Stephanie M. Merritt

Not all trust is created equal: dispositional and history-based trust in human-automation interactions.

Objective: We provide an empirical demonstration of the importance of attending to human user individual differences in examinations of trust and automation use. Background: Past research has generally supported the notions that machine reliability predicts trust in automation, and trust in turn predicts automation use. However, links between user personality and perceptions of the machine with trust in automation have not been empirically established. Method: On our X-ray screening task, 255 students rated trust and made automation use decisions while visually searching for weapons in X-ray images of luggage. Results: We demonstrate that individual differences affect perceptions of machine characteristics when actual machine characteristics are constant, that perceptions account for 52% of trust variance above the effects of actual characteristics, and that perceptions mediate the effects of actual characteristics on trust. Importantly, we also demonstrate that when administered at different times, the same six trust items reflect two types of trust (dispositional trust and history-based trust) and that these two trust constructs are differentially related to other variables. Interactions were found among user characteristics, machine characteristics, and automation use. Conclusion: Our results suggest that increased specificity in the conceptualization and measurement of trust is required, future researchers should assess user perceptions of machine characteristics in addition to actual machine characteristics, and incorporation of user extraversion and propensity to trust machines can increase prediction of automation use decisions. Application: Potential applications include the design of flexible automation training programs tailored to individuals who differ in systematic ways.

The use of background and ability profiles to predict college student outcomes

To determine whether profiles of predictor variables provide incremental prediction of college student outcomes, the authors 1st applied an empirical clustering method to profiles based on the scores of 2,771 entering college students on a battery of biographical data and situational judgment measures, along with SAT and American College Test scores and high school grade point average, which resulted in 5 student groups. Performance of the students in these clusters was meaningfully different on a set of external variables, including college grade point average, self-rated performance, class absenteeism, organizational citizenship behavior, intent to quit their university, and satisfaction with college. The 14 variables in the profile were all significantly correlated with 1 or more of the outcome measures; however, nonlinear prediction of these outcomes on the basis of cluster membership did not add incrementally to a linear-regression-based combination of these 14 variables as predictors.

I Trust It, but I Don’t Know Why Effects of Implicit Attitudes Toward Automation on Trust in an Automated System

Objective: This study is the first to examine the influence of implicit attitudes toward automation on users’ trust in automation. Background: Past empirical work has examined explicit (conscious) influences on user level of trust in automation but has not yet measured implicit influences. We examine concurrent effects of explicit propensity to trust machines and implicit attitudes toward automation on trust in an automated system. We examine differential impacts of each under varying automation performance conditions (clearly good, ambiguous, clearly poor). Method: Participants completed both a self-report measure of propensity to trust and an Implicit Association Test measuring implicit attitude toward automation, then performed an X-ray screening task. Automation performance was manipulated within-subjects by varying the number and obviousness of errors. Results: Explicit propensity to trust and implicit attitude toward automation did not significantly correlate. When the automation’s performance was ambiguous, implicit attitude significantly affected automation trust, and its relationship with propensity to trust was additive: Increments in either were related to increases in trust. When errors were obvious, a significant interaction between the implicit and explicit measures was found, with those high in both having higher trust. Conclusion: Implicit attitudes have important implications for automation trust. Application: Users may not be able to accurately report why they experience a given level of trust. To understand why users trust or fail to trust automation, measurements of implicit and explicit predictors may be necessary. Furthermore, implicit attitude toward automation might be used as a lever to effectively calibrate trust.

Affective Processes in Human–Automation Interactions

Objective: This study contributes to the literature on automation reliance by illuminating the influences of user moods and emotions on reliance on automated systems. Background: Past work has focused predominantly on cognitive and attitudinal variables, such as perceived machine reliability and trust. However, recent work on human decision making suggests that affective variables (i.e., moods and emotions) are also important. Drawing from the affect infusion model, significant effects of affect are hypothesized. Furthermore, a new affectively laden attitude termed liking is introduced. Method: Participants watched video clips selected to induce positive or negative moods, then interacted with a fictitious automated system on an X-ray screening task. At five time points, important variables were assessed including trust, liking, perceived machine accuracy, user self-perceived accuracy, and reliance. These variables, along with propensity to trust machines and state affect, were integrated in a structural equation model. Results: Happiness significantly increased trust and liking for the system throughout the task. Liking was the only variable that significantly predicted reliance early in the task. Trust predicted reliance later in the task, whereas perceived machine accuracy and user self-perceived accuracy had no significant direct effects on reliance at any time. Conclusion: Affective influences on automation reliance are demonstrated, suggesting that this decision-making process may be less rational and more emotional than previously acknowledged. Application: Liking for a new system may be key to appropriate reliance, particularly early in the task. Positive affect can be easily induced and may be a lever for increasing liking.

