The latest discouraging findings as to the lasting benefits to children of the federal Head Start program highlight the need to separate the wheat from the chaff, US commentators argue.

Disappointing as the latest results may be, say Isabel Sawhill and John Baron, de-funding Head Start, especially in these hard times, would be tantamount to giving up the fight against educational failure and poverty.

They write in Education Week that the experimental separation of individuals between the program and a control group masks the variation within "multiple models and strategies".

"Although evaluations may show that the program as a whole has little or no positive effect, certain specific interventions within it may indeed be effective.

"A far better alternative is to use rigorous evidence about 'what works' to evolve Head Start and other federal efforts into truly effective programs over time, and to use sophisticated models to trace their longer-term effects on children's life prospects."

John Baron is president of the Coalition for Evidence-based Policy in Washington DC, where Isabel Sawhill is a senior fellow at the Brookings Institution and co-director of its Center on Children and Families.

The findings to which they are responding have emerged from the Congressionally-mandated randomized controlled study of the 2002-3 Head Start cohort, which measured the cognitive and social/emotional development, health status and behavior of 5,000 three- and four-year-olds.

The message that clear improvements measured at the end of the first program year largely dissipated by the end of kindergarten brought an immediate announcement from the US Department of Health and Human Services of plans to strengthen the program as part of the Obama administration's focus on early learning through the first eight years of life. [See Head Start must hold its feet to the quality fire]

As an example of the successful strands sometimes concealed within "broad funding streams" Baron and Sawhill offer Project Upgrade, a Miami initiative that trains teachers of low-income preschoolers in language and literacy instruction.

"Its interventions were shown in a large randomized evaluation to increase the development of children's vocabulary and early reading skills by four to nine months over the course of a single school year, compared with the control group," they say, going on to recommend Success for All and Nurse Family Partnership as equivalent non-preschool successes.

They urge Congress to allocate a small portion of funds invested in social programs toward rigorous evaluations to grow the number of proven interventions, and then to provide strong incentives for recipients to adopt the proven interventions and put them into widespread use.

"The Obama administration has proposed such an approach in federal teen-pregnancy and home-visitation programs. It is clearly also needed in Head Start and other large federal programs that are not performing well," they say.

"The American public is increasingly concerned about the way their tax dollars are being spent. A clear shift in direction, based on proven-effective strategies, could turn programs such as Head Start into potent forces against the major problems facing the nation."