Participants

Four hundred and forty-four participants were obtained from the subject pool of students enrolled in introductory psychology courses, other class exercises, and on a voluntary basis at Loyola University Chicago, an ethnically diverse urban Midwestern university. Cash lotteries were conducted for volunteers not required to complete subject pool hours. Two cases were dropped from the path analyses because of missing data (N = 442), although these cases were included in other analyses. Participation folders bore the label �Do not sign up if you are an only child,� to reduce the possibility that participants could not properly respond to the target relationship manipulation and other possible confounds. The mean age was 21 years (MD = 20, SD = 4.77), 304 participants were female, 138 were male. Forty-six percent of participants were of Western European descent, others included Eastern European (16%), Hispanic (9%), East-Asian (9%), African (7%), Indian/Pakistani (7%), Middle-Eastern (3%) and multiethnic (3%) descent. Additional demographic data added during the survey administration was available for the final 235 participants. For these participants, the mean number of years of education completed by fathers was 14.73 (SD = 3.29), the mean number of years of education completed by mothers was 14.28 (SD = 3.03).

Materials

The instrument used in this study is found in the appendix, with items grouped by construct to simplify interpretation. Items were randomly arranged within each section of the questionnaire form via a random number generator. This arrangement was held constant throughout the study. Participants were asked to think of a target character (a sibling close in age or a close friend they have known for several years, depending on the form). Participants were asked a number of questions about this target, such as their estimated similarity to the target character, to activate their cognitive representations of that individual. Participants then read a scenario in which a target character was described as being in a possibly life-threatening situation. After reading the scenario, participants were asked to respond to items on perspective taking, self-other overlap, empathic concern, aversive arousal, sadness, involvement, constructs related to reciprocal altruism, and the likelihood that they would attempt to help rescue the target character. Participants read a second scenario and completed another set of items, which were not used in this analysis.

Measures of self-other overlap. Several validated scales measuring constructs of interest were found in the literature. To measure relationship closeness for example, Aron et al. (1992) developed the single item, pictographic Inclusion of Other in Self (IOS) scale, which exhibited convergent validity with longer relationship closeness measures, discriminant validity with measures of other constructs, and test-retest reliability. Although the usefulness of this scale has been demonstrated, there were at least two concerns with the IOS scale. One concern was that the size of the circles is not constant. Aron et al. (1992, p. 597) stated; �the total area of each figure is constant (thus as the overlap of the circles increases, so does the diameter)." This introduces a second dimension to the scale, in addition to the degree of overlap. In the modified IOS scale, the size of the circles remained constant across the range of overlap/separation. The other concern was that although the circles range from near complete overlap to adjacent circles with no overlap, there was no option where the circles become separate from each other. This may have restricted the range of the instrument. The modified IOS scale included the separation of self and other beyond an adjacent placement. Equal intervals of overlap and separation were also used, reaching one diameter of separation between the circles. Thus, the maximum degree of separation was equivalent to the maximum degree of overlap.

Another task asked participants to indicate the closeness of their relationship to the target character by placing a representation of themselves with a representation of the other, in the form of circles traced with a washer provided by the experimenter. A rectangular box containing the centered outline of a washer (approximately the size of a U.S. quarter) labeled �Other� was created for the instrument. Participants were asked to trace the outline of the washer labeled �Self� within the boundaries of the rectangle in order to represent the relationship between themselves and the target character. The rectangle was designed to allow a range from complete overlap of the images (one washer unit of overlap) to a separation of images as large as their diameters (one washer unit of separation). The distance between the most distant points on the outlines representing self and other was measured in millimeters.

To assess oneness, Cialdini et al. (1997) created a 7-point scale which participants used to indicate the extent to which they would use the term �we� to describe their relationship with a target character. Cialdini et al. (1997) combined this measure with Aron et al.�s (1992) IOS scale to create an oneness index. In Cialdini et al.�s (1997) first experiment, this oneness index was successfully affected by relationship closeness, and also explained 57.8% of the variance in a helping measure (empathic concern accounted for 20.3% of the variance). The unique variance explained by empathic concern after partialing out the variance in helping accounted for by self-other overlap was nonsignificant. The oneness index was used in further studies that replicated these results. The modified IOS scale, the convergence representation task, and evaluation of the term �we� as a relationship descriptor, were used to indicate the construct of �oneness,� the perceived overlap between self and other, in the present study. Distances on the convergence representation task were converted to a range from one to seven, scores from the three items were averaged to produce an index score. Note that Batson et al. (1997) used similarity of attribute ratings and a one-item similarity judgment along with the IOS scale to measure self-other overlap. The three measures did not correlate very highly, r ranged from -.03 to .20. The current study is believed to provide a more reliable assessment of self-other-overlap.

