The "Ideal" Amount of Government Assistance.

I recently had a discussion with some friends about how much to help the poor.  Do you give all you have, or do you give what you think they need?  Who determines how much is enough, how much is too much?
Most people believe in the principle of "Give a man a fish = feed him for a day, teach a man to fish = feed him for a lifetime."

In 2009 the U.S. spent $533 billion in government assistance for "the poor". 
That can be divided up into 2 main categories:
1. That which must be earned or worked for (pensions, unemployement) = $249 billion
2. That which has no earning or working requirement: $284 billion

So we pay for housing assistance, food stamps, WIC, welfare, foster care, shelters, etc...

How much should people receive?  I mean it, how do we determine how much assistance people should get?

Give a single adult homeless male $1 per day, or $100, or $10,000, or $1 million.  What is the right amount.  What helps him enough to keep him alive, functioning, having some hope, moving forward, progressing, getting skills, getting a job, and getting off government assistance?
How much makes him complacent, lazy, even destructive?  Where is the line?

Should we give people enough to get up to the poverty line?  150% of the poverty line?  What is the supposed "magic number?"

Here's the problem.  I don't think anyone knows.

I honestly don't think that study has ever been done.  I don't think we have ever randomized people and said this group of 10,000 people get X, this 10,000 get y, and this 10,000 get z.  Give them all the same guidance, same programs, some access to skills training, day care, etc.  Which group does best?  Which group is the most independent/ has the highest quality of life / doesn't need government assistance  in 5 years, 10 years etc..?

I'm guessing that study would be very hard to do because there would be outrage that some people were receiving more money because of random selection.

Have we looked at it retroactively?  Has anyone looked at which countries, states, cities, etc... are giving the most money compared to local cost of living?  Have we followed the people receiving assistance to see what the result is?

We are spending $284 billion dollars per year to help people who did not have to work or do anything to earn their assistance.  We are helping them because we care, because we think it's decent, it's humane, it's what we SHOULD do.

Instead of spending all that money and never really knowing the results other than people survived...why not put some of it into a study, or into at least following the results of those we help anyway?

I want to do this.  I want to design this study, this research project.  I want to figure out how to REALLY help people with the money we are already spending.  Any suggestions?