Announcement and recruitment of subjects. The announcement explained that on a particular day the class would be devoted to a demonstration of experimental methods that would also be part of an ongoing research program on "interpersonal closeness." The main part of the announcement was as follows:
You will be paired with another person in this class whom you don't know. (We will match you, based on the questionnaire [you are about to complete], with someone we think will like you and whom you will like.) During the first hour of class on this day you and the person we have paired you with will do a series of activities (such as talking about particular topics) designed to help you get close.
Students were not required to participate, and no record was made available to the instructor of who did and did not. About 80% of the students enrolled in the class completed the initial questionnaire; of these, about 90% came on the day of the study and took part. (These percentages were approximately the same in all three studies.)
Initial questionnaire. The initial questionnaire included a consent form, a brief written description of the project (restating the oral announcement), demographic items, an item asking subjects to list all other students they know in the class, 17 attitude questions, and an attachment-style measure. The attitude questions assessed attitudes and behaviors disagreement about which would make a person undesirable as a relationship partner (e.g., "Students should dress in conventional ways" and "I smoke"). The items were created based on results of an open-ended questionnaire on this theme administered to a separate sample at the same university. For each item, subjects indicated both their agreement- disagreement and how important-unimportant the issue was to them, using separate 7-point Likert-type scales. The attachment-style measure was a version of Hazan and Shaver's (1987) forced-choice attachment-style question, modified by Bartholomew and Horowitz (1991) for their fourfold classification. Subjects read a paragraph describing each style and then (a) selected the style most applicable to themselves and (b) rated how much each style applied to them on a 7-point scale.
Matching procedure and subjects. The matching procedure involved several steps and was quite elaborate and complex. The result was the random assignment of individuals to pairs and of pairs to conditions within constraints of sex and attachment style, all counterbalanced across conditions, attachment-style pairings, and cross- sex versus all-women pairings. In addition, subjects who knew each other, as indicated by having listed the other's first name on the initial questionnaire, were not matched. Also, subjects who disagreed on any item that either had rated as very important were not matched. (See Study 2 for more details on the matching for non- disagreement.) As in most psychology courses at this university, about 70% of the students were women. Thus we decided to use only cross-sex and all-women pairings (our preliminary studies had found no differences between all-women and all-men pairings2 but had found differences between cross-sex and same-sex pairings). We randomly assigned the women into two groups: One group, corresponding to the number of men, were put into cross-sex pairs (n = 33); the remaining were put into all-women pairs (n= 17).
Experimental procedure. After subjects were paired and seated, they were instructed to open the envelope with which they had been provided and begin. Each envelope contained an instruction sheet and three sets of slips. It was emphasized that 'This is a study of interpersonal closeness, and your task, which we think will be quite enjoyable, is simply to get close to your partner, with whom you've been matched." The instructions also explained the procedure they should follow in which, for each slip, one of them (in alternating order) reads it aloud, both carry out the activity, and then they go on to the next slip. (The full text of our standard instructions for this procedure is given in the appendix.)
After reading the instructions, they were to begin at once with the first Set I slip. After 15 min, the experimenter told the subjects to stop, put away the Set I slips, and begin Set II; after another 15 min, to begin Set III; and after a final 15 min, to stop, quickly move to another location in the room as far away as possible from their partners, and then complete the postinteraction questionnaire.
Tasks and experimental manipulation. Subjects were given one of two types of tasks. The closeness-condition tasks were based on the procedure developed in our preliminary research (Aron, Aron, Melinat, & Vallone, 1991). These tasks called for self-disclosure or other intimacy-associated behaviors; the intensity of these tasks gradually increased, both within sets and over the three sets. (We used three sets of slips so that even pairs that went very slowly through the tasks would do at least some of the fairly intense Set III tasks.) The small-talk- condition tasks involved minimal disclosure or focus on partner or relationship. The full set of tasks for each condition is given in the appendix.
Dependent measure: closeness. The postinteraction questionnaire included Aron et al.'s (1992) Inclusion of Other in the Self (IOS) Scale and Berscheid et al.'s (1989) Subjective Closeness Index (SCI). The IOS Scale consists of seven pairs of circles labeled Self and Other (in this study, Partner) that overlap to various degrees, creating a 7-point, interval scale. Subjects select the pair that best describes their relationship. The IOS Scale has shown high levels of test-re test and alternate-form reliability (.85 and .92, respectively, for friendships) and convergent and discriminant validity with appropriately related measures; it also predicts relational maintenance over 3 months—all as well as or better than several more elaborate, standard measures of closeness that Aron et al. also tested. The SCI consists of two items in which the subject rates on a 7-point scale his or her degree of closeness to another person (in this study, his or her partner). On one item the relationship is compared with all of one's other relationships; on the other item, the relationship is compared with what the subject knows about the closeness of other people's relationships. We included the SCI because it seems to tap very directly the feeling aspect of closeness (its loading on this latent variable in the Aron et al. confirmatory analysis was .99).
Further, the SCI is a short scale that provided a complement to the IOS Scale.
In the three studies reported in this article, the correlations between the two measures ranged from .69 to .83, with a median of .77. Thus, to simplify reporting of results and to maximize reliability, we combined the two measures into a single composite. In all three studies, standard deviations were very close for the two measures, so that we simply averaged raw scores. Treating this composite as a scale with two subparts (IOS and SCI) yielded a median alpha (over the three studies) of .88. Also, as would be expected from the high correlations, in all analyses in which there was a significant effect for the composite, both IOS and SCI individually showed the same pattern of results. Both of these scales correspond closely to the feeling of closeness as we have described it in the introduction.
Additional measures on postinteraction questionnaire. Both Studies 1 and 2 added two measures for the attachment- style analyses: (a) a version of the IOS Scale completed for "HOW YOU WISH your relationship with your partner had been at the end of the experiment" (to assess discrepancy between obtained and desired IOS Scale closeness) and (b) the same attachment-style scales as on the initial questionnaire (to assess change in reported attachment style).