No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Process Components of the Implicit Association Test: A Diffusion-Model Analysis
Abstract
The authors present a diffusion-model analysis of the Implicit Association Test (IAT). In Study 1, the IAT effect was decomposed into 3 dissociable components: Relative to the compatible phase, (a) ease and speed of information accumulation are lowered in the incompatible phase, (b) more cautious speed-accuracy settings are adopted, and (c) nondecision components of processing require more time. Studies 2 and 3 assessed the nature of interindividual differences in these components. Construct-specific variance in the IAT relating to the construct to be measured (such as implicit attitudes) was concentrated in the compatibility effect on information accumulation (Studies 2 and 3), whereas systematic method variance in the IAT was mapped on differential speed-accuracy settings (Study 3). Implications of these dissociations for process theories of the IAT and for applications are discussed.
... As we note below, there have been efforts to dissect performance into multiple dimensions with models like the diffusion model (Röhner & Lai, 2021;Klauer et al., 2007), but even these models suffer from an additional issue related to test-retest reliability. Specifically, they quantify performance on IATs using a difference score or comparison between conditions. ...
... We posit that the controversies surrounding the reliability and validity of IATs are intractable until researchers embrace modeling approaches that can decompose the individuallevel behaviors into unique components that are both reliably quantifiable and theoretically-grounded (e.g., Klauer et al., 2007). Doing so will allow researchers to gain new insight into which specific aspects of IAT task performance can be reliably captured across repeated measurements (i.e., testretest reliability), and whether the unique parameters can predict outcomes above and beyond typical D-scores (i.e., predictive validity). ...
... Ratcliff & McKoon, 2008). A basic version of the diffusion model (i.e., not including start point bias or cross-trial variability) estimates parameters related to three processes underlying IAT performance, including the speed at which information is gathered (drift), speed-accuracy tradeoffs (thresholds), and nondecision time (Röhner & Lai, 2021;Röhner & Ewers, 2016;Klauer et al., 2007). The accumulation speed parameter or drift represents how easily stimulus information is processed before deciding which key to press on each trial, and it is intended to reflect the relatively automatic associative processes between targets and responses that researchers typically hope to measure with IATs. ...
The Implicit Association Test (IAT), like many behavioral measures, seeks to quantify meaningful individual differences in cognitive processes that are difficult to assess with approaches like self-reports. However, much like other behavioral measures, many IATs appear to show low test-retest reliability and typical scoring methods fail to quantify all of the decision-making processes that generate the overt task performance. Here, we develop a new modeling approach for IATs based on the geometric similarity representation (GSR) model. This model leverages both response times and accuracy on IATs to make inferences about representational similarity between the stimuli and categories. The model disentangles processes related to response caution , stimulus encoding, similarities between concepts and categories, and response processes unrelated to the choice itself. This approach to analyzing IAT data illustrates that the unrelia-bility in IATs is almost entirely attributable to the methods used to analyze data from the task: GSR model parameters show test-retest reliability around .80-.90, on par with reliable self-report measures. Furthermore, we demonstrate how model parameters result in greater validity compared to the IAT D-score, Quad model, and simple diffusion model contrasts, predicting outcomes related to intergroup contact and motivation. Finally, we present a simple point-and-click software tool for fitting the model, which uses a pre-trained neural network to estimate best-fit parameters of the GSR model. This approach allows easy and instantaneous fitting of IAT data with minimal demands on coding or technical expertise on the part of the user, making the new model accessible and effective.
... Ratcliff & McKoon, 2008). The diffusion model estimates three processes underlying IAT performance including the speed at which information is gathered, speed-accuracy tradeoffs, and nondecision time (Röhner & Lai, 2021;Klauer et al., 2007). The accumulation speed parameter or drift represents how easily stimulus information is processed before deciding which key to press on each trial, and it is intended to reflect the relatively automatic associative processes between targets and responses that researchers typically hope to measure with the IAT. ...
... As with the ReAL model, most diffusion model analyses are too inefficient to be applied with less than 90 trials per condition (Röhner & Ewers, 2016;Klauer et al., 2007). Accordingly, diffusion models on standard IAT data are limited to estimating parameters that do not directly incorporate trial-level information (Röhner & Thoss, 2018;Röhner & Lai, 2021). ...
... This approach allows us to characterize both individual differences in performance (e.g., association strengths) alongside group-level trends (e.g., differences in response caution between conditions, general tendencies toward anti-Black associations), as well as to estimate the covariance and uncertainty in performance across multiple testing sessions. By virtue of using hierarchical Bayesian estimation, we did not require the large volume of data that other dynamic modeling approaches do (Röhner & Lai, 2021;Klauer et al., 2007). As a result, this model is widely applicable to the deep IAT literature whose foundation is built on traditional paradigms with only 60 total trials for the compatible or incompatible conditions (and only 20 or 40 per testing block within each condition), rather than restricting our inferences to a limited set of specific data sets featuring a large number of trials or heavy time pressure to induce mistakes (Calanchini & Sherman, 2013). ...
The Implicit Association Test [IAT], like many behavioral measures, seeks to quantify meaningful individual differences in cognitive processes that are difficult to assess with approaches like self-reports. However, much like other behavioral measures, the IAT appears to show low test-retest reliability and typical scoring methods fail to quantify all of the decision-making processes that generate the overt task performance. Here, we develop a new modeling approach for the IAT, the CAVEAT model, that leverages both response times and accuracy on the task to make inferences about representational similarity between the stimuli and categories, as in computational linguistic models of representation. The model disentangles processes related to cognitive control, stimulus encoding, associations between concepts and categories, and processes unrelated to the choice itself. This approach to analyzing IAT data illustrates that the unreliability in the IAT is almost entirely attributable to the methods used to analyze data from the task: the model parameters show test-retest reliability around .8-.9, on par with that of many of the most reliable self-report measures. Furthermore, we demonstrate how model parameters are better and more unbiased compared to the IAT D-score in predicting outcomes related to intergroup contact and motivation. Put together, the model provides much greater reliability, discriminant and predictive validity, and the ability to make inferences about processes like associations and response caution that are not otherwise possible. We conclude by reviewing new, model-based insights about the IAT related to awareness, strategic caution, faking, and the role of associations in decision-making.
... We used the R code provided by Röhner and Thoss (2019) to compute the D 2 algorithm suggested by Greenwald et al. (2003aGreenwald et al. ( , 2003b as a measure of the IAT effect. In addition, we calculated the diffusion-model-based IAT effect IAT v (Klauer et al., 2007) by subtracting parameter v of the compatible phase from parameter v of the incompatible phase. For diffusion modeling, we followed the tutorial by Röhner and Thoss (2018) and used the EZ software, which can be downloaded (http:// www. ...
... IAT a and IAT t Both indices were computed using the diffusion model analyses (e.g., Klauer et al., 2007;Röhner & Ewers, 2016) that we explained above. IAT a represents participants' speed-accuracy tradeoffs and was computed by subtracting parameter a of the compatible phase from parameter a of the incompatible phase, whereas IAT t 0 represents participants' non-decision-related processes and was computed by subtracting parameter t 0 of the compatible phase from parameter t 0 of the incompatible phase (Klauer et al., 2007). ...
... IAT a and IAT t Both indices were computed using the diffusion model analyses (e.g., Klauer et al., 2007;Röhner & Ewers, 2016) that we explained above. IAT a represents participants' speed-accuracy tradeoffs and was computed by subtracting parameter a of the compatible phase from parameter a of the incompatible phase, whereas IAT t 0 represents participants' non-decision-related processes and was computed by subtracting parameter t 0 of the compatible phase from parameter t 0 of the incompatible phase (Klauer et al., 2007). ...
Research has shown that even experts cannot detect faking above chance, but recent studies have suggested that machine learning may help in this endeavor. However, faking differs between faking conditions, previous efforts have not taken these differences into account, and faking indices have yet to be integrated into such approaches. We reanalyzed seven data sets (N = 1,039) with various faking conditions (high and low scores, different constructs, naïve and informed faking, faking with and without practice, different measures [self-reports vs. implicit association tests; IATs]). We investigated the extent to which and how machine learning classifiers could detect faking under these conditions and compared different input data (response patterns, scores, faking indices) and different classifiers (logistic regression, random forest, XGBoost). We also explored the features that classifiers used for detection. Our results show that machine learning has the potential to detect faking, but detection success varies between conditions from chance levels to 100%. There were differences in detection (e.g., detecting low-score faking was better than detecting high-score faking). For self-reports, response patterns and scores were comparable with regard to faking detection, whereas for IATs, faking indices and response patterns were superior to scores. Logistic regression and random forest worked about equally well and outperformed XGBoost. In most cases, classifiers used more than one feature (faking occurred over different pathways), and the features varied in their relevance. Our research supports the assumption of different faking processes and explains why detecting faking is a complex endeavor.
... We focus on a specific decision task that the diffusion model has repeatedly been applied to: the Implicit Association Test (IAT; Greenwald and Farnham 2000;Greenwald et al. 1998Greenwald et al. , 2003Klauer et al. 2007). In the IAT, participants make binary decisions, typically classifying presented stimuli into one of two categories. ...
