Downstream Effects of Dispositional Inferences on Confirmation Biases

Abstract

Although research has given substantial attention to understanding the antecedents of dispositional inferences, less attention has been directed at the consequences of these inferences, such that evidence linking dispositional inferences to downstream effects is relatively scarce. The present investigation examined whether dispositional inferences formed during initial observations elicited confirmatory processing of subsequent information about observed targets. Because confirmation biases influence a variety of information processing strategies, four experiments examined the extent to which dispositional inferences guided memory of new information (Experiment 1), interpretation of ambiguous information (Experiment 2), and information-seeking behavior (Experiments 3 and 4). Results indicated that biased processing of subsequent information was more likely when dispositional inferences were encouraged (i.e., impression formation objective) versus discouraged (i.e., narrative construction objective). This investigation highlights the role of causal inferences on confirmation biases and reveals the ease with which biases can be both bolstered and attenuated.

Keywords

confirmation bias interpersonal expectancies dispositional inferences narrative

Research on social perception processes has converged on the conclusion that inferring dispositions from observed behavior is a particularly prevalent and efficient method for organizing and interpreting information about others in the environment (Carlston & Skowronski, 1994; Todorov & Uleman, 2003). Although research has given substantial attention to understanding the antecedents of dispositional inferences (e.g., Kelley & Michaela, 1980), considerably less attention has been directed at the consequences of these inferences, such that the evidence linking dispositional inferences to their downstream effects is relatively scarce. In this research, we considered whether dispositional inferences formed during an initial observation of a target bias how perceivers process subsequent information about the target. The findings provide insight into the way in which causal inferences can set perceivers on different information processing trajectories that affect how they process social information across successive interactions with a target.

Confirmation Bias

Confirmation bias is the tendency for perceivers to seek out, remember, and interpret information in ways that conform to preexisting beliefs or expectations (Nickerson, 1988). Researchers in social and cognitive psychology have long acknowledged that the perception process is not veridical; rather, perceivers view the world through the lens of their backgrounds, experiences, and expectations (e.g., Bruner, 1957). This process allows perceivers to quickly identify and respond to the most relevant and salient aspects of a given situation. However, the rapid processing of persons and social situations via confirmation biases carries consequences. Confirmation bias has been implicated in wrongful convictions (Findley & Scott, 2006), academic underachievement (Rosenthal & Jacobson, 1968), perceptions of mental illness (Pennebaker & Skelton, 1978), and the misinterpretation of scientific evidence (see Loehle, 1987, for a review). In each case, perceivers’ expectancies interfered with accurate, unbiased processing of information.

Although the role of dispositional inferences in the confirmation bias process has not previously been explicitly articulated or systematically examined, two lines of research point to its potential influence. First, interpersonal expectancies are tightly bound with conceptions of a target’s general disposition and characteristics (e.g., F. Heider, 1958; Nisbett, 1987). Research suggests that both explicit dispositional inferences (Newman, 1996) and implicit dispositional inferences (McCarthy & Skowronski, 2011) function to provide an expectancy of a target’s future behavior. Second, several variables that moderate the confirmation bias process—that is, memory, cognitive resources, and personal values—can disrupt perceivers’ tendency to draw dispositional inferences (Allen, Sherman, Conrey, & Stroessner, 2009; Neuberg, 1989; Wyer, 2004). Thus, there is reason to believe that dispositional inferences may contribute to confirmation biases by influencing how perceivers process subsequent information about a target.

We posit that circumstances that encourage a perceiver to draw dispositional inferences about a target increase a perceiver’s tendency to process subsequent information in accordance with a disposition-based expectancy, thus resulting in confirmation biases. By contrast, circumstances that discourage perceivers from drawing dispositional inferences about a target reduce a perceiver’s tendency to process subsequent information in accordance with a disposition-based expectancy, thus attenuating confirmation biases. In proposing these effects, we do not distinguish between deliberate versus unintentional inferences, nor do we distinguish between a perceiver’s ability versus motivation to individuate a target. In all cases, we predict that dispositional inferences have downstream effects that can bias how perceivers process subsequent information. Accordingly, we propose that any method of reducing dispositional inferences would similarly reduce interpersonal expectancies and, therefore, disrupt the confirmation bias process. Moreover, given that confirmation biases have been found to influence a variety of information processing strategies (Nickerson, 1988), we further predict that the effects of dispositional inferences on subsequent information processing should be apparent across multiple outcomes associated with the confirmation bias.

Overview of Experiments

Four experiments tested the hypothesis that dispositional inferences encourage perceivers to engage in confirmation biases when making social judgments. These experiments manipulated participants’ tendency to form dispositional inferences via one of two processing objectives: impression formation or narrative construction. We chose these processing objectives, in particular, for two reasons. First, it is well established that perceivers naturally use both impression formation and narrative construction processing objectives to make sense of social information (Costabile, 2011) and that such objectives result in similar levels of cognitive attention, elaboration, causal thinking, and recall of events (Costabile, 2016; Costabile & Klein, 2008). Second, however, unlike impression formation objectives that encourage perceivers to draw dispositional inferences (Hamilton, Katz, & Leirer, 1980), narrative construction objectives encourage perceivers to draw situational inferences on the basis of context, thus reducing dispositional inferences (Costabile, 2016). To reduce concerns about the potential influence of alternative processes such as perceptual accuracy, prior experiences, and target responses, we manipulated perceivers’ information processing objectives at initial encoding of social information. The outcome variables that we examined across the four experiments were participants’ recall of behavioral information (Experiment 1), interpretation of ambiguous information (Experiment 2), and information-seeking strategies (Experiments 3 and 4).

We determined the sample size for all experiments on the basis of a power analysis that used the values reported in previous research that employed an experimental design and methodology that are similar to those used in the present studies (Experiment 4; Costabile, 2011; $η_{p}^{2}$ = .09, correlations of within-subject measures of r = .40, power = .95). Results indicated a minimum sample size of 42 participants per experiment (Faul, Erdfelder, Lang, & Buchner, 2007). Our sample sizes exceeded these minimums.

Experiment 1

Experiment 1 examined the downstream effects of dispositional inferences on the recall of behavioral information. During an initial exposure phase, participants viewed a series of target photographs each of which was paired with a trait-implying behavior ostensibly performed by the target in the photograph. After a brief distraction task, participants learned new behaviors performed by some of the targets shown during initial exposure. In this subsequent exposure phase, some behaviors were consistent with the trait implied by the behavior at initial exposure; other behaviors were unrelated to the trait implied by the behavior at initial exposure. To assess the presence of confirmation biases, we compared the rate of recall between the consistent and unrelated behaviors immediately after the subsequent exposure phase (Rothbart, Evans, & Fulero, 1979).

Method

Participants

Introductory psychology students (N = 101) participated in this experiment in exchange for partial course credit. Sixty-three percent of participants were women and 82% of participants identified as European American.

Design

This study used a 2 (processing objective: impression formation vs. narrative construction) × 2 (behavior recalled: consistent vs. unrelated) mixed-model experimental design with repeated measures on the second factor. Participants were randomly assigned to a processing objective that encouraged (impression formation) or discouraged (narrative construction) the formation of dispositional inferences at initial exposure. Behavior recalled examined percent recall of behaviors presented during the subsequent exposure phase that were consistent or unrelated to the trait implied by the behavior at initial exposure.

