Goldstein argues that discussions of education in the U.S. have repeatedly been framed in terms of moral panics. A moral panic, she says, occurs when "policymakers and the media focus on a single class of people … as emblems of a large, complex social problem." That single class of people is then systematically demonized, as politicians and pundits present "worst of the worst" cases as emblematic of the whole.
In fact, I think you could argue (though Goldstein does not quite) that moral panics do more than demonize a group of people. They serve in part to create a group of people—to delimit or describe a particular identity and mark it as deviant. In the ’80s and ’90s, moral panics around poverty created the "welfare mother"—an irresponsible, iconically black inner-city woman who supposedly abused drugs and had multiple children simply to receive hand-outs from the government. Similarly, Arun Kundnani, in his recent book The Muslims Are Coming!, argues that before the 9/11 attacks, Muslims were one assimilating ethnic group among many. Afterwards, the state began to view Islam as a dangerous identity and a problem to be studied, controlled, and policed. The control and policing is important. Moral panics create identities in order to regulate them.
That helps to explain the otherwise mystifying path that current school reform has taken. As Goldstein says, there are a couple of school reform strategies that have been successful. For instance, the Children's Literary Initiative, which provides prospective teachers with three years of training in early reading, has shown large gains in reading scores in Chicago, Philadelphia, and other areas. But rather than trying to implement such programs broadly, national attention has focused on testing to try to determine the "value-added" of each teacher through student testing. Value-added—looking at how individual teachers increase student test scores increase year over year—can be a useful diagnostic tool in some situations. However, value-added measures all have a wide margin of error, and are quite sensitive to manipulation (such as teaching to the test). As Goldstein says, "there is absolutely no reason to believe that value-added retains its legitimacy in high-stakes settings, when test scores are used to evaluate, pay, and fire teachers and administrators."
The focus on testing to evaluate teachers, then, is not based on a rational look at the research. Instead, one could argue, it's based on the logic of the moral panic, and the created identity of teachers. If welfare mothers are the cause of poverty, then you can solve the problem of poverty with tighter restrictions on welfare mothers. If teachers are the problem with education, then you can solve the problem of education with ever more vigorous control, and ever more constant evaluation, of teachers.