... When applying the diffusion model to the IAT, differences in performance can be decomposed into differences in speed of information processing (ν), differences in decision caution (a), and differences in non-decision time (τ). Previous studies have shown that the IAT effect can mostly be attributed to differences in ν that are strongly linked to the D scores usually employed to estimate the IAT effect (Klauer et al. 2007). At the same time, there were also differences in a and τ for the congruent and incongruent blocks (Klauer et al. 2007;van Ravenzwaaij et al. 2011). ...
... Previous studies have shown that the IAT effect can mostly be attributed to differences in ν that are strongly linked to the D scores usually employed to estimate the IAT effect (Klauer et al. 2007). At the same time, there were also differences in a and τ for the congruent and incongruent blocks (Klauer et al. 2007;van Ravenzwaaij et al. 2011). Thus, the IAT could be an interesting example to study the stability and change in diffusion model parameters. ...
In recent years, mathematical models of decision making, such as the diffusion model, have been endorsed in individual differences research. These models can disentangle different components of the decision process, like processing speed, speed–accuracy trade-offs, and duration of non-decisional processes. The diffusion model estimates individual parameters of cognitive process components, thus allowing the study of individual differences. These parameters are often assumed to show trait-like properties, that is, within-person stability across tasks and time. However, the assumption of temporal stability has so far been insufficiently investigated. With this work, we explore stability and change in diffusion model parameters by following over 270 participants across a time period of two years. We analysed four different aspects of stability and change: rank-order stability, mean-level change, individual differences in change, and profile stability. Diffusion model parameters showed strong rank-order stability and mean-level changes in processing speed and speed–accuracy trade-offs that could be attributed to practice effects. At the same time, people differed little in these patterns across time. In addition, profiles of individual diffusion model parameters proved to be stable over time. We discuss implications of these findings for the use of the diffusion model in individual differences research.
... Additionally, we repeated all relevant analyses using IAT scores based on a diffusion-model decomposition of the response-time data (Klauer et al., 2007). The latter exhaustively exploits the joint information in response times and classification errors by means of fitting a theoretical model of a decision process to the responsetime distributions. ...
... Therefore, our findings call into question the utility of the IAT to predict suicidal ideation or behavior in inpatient samples of moderate or high suicide risk. However, given the potential methodological problems with the IAT discussed below as well as in the literature (e.g., compatibility-order effects: Greenwald et al., 2003;Rath et al., 2018; task-switching ability, speed-accuracy trade-off: Klauer et al., 2007Klauer et al., , 2010, further research addressing these effects appears to be warranted. Furthermore, the low correlation of IAT scores at T 0 and T 3 should be discussed. ...
... In addition to compatibility order, several cognitive factors have been shown to bias IAT scores, including the task-switching ability (Back et al., 2005;Klauer et al., 2010) and individual differences in the speed-accuracy trade-off (i.e., sacrificing speed for the accuracy or vice versa; e.g., Klauer et al., 2007). To control for the latter, we reanalyzed data from both studies with the diffusion model, which yielded highly comparable results as those obtained with conventional IAT scores (see SOM). ...
Assessment of implicit self-associations with death, measured by a death Implicit Association Test (IAT), has shown promise for the prediction of suicide risk. The present study examined whether the performance on the death IAT is associated with lifetime, recent, or future suicide attempt status as well as self-report measures of suicide risk factors (e.g., perceived burdensomeness, thwarted belongingness) in two inpatient samples with low versus high severity of suicidality. Furthermore, we investigated whether explicit suicidal ideation and implicit associations with death predict recent and future suicide attempt status. Seventy-one depressed inpatients with recent/lifetime suicidal ideation (first sample) as well as 226 inpatients with a recent suicide attempt or a severe suicidal crisis (second sample) were interviewed on lifetime suicidal ideation and behavior, completed self-report measures (i.e., suicidal ideation, thwarted belongingness, perceived burdensomeness), and conducted the death IAT. The second sample was also interviewed and completed self-report measures longitudinally, 6, 9, and 12 months later. The IAT was conducted twice in this sample, at the beginning of the assessment (T₀) as well as 12 months later (T₃). Implicit associations with death neither differ between lifetime suicide ideators, single attempters, and multiple attempters, nor between recent and future nonattempters and attempters. IAT scores were unrelated to other suicide risk factors. Neither the IAT scores nor the interaction of IAT scores and explicitly stated suicidal ideation was predictive of recent or future suicide attempts. The present study points to a limited utility of the death IAT for the prediction of suicide risk. (PsycInfo Database Record (c) 2021 APA, all rights reserved).
... Not only is it complicated to infer operating principles from task performance, but each task engages multiple processes/operating principles. Though indirect measures are largely treated strictly as measures of associative processes, it is now clear that they reflect a variety of additional processes, including the inhibition of associative biases (Bartholow et al., 2006;Stahl and Degner, 2007;Sherman et al., 2008;Moskowitz and Li, 2011), the detection of appropriate responses (Payne, 2001;Correll et al., 2002;Amodio et al., 2004;Klauer et al., 2007;Sherman et al., 2008;Krieglmeyer and Sherman, 2012;Meissner and Rothermund, 2013), response biases (e.g., Klauer et al., 2007;Stahl and Degner, 2007;Sherman et al., 2008;Krieglmeyer and Sherman, 2012), bias correction processes (e.g., Krieglmeyer and Sherman, 2012), stimulus recoding (e.g., Rothermund and Wentura, 2004;Kinoshita and Peek-O'Leary, 2005;Chang and Mitchell, 2011;Meissner and Rothermund, 2013), misattribution processes (Payne et al., 2005;Payne et al., 2010), task-set shifts and task-set simplification Klauer, 2001, 2003), and speed-accuracy tradeoffs (e.g., Brendl et al., 2001;Klauer et al., 2007). Thus, outcomes on any indirect (and direct) measure reflect the ongoing interplay of a variety of cognitive processes, and those outcomes cannot, on their own, reveal the nature of the underlying processes that produced the outcomes. ...
... Not only is it complicated to infer operating principles from task performance, but each task engages multiple processes/operating principles. Though indirect measures are largely treated strictly as measures of associative processes, it is now clear that they reflect a variety of additional processes, including the inhibition of associative biases (Bartholow et al., 2006;Stahl and Degner, 2007;Sherman et al., 2008;Moskowitz and Li, 2011), the detection of appropriate responses (Payne, 2001;Correll et al., 2002;Amodio et al., 2004;Klauer et al., 2007;Sherman et al., 2008;Krieglmeyer and Sherman, 2012;Meissner and Rothermund, 2013), response biases (e.g., Klauer et al., 2007;Stahl and Degner, 2007;Sherman et al., 2008;Krieglmeyer and Sherman, 2012), bias correction processes (e.g., Krieglmeyer and Sherman, 2012), stimulus recoding (e.g., Rothermund and Wentura, 2004;Kinoshita and Peek-O'Leary, 2005;Chang and Mitchell, 2011;Meissner and Rothermund, 2013), misattribution processes (Payne et al., 2005;Payne et al., 2010), task-set shifts and task-set simplification Klauer, 2001, 2003), and speed-accuracy tradeoffs (e.g., Brendl et al., 2001;Klauer et al., 2007). Thus, outcomes on any indirect (and direct) measure reflect the ongoing interplay of a variety of cognitive processes, and those outcomes cannot, on their own, reveal the nature of the underlying processes that produced the outcomes. ...
... Not only is it complicated to infer operating principles from task performance, but each task engages multiple processes/operating principles. Though indirect measures are largely treated strictly as measures of associative processes, it is now clear that they reflect a variety of additional processes, including the inhibition of associative biases (Bartholow et al., 2006;Stahl and Degner, 2007;Sherman et al., 2008;Moskowitz and Li, 2011), the detection of appropriate responses (Payne, 2001;Correll et al., 2002;Amodio et al., 2004;Klauer et al., 2007;Sherman et al., 2008;Krieglmeyer and Sherman, 2012;Meissner and Rothermund, 2013), response biases (e.g., Klauer et al., 2007;Stahl and Degner, 2007;Sherman et al., 2008;Krieglmeyer and Sherman, 2012), bias correction processes (e.g., Krieglmeyer and Sherman, 2012), stimulus recoding (e.g., Rothermund and Wentura, 2004;Kinoshita and Peek-O'Leary, 2005;Chang and Mitchell, 2011;Meissner and Rothermund, 2013), misattribution processes (Payne et al., 2005;Payne et al., 2010), task-set shifts and task-set simplification Klauer, 2001, 2003), and speed-accuracy tradeoffs (e.g., Brendl et al., 2001;Klauer et al., 2007). Thus, outcomes on any indirect (and direct) measure reflect the ongoing interplay of a variety of cognitive processes, and those outcomes cannot, on their own, reveal the nature of the underlying processes that produced the outcomes. ...