Procedure

The experimental procedure consisted of four phases: initial exposure, distraction, subsequent exposure, and recall. For all phases, participants were seated at individual computer desks.

Initial exposure phase

Participants viewed 12 behavioral descriptions, each of which described a target performing a discrete behavior. Each behavioral description was accompanied by a headshot photograph that ostensibly depicted the target. Nine of the behavioral descriptions have previously been found to elicit trait inferences (e.g., stopped and motioned for the pedestrians to cross; Costabile, 2016; Todd, Molden, Ham, & Vonk, 2011), whereas the remaining three were fillers that did not imply a specific trait (e.g., walked around the block). Participants were instructed to either form an impression or construct a narrative about each target.

Each behavioral description was presented for 8 s after which participants were given 1 min to write an impression or narrative essay about the target. The next behavioral description was not presented until the 1-min period elapsed. Participants viewed one of two stimulus sets—either Set A or Set B—in the initial exposure phase. The two sets contained the same stimuli, except the behavior paired with the two key target photographs differed by set (e.g., in Set A, one photograph was paired with the description “Emily walked into the waitress and the food spilled onto the floor,” and in Set B, the same photograph was paired with the description “Emily stopped her car and motioned the pedestrians to cross”). This method controlled for familiarity with target photographs and provided a within-study control for the recall task such that behaviors consistent with the initial behavior from Set A would serve as unrelated behaviors for those initially exposed to Set B and vice versa.

To ensure that the processing objectives had the intended effects, impression and narrative essays were coded by two independent individuals for evidence of dispositional inferences. Each dispositional inference was counted, and responses were averaged across the 12 essays. Interrater reliability was high (r = .94, M difference between coders = 0.06). Discrepancies were resolved through discussion.

Distraction phase

Participants completed a 15-min distraction task, during which time, they read 60 behavioral descriptions of two new targets and indicated which target was more likable (e.g., Person A labeled the boxes of clothes to be stored for winter or Person B returned the extra change to the cashier), see Carlston and Skowronski (1994) for similar distraction procedures.

Subsequent exposure phase

Adapting procedures used by Rothbart et al. (1979), participants were asked to memorize 48 photograph–behavioral description pairs involving four targets who performed 12 behaviors each. The same headshot photograph was used for all behaviors performed by the same target. Each pairing was presented one at a time for 6 s on a computer monitor. The order in which the pairings were displayed to participants did not have a discernable pattern. Two of the targets depicted in the pairings had been presented during the initial exposure phase. These familiar targets each performed 12 new behaviors during the subsequent exposure phase; four that were consistent with the trait implied by Stimulus Set A at initial exposure, four that were consistent with the trait implied by Stimulus Set B at initial exposure, and four that were unrelated to any trait implied at initial exposure. The remaining two targets had not been presented during the initial exposure phase, and were, therefore, unfamiliar to participants. These unfamiliar targets were fillers, designed only to increase the difficulty of the task. Recall of behaviors performed by unfamiliar targets was not examined.

Recall phase

Participants were shown a photograph of each of the four targets from the subsequent exposure phase, one at a time, and were asked to list all behaviors that were paired with the target during that phase of the study. Responses could be recalled in any order. Responses were coded as correct if participants, at minimum, indicated the gist of the behavior. Two independent raters coded each response. Interrater reliability was high (rs range = .90-.96 across targets, M differences between coders range = 0.01-0.04). Discrepancies were resolved through discussion.

Results

Manipulation check

A one-way ANOVA that examined participants’ essay responses confirmed the effectiveness of the processing objective manipulation to influence the rate at which participants formed dispositional inferences about the targets. Participants included significantly more dispositional inferences in their essays when given an impression formation processing objective (M = 1.43, SD = 0.84) than when given a narrative processing objective (M = 0.12, SD = 0.17), F(1, 99) = 112.04, p < .001, $η_{p}^{2}$ = .53 (90% confidence interval [CI] = [0.41, 0.61]), 1 – β = 1.00.¹

Percent of behaviors recalled

According to the hypothesis, participants given an impression formation processing objective should evidence better recall of consistent versus unrelated behavioral information than participants given a narrative construction processing objective. We tested this hypothesis with a 2 (processing objective: impression vs. narrative) × 2 (behavior recalled: consistent vs. unrelated with initial behavior) × 2 (stimulus set: A vs. B) repeated-measures ANOVA, with behavior recalled serving as the within-subject factor. The results indicated no significant main effects for processing objective, for behavior recalled, or for stimulus set, all Fs < 0.50, ps > .40, $η_{p}^{2}$ s < .01.

However, the predicted interaction between processing objective and behavior recalled was significant, F(1, 97) = 6.76, p = .011, $η_{p}^{2}$ = .07, 90% CI = [0.01, 0.16], 1 – β = .73 (Figure 1). As expected, participants given an impression formation objective recalled more behaviors from the subsequent exposure phase that were consistent with the trait implied by the behavior presented at initial exposure than behaviors that were unrelated to the trait implied at initial exposure, F(1, 97) = 5.95, p = .017, $η_{p}^{2}$ = .06, 90% CI = [0.01, 0.14], 1 – β = .68. By contrast, participants given a narrative objective did not show enhanced recall for the consistent versus unrelated behaviors, F(1, 97) = 1.40, p = .24, $η_{p}^{2}$ = .01, 90% CI = [0.00, 0.07], 1 – β = .22. The difference between impression and narrative objectives on recall of consistent behaviors was marginally significant, F(1, 97) = 2.93, p = .090, $η_{p}^{2}$ = .03, 90% CI = [0.00, 0.10], 1 – β = .40.

Figure 1.

Effects of processing objectives on free recall of behavior that relate to a familiar trait or an unrelated trait (Experiment 1).

To examine whether dispositional inferences guide the confirmation bias process, we created a biased recall measure in which recall of unrelated behaviors was subtracted from recall of consistent behaviors and then regressed biased recall on the number of dispositional inferences provided in essays across both experimental conditions. The results indicated that dispositional inferences predicted biased recall of subsequent behaviors, β = .26, t(98) = 2.68, p = .007. Given the close relationship between the independent variables and manipulation checks, mediational analyses were not conducted (Fiedler, Schott, & Meiser, 2011).

Discussion

The experimental procedure used in Experiment 1 represented the building blocks of a social encounter. The initial exposure phase provided a brief observation of multiple target individuals. The subsequent exposure phase provided a second, more in-depth observation where participants could gain additional information about some of the original targets. Accordingly, Experiment 1 explored whether the inferences generated during initial observations of others prompted biased processing of information gleaned during a subsequent exposure.

The results indicated that, compared with processing objectives that discouraged dispositional inferences (i.e., narrative construction), objectives that encouraged dispositional inferences at initial exposure (i.e., impression formation) led to biased recall of information. This result, which is consistent with the research conclusions of Carlston and Skowronski (1994), suggests that dispositional inferences can have important downstream effects on the processing of subsequent information. In addition, because initial behavioral information, exposure time, and processing time were held constant across conditions, the present findings indicate that confirmatory processing of information was initiated not by the content of the initial behavior but rather by the dispositional inferences drawn by the perceiver. Thus, whereas researchers have typically initiated the confirmation bias process by providing perceivers with expectancy information (e.g., Neuberg, 1989), measuring perceivers’ expectancies about a target (e.g., Jussim, 1989), or relying on stereotypic thinking to enact the expectation (e.g., Bodenhausen, 1988), the results of this experiment show that confirmatory processing of information can be initiated by merely inducing a mindset that encourages perceivers to draw dispositional inferences during an initial social encounter.

