Richard P. Phelps, via a kind email:

The mainstream research that informs our country’s education policies is often more caricature than genuine research. Policy discussions tend to be dominated by the research that the ruling class inside education wishes to be true, rather than by that which is true.

Among the several falsehoods book author Daniel Koretz and his colleagues have peddled over the years is the claim that the evidence for the benefits of testing is “thin” (and the evidence for costs abundant). Largely in response to their claims, I several years ago published a meta-analysis of 100 years’ worth of research on the effects of testing on student achievement. I reviewed over 800 quantitative, experimental, survey, and qualitative studies. The weight of the latter two types of studies was overwhelmingly positive (e.g., 93% of qualitative studies found a positive result to a testing intervention and the average effect size for survey studies exceeded 1.0, a very high effect. The effect sizes for the quantitative and experimental studies—hundreds of mostly random assignment experiments dating back to the 1920s—ranged between moderately and highly positive.

Because I read and heard the same messages as everyone else from those prominent education researchers who receive press attention, I had expected to find clearly negative effects. Some of the most widely covered studies allegedly demonstrating that testing was, on balance, harmful were included in my meta-analysis. But also included in my meta-analysis were hundreds of studies that had received virtually no public attention. Testing experts, education practitioners, and psychologists performed most of those studies.

(True to form, not a single education journalist has ever asked me about the meta-analysis. Meanwhile, DC-based education journalists talk to anti-testing spokespersons thousands of times a year and often promote the single research studies conducted by celebrity researchers as hugely consequential to policy.)

Therein lies the chief secret of the success of the anti-testing forces in education research: they count (i.e., cite or reference) the research that reaches anti-testing conclusions and they ignore the abundance of research that contradicts. (For the few pro-testing studies that receive so much public attention they cannot simply ignore them, other information suppression methods may be used, such as dismissive reviews, tone policing, misrepresentation, or character assassination).