In this article, we describe four theoretical and methodological problems that have impeded implicit attitude research and the popular understanding of its findings. The problems all revolve around assumptions made about the relationships among measures (indirect vs. versus direct), constructs (implicit vs. explicit attitudes), cognitive processes (e.g., associative vs. propositional), and features of processing (automatic vs. controlled). These assumptions have confused our understandings of exactly what we are measuring, the processes that produce implicit evaluations, the meaning of differences in implicit evaluations across people and contexts, the meaning of changes in implicit evaluations in response to intervention, and how implicit evaluations predict behavior. We describe formal modeling as one means to address these problems, and provide illustrative examples. Clarifying these issues has important implications for our understanding of who has particular implicit evaluations and why, when those evaluations are likely to be particularly problematic, how we might best try to change them, and what interventions are best suited to minimize the effects of implicit evaluations on behavior.
... On incompatible trials, the difficulty of the task increases because cognitive processes drive decision processes in conflicting directions due to the co-activation of Black and Bad that pulls categorization to both Black+Good and White+Bad. As evidence of validity, parameter v was found to be related to explicit attitudes while the others were not (Klauer et al., 2007). ...
... The higher the a, the more information the participant sampled before deciding. Underpinning its ability to assess speed-accuracy tradeoffs, parameter a was found to be related to method-specific variance in the IAT (Klauer et al., 2007). ...
... Parameter t0 (non-decision-related processes). This parameter assesses all processes that precede and follow decision-making but are not involved in the actual decisionmaking process (Klauer et al., 2007). Thus, t0 reflects the encoding of stimuli, response output processes, and motor responses (Lerche & Voss, 2017;Ratcliff et al., 2016;Voss et al., 2004). ...
p>Performance on implicit measures reflects construct-specific and non-construct-specific processes. This creates an interpretive issue for understanding interventions to change implicit measures: change in performance could reflect changes in the constructs-of-interest or changes in other mental processes. We re-analyzed data from six studies ( N = 23,342) to examine the process-level effects of 17 interventions and one sham intervention to change race Implicit Association Test (IAT) performance. Diffusion models decompose overall IAT performance ( D -scores) into construct-specific (ease of decision-making), and non-construct-specific processes (speed-accuracy tradeoffs, non-decision-related processes like motor execution). Interventions that effectively reduced D- scores changed ease of decision-making on compatible and incompatible trials. They also eliminated differences in speed-accuracy tradeoffs between compatible and incompatible trials. Non-decision-related processes were impacted by two interventions only. There was little evidence that interventions had any long-term effects. These findings highlight the value of diffusion modeling for understanding the mechanisms by which interventions affect implicit measure performance.</p
... On incompatible trials, the difficulty of the task increases because cognitive processes drive decision processes in conflicting directions due to the co-activation of Black and Bad that pulls categorization to both Black+Good and White+Bad. As evidence of validity, parameter v was found to be related to explicit attitudes while the others were not (Klauer et al., 2007). ...
... The higher the a, the more information the participant sampled before deciding. Underpinning its ability to assess speed-accuracy tradeoffs, parameter a was found to be related to method-specific variance in the IAT (Klauer et al., 2007). ...
... Parameter t0 (non-decision-related processes). This parameter assesses all processes that precede and follow decision-making but are not involved in the actual decisionmaking process (Klauer et al., 2007). Thus, t0 reflects the encoding of stimuli, response output processes, and motor responses (Lerche & Voss, 2017;Ratcliff et al., 2016;Voss et al., 2004). ...
Performance on implicit measures reflects construct-specific and non-construct-specific processes. This creates an interpretive issue for understanding interventions to change implicit measures: change in performance could reflect changes in the constructs-of-interest or changes in other mental processes. We re-analyzed data from six studies (N = 23,342) to examine the process-level effects of 17 interventions and one sham intervention to change race Implicit Association Test (IAT) performance. Diffusion models decompose overall IAT performance (D-scores) into construct-specific (ease of decision-making), and non-construct-specific processes (speed-accuracy tradeoffs, non-decision-related processes like motor execution). Interventions that effectively reduced D-scores changed ease of decision-making on compatible and incompatible trials. They also eliminated differences in speed-accuracy tradeoffs between compatible and incompatible trials. Non-decision-related processes were impacted by two interventions only. There was little evidence that interventions had any long-term effects. These findings highlight the value of diffusion modeling for understanding the mechanisms by which interventions affect implicit measure performance.
... A promising route in this direction is formal modeling. In this line of research, Klauer et al. (2007) proposed a diffusion model (DM) analysis of the IAT, whereas other authors proposed models specifically designed for the IAT, namely the Quad model (Conrey et al., 2005), the ReAL model (Meissner & Rothermund, 2013), and the aforementioned DAM . ...
... This led to the deletion of 6.84% of all the responses (from 1.63% to 14.67% per participant). Following Studies 2 and 3 (Klauer et al., 2007), two DMs were estimated for each participant, one on the data from the two blocks Pepsi-bad/Coca-good and the other on the data from the two blocks Coca-bad/Pepsi-good. Parameter z was set equal to a/2, that is, equal to the position corresponding to the absence of response bias (Klauer et al., 2007). Maximumlikelihood estimates of the parameters of the DM were computed using the software Fast-DM , 2008. ...
... This led to the deletion of 6.84% of all the responses (from 1.63% to 14.67% per participant). Following Studies 2 and 3 (Klauer et al., 2007), two DMs were estimated for each participant, one on the data from the two blocks Pepsi-bad/Coca-good and the other on the data from the two blocks Coca-bad/Pepsi-good. Parameter z was set equal to a/2, that is, equal to the position corresponding to the absence of response bias (Klauer et al., 2007). Maximumlikelihood estimates of the parameters of the DM were computed using the software Fast-DM , 2008. ...
The discrimination-association model (DAM; Stefanutti et al. 2013) disentangles two components underlying the responses to the implicit association test (IAT), which pertain to stimuli discrimination (the strength of the association of the stimuli with their own category) and automatic association (the strength of the association between targets and attributes). The assumption of the DAM that these two components sum into a single process generates critical drawbacks. The present work provides a new formulation of the model, called DAM-4C, in which stimuli discrimination and automatic association are separate, independent, and competing processes. Results of theoretical and simulation studies suggest that the DAM-4C outperforms the DAM. The IAT effect is found to vary with the association rates of the DAM-4C and not with those of the DAM. The parameters of the DAM-4C fitted on data from a Coca-Pepsi IAT are found to account for variance in brand attractiveness, taste preference, and cola choice that is not accounted for by the D score and the diffusion model. In addition, the association rates estimated on data from a Black-White IAT are in line with expectations.
... Although the construct-related validity of IATs for measuring automatic associations has been documented in a number of studies (e.g., Bar-Anan & Nosek, 2014;Greenwald et al., 1998), confounds in IAT effects, such as task-switching costs (e.g., Mierke & Klauer, 2001), figure-ground asymmetries (e.g., Rothermund & Wentura, 2001), and items' cross-categories (e.g., Steffens & Plewe, 2001) have also been revealed. Such methodspecific variance is combined with construct-specific variance in traditional IAT effects, although both types of variances are actually based on different processes (e.g., Klauer et al., 2007). Further, IATs and explicit measures (e.g., self-reports) are highly correlated (Schimmack, 2021) and they can even be used as indicators of the same latent construct, a fact that seems to contradict the claim that IATs have discriminant validity regarding explicit measures. ...
... Several models have been suggested to disentangle the underlying processes in IATs. For example, the Quadruple process model (Quad model; Conrey et al., 2005), and the diffusion model (e.g., Klauer et al., 2007;Röhner & Ewers, 2016b) have been applied to IATs. 3 Disentangling IAT-related processes may help improve the validity of IATs, thereby helping to cure them further. ...
... Longer non-decision times have been observed in switch trials as compared to taskrepeat trials in several studies whenever the new task set cannot be prepared in advance, either because it cannot be predicted with certainty or because there is not sufficient preparation time (Boag et al. 2021; Ging-Jehli and Ratcliff 2020; Karayanidis et al. 2009;Karayanidis et al. 2010;Klauer et al. 2007;Madden et al. 2009;Miller et al. 2018;Schmi and Voss 2012). Further, the increase in t0 has been shown (Schmi and Voss 2014) to be indeed related to task-switch processes and is not confined to cue-retrieval processes, as postulated by the compound cueing account (Logan and Bundesen 2003;Schneider and Logan 2005). ...
... Further, inertia may exist in terms of persisting activation of inhibition (Schuch and Koch 2003) which could be described as differential levels of task readiness on a dimensional scale. Generally, the mean-effects for the diffusion model parameters can be reconciled with multiple component models of task switching (Mayr and Kliegl 2003;Ruthruff et al. 2001), thereby replicating previous research (Boag et al. 2021;Ging-Jehli and Ratcliff 2020;Karayanidis et al. 2009;Karayanidis et al. 2010;Klauer et al. 2007;Madden et al. 2009;Miller et al. 2018;Schmi and Voss 2012). ...
