Recent methods to improve generalizations from nonrandom samples typically invoke assumptions such as the strong ignorability of sample selection, which is challenging to meet in practice. Although researchers acknowledge the difficulty in meeting this assumption, point estimates are still provided and used without considering alternative…

Community colleges must often analyze and report rates for outcomes, such as transfer to four-year colleges. A single, summary rate may be an invalid measure of its achievement in the transfer goal if the summary rate ignores the real difference in enrollment composition at different institutions. California's community colleges embody a very…

Illustrates problems facing researchers trying to demonstrate causal relationships between types of nonparental care and differences between groups of Swedish children. Argues that efforts must be made to validate and interpret differences that are found. Indicates ways to avoid misinterpretation of differences that are attributable to…