Baseline levels in muscle relaxation training.

Variations in the baseline levels of physiological measures, a familiar problem in psychophysiological research, can affect the results of clinical applications and research in the self-control of bodily processes. In this presentation, the problem is illustrated within the context of skeletal muscle relaxation training using continuous biofeedback (BF) based on surface electromyographic (EMG) activity. In terms of the Law of Initial Values (LIV), higher EMG levels are expected to be associated with greater decreases during training. The combined results of two studies documented an LIV-like effect for pretraining baseline levels with greater EMG decreases after training for subjects with the higher pretraining baselines. Left uncorrected, such baseline differences were shown to lead to discrepant results between two identical studies, and therefore to conflicting conclusions about the effectiveness of these procedures. The available methods suggested to correct for the biasing effect of baseline differences in research are described, with particular emphasis on the analysis of covariance.