DR. MURNANE: Those of you who know me know that when I talk to big groups I have trouble with certain combinations of words. I stutter on certain words and I need to ask you to help me. When I get stuck on a word, say the word. If you get it right, I just move on. If you get it wrong that may be the start of a much more interesting conversation. So, I do ask for your help.

I want to talk about how research on and progress in the treatment of cystic fibrosis I think has direct relevance to how we think about improving educational practice and the role of research with multiple methods in improving educational practice.

To kind of set the scene let me talk a little bit about how I see the challenge superintendents across the country face; and with No Child Left Behind, superintendents are under enormous pressure to improve learning for all children.

How do you think about that work? I want to suggest there are at least two polar models that inform how educational leaders think about their work and it is probably the case that no superintendent subscribes totally to either of these polar models, but some are much closer to one than the other and these models in my view have quite different implications for research.

The one model is that you will look for the most effective curricula and instructional methods. You buy the most effective curricula and you train teachers to implement the curriculum in the appropriate fashion and you work to get fidelity of implementation. Now, seen with this perspective the What Works Clearinghouse is an extremely helpful resource. Now, this is the research and development paradigm that goes back to agricultural research.

A second model is that some children will not develop skills at an acceptable pace irrespective of what standardized curriculum and instructional methods a school or school district chooses. So, the challenge is to identify those children rapidly, to diagnose their learning difficulties and to develop an individualized improvement strategy and monitor progress closely; the challenge is to develop an organization that is good at that work.

Now, these two perspectives imply very different strategies. Research that would be helpful in forming these perspectives would both involve multiple methods but would be focused on different kinds of activities.

Now, I want to suggest this is not only true in education, and I want to turn to cystic fibrosis research. There are probably at least five people who know more about it than I do. Most of what I know about it comes from reading an article in the New Yorker of December 6, which is a week ago by Atul Gawande that some of you may have seen who is both a terrific writer and I am told a very good surgeon as well.

Now, there are really two themes in the Gawande piece which is called The Bell Curve interestingly enough for reasons that I will explain. The first theme is that the length of life of cystic fibrosis patients has increased markedly over the last 40 years, a more than 10-year increase and a key part of the explanation is that new treatment methods have been developed, evaluated. The improvements have been standardized and disseminated widely so all of the 100 or so centers for treating cystic fibrosis in this country adopt the best methods.

Now, this is very much akin to the research and development paradigm from agriculture and it is very similar to the first model for improving education and it fits very well with the research model that led to the What Works Clearinghouse.

For research based on this model multiple methods are clearly necessary to define treatments very carefully paying attention to detail and to identify the subgroups for which they are appropriate.

Now, the second theme in Gawande's article is that there are remarkable differences in the average length of life of patients who are treated at the different centers for cystic fibrosis research in this country.

This site-to-site variation of course is not only true in cystic fibrosis research. It is true in many other medical areas. It is true in welfare reform research. It is true really I think in every endeavor that I know about where there are for example experimental designs with many sites. So, the question is why are outcomes better in some sites than they are in other sites and in the case of the cystic fibrosis centers and the differences are remarkable again 7 or 8 years difference in length of life of patients and the centers that are good in one year tend to be better in subsequent years as well.

So, it is not that there is enormous switching. Well, according to Gawande, the explanation does not lie in differences in the mix of standardized treatments that are applied. All of those centers use, pay close attention to research. When a new standardized technique is found they apply it rapidly. Nor is it found in the observed qualifications of the staff. Those don't explain very much of the site-to-site variation at all.

So, where does the difference lie? According to Gawande the best centers are especially effective at monitoring key indicators of how patients are doing. In the case of cystic fibrosis it is lung function and monitoring regularly and identifying very early and the early is important because if you wait until patients are in trouble a lot of ground is really lost that is not recoverable. So, identifying early patients whose lung function is dropping, and then he says that the best centers are effective in training all staff to work with individual patients to diagnose the source of the problem which often stems from changes in the individual personal lives and to working with the patient to develop a proven strategy that is tailored to that particular patient and again monitoring subsequent progress very carefully.

Now, the research implications of this second part of Gawande's story are that you need to focus on the organization's ability to improve and understand the factors that contribute to an organization becoming a learning organization, that include a good patient tracking system, an early warning system of indicators, building the capacity of the staff, getting the incentives right, everybody to pay attention to these indicators and learning from failures.

Now, both of these models I would argue have value and the relevance of the model will vary depending on the setting. For example I was in a school out in the San Fernando Valley a number of years ago that served primarily kids whose first language was not English. Almost all of the staff had virtually no training in how to teach, almost no certified teachers and under those circumstances providing them with a structured curriculum and saying do it exactly this way made enormous sense particularly since you had very high turnover of staff. So, investing in developing the capacity of staff in a more broad range sort of way wouldn't have big returns given the high mobility of staff.

Another school that Frank Levy and I have written about is Boston a public school. Murphy Elementary School has made enormous progress in building a learning organization very much akin to this second model, but again they have a very stable staff. So, consequently the returns on professional development bear fruit over several years.

So I think these two alternative models for how to improve performance are very much present in education research but not only in education research. They are present in cystic fibrosis as well.

Both models imply the use of multiple research methods. However, what researchers study, how they do the research and how they write about what they find are very different and I suggest an awareness of this is really important in improving research and making it useful to educators.

(Applause.)