A qualitative study investigated the test-taking behaviors, knowledge, and perceptions of 20 urban, adult basic education students reading at third to fifth grade equivalency levels. The entire reading comprehension subtest of the Test of Adult Basic Education, levels E and M, was administered under standardized conditions. A combination of observation and self-report methodologies was used. The
... [Show full abstract] researcher observed the participants during the test and asked some questions raised by observations after the test. A few days later, the researcher asked the participants to give advice to a hypothetical test-taker and to read each passage aloud and choose answers orally without referring to previous answers. A comparative analysis was made of the data. Findings indicated that levels E and M lacked construct validity because the majority of test-takers were penalized for interacting with the text, interpreting it, and bringing their own knowledge and experience to bear upon the question-answering tasks. Two other factors also heavily influenced the validity of participants' scores--test-wiseness and speed. Problems of test design adversely affected participants' ability to demonstrate their reading comprehension skills. The test-takers demonstrated a high degree of metacognitive awareness and self-regulation. (YLB)