Experiment 2

Experiment 2 examined the downstream effects of dispositional inferences on interpretation of ambiguous information in an expectancy-consistent way (e.g., Srull & Wyer, 1980). We hypothesized that participants given an impression formation objective at initial exposure would be more likely to interpret ambiguous information as indicative of a previously implied trait than participants given a narrative construction objective at initial exposure. Moreover, we hypothesized that the biased interpretation of ambiguous information would be confined to the specific disposition inferred in the initial exposure task. Thus, processing objectives were not expected to differentially influence whether the ambiguous information was judged as indicative of traits that were not previously implied.

Method

Participants

Sixty-six introductory psychology students participated in this experiment in exchange for partial course credit. Fifty-two percent of participants were women and 90% of participants identified as European American.

Design

This study used a 2 (processing objective: impression formation vs. narrative construction) × 2 (behavior interpretation: consistent trait rating vs. unrelated trait rating) mixed-model experimental design with repeated measures on the second factor. Participants were randomly assigned to a processing objective.

Procedure

The procedures matched those used in Experiment 1 with the following exceptions: (a) Among the series of targets presented to participants at initial exposure in which participants wrote impression or narrative essays, one target was paired with the behavior “tripped over a crack in the sidewalk,” which implied the trait clumsy (Costabile, 2016). (b) This same target was presented to participants during the subsequent exposure phase along with a 100-word paragraph that described new information about the target in which she stepped on her dance partner’s feet when learning a new dance. The context contained in the paragraph provided ambiguity with respect to whether the new information was indicative of the target’s clumsiness (e.g., new student was assigned to be dance partner, target knew the steps well). (c) As a strong test of the hypothesis, participants were instructed to evaluate the target using only the information provided in the 100-word paragraph. To assess perceptions of clumsiness, participants indicated whether the target was clumsy, coordinated (reverse coded), and graceful (reverse coded) on scales ranging from 1 (not at all) to 7 (extremely), Cronbach’s α = .68. To assess perceptions of traits unrelated to the behavior presented during initial exposure, participants also indicated whether the target was patient and kind on a scale ranging from 1 (not at all) to 7 (extremely), Cronbach’s α = .66. (d) Only one stimulus set was used in this study.

Results

Manipulation check

Two independent coders counted the number of dispositional inferences provided in essays written at initial exposure. Scores were averaged across essays. Interrater reliability was high (r = .95, M difference between coders = 0.58). A one-way ANOVA confirmed the effectiveness of the processing objective manipulation to influence the rate at which participants formed dispositional inferences at initial exposure. Participants given an impression formation objective used more dispositional inferences in their essays (M = 1.53, SD = 0.95) than did those given a narrative processing objective (M = 0.13, SD = 0.20), F(1, 64) = 71.20, p < .001, $η_{p}^{2}$ = .53, 90% CI = [0.38, 0.62], 1 – β = 1.00.

Interpretation of ambiguous information

According to the hypothesis, participants given an impression formation processing objective at initial exposure should be more likely than participants given a narrative processing objective to interpret ambiguous information about the target in a manner consistent with the trait implied at initial exposure. By contrast, interpretation of the unrelated traits should be less affected by the processing objective manipulation. We tested this hypothesis with a 2 (processing objective: impression formation vs. narrative construction) × 2 (behavior interpretation: consistent trait rating vs. unrelated trait rating) mixed-model ANOVA with repeated measures on the second factor.

Results yielded a nonsignificant effect of processing objective, F(1, 64) = 0.96, p = .33, $η_{p}^{2}$ = .02, 90% CI = [0.00, 0.09], 1 – β = .16, and a significant effect of behavior interpretation, such that participants interpreted the ambiguous information as more indicative of unrelated traits (M = 4.33, SD = 1.13) than consistent traits (M = 3.75, SD = 1.04), F(1, 64) = 12.83, p < .001, $η_{p}^{2}$ = .17, 90% CI = [0.05, 0.31], 1 – β = .94. More important, and consistent with the hypothesis, there was also a significant interaction between processing objective and behavior interpretation, F(1, 64) = 6.99, p = .010, $η_{p}^{2}$ = .10, 90% CI = [0.01, 0.25], 1 – β = .74.² Participants given an impression formation objective at initial exposure interpreted the ambiguous information as more indicative of the trait related to the behavior at initial exposure (M = 4.09, SD = 1.18) than did participants given a narrative construction objective (M = 3.46, SD = 0.81), F(1, 64) = 6.48, p = .013, $η_{p}^{2}$ = .09, 90% CI = [0.01, 0.22], 1 – β = .71. No such effect emerged for the trait that was unrelated to the behavior at initial exposure (M_{impression formation objective} = 4.21, SD = 1.23; M_{narrative objective} = 4.43, SD = 1.04), F(1, 64) = 0.61, p = .44, $η_{p}^{2}$ = .01, 90% CI = [0.00, 0.08], 1 – β = .12.

Although not directly relevant to our hypothesis, we conducted additional pairwise comparisons to better explore the relationships among all conditions. Results indicated that participants given an impression objective at initial exposure were just as likely to interpret the ambiguous information as indicative of the trait related to the behavior at initial exposure as they were to interpret it as indicative of a trait unrelated to the behavior at initial exposure, F(1, 64) = 0.28, p = .598, $η_{p}^{2}$ < .01, 90% CI = [0.00, 0.06], 1 – β = .08. By contrast, participants given a narrative objective were more likely to interpret the ambiguous information as indicative of the trait unrelated to the behavior at initial exposure than of the trait related to the behavior at initial exposure, F(1, 64) = 19.54, p < .001, $η_{p}^{2}$ = .23, 90% CI = [0.10, 0.37], 1 – β = .99. A regression analysis using the full dataset indicated that the number of dispositional inferences in the initial essay predicted the extent to which participants interpreted the ambiguous information as indicative of the trait related to the behavior at initial exposure, β = .45, t(64) = 4.01, p < .001.

Discussion

Experiment 2 examined the downstream effects of dispositional inferences on the interpretation of ambiguous social information. Results indicated that participants who were encouraged to form dispositional inferences via an impression formation processing objective at initial exposure were more likely than participants who were discouraged from forming dispositional inferences via a narrative construction processing objective at initial exposure to interpret ambiguous information about a target in line with a trait that was previously implied by the target’s behavior. This effect did not occur when the trait was unrelated to the target’s previous behavior. Discouraging dispositional inferences increased participants’ tendency to interpret the ambiguous information in line with a trait that was unrelated (i.e., kind) versus related (i.e., clumsy) to the target’s previous behavior. Although this difference might be the result of participants’ willingness to endorse positive-valenced versus negative-valenced traits, baseline judgments of the unrelated and related traits were not obtained; thus, it is difficult to draw conclusions about this effect. However, it is clear that ratings of the target on the unrelated trait did not differ by initial processing objective, suggesting that a narrative processing objective may not lead perceivers to generally endorse unrelated traits more often than related traits. In sum, the results suggest that circumstances that encourage early dispositional inferences can nudge perceivers onto a processing trajectory that leads to biased interpretation of subsequently encountered ambiguous social information.