Measures of sadness. Other measures of egoistic influence included sadness (feeling low, low-spirited, heavy hearted) and personal distress/aversive arousal (disturbed, troubled, uneasy), as measured by items suggested by Fultz et al. (1988). In a factor analysis of responses to three scenarios where female participants were presented with a woman in a state of need, �feeling low� loaded between .78 and .80, �low-spirited� loaded between .74 and .86, and �heavy hearted� loaded between .67 and .79 on the factor interpreted as sadness. None of these items loaded above .38 on factors interpreted as empathy and distress. Cialdini et al. (1997) used �sad� instead of �feeling low.� In Fultz et al.�s analysis, �sad� had factor loadings between .47 and .63 for the three scenarios, and had a factor loading of .56 on empathy in one of the scenarios.

Measures of aversive arousal. In the same factor analysis, �disturbed� loaded between .78 and .87, �troubled� loaded between .78 and .80, and �uneasy� loaded between .69 and .84 on the factor interpreted as distress. None of these items loaded above .39 on factors interpreted as empathy and sadness. In their second study, Fultz et al. (1988) reported factor loadings of .78 (disurbed), .74 (troubled), and .78 (uneasy) for the factor interpreted as distress. Although �troubled� had a factor loading of .52 on sadness, none of the other items loaded above .36 on the factors of empathy and sadness. Fultz et al. (1988) reported factor loadings of .72 (feeling low), .85 (low-spirited), and .75 (heavy hearted) on the factor interpreted as sadness. Although �feeling low� had a factor loading of .48 on distress, none of the other items loaded above .41 on the factors of empathy and distress. These measures have been successfully used in subsequent studies, although Cialdini et al. (1997) used �alarmed� and �worried� instead of �disturbed� and �troubled.� Factor loadings for �alarmed� on the construct interpreted as distress ranging from .34 to .74, and had a loading of .52 on empathy for one of the scenarios. Factor loadings for �worried� on the construct interpreted as distress ranging from .42 to .73. However, factor loadings for �worried� hovered between .35 and .37 on sadness, and �worried� loaded .39 on empathy in one of the scenarios. This experiment used the items �disturbed,� �troubled,� and �uneasy.�

Measures of empathic concern. To measure empathic concern, Batson et al. (1995) developed a four-item index, where participants rated themselves on the terms sympathetic, compassionate, softhearted and tender, after completing a helping measure. In the previously cited three-scenario experiment by Fultz et al. (1988), �sympathetic� loaded between .50 and .79, �compassionate� loaded between .66 and .86, �softhearted� loaded between .62 and .78, and �tender� loaded between .63 and .66 on the factor interpreted as empathy. Factor complexity was suggested for some items. In one scenario, �sympathetic� had nearly the same loading on empathy (.50) and distress (.51). �Softhearted� had loadings of .48 on sadness for two scenarios. �Tender� had a loading of .45 on sadness in one scenario.

Fultz et al. (1988) included other items not used by Batson et al. (1995, 1997). �Moved� had factor loadings between .69 and .79 on empathy, and did not have a loading above .27 on any other factor. �Touched� had factor loadings between .73 and .85 on empathy, and did not have a loading above .33 on any other factor. �Warm� had factor loadings between .64 and .77 on empathy, and did not have a loading above .38 on any other factor. Items for �moved,� �warm,� and �tender� were not included in the second experiment. Fultz et al. (1988) reported factor loadings of .91 (softhearted), .84 (compassionate), .75 (touched), and .71 (sympathetic) for empathy. �Sympathetic� and �touched� both had loadings of .49 on sadness in the second experiment. Due to these findings, the items used to indicate empathic concern differed slightly from Batson et al.�s (1995, 1997) analyses. An item for �moved� replaced the item for �sympathetic.�

Measures of cognitions related to reciprocal altruism. In addition to the inclusion of scales from the literature, this study generated items measuring the likelihood of a helping behavior and constructs related to reciprocal altruism. Two sets of items were created to measure the influence of reciprocal altruism. One set asked participants to estimate the likelihood that the target would help the participant if positions in the scenario were reversed. The other set asked participants to rate the degree of obligation or indebtedness targets are expected to have to the donor after being helped. Three items were used to construct an index score for each of these measured variables. These constructs were seen as related but distinct cognitive components characteristic of reciprocal altruism, and were expected to differ according to the genetic relationship to the target in need.

Measures of the likelihood of helping. Helping intention was assessed with three items targeting the likelihood of the respondents making a risky attempt to rescue the target character. These included an estimate of how likely participants would be to try and save the target character, the extent to which participants would risk their own lives in a rescue attempt, and an estimate of the probability (from 0-100%) that participants would take action if they were actually in the situation. These items were arranged in randomized order along with the empathic concern, perspective taking, sadness and aversive arousal items.

Other measures: Involvement and social desirability. Three involvement items provided an assessment of the effectiveness of the scenario in kindling the imagination of the respondents. Involvement was hypothesized to be a measure of general cognitive arousal and the values on these items were expected to covary with the levels other psychological variables, empathy, oneness, sadness, aversive arousal, and the reciprocal altruism indicators. The Marlowe-Crowne, Form C Social Desirability measure (Reynolds, 1982; Silverstein, 1983) was included to determine if socially desirable responding confounded the experimental results. Respondents might inflate estimates for the likelihood they would make a risky rescue attempt in order to project a positive image of themselves.