Response Scaling: Night-time Baselines, Resting Baselines, and Initial Value Dependencies

Ideally, response scores that are derived to assess individual differences in responsiveness should be independent of the researcher's subjective preferences for any specific baseline condition, for instance, resting (pre-task) values or night-time baselines. A methodological investigation was designed to evaluate three issues: namely, baseline bias, initial-value dependency bias, and response score bias. 51 hypertensive patients participated in three 24-hour periods of psychophysiological monitoring. The monitoring consisted of measurements taken during rest periods and tasks in the laboratory, as well as measurements during daytime activities and during night-time. The assessment included the cold pressor test, mental load, active relaxation, and stair climbing as a physically demanding task. Blood pressure and heart rate were employed in exploring specific methods of response scaling. The findings indicated that the assessment of individual differences in responsiveness, i.e. rank ordering of subjects, was biased substantially. (1) Inconsistencies exist between response scores that were derived by relating task level to night-time baselines, rather than to resting baseline in the laboratory. This deviation was obvious for task-baseline differences but less evident for residualized change scores and true scores. (2) Positive initial-value dependency for BP and HR, i.e. association of higher baseline and higher response magnitude, was found when resting baselines were used. However, negative initial-value dependency was found in several instances when night-time baselines were employed. (3) Inconsistencies were also evident among various methods of response scaling, whereby a discordance seems to exist between the simple difference, on the one hand, and residualized change score and true score models, on the other. A strategy of response scaling is suggested that disregards simple difference scores and employs (1) a residualized change score to represent incremental change and (2) a design-specific true score model that allows for both a test of initial-value dependency and an estimation of true difference. Baseline bias, initial-value dependency bias, and response score bias may be responsible for some of the inconsistencies in outcomes of psychophysiological research.