Experiment 3

Experiments 1 and 2 supported predictions regarding the downstream effects of causal inferences. Processing objectives that encouraged participants to draw dispositional inferences about a target at initial exposure led to biased memory and biased interpretation of ambiguous social information relative to processing objectives that encouraged participants to construct a narrative about the target. We designed Experiment 3 to test whether biased processing of information extends to perceivers’ information-seeking behaviors.

When evaluating whether one’s expectations are true, people tend to use a positive test strategy, whereby they test cases that are expected to have the characteristic of interest, and neglect to test cases that are expected to lack that characteristic (e.g., Klayman & Ha, 1987; Wason, 1960). This strategy is also employed in interpersonal interactions. Snyder and Swann (1978), for example, asked undergraduate students to interview a fellow student to uncover whether the individual had characteristics described by the experimenter (e.g., extraverted). The researchers found that interviewers were more likely to ask questions that would confirm (e.g., What would you do to liven things up at a party?) than disconfirm their hypothesis (e.g., In what situations do you wish you were more outgoing?). Given the general tendency of individuals to acquiesce to an interviewer’s questions, this strategy has the potential to lead to behavioral confirmation (Snyder, Tanke, & Berscheid, 1977; Zuckerman, Knee, Hodgins, & Miyake, 1995).

Experiment 3 examined whether processing objectives at initial exposure to trait-implying behaviors caused participants to ask confirmatory questions in an ersatz interview with the target individuals. Participants were first given a processing objective (i.e., form an impression or construct a narrative) when learning about multiple targets each of whom performed a trait-implying behavior. Next, participants were asked to engage in an ersatz interview with four targets about whom participants would select questions and receive answers. Two of the targets had been presented during the initial exposure phase and were, therefore, familiar to participants. The other two had not been presented during the initial exposure phase and were, therefore, unfamiliar to participants. We hypothesized that participants given impression formation objectives at initial exposure would be more likely to select trait confirmatory questions for the familiar targets than for unfamiliar targets; however, this tendency would be attenuated for participants given narrative construction objectives.

Method

Participants

Ninety-three psychology students participated in this experiment in exchange for partial course credit. Seventy-two percent of participants were women and 84% of participants identified as European American.

Design

This study used a 2 (processing objective: impression formation vs. narrative construction) × 2 (target familiarity: familiar vs. unfamiliar) mixed-model experimental design with repeated measures on the second factor. Participants were randomly assigned to processing objective.

Procedure

The procedures matched those used in Experiment 1 with an information-seeking task replacing the subsequent exposure and recall tasks. As in Experiment 1, the initial exposure phase consisted of two stimulus sets (A and B) to compare information-seeking behavior for familiar and unfamiliar targets during the interview phase.

Information-seeking task

Following the distraction phase, all participants were told that they would participate in an ersatz interview in which they would ask a target questions and receive answers. Following Snyder and Swann (1978), participants were instructed to select questions to see whether their general impression of the target was accurate. Participants completed ersatz interviews with four targets, two from Stimulus Set A and two from Stimulus Set B, such that all participants interviewed two targets who were familiar and two who were unfamiliar.

The procedures employed a serial presentation task (Jonas, Shulz-Hardt, Frey, & Thelen, 2001). For each target, participants received a series of question trials, presented sequentially. Each question trial consisted of two question options from which participants selected one to ask the target. After selecting the question, the target’s ostensible answer was displayed. Then, participants were presented with the next trial.

Critical trials were those in which question options related to the trait implied in the initial exposure phase. In these trials, one question option was consistent and the other inconsistent with the trait implied in the initial exposure phase. Critical trials were selected from pilot testing in which participants (N = 28) who were not exposed to the initial exposure phase completed the interview task. Trials were selected for Experiment 3 if each question option in the trial was selected by at least 25% of pilot participants and not more than 75% of pilot participants. On these 11 critical trials, pilot participants selected, on average, 48% of the questions identified as confirmatory, a value that did not differ from chance, t(27) = −0.55, p = .59. Critical trials were interspersed among nine filler trials in which both question options were unrelated to the previously implied trait.

Results

Manipulation check

Due to a programming error, essays written in the initial exposure phase were saved for only 33 participants. Available essays were coded for the number of dispositional essays by two independent coders and scores were averaged across essays. Interrater reliability was high (r = .97, M difference between coders = .09). An ANOVA indicated that individuals given an impression objective provided more dispositional inferences in their essays (M = 1.81, SD = 0.84) than did those given a narrative objective (M = 0.36, SD = 0.38), F(1, 31) = 41.65, p < .001, $η_{p}^{2}$ = .57, 90% CI = [0.36, 0.69], 1 – β = 1.00.

Information-seeking bias

We hypothesized that participants who received an impression formation objective during initial exposure would be more likely to select questions consistent with the trait implied in the initial exposure phase for familiar targets than for unfamiliar targets; however, this tendency would be attenuated for those given narrative objectives. A 2 (processing objective: impression formation vs. narrative construction) × 2 (target familiarity: familiar vs. unfamiliar) × 2 (stimulus set: A vs. B) ANOVA on number of questions selected that were consistent with the trait implied in the initial exposure task did not indicate a main effect of processing object, target familiarity, or stimulus set, all Fs < 1.20, ps ⩾ .30, $η_{p}^{2}$ s = .01, 90% CI = [0.00, 0.07].

However, results did yield the predicted interaction between processing objective and target familiarity on question selection, F(1, 89) = 6.25, p = .014, $η_{p}^{2}$ = .07, 90% CI = [0.01, 0.16], 1 – β = .70. As shown in Figure 2, participants given an impression formation processing objective during initial exposure selected questions consistent with the implied trait more often for familiar than unfamiliar targets, F(1, 89) = 5.76, p = .018, $η_{p}^{2}$ = .06, 90% CI = [0.01, 0.15], 1 – β = .66. Participants given a narrative construction processing objective during initial exposure selected similar questions for both familiar and unfamiliar targets, F(1, 89) = 1.27, p = .263, $η_{p}^{2}$ = .01, 90% CI = [0.00, 0.08], 1 – β = .20. Additional pairwise comparisons indicated that for familiar targets, participants given an impression formation processing objective were more likely to select questions consistent with the implied trait than were participants given a narrative processing objective, F(1, 89) = 6.98, p = .010, $η_{p}^{2}$ = .07, 90% CI = [0.01, 0.17], 1 – β = .74. Question selection for unfamiliar targets did not differ by initial processing objective, F(1, 89) = 1.13, p = .291, $η_{p}^{2}$ = .01, 90% CI = [0.00, 0.07], 1 – β = .18.

Figure 2.

Effects of processing objectives on the percentage of questions selected that were consistent with trait implied in initial exposure phase (Experiment 3).