The task-switching paradigm is deemed a measure of cognitive flexibility. Previous research has demonstrated that individual differences in task-switch costs are moderately inversely related to cognitive ability. However, current theories emphasize multiple component processes of task switching, such as task-set preparation and task-set inertia. The relations of task-switching processes with cognitive ability were investigated in the current study. Participants completed a task-switching paradigm with geometric forms and a visuospatial working memory capacity (WMC) task. The task-switch effect was decomposed with the diffusion model. Effects of task-switching and response congruency were estimated as latent differences using structural equation modeling. Their magnitudes and relations with visuospatial WMC were investigated. Effects in the means of parameter estimates replicated previous findings, namely increased non-decision time in task-switch trials. Further, task switches and response incongruency had independent effects on drift rates, reflecting their differential effects on task readiness. Findings obtained with the figural tasks employed in this study revealed that WMC was inversely related to the task-switch effect in non-decision time. Relations with drift rates were inconsistent. Finally, WMC was moderately inversely related to response caution. These findings suggest that more able participants either needed less time for task-set preparation or that they invested less time for task-set preparation.
... Drift rate was our primary parameter of interest since category advantage in terms of accuracy and response time gains in tasks with a memory component can be most readily attributed to changes in drift rate. However, although conventional accounts of nondecision time have mostly focused on its relevance for perceptual encoding and motor preparation processes, more recent studies showed that nondecision time can also be modulated as a response to Stroop-like interference, task switching, increased working memory demands, or a general increase in task difficulty (see Klauer, Voss, Schmitz, & Teige-Mocigemba, 2007;Schmitz & Voss, 2012;Voss, Rothermund, Gast, & Wentura, 2013). Thus, the manifestation of categorical perception effects when the memory component is eliminated solely in terms of response time gains could potentially be explained by nondecision delays as well. ...
... Furthermore, if drift rates capture typicality effects in color naming, the category boundary (i.e., the point of subjective equality [PSE]) should correspond to the point at which the drift rate is zero given that there is no information favoring either one of the hypotheses (green or blue). In addition, we also tested how nondecision time changes as a function of color typicality to see whether the typicality of the color also influences nondecisional components of behavior, such as preparation and execution of response or encoding of stimuli (see Klauer et al., 2007;Schmitz & Voss, 2012;Voss et al., 2013 for relations of this parameter to increased working memory demands or task difficulty). ...
Cross‐category hues are differentiated easier than otherwise equidistant hues that belong to the same linguistic category. This effect is typically manifested through both accuracy and response time gains in tasks with a memory component, whereas only response times are affected when there is no memory component. This raises the question of whether there is a common generative process underlying the differential behavioral manifestations of category advantage in color perception. For instance, within the framework of noisy evidence accumulation models, changes in accuracy can be readily attributed to an increase in the efficacy of perceptual evidence integration (after controlling for threshold setting), whereas changes in response time can also be attributed to shorter nondecisional delays (e.g., due to facilitated signal detection). To address the latent decision processes underlying category advantage across different behavioral demands, we introduce a decision‐theoretic perspective (i.e., diffusion decision model) to categorical color perception in three complementary experiments. In Experiment 1, we collected data from a binary color naming task (1) to determine the green–blue boundary in our sample and (2) to trace how parameter estimates of interest in the model output change as a function of color typicality. In Experiments 2 and 3, we used same‐different task paradigms (with and without a memory component, respectively) and traced the category advantage in color discrimination in two parameters of the diffusion decision model: nondecision time and drift rate. An increase in drift rate predominantly characterized the category advantage in both tasks. Our results show that improved efficiency in perceptual evidence integration is a common driving force behind different manifestations of category advantage.
... Este incremento genera fatiga, penalizando la validez ecológica. Dicho esto, los Random Walk se consideran modelos a explorar dada su riqueza semántica y, aunque utilizados en los IAT(Klauer et al., 2007), su aceptación no ha sido del todo unánime(De Houwer et al., 2009). Finalmente, se han estudiado otras formas de calcular los parámetros mediante estadística Bayesiana que podrían resultar de interés futuro(Lin & Strickland, 2020) Meissner et al. (2019, de manera similar aKlauer et al., (2007), recomiendan utilizar el modelo de procesamiento multinomial ReAL(Meissner et al., 2019), el cual permite identificar de forma separada la contribución del constructo a medir (por ejemplo, actitudes) y de los procesos de recodificación en un IAT. ...
... Dicho esto, los Random Walk se consideran modelos a explorar dada su riqueza semántica y, aunque utilizados en los IAT(Klauer et al., 2007), su aceptación no ha sido del todo unánime(De Houwer et al., 2009). Finalmente, se han estudiado otras formas de calcular los parámetros mediante estadística Bayesiana que podrían resultar de interés futuro(Lin & Strickland, 2020) Meissner et al. (2019, de manera similar aKlauer et al., (2007), recomiendan utilizar el modelo de procesamiento multinomial ReAL(Meissner et al., 2019), el cual permite identificar de forma separada la contribución del constructo a medir (por ejemplo, actitudes) y de los procesos de recodificación en un IAT. Con el fin de explicar el patrón de respuestas correctas y erróneas en un IAT, los autores identifican tres procesos implicados: recodificación de las categorías target y de las categorías atributo en una representación binaria en el bloque compatible (Re), asociaciones evaluativas de las categorías target (A) y la identificación de la respuesta correcta basada en la etiqueta de consumo de recursos (L). ...
https://www.tdx.cat/handle/10803/672808#page=1
The need to delve deeper into the study of implicit cognition lies in a fact that is widely accepted in the current academic landscape: human beings are not as "sapiens" as is often assumed. A large part of our judgments and decisions are supported by cognitive processes that Kahneman (2011) describes as System 1: automatic, intuitive, unconscious and demanding of low resources, answering questions quickly through associations and similarities, non-statistical, credulous and heuristic. In this line, the development of implicit measures arose as a response to the need to develop tools that would allow better study of System 1. 25 years ago, the irruption of the Implicit Association Test (IAT) was received with particular enthusiasm in the academic community for two reasons. First, the IAT was interpreted as the solution to reduce the effects of the desirability bias inherent in self-reports. Second, it was assumed that the IAT would allow the measurement of implicit psychological constructs, which are difficult to access introspectively (Payne & Gawronski, 2010). However, implicit measures are currently the focus of debate, accumulating criticism at both theoretical and methodological levels (Machery, 2016; Richetin et al., 2015). In this sense, (Corneille & Hütter (2020)go so far as to suggest that, in view of the conceptual confusion linked to the term "implicit", this term should be replaced by one of the terminological alternatives they propose. However, the majority of authors advocate a constructive critical stance, which would optimize the application, treatment and interpretation of implicit measures (Brownstein et al., 2019; Gawronski et al., 2020; Meissner et al., 2019; Van Dessel et al., 2020). It is along these lines that this doctoral work is conceived, which pursues a twofold strategy: (a) to introduce the implicit measures approach in the study of new domains of the study of cognition; and (b) to apply implicit measures in domains of human cognition in which, due to their own characteristics, they may particularly benefit from an implicit approach. To this end, through a series of experiments we examine the potential of these methodologies in areas of psychology that are considered to be of particular interest and novelty: moral cognition, personality disorders, the Uncanny Valley (UV) and cyberpsychology. To the best of our knowledge, our proposal constitutes the first application of implicit measures to cyberpsychology, one of the areas of greatest academic interest in recent years. In response to the current academic debate on the validity of implicit measures, the present work incorporates a critical perspective, with the aim of identifying aspects to be improved in the methodological treatment of data obtained through implicit measures procedures (Van Dessel et al., 2020). Likewise, an exploratory approach is adopted by applying these measures to situations in which implicit preferences are less antagonistic, a strategy that allows us to better explore the empirical limits of these methodologies.
... Our recent work incorporated mouse tracking, deception decisions with DDM, and found correlations between DDM parameters and MT indices in the the decisions . However, the IAT has only been modeled with the traditional DDM (Klauer et al., 2007;Van Ravenzwaaij, van der Maas and Wagenmakers, 2011), which is not able to connect with mouse tracking trajectories, because it mainly focuses on simulating the distribution of RTs and the fitted parameters do not have temporal dynamics. Following Wong's (2007) work, we focuses on temporal neural dynamics of decision processes and accurately simulated the IAT effect, allowing us to combine it with the mouse trajectories and disassociate the sensory information and motor responses. ...
... Behavioral performance on the IAT can be modeled with DDM (Wong, Huk, Shadlen and Wang, 2007;Klauer et al., 2007;Van Ravenzwaaij et al., 2011) consisting of two self-excitatory but mutually inhibitory neural populations (Bedder et al., 2019). These two neural populations code for left and right motor outputs, respectively(see Figure 2(a)). ...
