I see two main problems, one with the philosophy and one with the means of learning assessment.
As for philosophy, it sounds good, but like many social engineering efforts by the government, there are major unintended consequences that are too destructive. In this case, leaving no child behind has the effect of "no child pushed forward." In order to save kids who don't care about learning or whose parents don't care, we manipulate the whole system so that kids who are conscientious and who have talent are neglected. These kids don't pull down the schools' average scores on high-stakes testing, so they are left mostly to fend for themselves. The emphasis, which borders on compulsive, is on bringing up the bottom, so the school and teachers won't look so bad.
Worse yet is the unavoidable tendency to teach to the state standards and the test that is based on them. This not only engenders a "drill and kill" negative attitude among students, but many teachers just leave out most other enriching instruction that is not likely to get tested. For years, all of science was largely ignored in my state of Texas, because the high-stakes testing was restricted to English and math.
Also, the state standards are not infallible. In Texas, I know many educators who think the standards lack adequate rationale and coherence, especially across grade levels. States develop their standards by putting together a committee to write them. Such committees can be very opinionated, driven as much by ideology as by logic.
Then there is the assessment process of high-stakes testing. The President of the U.S. National Academy of Education, Professor Lorrie Shepard, recently argued that scores on these tests can be increased without any corresponding increase in learning or skills. This has been verified by use of other independent measures of the same content.
There is also the assumed necessity of using multiple-choice testing because so many schools and kids are involved. As an expert on memory, I can assure you that multiple-choice tests are the least reliable way to assess knowledge and understanding. More complex indicators of learning are needed, and this is recognized by the new, but very limited new program of "Race to the Top."
Shepard says if you really want to know whether education is being improved, especially in math and science, you have to evaluate such things as solving non-routine problems, to assess the reliability and meaning of evidence, apply knowledge in different contexts, and to communicate their learning effectively, both orally and in writing, In general, our schools don't do that.
Good teachers know how to assess learning without multiple-choice testing. They know how to structure student work requirements so that meaningful assessment is possible. In the old days, that is what teachers were expected to do. Today, they are expected to make sure the class scores high-enough on the state tests. A better approach, I submit, is to require students to create portfolios that reflect their ability to solve non-routine problems, evaluate evidence, and to apply and communicate their understanding. Then, these portfolios could be reviewed anonymously by an outside group, perhaps by educators in other schools, who in turn subject in reciprocal manner the portfolios of their own students to similar review.
Source:
Shepard, L. A. 2010. Next-generation assessments. Science. 330: 890.