228 HODGES.MEAGHER.NORTON.McBAIN.AND SROUBEK stan osion o be nof be also believ e that the exper is expec ting ther sneakins-o-difernt ef hen this set of attributic self-presentation effect(e.g.drawing attention to oneself as vere the Regarding the ossibility that n agreed more than they disagreed furthe ore,it is no might have be ngry abo g been placed in Agreers believ the e wha wers.there vas no diffe the nge index bet giving disagreeing answers at Position 6 was negligible (r07 One more factor that might have inf ced decision were male and real participan was fem L The ter kno unlikely,given that 27 of 46 g mixed on es emal ght the have beer gain ing the 40 ns (Ms =4.69 and 4 vely)but n significant dif mixe and Disagreers,(44)=0.66.p=51.and (44)=1.67.p 10 and dis Packer (2008)and H and hi y thought the ted them to do.Agn loval dissent as well as a when individual me t rating wasrelatively utral (M others in the group are not vi the otne ed the st argue that Agreers st guess I coul ven if it was inc 44) lisappointed in the group but they were these two item 1.44) 857,p-005. More specific y.for others ile ther as significant differencefor Disa 200 dence in support of the ount of SFI effects offered by 06.P 0 pragma d By predi t of the time.that m better position ntpositions and from measures failed to provid 5 and 6 do nder ere relatively weak pared to othe Both So at Position 6 and Experiment 3 022 he focus of Ex given that there effect.To do this we mpared prime to a no-prim nt with we others. incentive to agree with the correct answers for both truth-primeand are refusing to imitate to avoid being viewed as a copycat. What seems crucial to the SFI effect is not merely being in a different position or being the final person to answer on each trial but one’s epistemic and pragmatic position relative to others. It is an SFI effect, not a speaking-last effect, a speaking-from-a￾different-position effect, a speaking-to-differentiate effect, or a self-presentation effect (e.g., drawing attention to oneself as unique or creative). Anger or reactance. Regarding the possibility that partici￾pants might have been angry about having been placed in a position of ignorance, leading them to give disagreeing, incorrect answers, there was no difference on the Anger index between Agreers (M 4.56) and Disagreers (M 4.62), t(44) 0.17, p .86. Furthermore, the correlation between the Anger index and giving disagreeing answers at Position 6 was negligible (r .07, p .66). Experimenter and participant expectancies. Earlier, we indi￾cated that the SFI effect did not depend on experimenters’ knowl￾edge about values-pragmatics theory or specific hypotheses being tested. There were no effects of experimenter knowledge on any of the motivational indexes or other indexes (e.g., Suspicion, Trust) either (ps ranged from .24 to .99). When participants were asked whether they thought the experiment might “really” have been about the “courage not to conform” or having the “humility to depend on others,” there was very modest agreement with both items (Ms 4.69 and 4.40, respectively) but no significant dif￾ference between items, t(45) 1.10, p .28, nor between Agreers and Disagreers, t(44) 0.66, p .51, and t(44) 1.67, p .10, respectively. Finally, two items asked about participants’ beliefs about what they thought the experimenter expected them to do. Agreers agreed more often than Disagreers that “the experimenter wanted me to agree with other people,” t(44) 2.65, p .011, although even the Agreers’ mean rating was relatively neutral (MAgreers 4.20, MDisagreers 2.48). On the other hand, Disagreers more than Agreers endorsed the statement “the experimenter wanted me to make the best guess I could, even if it was incorrect,” t(44) 2.21, p .032, MAgreers 5.08, MDisagreers 6.24. The interaction between these two items and Agreers/Disagreers was significant, F(1, 44) 8.57, p .005. More specifically, for Agreers, there was no difference between these items, t(24) 1.20, p .24, while there was a significant difference for Disagreers, t(20) 6.06, p  .001. Overall, participants tended to think the experi￾menter wanted them to make their best guess, and for most of them most of the time, that meant agreeing with better positioned others. Independence. Differing amounts of disagreement at Posi￾tions 5 and 6 do not rule out the possibility that people who disagree are motivated to act independently of others. However, results from the motivational indexes suggest that any motives for independence were relatively weak compared to other motives. Both Social Solidarity and Pragmatics indexes are as strong as or stronger than the Independence index, even for Disagreers. There was a positive correlation between disagreements at Position 6 and the Independence index (r .34, p .022), so concerns for independence may play a role for some participants. However, given that there was not a significant difference in the Indepen￾dence index for Agreers and Disagreers, it is difficult to rationalize how independence would lead so frequently to agreement with others. Attribution. It might be suggested that participants who start at Position 6 become focused on the strangeness of being asked by the experimenter to identify words they cannot see. If participants also believe that the experimenter is expecting them to guess at answers without benefit of the information provided by A and B, then this set of attributions might yield disagreements. However, if this were the case, it is not easy to understand why so many participants at Position 6 always agreed and why one third of Disagreers agreed more than they disagreed. Furthermore, it is not clear why Agreers believed that the experimenter wanted them to “make the best guess about what I saw, even if it was incorrect” (M 5.08) as strongly (p .24) as they believed that the experimenter wanted them to agree with others who were in a better position (M 4.20). Gender. One more factor that might have influenced decisions to agree or disagree is the gender composition of the participants. If gender composition was incongruent (i.e., Participants A and B were male and the real participant was female, or vice versa), would this lead to an increase in disagreeing answers? This is unlikely, given that 27 of 46 groups were mixed (one male and one female confederate), but to address this concern, we compared gender-congruent, gender-incongruent, and mixed-gender groups, again using zero-inflated negative binomial regression. Neither gender-congruent groups (b .11, p .83) nor gender￾incongruent groups (b .40, p .53) differed significantly from mixed-gender groups in terms of rates of disagreements. Dissent. A number of authors have discussed the possibility of anticompliance and dissent. Packer (2008) and Hornsey and his colleagues (Hornsey, Majkut, Terrie, & McKimmie, 2003; Horn￾sey, Smith, & Begg, 2007) proposed that individuals might engage in loyal dissent, as well as uneasy conformity, when individuals identify with a group but are disappointed with the group’s actions. Neither of these interesting proposals seems to account for the SFI effect because others in the group are not viewed as wrong. One could argue that Agreers often acted as if uneasy conformity characterized their actions, but it was not because they were disappointed in the group but because they were frustrated with their own ignorance. In short, nothing in our experimental results suggests that the SFI effect is due to dissent or a refusal to comply with others. Summary. The results of Experiment 2 provided considerable evidence in support of the account of SFI effects offered by values-pragmatics theory. Virtually all the predictions made by the theory were supported. By contrast, the available evidence from different positions and from self-report measures failed to provide clear support for any of the several alternative hypotheses consid￾ered. This does not mean that these alternatives have no explana￾tory power, at least for some people some of the time, but none of them appears to capture the broad ecological dynamics of the SFI situation in the way the values-pragmatics account does. Experiment 3 The focus of Experiment 3 was to test whether manipulating participants’ concern for speaking truthfully would affect the SFI effect. To do this we compared a truth-prime to a no-prime condition, predicting a larger SFI effect for the former. In order to create a demanding test of this prediction, we added a monetary incentive to agree with the correct answers for both truth-prime This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly. 228 HODGES, MEAGHER, NORTON, MCBAIN, AND SROUBEK