A significant share of education and development research uses data collected by workers called "enumerators." It is well-documented that "enumerator effects"--or inconsistent practices between the individual people who administer measurement tools-- can be a key source of error in survey data collection. However, it is less…

Ainsworth et al.'s paper "Sources of Bias in Outcome Assessment in Randomised Controlled Trials: A Case Study" examines alternative accounts for a large difference in effect size between 2 outcomes in the same intervention evaluation. It argues that the probable explanation relates to masking: Only one outcome measure was administered by…

Large-scale assessments (LSAs) use Mislevy's "plausible value" (PV) approach to relate student proficiency to noncognitive variables administered in a background questionnaire. This method requires background variables to be completely observed, a requirement that is seldom fulfilled. In this article, we evaluate and compare the…

Meta-analysis of research on a topic may exclude reports that do not present their results in a statistical form amendable to summarizing procedures used. This article illustrates how researchers can consider this aspect in computing statistics from their data. An illustration of how that can sometimes be done is presented. (Author/JAZ)

Factors contributing to American students' declining scores on the TIMSS (Third International Mathematics and Science Study) between grades 4 and 8 include textbook size and U.S. educators' traditional views of middle-school years. Bad data probably invalidate claims about the grade 8 to 12 decline. International reading scores favor Americans.…

Discusses the use of standardized achievement scores to measure the quality of practice-oriented school improvement activities in school districts. Focuses on four areas of concern: the multiple functions of schooling; documentation and data collection; real world complexities; and ethics. Recommends that cautions be taken against exclusive…

Strengths and weaknesses of three types of semiquantitative methods used by the federal government in research impact assessment are presented, and examples of their use are reviewed. These include the classic retrospective method, another retrospective approach, and accomplishments books used by selected research-sponsoring organizations. (SLD)

This study uses a production function approach to identify the impact of student, parent, teacher, and school policy variables on student performance as measured by test scores. Our statistical analysis is conducted in a comparative institutional context that includes government, private, and non-governmental organization schools. We find that, at…

Having low literacy skills could impact everything from reading a newspaper and balancing a checkbook to wise consumer decisions and job advancement. This study estimates adult literacy measures for Cuyahoga County, Ohio and its municipalities. It is also important to have a sense of how many children are at risk for developing literacy problems…