To examine whether dispositional inferences guide information-seeking bias, we constructed a bias measure in which we subtracted the number of consistent questions selected for unfamiliar targets from the number of consistent questions selected for familiar targets. A regression analysis using data from both experimental conditions indicated a positive association between dispositional inferences written in essays at initial exposure and information-seeking bias, though the effect did not attain statistical significance, β = .27, t(31) = 1.55, p = .131. Even so, the strength of the relationship was consistent with that observed in Experiment 1, suggesting that the nonsignificant result was likely due to low power as analyses were based on those few participants for whom essays written at initial exposure were not affected by the programming error.

Discussion

The findings of Experiment 3 indicated that the downstream effect of early dispositional inferences affect perceivers’ information-seeking behaviors. Participants given a processing objective that encouraged dispositional inferences at initial exposure were more likely to select confirmatory questions at a subsequent interview of the target than were participants given a processing objective that discouraged dispositional inferences. This result suggests that dispositional inferences can serve as expectancies for perceivers, and when given an opportunity to test whether these expectancies are true, perceivers use a positive test strategy in which they are more likely to select questions that will confirm than disconfirm their expectancies. As noted previously, this strategy has important implications for social interactions because positive test strategies can result in behavioral confirmation (Zuckerman et al., 1995). Accordingly, the present research identifies dispositional inferences as an important mechanism through which such downstream consequences can be both bolstered and attenuated.

Experiment 4

Results from Experiments 1 to 3 provided evidence that perceivers who receive processing objectives that encourage dispositional inferences at initial exposure were more likely to demonstrate confirmatory processing of subsequent information. To explicitly examine the role of dispositional inferences in the confirmation process, Experiment 4 examined whether dispositional inferences made for observed behaviors mediate the association between processing objective at initial exposure and downstream confirmation biases. To do so, we adapted the two-phase procedure pioneered by Word, Zanna, and Cooper (1994) in their investigation of the mediating mechanism underlying behavioral confirmation of racial stereotypes. In their study, Word and colleagues (1994) first manipulated a job applicant’s race to examine whether stereotypes influenced perceivers’ interview style. Second, they manipulated perceivers’ interview style to examine whether it influenced the interview performance of job applicants. This method is referred to as a double-randomization design because it involved the sequential manipulation of two variables (MacKinnon, Fairchild, & Fritz, 2007). This design is particularly useful because it experimentally tests the effect of a mediator variable, thereby ruling out alternative explanations for the association between a manipulation and measured dependent variable.

We used a double-randomization design in Experiment 4 to examine whether dispositional inferences mediate the link between processing objective and biased information-seeking behaviors. Specifically, in Experiment 4a, we examined whether processing objectives influenced the extent to which participants formed disposition-based versus situation-based judgments about a target. Then, in Experiments 4b and 4c, we presented a new group of participants with the modal disposition-based or situation-based responses reported by participants in Experiment 4a to examine their effects on information-seeking behaviors.

Experiment 4a

Experiment 4a, which constituted the first phase of our double-randomization design, examined whether processing objective influences dispositional inferences drawn from observed behaviors. Participants were asked to form an impression or create a narrative from the behavior–photograph pairs used in Experiment 3. Then, causal inferences were assessed.

Participants, design, and procedure

Psychology students (N = 100) participated in this experiment in exchange for partial course credit. Seventy-six percent of participants were women and 87% identified as European American. Participants were randomly assigned to a level of processing objective (impression formation vs. narrative construction). Sample size was determined on the basis of a power analysis that used the values obtained on the manipulation check measure in Experiment 3 with power = .80. Results indicated a minimum sample size of only 18 participants (Faul et al., 2007). Given alterations to the paradigm, we increased our sample size substantially to ensure adequate power.

The exposure phase mirrored that of Experiment 3, with the following exceptions: (a) Participants were asked to form impressions or create narratives, but they were not given the opportunity to write their impressions or narratives until after causal inferences were assessed. Assessing inferences before the essay writing task eliminated the possibility that essay writing in Experiments 1 to 3 increased trait accessibility for those who wrote impressions, and that the increased trait accessibility, then, resulted in the observed confirmation biases. (b) We created a single stimulus set that combined all targets used in Experiment 3’s interview task.

Immediately after the exposure phase, causal inferences were assessed. Participants were asked a free-response question assessing explicit causal inferences for each target (e.g., based on what you know about Jenny, why do you think she left her mom with 20 pounds of laundry?, Costabile, 2016). Two independent coders, blind to condition and hypotheses, independently coded each response for a dispositional explanation (1 = present, 0 = absent) and for a situation explanation (1 = present, 0 = absent). Interrater reliability was high (κs range = .82-.98, percent agreement range = 93%-99%). Discrepancies were resolved through discussion. Responses were averaged across targets for each type of inference.

To ensure that results were due to differences in behavior construal, rather than to differences in reporting of causal inferences, participants were also asked two generalization questions for each target. Behavior generalization is an indirect indication of dispositional inferences that is less prone to concerns of demand characteristics than explicit measures (Locksley et al., 1980; Newman, 1996). Participants were asked to estimate the likelihood that the target would perform an identical behavior in the future on a scale ranging from 1 (not at all or not often) to 7 (extremely likely or extremely often). Responses were averaged across targets.

Results

As expected, a series of one-way ANOVAs of processing objective on inferences indicated that participants with an impression formation objective generated more dispositional inferences, fewer situation inferences, and were more likely to generalize observed behavior to occurring again in the future than those with a narrative construction objective, see Table 1. Thus, processing objective guided the degree to which participants inferred dispositions from the target’s observed behaviors.

Table 1.

Causal Inferences as a Function of Processing Objective (Experiment 4a).

Variable	Possible scale range	Impression processing objective		Narrative processing objective		90% CI		90% CI		Power
Variable	Possible scale range	M	SD	M	SD	F(1, 98)		LL	UL	Power
Disposition-based explanation	0-1	0.64	0.30	0.16	0.18	91.45	0.48	0.36	0.57	1.00
Situation-based explanation	0-1	0.58	0.28	0.89	0.18	41.30	0.30	0.17	0.40	1.00
Behavior generalization	1-7	5.08	0.71	4.58	0.90	15.95	0.14	0.05	0.24	0.98

Note. CI = confidence interval; LL = lower limit; UL = upper limit.

Experiment 4b

Experiment 4b constituted the second phase of our double-randomization design. Mirroring the procedures of Word and colleagues (1994), we presented the most common dispositional and situational inferences provided by participants in Experiment 4a to participants in the current experiment during the initial exposure phase to examine their effect on these participants’ information-seeking behaviors. We hypothesized that participants who received a dispositional inference during initial exposure would be more likely to select questions consistent with the trait implied by the behavior in the initial exposure phase for familiar targets than for unfamiliar targets; this tendency was expected to be reduced among participants given a situational inference during initial exposure.

Participants, design, and procedure

Psychology students (N = 113) participated in this experiment in exchange for partial course credit. Sixty-five percent of participants were women and 83% identified as European American. This study used a 2 (inference type: dispositional vs. situational) × 2 (target familiarity: familiar vs. unfamiliar) mixed-model experimental design with repeated measures on the second factor. Participants were randomly assigned to inference type. Sample size determination was based on a power analysis that used the values found for the information-seeking task used in Experiment 3 with power = .80. Results indicated a minimum sample size of 70 participants (Faul et al., 2007). Given alterations to the paradigm, we increased our sample to ensure adequate power.

