![Sample our Communication Studies journals, sign in here to start your access, 2013 & 2014 volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5927/TOC-Subject-Sample-2021-Information-Science2.jpg"})
Abstract
This study evaluated the effect of support matching, perceived support adequacy, and support invisibility on physiological stress recovery. Participants (N = 103) received supportive messages from a dating partner after completing a series of stressful tasks and receiving negative performance feedback. Participants reported their general preferences for emotional support before the interaction, and afterward they evaluated support adequacy and their partner's supportiveness during the interaction. Observers rated the emotional support provided, and salivary cortisol indexed changes in stress. As predicted, the positive association between a partner's provision of emotional support and rate of stress recovery was greater for people who reported a greater preference for emotional support, in general, and people who evaluated their partner's support during the interaction as more, compared to less, adequate. Contrary to expectations, support invisibility did not attenuate emotional support benefits.
Acknowledgments
The services provided by the General Clinical Research Center of the Pennsylvania State University are appreciated. The study was supported by NIH grant [M01 RR 10732] awarded to the first author.
Notes
[1] The support matching hypothesis emphasizes the type, not the level, of support provided. For example, the effectiveness of a supportive response to a person seeking informational support (i.e., advice) depends on whether the provider gives the desired information. Alternative types of support (e.g., “you're so smart; I’m sure you'll figure it out” or “I can empathize with how frustrated you are”) are less likely to facilitate stress recovery.
[2] Details on collecting the cortisol samples can be obtained from the first author.
[3] The pre-interaction survey included a single item measure of relationship satisfaction (“All things considered, how satisfied or dissatisfied have you been with your relationship with this person over the last two months”). Responses on a 7-point scale (1 = completely satisfied, 4 = neutral, 7 = completely dissatisfied) were reversed coded, and descriptive statistics indicated a highly satisfied sample, (M = 6.16, SD = 1.10). This variable was not significant in the models that evaluated our hypotheses, and covarying this measure did not change the results we report.
[4] On scales completed subsequent to the procedures, participants reported that they felt that the tasks made them question their abilities (M = 4.70, SD = 1.99 on a 7-point scale), were overwhelming (M = 5.15, SD = 1.75), were frustrating (M = 5.49, SD = 1.70), and posed a challenge they could not overcome (M = 4.25, SD = 1.53). Thus, we believe participants were receptive to the negative feedback they received.
[5] We recognize that the instructions were unlikely to produce a lasting change in the partners' supportive communication behavior, and studies using confederates to deliver person-centered support typically provide hours of training. Because training was provided immediately prior to the interaction in which it was to be implemented, we anticipated that the instructions to partners would be sufficient to effect the manipulation of interest.
[6] We did not observe significant sex differences in the cortisol variables, which suggests that the stressor activities were equivalent for males and females.
[7] Our hypotheses focus on observable emotional support, as measured by the judges' ratings, and testing H3 requires the measure of the participants' evaluations of supportiveness. Because these measures provide continuous measures of enacted and perceived support, we used them, rather than the dichotomous manipulation of support provision, in tests of our hypotheses. Including a dichotomous variable reflecting assignment to the neutral vs. supportive communication conditions as a control variable in our analyses did not alter the findings we report subsequently.
[8] Cortisol samples collected from the same individual are highly correlated, because people who have a pronounced increase in cortisol in reaction to the stimulus activities tend to have higher cortisol levels over the course of six samples (and vice versa). Consistent with this expectation, we observed that baseline cortisol measures were only moderately correlated with subsequent cortisol measures (rs ranged from .30 to .43, df = 101, p < .01), but correlations among the five cortisol measures indexing reactions to the study procedures were strong (rs > .68, df = 101, p < .001) with the highest correlations between the final two measurements (r =. 97, df = 101, p < .001). Multi-level modeling statistically accounts for dependency in linked observations and allowed us to evaluate the extent to which the participant-level variables significantly influenced the slope of the recovery portion of the growth curve, over and above the trends created by tendency for the cortisol measures to covary.
[9] We also evaluated relationship length on the intercept and both slopes. Because it was not a significant predictor of cortisol, it was removed from the models.