This study assesses the validity of a newly integrated memory detection method, MT-aIAT, which is a combination of the autobiographical Implicit Association Test (aIAT) and the mouse-tracking method. Participants were assigned to steal a credit card and then performed the aIAT while mouse tracker was recording their motor trajectories. Replicating previous work, we found a RT congruency effect. Critically, the mouse trajectories indicate a congruency effect and a block order effect, suggesting the validity of mouse-tracking technique in unraveling real-time measurement of the IAT congruency effect. Lastly, to test the computational modeling in MT-aIAT, we posited a connectionist model combined with the drift-discussion model to simulate participants’ behavioral performance. Our model captures the ubiquitous implicit bias towards the autobiographical event. Implications of the MT-aIAT in identifying autobiographical memories, the combination of MT-aIAT with computational modeling approach were discussed.
... The Diffusion Model (DM; Ratcliff, 1978Ratcliff, , 2013 allows for modelling both response time and accuracy of fast human decisions (Forstmann et al., 2016;Ratcliff et al., 2016). Typical applications are found in experimental psychology, especially in the context of the Two-Alternative-Forced-Choice (2AFC) paradigm (Laming, 1968;Arnold et al., 2015;Aschenbrenner et al., 2016;Dirk et al., 2017;Klauer et al., 2007;Mayerl et al., 2019;Mulder et al., 2010;Park and Starns, 2015;Schubert et al., 2019;Schuch, 2016;Voss et al., 2013;Yap et al., 2015). It assumes a decision process based on the accumulation of evidence triggered by a stimulus until one of two decision boundaries reflecting the two decision options is reached. ...
... The EZ method may be more appropriate in settings, in which the DM's boundaries represent correct and incorrect responses and respondents are unaware of which key is correct and which not and therefore cannot be a priori biased. This might be the case, e.g., in DM applications to the Implicit Association Test as in Klauer et al. (2007). If, in contrast, responses reflect substantive categories, such as "word" and "non-word" in a lexical decision task (Wagenmakers et al., 2008a), or "shoot" and "not shoot" in a first person shooter task (Correll et al., 2015), the presence of a priori bias cannot be excluded. ...
The Ratcliff Diffusion Model has become an important and widely used tool for the evaluation of psychological experiments. Concurrently, numerous programs and routines have appeared to estimate the model's parameters. The present study aims at comparing some of the most widely used tools with special focus on freely available routines (i.e., open source). Our simulations show that (1) starting point and non-decision time were recovered better than drift rate, (2) the Bayesian approach outperformed all other approaches when the number of trials was low, (3) the Kolmogorov-Smirnov and χ 2 approaches revealed more bias than Bayesian or Maximum Likelihood based routines, and (4) EZ produced substantially biased estimates of threshold separation, non-decision time and drift rate when starting point z ≠ a/2. We discuss the implications for the choice of parameter estimation approaches for real data and suggest that if biased starting point cannot be excluded, EZ will produce deviant estimates and should be used with great care.
... One important implication is that if people can be aware of their implicit self-concepts, the simple assumption that indirect measures reflect only automatically activated and introspectively inaccessible mental associations becomes dubious. There is already substantial empirical evidence showing that indirect measures are not process pure (e.g., Klauer, Voss, Schmitz, & Teige-Mocigemba, 2007), and there is initial evidence that what are called associations could be at least in part simple propositions that are automatically activated (De Cuyper et al., 2017). In turn, this implies that indirect measures can be sensitive to simple relational qualifiers, a property that is quite relevant for personality self-concepts. ...
... Second, current multinomial models are focused on accuracy (i.e., errors) but arguably a key portion of the valid variance in a paradigm such as the IAT is reflected in latencies (i.e., response times). Diffusion models instead focus on latencies but require a large number of trials to provide reliable parameter estimates (Klauer et al., 2007). Recent mathematical models have been developed extending the multinomial models to reaction times (Heck & Erdfelder, 2016;Klauer & Kellen, 2018) and a hierarchical Bayesian approach to diffusion models requiring less trials for estimating parameters (Johnson, Hopwood, Cesario, & Pleskac, 2017). ...
During the last decades, dual process models concerning reflective and impulsive pathways to behavior have been applied in many psychological domains, including personality psychology. In this chapter, we review the literature on dual processes approaches and models for the conceptualization and assessment of two broad domains of personality, self-concepts and motives. We first distinguish explicit and implicit constructs as assessed with direct and indirect measures, respectively. We then focus on measures to assess implicit representations of self-concepts and motives, with special attention to reliability and validity. Some advanced issues will also be examined, specifically novel assessment methods and scoring systems for indirect measures, developmental aspects of implicit personality characteristics, and interpersonal extensions of dual process approaches to personality. To conclude, we share some reflections on controversial issues in dual-process personality research, that is the convergence (or lack thereof) among indirect measures and between indirect and direct measures and the debate on unitary versus dualistic theories.
... Readers interested in an extended tutorial can refer elsewhere for descriptions of such models (Guest & Martin, 2020 One extension of the response time models presented here is to add mechanisms to account for not only response times but also response accuracy. Estimates of individual differences related to paradigms such as the IAT can be informed by also quantifying joint distributions of correct or incorrect responses and corresponding response times (e.g., Conrey et al, 2005;Klauer et al., 2007). The diffusion decision model has been leveraged to this end (Klauer et al, 2007;Ratcliff et al, 2016), but precisely estimating the full model requires far more data than is ordinarily collected in the IAT. ...
... Estimates of individual differences related to paradigms such as the IAT can be informed by also quantifying joint distributions of correct or incorrect responses and corresponding response times (e.g., Conrey et al, 2005;Klauer et al., 2007). The diffusion decision model has been leveraged to this end (Klauer et al, 2007;Ratcliff et al, 2016), but precisely estimating the full model requires far more data than is ordinarily collected in the IAT. One way to sidestep this problem for practical applications is to use a simpler model, such as the EZ diffusion model , and then compare parameters between conditions (congruent, incongruent). ...
Behavioral tasks (e.g., Stroop task) that produce replicable group-level effects (e.g., Stroop effect) often fail to reliably capture individual differences between participants (e.g., low test-retest reliability). This “reliability paradox” has led many researchers to conclude that most behavioral tasks cannot be used to develop and advance theories of individual differences. However, these conclusions are derived from statistical models that provide only superficial summary descriptions of behavioral data, thereby ignoring theoretically-relevant data-generating mechanisms that underly individual-level behavior. More generally, such descriptive methods lack the flexibility to test and develop increasingly complex theories of individual differences. To resolve this theory-description gap, we present generative modeling approaches, which involve using background knowledge to specify how behavior is generated at the individual level, and in turn how the distributions of individual-level mechanisms are characterized at the group level—all in a single joint model. Generative modeling shifts our focus away from estimating descriptive statistical “effects” toward estimating psychologically meaningful parameters, while simultaneously accounting for measurement error that would otherwise attenuate individual difference correlations. Using simulations and empirical data from the Implicit Association Test and Stroop, Flanker, Posner Cueing, and Delay Discounting tasks, we demonstrate how generative models yield (1) higher test-retest reliability estimates, and (2) more theoretically informative parameter estimates relative to traditional statistical approaches. Our results reclaim optimism regarding the utility of behavioral paradigms for testing and advancing theories of individual differences, and emphasize the importance of formally specifying and checking model assumptions to reduce theory-description gaps and facilitate principled theory development.
... The temporal separation of control processes, proposed by the MCT approach, can be used to link diffusion model parameters to the distinct phases of task switching. Based on previous diffusion model applications to task-switch paradigms (e.g., Karayanidis et al., 2009;Klauer, Schmitz & Voss, 2012;Voss, Schmitz, & Teige-Mocigemba, 2007), the diffusion model involves three main parameters presumed to be most informative for the study of task switching. In the following paragraphs, we will introduce these three model parameters (Ter, v, and a) as well as reasons for the proposed associations between model parameters and control processes. ...
... In the context of task switching, processes of basic encoding and response execution contribute to the nondecision time component of all trial types in a task-switch paradigm (i.e., pure, no-switch, and switch trials). Moreover, several studies found that differences in the nondecision time components (between no-switch and switch trials) additionally contributed to the local switch costs (Karayanidis et al., 2009;Klauer et al., 2007;Madden et al., 2009;Schmitz & Voss, 2012). For instance, Karayanidis et al. (2009) paired characters such as letters, digits, and nonalphanumeric symbols (either in gray or in color) for a task-switch paradigm that involved three tasks (classifying the compound stimuli either according to the letter, the digit, or the color) and four cues (i.e., repeat, switch-to, switch-away, and a cue not specifying whether a task-switch would occur). ...