In the initial exposure phase, participants viewed the critical targets used in Set A of Experiment 3. Participants then completed the interview task used in Experiment 3, in which they selected questions to interview familiar targets (those presented in the initial exposure phase) and unfamiliar targets.

Inference type manipulation

Inference information was added to the behavioral description provided for each target (e.g., Jim is an organized person and he carefully recorded the meeting in his appointment book). Inferences were constructed using the most common dispositional and situational explanations provided by participants in Experiment 4a. For example, organized was listed in 51% of responses from those given an impression objective as an explanation for why Jim carefully recorded the meeting in his appointment book, and to not miss an important meeting was listed in 49% of explanations from those with a narrative objective.

Results

A 2 (inference type: dispositional vs. situational) × 2 (target familiarity: familiar vs. unfamiliar) ANOVA on question selection indicated main effects of inference type, F(1, 111) = 13.76, p < .001, $η_{p}^{2}$ = .11, 90% CI = [0.03, 0.20], 1 – β = .96, and target familiarity, F(1, 111) = 4.32, p = .04, $η_{p}^{2}$ = .04, 90% CI = [0.00, 0.11], 1 – β = .54. These main effects were qualified by the predicted interaction between inference type and target familiarity on question selection, F(1, 111) = 5.09, p = .026, $η_{p}^{2}$ = .04, 90% CI = [0.00, 0.12], 1 – β = .61. Participants given disposition information during initial exposure selected questions consistent with the implied trait more often for familiar targets (M = 62.29%, SD = 26.82) than for unfamiliar targets (M = 47.70%, SD = 19.42), F(1, 111) = 13.27, p < .001, $η_{p}^{2}$ = .11, 90% CI = [0.03, 0.20], 1 – β = .95. However, participants given situation information during initial exposure selected similar questions for both familiar (M = 43.06%, SD = 29.31) and unfamiliar targets (M = 43.65%, SD = 21.03), F(1, 111) = 1.13, p = .290, $η_{p}^{2}$ = .01, 90% CI = [0.00, 0.06], 1 – β = .18. These results indicate that dispositional inferences did in fact guide the degree to which participants engaged in confirmatory information-seeking behaviors.

Experiment 4c

We conducted a follow-up study to ensure that results were not due to (a) increased trait accessibility that resulted from receiving disposition information during initial exposure or (b) reinterpretation of initial behavioral information. Experiment 4c followed the procedures of Experiment 4b with the following exceptions: (a) Inference type was manipulated within subjects and (2) no behavioral descriptions were provided. Presentation order and target–inference type pairing were counterbalanced across participants.

We hypothesized that participants (N = 101) would be more likely to select questions consistent with the disposition information when it was provided in the initial exposure phase than when situation information was provided. A repeated-measures ANOVA indicated a main effect of inference type, F(1, 100) = 6.36, p < .013, η² = .06, 90% CI = [0.01, 0.15], 1 – β = .70. Participants were more likely to ask questions consistent with a dispositional inference when dispositional information was provided during initial exposure (M = 55.30%, SD = 24.45) than when situational information was provided (M = 46.71%, SD = 23.60). Together, the results of Experiment 4 suggest that dispositional inferences made for observed behaviors mediate the association between processing objective at initial behavior observation and confirmatory processing of subsequent information.

General Discussion

As compared with understanding the antecedents of dispositional inferences (e.g., Kelley & Michaela, 1980), social cognitive researchers have given relatively less attention to the consequences of these inferences. To redress this limitation of the literature, the present investigation tested whether dispositional inferences initiate confirmatory processing of subsequent information about a target. In four experiments, we examined the effects of processing objectives at initial exposure on the three components that reflect confirmation biases: memory for consistent evidence, interpretation of ambiguous information, and information-seeking behavior. We found that a processing objective that encouraged participants to form dispositional inferences led to greater confirmation bias effects on all three outcome variables than did a processing objective that discouraged participants from making dispositional inferences. The double-randomization procedure used in Experiment 4 confirmed that dispositional inferences can operate as a mechanism through which interpersonal confirmation biases occur. This work highlights the pervasive nature of confirmation biases by demonstrating that these biases occurred with little provocation even when stimuli were weak facsimiles of social encounters. It also demonstrates that a relatively minor instruction to construct a narrative proved sufficient to dislodge the automatic and stubborn cognitive biases often observed in experimental studies.

By blending theory and procedures in social perception and confirmation bias, the present investigation provided a strong test of the hypothesis that dispositional inferences guide perceivers onto a trajectory that culminates in interpersonal confirmation biases. If dispositional inferences were not the catalyst of confirmation bias processes, confirmation biases would not have been observed for participants in any of our experimental conditions because the procedures used in our experiments did not provide interpersonal expectancies for the target’s future behaviors. In addition, if exposure to trait-implying information was sufficient to initiate confirmatory biases, we would not have observed differences between experimental conditions as participants were exposed to the same trait-implying information in all conditions. Furthermore, because results indicated processing differences with regard to familiarity (i.e., of traits in Experiments 1 and 2 and of targets in Experiment 3), we were able to ensure that observed differences were due to dispositional inferences, and were not merely a by-product of manipulated objectives at initial exposure. Finally, the use of three distinct confirmation bias measurements demonstrates the broad and widespread nature of the downstream effects of dispositional inferences on the cognitive processing of subsequent information.

The present experiments build on previous work to articulate a parsimonious explanation for when interpersonal confirmation biases are likely to occur. In the domain of interpersonal perception, confirmation biases can be particularly insidious because biases facilitate behavioral confirmation in which the treatment of a target individual leads the target to respond in a manner that confirms a perceiver’s expectations (Rosenthal & Jacobson, 1968). This confirming behavior strengthens a perceiver’s confidence in the original expectation, thus reinforcing an impression that resists change even in the face of disconfirming evidence (Snyder & Swann, 1978). The present investigation suggests that reducing initial dispositional inferences might be one strategy to break this cycle. By uncovering circumstances that initiate confirmatory processes, these findings may provide the key to circumventing biased processing of events in those situations in which biases could be most detrimental to the involved parties.

Recent work has called into question the robustness of the confirmation bias in naturalistic settings (e.g., Jussim & Stevens, 2016), and a comprehensive review indicates relatively modest effects (Richard, Bond, & Stokes-Zoota, 2003). The modest effects observed in real-world studies might be due, in part, to perceivers’ general accuracy (Jussim, 2012) or to the general tendency of perceivers to avoid dispositional inferences when explaining observed events. Indeed, Malle and colleagues (Malle, 1999; Malle & Holbrook, 2012) reported that participants were more likely to explain observed events as due to a target’s current motives than to enduring traits. The present experiments indicate that interpersonal confirmatory biases occurred with relative ease when perceivers were encouraged to make dispositional inferences about the target, but that confirmatory processes were disrupted when perceivers were discouraged from drawing firm dispositional judgments of targets. Accordingly, we argue that the modest effects observed in some real-world studies on confirmation biases might be explained, in part, by lack of dispositional inferencing in these naturalistic contexts.

