A video clip for the long weekend

Did you play imaginary games with yourself, staring out the car window? I do, too. And here's a video suggesting that, well, we all do it.

Jack Kerouac probably wasn't the first, either. What does this have to do with my numbers blog? Well, we all think we're unique. It's one of the arguments I get when I work with staff in programs that are starting to use outcome measures (and sometimes when they've been using them for a while): we're all different. Lumping us all together to count things overlooks our differences. Well, yes. But we have some commonalities too, and it's occasionally helpful to be reminded of that. The Atlantic posted a brief interview with the creator, Ian Bennett, here. Enjoy the long weekend!


Calorie restricted diets.

How is the thoughtful reader supposed to interpret Gina Kolata's New York Times story about the newly reported findings that calorie restricted (20% or more below typical intake) diets do not extend life, at least not in monkeys? Earlier studies of mice and other primates had shown promising results, and there is at least one study of the effects of calorie restricted diets in humans underway.

The article includes several useful hints:
  •  Replication is key to the scientific method, and failure to replicate a result means that the original study may have had flaws. In fact, as Kolata reports, the original study had several flaws, notably that deaths among younger monkeys in the earlier study were excluded from the study.
  • The new study suggests that restricted calorie diets may have some beneficial health results: older monkeys put on the diet had lower levels of triglycerides. Monkeys put on the diet when they were younger had less cancer. That's an interesting result, but might it result from what the monkeys were fed rather than the amount? According to the news reports, that question remains unanswered.

Some people have decided to start eating less as a result of early reports. And others simply refuse to believe the results of the new studies. I think that it's too soon to tell, though the failure to replicate the results is suggestive. I'd also like to know more about the content of the diets beyond the restriction in calories. It will be interesting to see what the human study reports.

What do you think? I'd be interested in learning others' thoughts in the comments.


Social Impact Bonds: A Primer

Now that the first social impact bond-funded program in New York City (and in the US) has been announced, I realized it was time to start learning about social impact bonds. A primer with some helpful links follows.

1. What are social impact bonds? How do they work?
Social impact bonds are a financing model in which government contracts with a bond issuer to pay for services based on outcomes or achieving performance targets. The bond issuer raises operating funds by issuing bonds, and contracts with service providers to deliver the services. Here's a diagram, based on social impact bond-funded work in the United Kingdom, showing how the money flows:

The Center for American Progress has a good article describing social impact bonds in more detail here.

2. Where have social impact bonds been tried?
Peterborough Prison, outside of London, is the site of one effort already underway. The bond issuing organization is Social Finance in the UK. The first US program, in which Goldman Sachs will fund an MDRC program serving jailed adolescents, has just been announced. Social Finance's US arm has identified some additional promising programs.

3. How has the Peterborough program worked out?
It's too soon to know whether the Peterborough prison program will generate a return for Social Finance investors - that information will first come in Year 4 - but you can read an interesting report about the program's promising first year here.

4. What are the benefits of using social impact bonds?
There are several. First is the improvement in evaluation - in order to show that their programs have worked, bond issuers and service providers must undertake rigorous evaluation. At present, there's rarely enough money to do that. Then there's the fact that approaches that appear to work can be ramped up to serve more people faster. That's good for everyone, for taxpayers as well as anyone receiving the services. Furthermore, as the Center for American Progress report puts it:
Government agencies, which might otherwise continue to fund the same old approaches they have funded in the past, would have an incentive to invest in promising new strategies, including preventive services. That’s because the risk of wasting taxpayer dollars if the new approaches fail is transferred to the private sector.

5. Are there barriers or concerns about social impact bonds?
Jeffrey B. Liebman, author of the Center for American Progress report, has identified five:
 a. Not all the interventions that are funded this way will succeed. Those that do must have a net payoff high enough to provide investors in the bonds with a return on their investments.
b. Program outcomes must be clear, measurable, and directly and comprehensively related to the program. And there must be some way of showing both of those considerations. That might require an independent entity to do the reviewing. (I related Jim Manzi's suggestion of a government agency to oversee the design and interpretation of randomized social policy experiements in my review of his book.)
c. The treatment and comparison populations must be defined at the outset in order to ensure that the program does not serve only those easiest to serve.
d. Outcome assessments must include an assessment of what might have happened without the program.
e. Funding agreements should include contingency planning for the shutdown of under-performing programs so as not to hurt the treatment population - or the service provider - if a program is not successful.

Social impact bonds have promise to shake up what can be a slow changing sector. I'll be watching the MDRC and Peterborough programs with interest. 

Popular Posts