We investigated aging effects in a task-switch paradigm with degraded stimuli administered to college students, 61-74 year olds, and 75-89 year olds. We studied switch costs (the performance difference between task-repeat and task-switch trials) in terms of accuracy and mean reaction times (RTs). Previous aging research focused on switch costs in terms of mean RTs (with accuracy at ceiling). Our results emphasize the importance of distinguishing between switch costs indexed by accuracy and by RTs because these measures lead to different interpretations. We used the Diffusion Decision Model (DDM; Ratcliff, 1978) to study the cognitive components contributing to switch costs. The DDM decomposed the cognitive process of task switching into multiple components. Two parameters of the model, the quality of evidence on which decisions were based (drift rate) and the duration of processes outside the decision process (nondecision time component), indexed different sources of switch costs. We found that older participants had larger switch costs indexed by nondecision time component than younger participants. This result suggests age-related deficits in preparatory cognitive processes. We also found group differences in switch costs indexed by drift rate for switch trials with high stimulus interference (stimuli with features relevant for both tasks). This result suggests that older participants have less effective cognitive processes involved in resolving interference. Our findings show that age-related effects in separate components of switch costs can be studied with the DDM. Our results demonstrate the utility of using discrimination tasks with degraded stimuli in conjunction with model-based analyses. (PsycInfo Database Record (c) 2020 APA, all rights reserved).
... To disentangle the contributions of multiple qualitatively distinct processes to implicit measures, theorists have developed formal models that provide quantitative estimates of these processes, including applications of process dissociation (Payne andBishara, 2009), multinomial modelling (Conrey et al., 2005;Meissner and Rothermund, 2013;Stahl and Degner, 2007), and diffusion modelling (Klauer et al., 2007). Overall, using implicit measures to detect implicit schemata is problematic, particularly when a direct link is drawn to action. ...
Cognitive sociology has made significant strides in recent years, prompting an exploration into the micro-level analysis of social phenomena. This article focuses on three aspects of cognitive sociology that remain problematic. A central theme involves the role of decision-making, mainly how cognitive schemata are linked to action and behaviour. Concurrently, there's an intensified focus on the representational forms of implicit and explicit cognition. However, there's a notable gap in aligning these studies with the extensive body of knowledge from behavioural science researchers, raising concerns about the caveats of using implicit measures for implicit schemata. In particular, we will discuss metrics issues, how well they capture unconscious processes, their automaticity, time-resistance, and purity.Additionally, the article delves into the complexities of the group detection problem, emphasising the challenges of identifying the heterogeneity of cognitive schemata and their attribution to groups. On the latter topic, the article introduces CART-based methods and a bottom-up and ecological use of heterogeneity in group detection.
... Other models that can concurrently account for accuracy and time responses have been applied to the IAT data, namely the Diffusion Model (DM; Klauer et al., 2007) and the Discrimination Association Model (DAM; Stefanutti et al., 2013, see also the four-counter DAM; Stefanutti et al., 2020). DM and DAM consider the performance of the respondents at the IAT as the result of different processes, each of which is expressed by its own parameter. ...
The measure obtained from the Implicit Association Test (IAT; Greenwald et al., 1998. DOI: 10.1037/0022-3514.74.6.1464) is often used to predict people’s behaviors. However, it has shown poor predictive ability potentially because of its typical scoring method (the D score), which is affected by the across-trial variability in the IAT data and might provide biased estimates of the construct. Linear Mixed-Effects Models (LMMs) can address this issue while providing a Rasch-like parametrization of accuracy and time responses. In this study, the predictive abilities of D scores and LMM estimates were compared. The LMMs estimates showed better predictive ability than the D score, and allowed for in-depth analyses at the stimulus level that helped in reducing the across-trial variability. Implications of the results and limitations of the study are discussed.
... This finding was further supported by the drift-diffusion model analyses, showing larger non-decision times in task cueonly trials. Non-decision times were shown to reflect switching costs (Ging-Jehli & Ratcliff, 2020;Klauer et al., 2007;Schmitz & Voss, 2012. Accordingly, larger non-decision times in task cue-only trials could indicate an increased effort to get rid of the cued task set compared to induction-task trials, in which the cued task set had to be applied. ...
Recent research demonstrated that mere presentation of a task cue influences subsequent unconscious semantic priming by attentional sensitization of related processing pathways. The direction of this influence depended on task-set dominance. Dominant task sets with a compatible cue-task mapping were supposed to be rapidly suppressed, while weak task sets showed more sustainable activation. Building on this research, we manipulated cue-task compatibility as instance of task-set dominance in two experiments and tested how masked semantic priming was influenced by actually performing the cued task (induction-task trials) or by mere cue presentation (task cue-only trials). In induction-task trials, the results of earlier research were replicated; semantic priming was larger following a semantic induction task compared to a perceptual induction task. In task cue-only trials, priming effects were reversed compared to induction-task trials in both experiments. Priming was larger for a perceptual compared to a semantic task set in task cue-only trials, indicating suppression of task sets following mere cue presentation in preparation for the upcoming lexical decision task. This notion of an inhibition of task sets after mere cue presentation was further supported by switching-related costs and changes of task-set implementation throughout the experiment. The absence of a moderator role of cue-task compatibility for task cue effects on priming in the present study suggests that the precise time course of task-set activation and inhibition in response to task cues as a function of cue-task compatibility might depend on specific experimental settings.
... That said, accuracy-versus latency-based operationalizations of IAT compatibility effects often reveal the same pattern of results (Meissner & Rothermund, 2013). Nevertheless, future research should explore the generalizability of our findings with modeling approaches that rely solely on response latency (Haines et al., 2020), or incorporate both response latency and accuracy (Heck et al., 2018;Klauer et al., 2007;Klauer & Kellen, 2018). The present research is also limited in our reliance only on the race version of the IAT. ...
Implicit measures were initially assumed to assess stable individual differences, but other perspectives posit that they reflect context-dependent processes. This pre-registered research investigates whether the cognitive processes contributing to responses on the race Implicit Association Test are stable over time and reliably measured using multinomial processing tree modeling. We applied two models – the Quad model and the Process Dissociation Procedure – to six datasets (N = 2,036), each collected over two occasions, examined the within-measurement reliability and between-measurement stability of model parameters, and meta-analyzed the results. Parameters reflecting accuracy-oriented processes demonstrate adequate reliability and stability, which suggests these processes are relatively stable within individuals. Parameters reflecting evaluative associations demonstrate poor stability but modest reliability, which suggests that associations are either context-dependent or stable but noisily-measured. These findings suggest that processes contributing to implicit racial bias differ in temporal stability, which has practical implications for predicting behavior using the Implicit Association Test.
... This could be possible with methods such as diffusion modeling. Although not developed to indicate faking, Klauer et al. (2007) suggested that faking affects indices derived from diffusion model analyses. Notably, Röhner and Thoss (2018) and Röhner and Lai (2021) showed that faking was related to changes in participants' speed-accuracy setting (i.e., IAT a ) and in non-decision components such as taskswitching or motor responses (i.e., IAT t 0 ; e.g., Schmitz & Voss, 2012). ...
Research demonstrates that IATs are fakeable. Several indices [either slowing down or speeding up, and increasing errors or reducing errors in congruent and incongruent blocks; Combined Task Slowing (CTS); Ratio 150-10000] have been developed to detect faking. Findings on these are inconclusive but previous studies have used small samples suggesting they were statistically underpowered. Further, the results’ stability, the unique predictivity of indices, the advantage of combining indices, and the dependency of how faking success is computed have still to be examined. Therefore, we reanalyzed a large data set (N = 750) of fakers and non-fakers who completed an extraversion IAT. Results showed faking strategies depend on the direction of faking. It was possible to detect faking of low scores due to slowing down on the congruent block, and somewhat less with CTS – both strategies led to faking success. In contrast, the strategy of increasing errors on the congruent block was observed, but not successful in altering the IAT effect in the desired direction. Fakers of high scores could be detected due to slowing down on the incongruent block, increasing errors on the incongruent block, and with CTS – all three strategies led to faking success. The results proved stable in subsamples and generally across different computations of faking success. Using regression analyses and machine learning, increasing errors had the strongest impact on the classification. Apparently, fakers use various goal-dependent strategies and not all are successful. To detect faking, we recommend combining indices depending on the context (and examining convergence).
... Here, v 1 and v 2 denote the speed of information processing (drift rates) in the two experimental conditions, a 1 and a 2 denote the decision thresholds (boundary separation), and τ c and τ n denote the additive non-decision time constants for correct and incorrect responses. We estimate separate drift rates and boundary separations for the congruent and incongruent conditions, because these parameters have been shown to differ across the IAT conditions in previous studies 59 . In the congruent condition (where, for example, the response categories 'European American' and 'Good' are mapped to the same response button), participants have been found to show higher drift rates and lower boundary separations than in the incongruent condition (mapping ' African American' and 'Good' to the same response button). ...
Response speeds in simple decision-making tasks begin to decline from early and middle adulthood. However, response times are not pure measures of mental speed but instead represent the sum of multiple processes. Here we apply a Bayesian diffusion model to extract interpretable cognitive components from raw response time data. We apply our model to cross-sectional data from 1.2 million participants to examine age differences in cognitive parameters. To efficiently parse this large dataset, we apply a Bayesian inference method for efficient parameter estimation using specialized neural networks. Our results indicate that response time slowing begins as early as age 20, but this slowing was attributable to increases in decision caution and to slower non-decisional processes, rather than to differences in mental speed. Slowing of mental speed was observed only after approximately age 60. Our research thus challenges widespread beliefs about the relationship between age and mental speed.
