Reliability of Slope Scores for Individuals.
Abstract
Slope scores representing the rate of human information processing have often been used as dependent variables in experiments and in correlation studies. The reliability of the slope scores for individuals is an important consideration because it affects the power of experiments and the maximum expected correlation in correlation studies. This article examines the reliabilities (interday correlations) across 15 days of repeated measurements for each of six prominent human information processing tasks: high speed memory scanning, 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. This is remarkable because the slopes include more data than each mean response time. Reasons for the relative unreliability of slope scores are discussed. Strategies for improving the reliability of slope estimates are suggested. Finally, it is argued that in applied experimental 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. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Apr 01, 1983
- Accession Number
- ADA130252
Entities
People
- Michele Krause
- Robert C. Carter