucumber) and brownie-mix groups ("religion" inseparable from the entire product). What if brownie-mix groups are often more effective in fighting addiction than salad groups? Which I happen to believe.

As do I-yet government, in terms of direct grant-making, can fund salad groups that are less effective, but not the brownie-mix groups. Yes. That's why vouchers are the answer. You get to use taxpayer resources but in a manner that enables consumer choice. If a person receives a voucher and can choose any provider he wishes, there's no constitutional problem.

It seems to me that the grant system, which centralizes power in Washington, is in and of itself a problem. The government has subcontracted services in increasing proportions since World War II. Many of those grant-receiving organizations were faith-based, but in the '60s and '70s discrimination arose and choked the participation of many of these organizations. Faith-based groups and smaller, grass-roots organizations suffered the most.