Another way of estimating the degree of closeness achieved by subjects in these studies is to consider the absolute score on the SCI. On this scale subjects were explicitly instructed to rate their closeness "relative to all your other relationships" (Question 1) and "relative to what you know about other people's relationships" (Question 2). That is, on the SCI subjects were explicitly told to evaluate the closeness to their partners with real-world standards. For each question on the SCI, 4 is the midpoint on the scale from Not at all Close to Extremely Close. In the three studies reported here (excluding the small-talk condition in Study 1), the mean was about 4 for each question. This suggests that subjects rated their relationship to their partners of less than an hour to be about as close as the average relationship in their lives and in other people's lives.6 (Of course, it is still possible that subjects ignored the literal meaning of the instructions and used the midpoint to mean about average in some general sense. Also, "all other relationships" may have been taken to include casual acquaintanceships.)
A third indication that these studies actually produced some degree of felt closeness is that many subjects maintained some relationship with their partners afterwards. In Study 3 we were able to administer brief follow- up questionnaires about 7 weeks after the study (at the final exam). We did not report these data as part of the presentation of Study 3 because subjects had already been debriefed immediately following the studies, so any differences across conditions would be contaminated. The key finding was that of the 58 pairs represented in the follow-up questionnaires, 57% had had at least one subsequent conversation, 35% had done something together, and 37% had subsequently sat together in class.7 (It should be emphasized that the goal of the procedure as used in these studies was not to produce closeness beyond the context of the subjects' feelings immediately at the end of the interaction. That there was any carryover at all beyond the study, including several weeks later, is significant in indicating the power of the procedure. On the other hand, without a control group, it remains possible that there could have been this much closeness without the procedure.)
A fourth indication that the closeness produced in these studies is comparable in important ways to closeness in naturally occurring relationships is the parallel outcomes, when similar issues were tested, between our results and those obtained in previous research using naturally occurring pairs. For example, the difference in closeness between all-secure pairs and all-avoidant pairs in Studies 1 and 2 was just what would be expected from studies of secure versus avoidant individuals in naturally occurring relationships.
So are we producing real closeness? Yes and no. We think that the closeness produced in these studies is experienced as similar in many important ways to felt closeness in naturally occurring relationships that develop over time. On the other hand, it seems unlikely that the procedure produces loyalty, dependence, commitment, or other relationship aspects that might take longer to develop. Certainly, there is minimal shared history and minimal behavioral closeness in the ways measured by Berscheid et al. (1989). Thus the procedure is like other experimental paradigms such as mood- induction procedures, the minimal group paradigm, or methods for temporarily lowering self-esteem: It is useful as a means of creating a similar although not completely identical state, but under controlled conditions permitting experimental tests of causal hypotheses and theoretical issues. For these purposes the absolute level of the impact on the individual is less important than the relative level across experimental conditions.
Implications for Features of the Procedure and Its Application in Research
The procedure itself, in addition to putting pairs together to interact for 45 min, was initially developed to include four key elements: (a) gradually escalating reciprocal self-disclosure and intimacy-related behaviors, (b) matching by nondisagreement on important attitude issues, (c) expectations of mutual liking, and (d) making closeness an explicit task. In this series of studies, we systematically examined each of these elements and found that only the first, the nature of the tasks themselves (self-disclosure, etc., vs. small talk), made a significant overall difference. Also, Study 3 indicated that making closeness an explicit goal may be of importance for some subjects, such as introverts. Just how other aspects of these procedures have an impact on closeness (and how these may interact with personality or other variables) is a ripe subject for further research.
More generally, we have tried to demonstrate the practicality and flexibility of these procedures for examining hypotheses in the close-relationships and related research areas. In particular, the classroom version of the procedure we have used in these studies is relatively easy for most researchers to implement. The most time-consuming aspect of the original process was matching subjects on nondisagreement on critical attitudes, but the results of Study 2 suggest that neither the matching nor subjects' belief that they have been matched makes much difference in the closeness obtained. Indeed, if a planned study does not involve subject variables (i.e., it is manipulating only instructional, task, or situational variables), then a pretest/initial-questionnaire procedure can be eliminated entirely.
To demonstrate the practicality of this simplified approach, we conducted an additional study of222 subjects attending a large class at a different university. We gave no pretests at all but simply announced the planned study on the preceding class day (to allow students not to participate if they so chose—in fact, attendance on the day of the task was greater than usual). On the class day, we separated the students into two rooms (one for women, one for men) and randomly paired them on the spot, reassigning members of any pair who already knew each other. We then gave them envelopes containing the closeness tasks and the closeness-as-a-task instructions and proceeded in the usual way. The result was a mean closeness score of 4.02, a figure quite comparable to those in the previous studies that included pretests and matching.
One other practical boon of this particular research paradigm is that participants report enjoying it a great deal. This makes it easy to obtain access from instructors to carry out the procedure during a class session (which provides excellent opportunities for discussing research methods issues as well as relationship and personality material) and to obtain follow-up data or repeated participation. As a check on subjects' enjoyment of the procedure, in the study just described, which lacked the pretest/matching procedures, we included in the postin- teraction questionnaire an item about how much the subjects enjoyed their participation. The mean rating was 5.78, well above the midpoint on the scale, which ranged from 1 (Not Very Much) to 7 (Very Much). This item was highly correlated (r = .52) with the closeness composite. (In this course and in the other classes in which we used this procedure, instructors told us that it was frequently mentioned on student evaluations at the end of the term as a highlight of the course.)