... Table 1 in Greenwald and Lai (2020) summarizes six such alternative methods. Additionally, there have been multiple proposals for alternative approaches to scoring the data produced by IAT measures (e.g., Conrey et al., 2005;Klauer et al., 2007) or by variants of IAT measures (e.g., Meissner & Rothermund, 2013). Although none of these alternatives is yet established as an improvement over the standard form of the IAT (see Appendix A) or the currently standard scoring algorithm (see Appendix B), there is no reason to conclude that continued or further efforts to improve IAT procedures and scoring methods will be futile. ...
Interest in unintended discrimination that can result from implicit attitudes and stereotypes (implicit biases) has stimulated many research investigations. Much of this research has used the Implicit Association Test (IAT) to measure association strengths that are presumed to underlie implicit biases. It had been more than a decade since the last published treatment of recommended best practices for research using IAT measures. After an initial draft by the first author, and continuing through three subsequent drafts, the 22 authors and 14 commenters contributed extensively to refining the selection and description of recommendation-worthy research practices. Individual judgments of agreement or disagreement were provided by 29 of the 36 authors and commenters. Of the 21 recommended practices for conducting research with IAT measures presented in this article, all but two were endorsed by 90% or more of those who felt knowledgeable enough to express agreement or disagreement; only 4% of the totality of judgments expressed disagreement. For two practices that were retained despite more than two judgments of disagreement (four for one, five for the other), the bases for those disagreements are described in presenting the recommendations. The article additionally provides recommendations for how to report procedures of IAT measures in empirical articles.
... It is possible that (in accordance to the lower drift rates) the task was generally more demanding for Met+-carriers, therefore increasing frustration throughout the experiment. To sum up, according to their hypothesized superficial processing style, Met+ carriers accumulated lexical information including concept meaning more slowly, indicated by decreased drift rates (Voss, Rothermund, et al., 2013;Wentura et al., 2020; for preactivation effects on ν in other domains also see Klauer et al., 2007). Additionally, Met+ carriers formed a decision based on insufficient evidence, possibly due to frustration, indicated by lowered decision thresholds compared with ...
Automatic and strategic processes in semantic priming can be investigated with masked and unmasked priming tasks. Unmasked priming is thought to enable strategic processes due to the conscious processing of primes, while masked priming exclusively depends on automatic processes due to the invisibility of the prime. Besides task properties, interindividual differences may alter priming effects. In a recent study, masked and unmasked priming based on mean response time (RT) and error rate (ER) differed as a function of the BDNF Val66Met polymorphism (Sanwald et al., 2020). The BDNF Val66Met polymorphism is related to the integrity of several cognitive executive functions and might thus influence the magnitude of priming. In the present study, we reanalyzed this data with drift-diffusion models. Drift-diffusion models conjointly analyze single trial RT and ER data and serve as a framework to elucidate cognitive processes underlying priming. Masked and unmasked priming effects were observed for the drift rates ν, presumably reflecting semantic pre-activation. Priming effects on nondecision time t0 were especially pronounced in unmasked priming , suggesting additional conscious processes to be involved in the t0 modulation. Priming effects on the decision thresholds a may reflect a speed-accuracy tradeoff. Considering the BDNF Val66Met polymorphism, we found lowered drift rates and decision thresholds for Met allele carriers, possibly reflecting a superficial processing style in Met allele carriers. The present study shows that differences in cognitive tasks between genetic groups can be elucidated using drift-diffusion modeling.
... In this way, the models are, in fact, specified theories of the mechanisms underlying priming phenomena. The most commonly used types of models in social (and nonsocial) psychological research are signal detection (e.g., Correll et al., 2002;Green & Swets, 1966), process dissociation (e.g., Jacoby, 1991;Payne, 2001), mutinomial processing trees (e.g., Krieglmeyer & Sherman, 2012;Meissner & Rothermund, 2013;Payne, Hall, Cameron, & Bishara, 2010;Riefer & Batchelder, 1988;Sherman et al., 2008), and drift diffusion models (e.g., Klauer, Voss, Schmitz, & Teige-Mocigemba, 2007;Pleskac, Cesario, & Johnson, 2018;Ratcliff, 1978). ...
Failures to replicate high-profile priming effects have raised questions about the reliability of so-called “social priming” phenomena. However, not only are many of the relevant studies not particularly social in nature, but other robust priming effects that are clearly social in nature do not count as social priming. Most importantly, the focus on the supposedly social aspect of the work has obscured factors that help to account for the relative reliability of priming effects. Here, we examine the construct of social priming, describe some simple demonstrations on the role of experimental design in priming reproducibility, and discuss future avenues for building a better understanding of priming. We conclude that the term “social priming” should be laid to rest, and that it is time to move past arguments about the reliability of specific effects and shift our energy to building theories that help us better understand the mechanisms underlying priming effects.
... trial, allowing for trial-based analyses and thereby avoiding information loss. They have been previously applied in the mechanistic analysis of binary choice tasks, e.g., the Implicit Association Test (Klauer et al. 2007;van Ravenzwaaij et al. 2011), and other cognitive bias assessments, e.g., the dotprobe paradigm (Price et al. 2019). ...
Background
Interpretation biases are suggested to be transdiagnostic phenomena, but have rarely been compared across different disorders and current concerns.
Methods
We investigated explicit, decision-based, and more implicit, reaction time-based interpretation bias in individuals with body dysmorphic disorder (BDD; N = 29), social anxiety disorder (SAD; N = 36), generalized anxiety disorder (GAD; N = 22), and non-clinical controls (NC; N = 32), using an adapted Word Sentence Association Paradigm (WSAP).
Results
Results indicated that interpretation bias occurred transdiagnostically, while content-specific bias patterns varied meaningfully across groups. BDD and SAD shared explicit and, more inconsistently, implicit interpretation biases for appearance-related and social situations. The GAD group exhibited an explicit and implicit negative interpretation bias for general situations, and an additional implicit lack of positive bias. Mechanistic Wiener diffusion model analyses revealed that interpretation bias patterns were mainly driven by speeded information uptake, potentially mirroring disorder-specific associative memory organization.
Conclusions
These findings have important implications for understanding interpretation biases as both etiological and treatment factors.
... A valuable development in research using implicit measures is the ongoing trend toward formal models. Whereas early models were designed to disentangle the contribution of multiple distinct processes to responses on particular instruments (e.g., Conrey, Sherman, Gawronski, Hugenberg, & Groom, 2005;Klauer, Voss, Schmitz, & Teige-Mocigemba, 2007;Meissner & Rothermund, 2013;Payne, Hall, Cameron, & Bishara, 2010; for a review, see Sherman, Klauer, & Allen, 2010), the network theory of attitudes provides a broader model of attitudinal processes and representations that goes beyond responses on particular instruments (Dalege et al., 2016;Dalege, Borsboom, van Harreveld, & van der Maas, 2018). Inspired by the notion of entropy in thermodynamics, a key concept of the theory is entropy reduction, in that activation of attitudinal representations is assumed to transition from high entropy states (i.e., unstable, inconsistent) to low entropy states (i.e., stable, consistent). ...
The year 2020 marks the 25th anniversary of two seminal publications that have set the foundation for an exponentially growing body of research using implicit measures: Fazio, Jackson, Dunton, and Williams's (1995) work using evaluative priming to measure racial attitudes, and Greenwald and Banaji's (1995) review of implicit social cognition research that served as the basis for the development of the Implicit Association Test. The current article provides an overview of (1) two conceptual roots that continue to shape interpretations of implicit measures, (2) conflicting interpretations of the term implicit, (3) different kinds of dissociations between implicit and explicit measures, (4) theoretical developments inspired by these dissociations, and (5) research that used implicit measures to address domain-specific and applied questions. We conclude with a discussion of challenges and open questions that remain to be addressed, offering guidance for the next generation of research using implicit measures.
[Open Access: https://guilfordjournals.com/doi/pdf/10.1521/soco.2020.38.supp.s1]
... We will attempt to answer this question in two ways: methodologically and conceptually. Methodologically, much research has shown that implicit evaluations result from a multitude of processes, many of them unique to the method that is used (Conrey, Sherman, Gawronski, Hugenberg, & Groom, 2005;Klauer, Voss, Schmitz, & Teige-Mocigemba, 2007;Rothermund & Wentura, 2004). Testing the contribution of method-specific variance to accuracy in IAT score predictions experimentally, Hahn and colleagues (2014) found that predictions were if anything non-significantly most accurate when participants received no explanation on how the IAT works and had no experience with it. ...
Implicit evaluations are often assumed to reflect "unconscious attitudes". We review data from our lab to conclude that the truth of this statement depends on how one defines "unconscious". A trait definition of unconscious according to which implicit evaluations reflect cognitions that are introspectively inaccessible at all times appears to be inaccurate. However, when unconscious is defined as a state in which cognitions can be in at specific times, some data suggest that the cognitions reflected on implicit evaluations may sometimes unfold without direct awareness in that people seem to rarely pay attention to them. Additionally, people appear to be miscalibrated in their reports in that they construe even conscious biases in self-serving ways. This analysis suggests that implicit evaluations do not reflect unconscious cognitions per se, but awareness-independent cognitions that are often preconscious and miscalibrated. Discussion centers on the meaning of this analysis for theory and application.
