8 GENERAL DISCUSSION
Three experiments using different populations and domains investigated the roles of perspective taking and forecast type in impacting WT effects. Overall, our results revealed that instructions encouraging participants to consider the perspective of someone with an opposing viewpoint and the use of deliberative forecast measures were both associated with lower WT effects. Moreover, we found that this may have been due, in part, to changes in how people evaluated evidence about the two teams. Below, we describe the results in more detail and provide our interpretations.
8.1 Results overview and explanations
8.1.1 Impact of perspective taking on WT effects
The evidence across our experiments suggests WT effects can be reduced (or even eliminated) through the use of perspective taking. For example, when Trump (Clinton) supporters were asked to think about how Clinton (Trump) supporters might view the election outcome immediately before providing their own forecasts, this reduced WT effects about the election (Experiments 1 and 2). Likewise, when sports fans provided forecasts about an upcoming game between a favorite and rival team (Experiment 3), participants demonstrated reduced WT effects when thinking about how a rival fan might see the game going (at least when intuitive‐based measures came first; see below for discussion). Moreover, in Experiment 3, we provided evidence that considering the perspective of a rival fan shifted participants' weighting of evidence when making forecasts. Indeed, forecasts of those who took the perspective of a rival fan tended to be as related to their own preferences as they were to perceptions of the relative strengths/weaknesses of the teams (which is critical for accurate judgments). Contrariwise, forecasts of participants in the control condition tended to be more related to their preferences than to their perceptions of the relative strengths/weaknesses of the teams.
Overall, this evidence is consistent with prior research showing that focalism in referent‐dependent judgments (such as those used in the present study) can be debiased through changes in contextual features (e.g., salience in a questionnaire and information sharing; see Bar‐Hillel et al., 2008; Koriat et al., 1980; Pahl, 2012; Rose & Windschitl, 2008; Rose et al., 2012; Windschitl et al., 2008). Of course, there may be other potential mechanisms to explain changes in WT effects as a function of perspective‐taking instructions, such as shifts in anchoring and adjustment processes (Epley, Keysar, Van Boven, & Gilovich, 2004; Mussweiler, Ruter, & Epstude, 2004), changes in memory sampling procedures used when forming judgments (Hollander, 2004; Mutz & Martin, 2001), or simulation of worst case scenarios (Sharot, Korn, & Dolan, 2011). Follow‐up studies could be designed to test such alternative mechanisms.
Our research also revealed that taking the perspective of a person with a similar preference did not increase or decrease WT effects compared with the control condition. Although exposure to similar others can alter psychological and behavioral experiences under some conditions (Lewis & Neighbors, 2004; Mussweiler, 2003a, 2003b; Neighbors et al., 2010), thinking about a similar other does not appear to alter how participants go about forecasting future outcomes. This may be due to overlap in how participants viewed the self vis‐à‐vis a peer with the same political preferences (Mussweiler, 2003a, 2003b).
8.1.2 Impact of forecast type on WT effects
Across experiments, we found that participants demonstrated less WT effects when prompted to think deliberatively (vs. intuitively) when forecasting future outcomes. That is, when participants were asked to think about evidence statistically, judge with their heads, or provide precise probability judgments, forecasts were relatively less aligned with their preferences than when participants were asked to go with their gut impression, judge with their hearts, or make an outcome prediction (see also Windschitl et al., 2010). One explanation for this result is that, when prompted to think deliberatively, participants considered evidence in a more balanced manner such that they weighted both focal (about the target/preferred team) and nonfocal evidence (about the referent/nonpreferred team). Indeed, Experiment 3 showed that deliberative forecasts were less correlated with preferences and more correlated with perceptions of the relative strengths of the teams. Intuitive‐based judgments showed the opposite pattern. Aside from this core explanation, other potential reasons that deliberative measures may be less impacted by desirability could be that such measures disrupt the link between positive affect and forecasting desirable outcomes (Lench, 2009; Lench & Bench, 2012) and are more limited by reality constraints (Windschitl et al., 2010). Follow‐up studies could be conducted to test such alternative mechanisms.
As a final note, it is important to mention that instructions to make forecasts in a deliberative fashion did not completely eliminate WT effects. Rather, the impact of such measures resulted in a relative shift in biases, such that the effect dropped in magnitude but not direction (and WT effects were still present). The limiting impact for deliberative measures to completely debias WT effects could have to do with the fact that participants were considering nonstochastic outcomes (e.g., sporting events and political elections) that involved nonrandom processes, hence keeping open the possibility that anything could happen (Windshcitl et al., 2010; for review, see Krizan & Windschitl, 2007).
8.2 Limitations and future directions
There are several limitations to this research. First, rather than manipulating preferences, we assessed natural preferences for candidates and sports teams. This approach was used to capture ecologically valid beliefs about important real‐world outcomes. However, we are unable to establish that preferences were responsible for guiding expectations (vs. the reverse causal path or third variables). It is notable, though, that prior research examining the preference–expectation link experimentally (Windschitl et al., 2013; for review see Krizan & Windschitl, 2007, 2009) and longitudinally (Krizan & Sweeny, 2013) has already established that preferences guide expectations rather than the other way around or that third variables (e.g., knowledge) account for the effect. Moreover, regardless of whether preferences were manipulated, we did manipulate other variables in the design and elucidated their impact as a function of preferences. Nevertheless, future research examining the impact of perspective taking and forecast measures in WT effects following manipulated preferences would be informative.