Measuring Individual Differences in the Perfect Automation Schema

Objective A self-report measure of the perfect automation schema (PAS) is developed and tested. Background Researchers have hypothesized that the extent to which users possess a PAS is associated with greater decreases in trust after users encounter automation errors. However, no measure of the PAS currently exists. We developed a self-report measure assessing two proposed PAS factors: high expectations and all-or-none thinking about automation performance. Method In two studies, participants responded to our PAS measure, interacted with imperfect automated aids, and reported trust. Results Each of the two PAS measure factors demonstrated fit to the hypothesized factor structure and convergent and discriminant validity when compared with propensity to trust machines and trust in a specific aid. However, the high expectations and all-or-none thinking scales showed low intercorrelations and differential relationships with outcomes, suggesting that they might best be considered two separate constructs rather than two subfactors of the PAS. All-or-none thinking had significant associations with decreases in trust following aid errors, whereas high expectations did not. Results therefore suggest that the all-or-none thinking scale may best represent the PAS construct. Conclusion Our PAS measure (specifically, the all-or-none thinking scale) significantly predicted the severe trust decreases thought to be associated with high PAS. Further, it demonstrated acceptable psychometric properties across two samples. Application This measure may be used in future work to assess levels of PAS in users of automated systems in either research or applied settings.

Are well-calibrated users effective users? Associations between calibration of trust and performance on an automation-aided task.

Objective: We present alternative operationalizations of trust calibration and examine their associations with predictors and outcomes. Background: It is thought that trust calibration (correspondence between aid reliability and user trust in the aid) is a key to effective human–automation performance. We propose that calibration can be operationalized in three ways. Perceptual accuracy is the extent to which the user perceives the aid’s reliability accurately at one point in time. Perceptual sensitivity and trust sensitivity reflect user adjustment of perceived reliability and trust as the aid’s actual reliability changes over time. Method: One hundred fifty-five students completed an X-ray screening task with an automated screener. Awareness of the aid’s accuracy trajectory and error type was examined as predictors, and task performance and aid failure detection were examined as outcomes. Results: Awareness of accuracy trajectory was significantly associated with all three operationalizations of calibration, but awareness of error type was not when considered in conjunction with accuracy trajectory. Contrary to expectations, only perceptual accuracy was significantly associated with task performance and failure detection, and combined, the three operationalizations accounted for only 9% and 4% of the variance in these outcomes, respectively. Conclusion: Our results suggest that the potential importance of trust calibration warrants further examination. Moderators may exist. Application: Users who were better able to perform the task unaided were better able to identify and correct aid failure, suggesting that user task training and expertise may benefit human–automation performance.

The Red Effect, Anxiety, and Exam Performance A Multistudy Examination

Laboratory studies have established a negative relationship between the color red and academic performance. This research examined whether this effect would generalize to classroom performance and whether anxiety and negative affect might mediate the effect. In two studies, students taking classroom exams were randomly assigned an exam color. We found no significant effects for color on performance or expected performance and no evidence supporting a significant link between red and either anxiety or affect. We found no significant moderation effects for perceived exam difficulty, actual item difficulty, or anxiety. These results suggest that the color effects may account for only 2–4% of the variance in exam performance. Nevertheless, small effects may have large-scale implications across time. We provide recommendations for research and teaching practice.

Idiographic Anger Ratings by Female Dementia Family Caregivers: Generalizability and Preliminary Validity

The generalizability and validity of idiographic anger ratings were investigated among female dementia family caregivers (N = 65). The longitudinal (2 weeks) and multi-method measurement approach included standardized measures of anger/hostility, affect ratings elicited by caregiving events, and criticism in audiotaped speech samples. In an application of generalizability theory, variance components of idiographic anger ratings were examined. Ratings of anger elicited by specific caregiving events showed generalizability across time, event, and mood adjectives and were related to verbal criticisms displayed towards the patient and use of psychoactive medications. Intercorrelations among standardized anger measures were quite low. Results from this study indicate a strong need for further measurement development related to the experiences and expressions of anger in family caregivers, and caution by clinicians in selecting tools for practice settings.

Perceptions of family caring and its impact on peer associations and drug involvement among rural dwelling African American and White adolescents.

The purpose of this study was to investigate the premise that adolescent perceptions of family caring are a precipitating source of substance use deterrence. More specifically, this study examined the role of family caring on communication of substance use harm and sanctions of use and the effect of these on peer substance involvement and individual use outcomes. A sample of rural dwelling African American and White 7th and 8th grade students (N = 1780) was assessed through self-report. It was anticipated that family caring would be positively related to harm communication and sanctions of use, and that these would be negatively related to peer substance involvement and individual use. Results suggest that family caring was positively linked to harm communication and sanctions of use, and that these were both negatively related to peer substance involvement and individual use. Several significant race differences were noted, which suggest differential associations between some variables. Results are discussed in terms of these race differences, as well as in terms of rural residency.

Differential Item Functioning in Biodata: Opportunity Access as an Explanation of Gender- and Race-Related DIF

Investigations of differential item functioning (DIF) have been conducted mostly on ability tests and have found little evidence of easily interpretable differences across various demographic subgroups. In this study, we examined the degree to which DIF in biographical data items referencing academically relevant background, experiences, and interests was related to differences in judgments about access to these experiences by members of different gender and race subgroups. DIF in the location parameter was significantly related (r = –.51, p < .01) to gender differences in perceived accessibility to experience. No significant relationships with accessibility were observed for DIF in the slope parameter across gender groups or for the slope and location parameters associated with DIF across Black and White groups. Practical implications for use of biodata and theoretical implications for DIF research are discussed.