The stability of recovery from the foot cold pressor: An application of multivariate generalizability theory

Date of Award




Degree Name

Doctor of Philosophy (Ph.D.)

First Committee Member

Maria L. Llabre, Committee Chair


Cardiovascular reactivity has been intensively researched during the past decade, but less emphasis has been focused on the recovery period following the task. The research literature has, however, provided some evidence of racial and gender differences in cardiovascular recovery. This study used a novel approach, namely multivariate generalizability theory, to assess the stability of baseline measurements recorded prior to a forehead cold pressor task and recovery measurements recorded following the task. A multivariate approach was reasoned to be an appropriate methodology to determine the stability of the measurements because the cardiovascular system is an integrated, dynamic, multiparameter system. Multivariate generalizability provides not only the same results that can be obtained from a univariate study, but also information regarding the errors in the measurement of covariances that is only available with the multivariate approach. In addition, the maximally stable linear composite of the variables can be determined with this approach. Furthermore, this methodology is flexible, allowing for the recovery period to be conceptualized as a profile which can be separated into level and shape components. The baseline analysis used 42 normotensive, white males who had no known cardiovascular disease and were free of any medications with known cardiovascular effects. The recovery analysis used a subset of 14 of the 42 subjects who were randomly assigned to a foot cold pressor task. Cardiovascular recovery was defined as the fifteen minute period following removal of the foot from the cold pressor. A variety of hemodynamic variables was collected using impedance cardiography.Variables of interest in this study were SBP, DBP, HR, stroke volume (SV), cardiac output (Q), total peripheral resistance (TPR), and Heather Index (HI). The multivariate analyses were applied to the baseline, the recovery mean, the recovery delta (mean-baseline), and the recovery profile. Results demonstrated that when the cardiovascular variables were combined into a linear composite, the measurement was more stable than that of any individual variable. This was particularly true for the deltas, which are typically less reliable than absolute levels. When recovery was regarded as a profile, the shape of the profile was less stable than the mean level of the profile. Finally, this study demonstrated that the reliability of the crossproducts could be used to correct observed correlations which contain error. An updated form of the Spearman's Correction for attenuation was also provided; this can be used when reliabilities based on covariances are known. This study has demonstrated the use of a methodology which may allow researchers to obtain more reliable measures. Reliable measures are critical in cardiovascular behavioral medicine if true links between reactivity, recovery and disease are to be found.


Psychology, Psychometrics; Psychology, Physiological

Link to Full Text