Second, our results may not apply to all contexts in which WT occurs. Indeed, we used two specific real‐world events: the U.S. Presidential election and an American football game involving two specific college teams. Although our use of two different contexts helps with generalizability, it still could be the case that there is something idiosyncratic about these events. Further, these specific examples of the 2016 U.S. President election and a football game in 1 year between two universities may not be fully representative of all elections or sporting event outcomes. Additionally, and more broadly, taking the perspective of someone with an opposing preference works well in political and sports contexts because there are clear, opposing viewpoints. However, not all contexts are like this. For example, when thinking about the likelihood of a self‐relevant outcome (e.g., getting cancer and owning a home), perspective taking would not make sense given that there are no other people hoping for the opposite, nonpreferred outcome (at least we hope not!). However, it could be argued that a conceptually similar idea would involve a “consider‐the‐opposite” strategy wherein a person thinks about evidence for alternatives to their preferred outcome (Hirt & Markman, 1995; Koriat et al., 1980; Mussweiler, Strack, & Pfeiffer, 2000). Future research should examine other types of events and contexts to see whether our results replicate.
Third, it is notable that the timing in which forecasts were made varied across Experiments 1 and 3, such that the forecasts made for elections tended to be somewhat closer in time to the target event than forecasts made about the sporting events. Prior research suggests that the immediacy of an event can sometimes dampen optimism about the event (Shepperd, Ouellette, & Fernandez, 1996; for review see Sweeny & Krizan, 2013). As we did not systematically vary this feature across our experiments, it is unclear whether forecast timing would interact with our perspective taking and forecast type factors. This would make for interesting follow‐up research.
Fourth, although most of the main results were robust and consistent, in Experiment 3 we found that perspective taking only decreased WT effects when intuitive‐based measures came first in the sets of judgments (Experiments 1 and 2 always assessed measures in this order). Whereas this result was unexpected and would need replication, our research and that of others have already established that intuitive‐based and deliberative measures are not always correlated, trigger unique judgment processes, and are differentially related to certain outcomes (Portnoy et al., 2014; Sloman, 1996; Suls et al., 2013; Windschitl et al., 2010; Windschitl & Wells, 1996). One possibility for the differences across order conditions could be that the perspective‐taking instructions had a different meaning for participants depending upon whether they received an intuitive‐based versus deliberative forecast first. For instance, when an intuitive‐based forecast appeared first, participants might have more easily recognized the value of the perspective‐taking advice and applied this when making forecasts. That is, because intuitive‐based assessments reveal stronger default evidence of WT effects, participants who provided such forecasts first may have had the heightened experience (relative to when deliberative measures came first) of wanting to go with their preference, which then simultaneously imbued the perspective‐taking instructions with inherent validity. On the other hand, when a deliberative forecast appeared first, participants might have balked at the need for such instructions given that they were already (at a default) attempting to provide more balanced forecasts. Then, upon completing an intuitive‐based measure, participants may have felt the need to inflate (in a preference‐consistent direction) their responses in order contrast to the already less biased deliberative forecasts. Additionally, speculation about this effect is further complicated by the fact that the specific items used for intuitive‐based and deliberative forecasts were always presented in the same order within sets rather than completely randomized. Future research would need to follow‐up on these findings and explanations.
8.3 Implications and conclusions
The current research has both theoretical and applied implications. In relation to theoretical implications, our findings have relevance to at least two distinct research literatures: (a) WT, optimism, and uncertainty and (b) perspective taking. First, our research is broadly relevant to the literature on optimism and uncertainty where people develop expectations for preferred and nonpreferred outcomes in a variety of contexts (e.g., health, academics, and sports). Indeed, our research provides further evidence that people form expectations that are consistent with their preferences (for reviews, see Krizan & Windschitl, 2007, 2009). Moreover, whereas past research has yielded inconsistent or negligible debiasing effects, our experiments suggest that perspective taking and making deliberative forecasts reduces optimism‐based biases. Additionally, our research provides some support for at least one nonmotivational mechanism underlying WT (see Krizan & Windschitl, 2007, 2009). Specifically, we showed that undue focus on focal entity information, and insufficient attention to referent entity information (which inhibits a balanced evaluation of the relative strengths/weaknesses of the entities), can inflate WT effects and factors that undercut this can reduce them (Windschitl et al., 2013). Finally, our work can be positioned alongside a growing literature highlighting that not all types of uncertainty forecasts should be treated equally (Portnoy et al., 2014; Sloman, 1996; Suls et al., 2013; Windschitl et al., 2010; Windschitl & Wells, 1996).
Second, our research also has relevance to the perspective‐taking literature. Past research demonstrates that taking the perspective of others' viewpoints has a range of positive benefits (for review, see Hodges, Clark, & Myers, 2011), including greater empathy (Batson, 1987), reduced stereotypes and discrimination (Galinsky & Moskowitz, 2000; Wang et al., 2014), and more rational judgments (Yaniv & Choshen‐Hillel, 2012). Our research suggests that an additional benefit is reduced WT when forecasting preferred future outcomes that involve others. Interestingly, our research also suggests that taking the perspective of someone who has the same opinion as us does not appear to change WT effects for better or worse (Experiment 2).
From an applied perspective, researchers and practitioners are often interested in examining people's beliefs about future outcomes. As expectations can impact behavior, researchers may wish to design easy, portable, and effective interventions to modify overly optimistic forecasts to be more realistic. Our research offers two potential approaches that might be fruitful in allowing people to form judgments in ways that rely on more than just their preferences when formulating future forecasts. Ideally, these approaches will be useful in reducing WT, biased information processing, and potential downstream maladaptive consequences in such diverse contexts as health, academics, business, sports, and politics.
Posts
No comments:
Post a Comment