A Railroad Switch

The present investigation demonstrates that a processing objective that encouraged perceivers to form dispositional inferences at initial exposure led to confirmatory processing of subsequent information; however, these effects were eliminated when perceivers were given a processing objective that discouraged dispositional inferences. These findings suggest that circumstances that encourage dispositional inferences operate similar to that of a railroad switch. As a railroad switch guides a train from one track to another and, thus, determines its trajectory and eventual destination, circumstances that encourage dispositional inferences guide perceivers onto a track that results in confirmatory processing of subsequent information, whereas circumstances that discourage dispositional inferences guide the perceiver toward another track that culminates in evenhanded processing of information. This dichotomous characterization is, perhaps, too blunt to accurately capture the complexity of social perception. It is likely that expectancies are based on the degree to which a perceiver evaluates an inferred disposition to be relevant, accurate, and diagnostic of the person in his or her given situation. That is, dispositional information itself may not be sufficient to trigger confirmatory processing if a perceiver does not endorse the information, does not evaluate it to be accurate, or is not motivated to use the information (Copeland & Snyder, 1995; Snyder & Haugen, 1994). Moreover, disposition judgments vary along a number of dimensions (e.g., intentionality, expectedness, centrality, valence), each of which might influence how perceivers weight its value when forming interpersonal expectations. More work is needed to examine how these dimensions might moderate the observed effects.

The Role of Narrative

The present findings are consistent with a growing body of research suggesting that narrative may be an information processing strategy that has the potential to override automatic cognitive tendencies. Across experiments, we observed that perceivers who were given a narrative objective at initial exposure demonstrated a slight tendency in memory and information-seeking behavior toward novel, unfamiliar information as compared with familiar information. Recent work on social perceptions presents a number of possible of explanations for this attention to novel stimuli. First, given the ease with which spontaneous trait inferences occur (Todorov & Uleman, 2003), it is likely that, in addition to forming situation inferences, those with a narrative objective also inferred trait inferences at initial exposure. Situational inferences are often considered more satisfying explanations than dispositional inferences (Reeder & Trafimow, 2005); thus, narrative perceivers may have given unfamiliar information more attention because they considered this information more useful than information that was consistent with a dismissed inference. Second, narrative construction entails active creation of a target’s circumstances (Costabile, 2016). Active imagination tasks encourage perceivers to process information outside of their default mode and, thus, may encourage perceivers to seek out unfamiliar rather than familiar information (Todd, Simpson, & Tamir, 2016). Finally, narrative construction encourages predictive inferences (Costabile & Klein, 2008), and research finds that a prediction-focused mindset increases attention to inconsistent information (Erber & Fiske, 1984).

Limitations and Future Directions

Given the aim of our investigation, experiments provided explicit processing objectives to perceivers. Accordingly, it is not known the degree to which perceivers would have engaged in confirmatory processes absent explicit instructions. Also, the present results suggest that processing objectives that encourage dispositional inferences would, likewise, increase subsequent confirmation biases. Future research should examine different procedures for reducing dispositional inferences to provide further support for the proposition.

In addition, an alternative explanation of the present findings could be that the results were due to cognitive processing differences elicited by impression formation or narrative construction processing objectives that were not measured in the present investigation, such as visual perspective taking, causal thinking, or memory of the initial behavior. However, research using a paradigm identical to the initial exposure phase used in this research suggests that perspective taking, causal thinking, and memory for behaviors were comparable for individuals given impression formation and narrative construction objectives (Costabile, 2016; Costabile & Klein, 2008), suggesting that these variables were unlikely to be driving the observed effects. In addition, people shift processing strategies dynamically to cope with changing situations (Reeder & Trafimow, 2005), suggesting that future work must explore the social inference process in dynamic social interactions.

We broadly conceptualized dispositional inferences to include all inferences in which behavior is attributed to chronic, invariant causes that remain with the target indefinitely (Winter, John, Stewart, Klohnon, & Duncan, 1998). The present experiments used materials found to elicit trait inferences; however, research suggests that findings would be similar had we used stereotype-eliciting materials (J. D. Heider et al., 2007). Given the centrality of causal inferences to the social perception process (F. Heider, 1958), we would anticipate that similar results would be obtained with attitude- or value-based expectancies; however, future work using different bases of expectancies is needed. Along these lines, it is possible that situational inferences could also serve as interpersonal expectancies if the situational constraints of the expected behavior were consistent with those of the original behavior (Costabile, 2016).

Concluding Comments

The present investigation demonstrated that a perceiver’s information processing objective could encourage or attenuate the likelihood of confirmatory processing of information. Interpersonal confirmation biases have caused a host of social problems such as wrongful convictions (Kassin, Dror, & Kukucka, 2013), academic underachievement (Rosenthal & Jacobson, 1968), and the misinterpretation of scientific evidence (Loehle, 1987), but these biases are not inevitable. Given the insidious and often consequential nature of confirmation biases in interpersonal perception, it is critical to identify those circumstances that can guide perceivers onto the track that leads to unbiased processing of information.

Supplemental Material

Costabile_onlineappendix – Supplemental material for Downstream Effects of Dispositional Inferences on Confirmation Biases

Supplemental material, Costabile_onlineappendix for Downstream Effects of Dispositional Inferences on Confirmation Biases by Kristi A. Costabile and Stephanie Madon in Personality and Social Psychology Bulletin

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Supplemental Material

Supplemental material is available online with this article.

References

Allen

T. J.

Sherman

J. W.

Conrey

F. R.

Stroessner

S. J.

(2009). Stereotype strength and attentional bias: Preference for confirming versus disconfirming information depends on processing capacity. Journal of Experimental Social Psychology, 45, 1081-1087.

Bodenhausen

G. V.

(1988). Stereotypic biases in social decision making and memory: Testing process models of stereotype use. Journal of Personality and Social Psychology, 55, 726-737.

Bruner

J. S.

(1957). On perceptual readiness. Psychological Review, 64, 123-152.

Carlston

D. E.

Skowronski

J. J.

(1994). Savings in relearning of trait information as evidence for spontaneous inference generation. Journal of Personality and Social Psychology, 66, 840-856.

Copeland

Snyder

(1995). When counselors confirm: A functional analysis. Personality and Social Psychology Bulletin, 21, 1210-1220.

Costabile

K. A.

(2011). Rude or rushed? Effects of narrative and impression processing objectives on person perception. Social Cognition, 29, 445-475.

Costabile

K. A.

(2016). Narrative construction, social perceptions, and the situation model. Personality and Social Psychology Bulletin, 42, 589-602.

Costabile

K. A.

Klein

S. B.

(2008). Understanding and predicting social events: The effects of narrative construction on inference generation. Social Cognition, 26, 455-472.

Erber

Fiske

S. T.

(1984). Outcome dependency and attention to inconsistent information. Journal of Personality and Social Psychology, 47, 709-726.

10.

Faul

Erdfelder

Lang

A.-G.

Buchner

(2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39, 175-191.

11.

Fiedler

Schott

Meiser

(2011). What mediation analysis can (not) do. Journal of Experimental Social Psychology, 47, 1231-1236.

12.

Findley

K. A.

Scott

M. S.

(2006). Multiple dimensions of tunnel vision in criminal cases. Wisconsin Law Review, 2, 291-397.

13.

Hamilton

D. L.

Katz

L. B.

Leirer

V. O.

(1980). Cognitive representation of personality impressions: Organizational processes in first impression formation. Journal of Personality and Social Psychology, 39, 1050-1063. doi:10.1037/h0077711

14.

