The goal of the Performance Evaluation Tests for Environmental Research (PETER) Program was to identify a set of measures of human capabilities for use in the study of environmental and other time-course effects. 114 measures studied in the PETER Program were evaluated and categorized into four groups based upon task stability and task definition. The Recommended category contained 30 measures that clearly obtained total stabilization and had an acceptable level of reliability efficiency. The Acceptable-But-Redundant category contained 15 measures. The 37 measures in the Marginal category, which included an inordinate number of slope and other derived measures, usually had desirable features which were outweighed by faults. The 32 measures in the Unacceptable category had either differential instability or weak reliability efficiency. It is our opinion that the 30 measures in the Recommended category should be given first consideration for environmental research applications. Further, it is recommended that information pertaining to preexperimental practice requirements and stabilized reliabilities should be utilized in repeated-measures environmental studies.

This article examines the reliabilities (interday correlations) across 15 days of repeated measurements for each of five prominent human information-processing tasks: proactive memory interference, semantic reasoning, letter search, typographic error search, and choice reaction time. In each case, the reliability of the slope scores is less than the reliabilities of the mean response times from which the slopes were calculated. Reasons for the relative unreliability of slope scores are discussed. Finally, it is argued that in applied research it is usually unnecessary to calculate slope scores for individuals because the more reliable mean response times are sufficient to answer common applied questions.

No abstract

Item Recognition (Sternberg, 1966) is a task which reflects the operation of human memory. This task was considered as a candidate for use in a battery of Performance Evaluation Tests for Environmental Research (PETER). Environmental research involves comparison of performances in a baseline environment and in a novel environment. It is desirable that scores be stable at different occasions in the baseline environment, so that changes due to the novel environment will be clear if they occur. It was found that item recognition results were similar to those obtained by other investigations, although the traditional item recognition score (slope) was unreliable across repeated measurements. The response time (RT) was stable for each of the four memory set sizes (1, 2, 3 & 4 items), from the standpoint of reliability, after the fourth session.

No abstract