Identifying the Distress Cues that Influence Support Provision: A Paired Comparison Approach

ABSTRACT

Just as chemistry has methods for determining the composition of unknown substances, psychology needs, but has lacked, effective methods for identifying the “active ingredients” of social stimuli. We describe such a method and apply it to identify the verbal and nonverbal distress cues that serve as signals prompting others to provide social support. This method allowed us to identify 55 such cues. Furthermore, the cues that were salient in the minds of potential support providers depended on their goals. For example, when deciding whether someone needed support, negative nonverbal cues (e.g., distressed persons' fidgeting) were more salient, but when deciding whether one was willing to provide support, cues that suggested a positive outcome for the support seeker (e.g., distressed persons' positive attitude and open-mindedness) were more salient.

Keywords:

• within-subject
• individual differences
• situation features
• helping
• nonverbal
• decision making
• goals
• emotion

Acknowledgments

This article is based on the first author's dissertation, which was supported in part by NIMH Grant MH39349 to Yuichi Shoda. We wish to thank Brian Flaherty, who provided invaluable statistical consultation, and Vivian Zayas, who offered helpful comments on initial drafts.

Notes

1. Johnson (2006) used multiple methods of inducing anxiety, one of which involved a “taste-testing study” in which participants anticipated and then actually tasted a dried worm. The goal of Johnson's research was not to create stimuli for a social support study but to show how the anticipation of a negative event affects the actual experience of that event. As discussed, for the current research, the idea of anticipating the aversive worm-eating taste test was borrowed from Johnson in order to induce mild anxiety in our stimulus participants. We modified her method somewhat, as participants in the current study did not eat any worms, but were merely led to believe they would do so.

2. Because we were principally concerned with examining potential support providers' use of emotion cues in making support decisions, we wanted to ensure variability of emotion expression and general disclosure style among support seekers. To accomplish this, support seekers were asked to follow one of four sets of instructions during the interview: (1) suppress negative emotions, (2) express negative emotions, (3) mask negative emotions with positive expressions, and (4) express emotions normally. A fifth group was given no instructions. The instruction sets resulted in a range of emotion expressivity among support seekers; however, analyses specific to the manipulation are not a focus of this article. More details regarding stimulus development can be found in Zayas et al. (2008).

3. Because the manipulation check was added to the procedures halfway through stimulus production, anxiety ratings are only available for 35 of the 65 support seeker videos.

4. For two reasons, construction of the coding system required knowledge of the decision that support providers were making. First, responses often explicitly contained references to the particular decision a participant was assigned to (e.g., “I thought the second person needed more help, because…”). Second, construction of the coding system and interpretation of responses required knowledge of the context of each response. For example, for the response, “Person one seemed more distressed because he was fidgeting,” the word “distressed” would not be coded for the “Convey Distress” condition, but it would be coded for the other three conditions. Even so, coders had no a priori hypotheses concerning the types of cues that would be reported.

5. The inter-rater reliability could not be calculated for three of the cues, “timid,” “eye movements in general indicate a negative state,” and “consistently negative,” because ratings for one coder were constant. In the spirit of inclusiveness, these cues were nevertheless selected for the final list of cues because they met the second criterion of being endorsed by at least 10% of participants.

6. At level 2, dummy variables represent two of the conditions, and the intercept represents the third condition. Each model was run twice in order to determine the p-values for all three pairwise comparisons.