Heider

(1958). The psychology of interpersonal relations. Hoboken, NJ: John Wiley.

15.

Heider

J. D.

Scherer

C. R.

Skowronski

J. J.

Wood

S. E.

Edlund

J. E.

Hartnett

J. L.

(2007). Trait expectancies and stereotype expectancies have the same effect on person memory. Journal of Experimental Social Psychology, 43, 265-272.

16.

Jonas

Shulz-Hardt

Frey

Thelen

(2001). Confirmation bias in sequential information search after preliminary decisions: An expansion of dissonance theoretical research on selective exposure to information. Journal of Personality and Social Psychology, 80, 557-571. doi:10.1037/0022-3514.80.4.557

17.

Jussim

(1989). Teacher expectations: Self-fulfilling prophecies, perceptual biases, and accuracy. Journal of Personality and Social Psychology, 57, 469-480.

18.

Jussim

(2012). Social perception and social reality: Why accuracy dominates bias and self-fulfilling prophecy. New York, NY: Oxford University Press.

19.

Jussim

Stevens

S. T.

(2016). Why accuracy dominates self-fulfilling prophecies and bias. In Trusz

Babel

(Eds.), Interpersonal and intrapersonal expectancies (pp. 110-116). New York, NY: Routledge.

20.

Kassin

S. M.

Dror

I. E.

Kukucka

(2013). The forensic confirmation bias: Problems, perspectives, and proposed solutions. Journal of Applied Research in Memory and Cognition, 2, 42-52.

21.

Kelley

H. H.

Michaela

J. L.

(1980). Attribution theory and research. Annual Review of Psychology, 31, 457-501.

22.

Klayman

Y.-W.

(1987). Confirmation, disconfirmation, and information in hypothesis testing. Psychological Review, 94, 211-228. doi:10.1037/0033-295X.94.2.211

23.

Locksley

Borgida

Brekke

Hepburn

(1980). Sex stereotypes and social judgment. Journal of Personality and Social Psychology, 39, 821-831.

24.

Loehle

(1987). Hypothesis testing in ecology: Psychological aspects and the importance of theory maturation. The Quarterly Review of Biology, 62, 397-409.

25.

MacKinnon

D. P.

Fairchild

A. J.

Fritz

M. S.

(2007). Mediation analysis. Annual Review of Psychology, 58, 593-614. doi:10.1146/annurev.psych.58.110405.085542

26.

Malle

B. F.

(1999). How people explain behavior: A new theoretical framework. Personality and Social Psychology Review, 3, 21-43. doi:10.1207/s15327957pspr0301_2

27.

Malle

B. F.

Holbrook

(2012). Is there a hierarchy of social inferences? The likelihood and speed of inferring intentionality, mind, and personality. Journal of Personality and Social Psychology, 102, 661-684. doi:10.1037/a0026790

28.

McCarthy

R. J.

Skowronski

J. J.

(2011). What will Phil do next? Spontaneously inferred traits influence predictions of behavior. Journal of Experimental Social Psychology, 47, 321-332.

29.

Neuberg

S. L.

(1989). The goal of forming accurate impressions during social interactions: Attenuating the impact of negative expectancies. Journal of Personality and Social Psychology, 56, 374-386.

30.

Newman

L. S.

(1996). Trait impressions as heuristics for predicting future behavior. Personality and Social Psychology Bulletin, 24, 496-504.

31.

Nickerson

R. S.

(1988). Confirmation bias: A ubiquitous phenomenon in many guises. Review of General Psychology, 2, 175-220.

32.

Nisbett

R. E.

(1987). Lay personality theory: Its nature, origin, and utility. In Grunberg

N. E.

Nisbett

R. E.

Rodin

Singer

E. E.

(Eds.), A distinctive approach to psychological research: The influence of Stanley Schachter (pp. 87-117). Hillsdale, NJ: Lawrence Erlbaum.

33.

Pennebaker

J. W.

Skelton

J. A.

(1978). Psychological parameters of physical symptoms. Personality and Social Psychology Bulletin, 4, 524-530.

34.

Reeder

G. D.

Trafimow

(2005). Attributing motives to other people. In Malle

B. F.

Hodges

S. D.

(Eds.), Other minds: How humans bridge the divide between self and others (pp. 106-123). New York, NY: Guilford Press.

35.

Richard

F. D.

Bond

C. F.

Jr. Stokes-Zoota

J. J.

(2003). One hundred years of social psychology quantitatively described. Review of General Psychology, 7, 331-363.

36.

Rosenthal

Jacobson

(1968). Teacher expectations for the disadvantaged. Scientific American, 218, 19-23.

37.

Rothbart

Evans

Fulero

(1979). Recall for confirming events: Memory processes and the maintenance of social stereotypes. Journal of Experimental Social Psychology, 15, 343-355.

38.

Snyder

Haugen

J. A.

(1994). Why does behavioral confirmation occur? A functional perspective on the role of the perceiver. Journal of Experimental Social Psychology, 30, 218-246.

39.

Snyder

Swann

W. B.

Jr. (1978). Hypothesis-testing processes in social interaction. Journal of Personality and Social Psychology, 11, 1202-1212.

40.

Snyder

Tanke

E. D.

Berscheid

(1977). Social perception and interpersonal behavior: On the self-fulfilling nature of social stereotypes. Journal of Personality and Social Psychology, 35, 656-666.

41.

Srull

T. K.

Wyer

R. S.

(1980). Category accessibility and social perception: Some implications for the study of person memory and interpersonal judgments. Journal of Personality and Social Psychology, 38, 841-856.

42.

Todd

A. R.

Molden

D. C.

Ham

Vonk

(2011). The automatic and co-occurring activation of multiple social inferences. Journal of Experimental Social Psychology, 47, 37-49.

43.

Todd

A. R.

Simpson

A. J.

Tamir

D. I.

(2016). Active perspective taking induces flexible use of self-knowledge during social inference. Journal of Experimental Psychology: General, 145, 1583-1588.

44.

Todorov

Uleman

J. S.

(2003). The efficiency of binding spontaneous trait inferences to actors’ faces. Journal of Experimental Social Psychology, 39, 549-562. doi:10.1016/S0022-1031(03)00059-3

45.

Wason

P. C.

(1960). On the failure to eliminate hypotheses in a conceptual task, quarterly. Journal of Experimental Psychology, 12, 129-140. doi:10.1080/17470216008416717

46.

Winter

D. G.

John

O. P.

Stewart

A. J.

Klohnon

E. C.

Duncan

L. E.

(1998). Traits and motives: Toward an integration of two traditions in personality research. Psychological Review, 105, 230-250.

47.

Word

C. O.

Zanna

M. P.

Cooper

(1974). The nonverbal mediation of self-fulfilling prophecies in interracial interaction. Journal of Experimental Social Psychology, 10, 109-120.

48.

Wyer

N. A.

(2004). Not all stereotypic biases are created equal: Evidence for a stereotype-disconfirming bias. Personality and Social Psychology Bulletin, 30, 706-720.

49.

Zuckerman

Knee

C. R.

Hodgins

H. S.

Miyake

(1995). Hypothesis confirmation: The joint effect of positive test strategy and acquiescence response set. Journal of Personality and Social Psychology, 68, 52-60. doi:10.1037/0022-3514.68.1.52

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.12 MB