... A detailed review of the empirical evidence on the construct representation of IATs is beyond the scope of this commentary. However, we note that evidence on historical precursors of the IAT (e.g., Carlston & Skowronski, 1994;Fazio et al., 1986;Meyer & Schvaneveldt, 1971;Neely, 1976;Nuttin, 1985;Stroop, 1935), formal process models of task performance (e.g., Calanchini, Sherman, Klauer, & Lai, 2014;Conrey, Sherman, Gawronski, Hugenberg, & Groom, 2005;Klauer, Voss, Schmitz, & Teige-Mocigemba, 2007;Meissner & Rothermund, 2013), theoretically relevant manipulations that modulate responding on IATs (e.g., Cone, Mann, & Ferguson, 2017;De Houwer et al., 2020;Gawronski & Bodenhausen, 2006, and known group differences (e.g., Barnes-Holmes et al., 2009;Lindgren et al., 2012;van Harmelen et al., 2010) should also be seen as part of the evidence on construct validity. Crucially, as pointed out above, even if every individual in the world showed the exact same difference in response latency across the two critical blocks, some IATs may still be able to provide a "window into the unconscious" (Schimmack, 2020, proof p. 17♦♦♦). ...
Much of human thought, feeling, and behavior unfolds automatically. Indirect measures of cognition capture such processes by observing responding under corresponding conditions (e.g., lack of intention or control). The Implicit Association Test (IAT) is one such measure. The IAT indexes the strength of association between categories such as "planes" and "trains" and attributes such as "fast" and "slow" by comparing response latencies across two sorting tasks (planes-fast/trains-slow vs. trains-fast/planes-slow). Relying on a reanalysis of multitrait-multimethod (MTMM) studies, Schimmack (this issue) argues that the IAT and direct measures of cognition, for example, Likert scales, can serve as indicators of the same latent construct, thereby purportedly undermining the validity of the IAT as a measure of individual differences in automatic cognition. Here we note the compatibility of Schimmack's empirical findings with a range of existing theoretical perspectives and the importance of considering evidence beyond MTMM approaches to establishing construct validity. Depending on the nature of the study, different standards of validity may apply to each use of the IAT; however, the evidence presented by Schimmack is easily reconcilable with the potential of the IAT to serve as a valid measure of automatic processes in human cognition, including in individual-difference contexts.
... However, MPTs are not the only analytic option available to researchers interested in formally quantifying the contributions of multiple cognitive processes. In addition to MPTs, several other classes of formal models have been profitably applied to response conflict tasks within cognitive and social psychology, including signal detection (e.g., Correll, Park, Judd, & Wittenbrink, 2002;Nosek & Banaji, 2001;Yonelinas, Dobbins, Szymanski, Dhaliwal, & King, 1996), diffusion models (e.g., Klauer, Voss, Schmitz, & Teige-Mocigemba, 2007;Ratcliff, Thapar, Gomez, & McKoon, 2004;Ulrich, Schr€ oter, Leuthold, & Birngruber, 2015;White, Ratcliff, & Starns, 2011), and computational models (e.g., Cohen, Dunbar, & McClelland, 1990;Gilbert & Shallice, 2002;Logan & Cowan, 1984). ...
... As a multinomial model, the ReAL model is able to disentangle multiple cognitive processes accounting for the same observable response (Batchelder and Riefer, 1999). First and foremost, the ReAL model controls for the effects of recoding by measuring them in a separate model parameter (which clearly represents a unique feature as compared to other mathematical models for the IAT; e.g., the quad model, Conrey et al., 2005; or the diffusion model, Klauer et al., 2007). Besides addressing the problem of recoding, the ReAL model comes with another advantage: While IAT scores only reflect relative preferences (which could be problematic; for an overview, see Teige-Mocigemba et al., 2010), the ReAL model provides separate association parameters for each of the two target categories. ...
Two decades ago, the introduction of the Implicit Association Test (IAT) sparked enthusiastic reactions. With implicit measures like the IAT, researchers hoped to finally be able to bridge the gap between self-reported attitudes on one hand and behavior on the other. Twenty years of research and several meta-analyses later, however, we have to conclude that neither the IAT nor its derivatives have fulfilled these expectations. Their predictive value for behavioral criteria is weak and their incremental validity over and above self-report measures is negligible. In our review, we present an overview of explanations for these unsatisfactory findings and delineate promising ways forward. Over the years, several reasons for the IAT’s weak predictive validity have been proposed. They point to four potentially problematic features: First, the IAT is by no means a pure measure of individual differences in associations but suffers from extraneous influences like recoding. Hence, the predictive validity of IAT-scores should not be confused with the predictive validity of associations. Second, with the IAT, we usually aim to measure evaluation (“liking”) instead of motivation (“wanting”). Yet, behavior might be determined much more often by the latter than the former. Third, the IAT focuses on measuring associations instead of propositional beliefs and thus taps into a construct that might be too unspecific to account for behavior. Finally, studies on predictive validity are often characterized by a mismatch between predictor and criterion (e.g., while behavior is highly context-specific, the IAT usually takes into account neither situation nor domain). Recent research, however, also revealed advances addressing each of these problems, namely (1) procedural and analytical advances to control for recoding in the IAT, (2) measurement procedures to assess implicit wanting, (3) measurement procedures to assess implicit beliefs, and (4) approaches to increase the fit between implicit measures and behavioral criteria (e.g., by incorporating contextual information). Implicit measures like the IAT hold an enormous potential. In order to allow them to fulfill this potential, however, we have to refine our understanding of these measures, and we should incorporate recent conceptual and methodological advancements. This review provides specific recommendations on how to do so.
... The DDM has become increasingly popular for modelling RTs in in a wide variety of psychological research (Voss et al., 2013), including purchasing decisions (Krajbich et al., 2012), decisions under high and low pressure (Milosavljevic, Malmaud, Huth, Koch & Rangel, 2010), speed-accuracy trade-offs in visual perception (Zhang & Rowe, 2014), memory retrieval (Ratcliff, 1978;Pearson, Raškevičius, Bays, Pertzov & Husain, 2014), semantic categorisation (Klauer, Voss, Schmitz & Teige-Mocigemba, 2007), visual word-recognition in dyslexia (Zeguers et al., 2011), and inference in ageing populations (McKoon & Ratcliff, 2013). But although the DDM has also been used to model RTs in visual flanker tasks (e.g., Merkt et al., 2013), at the time of writing the work presented below appears to be the first attempt to model RTs in an auditory flanker task. ...
If listening to speech against a background of noise increases listening effort, then the effectiveness of a speech technology designed to reduce background noise could be measured by the reduction in listening effort it provides. Reports of increased listening effort in environments with greater background noise have been linked to accompanying decreases in performance (e.g., slower responses and more errors) which are commonly attributed to the increased demands placed on limited cognitive resources in these challenging listening environments, particularly when performing more than one task. As these cognitive resources are also implicated in maintaining attention and reducing distraction, the work reported here proposes to measure listening effort by measuring changes in distraction while listening to noisy and digitally-noise-reduced speech using an auditory flanker task designed to simulate an everyday situation: listening on the telephone. Over a series of experiments this novel listening effort measure is enhanced by the inclusion of a simultaneous memory task and contrasted with listening effort ratings and conventional speech technology evaluation measures (intelligibility and speech quality). However, while there are indications that increased background noise can increase listening effort and digital noise reduction fails to reverse this effect, the results are not consistent. These equivocal results are discussed in light of the recent surge of interest in listening effort research.
Implicit measures were initially assumed to assess stable individual differences, but other perspectives posit that they reflect context-dependent processes. This pre-registered research investigates whether the processes contributing to responses on the race Implicit Association Test are temporally stable and reliably measured using multinomial processing tree modeling. We applied two models—the Quad model and the Process Dissociation Procedure—to six datasets (N = 2,036), each collected over two occasions, examined the within-measurement reliability and between-measurement stability of model parameters, and meta-analyzed the results. Parameters reflecting accuracy-oriented processes demonstrate adequate stability and reliability, which suggests these processes are relatively stable within individuals. Parameters reflecting evaluative associations demonstrate poor stability but modest reliability, which suggests that associations are either context-dependent or stable but noisily measured. These findings suggest that processes contributing to racial bias on implicit measures differ in temporal stability, which has practical implications for predicting behavior using the Implicit Association Test.
Evidence accumulation models are a series of computational models that provide an account for speeded decision making. These models have been used extensively within the cognitive psychology literature to great success, allowing inferences to be drawn about the psychological processes that underlie cognition that are sometimes not available in a traditional analysis of accuracy or reaction time. Despite this, there have been only a few applications of these models within the domain of social cognition. In this paper we explore several ways in which the study of human social information processing would benefit from application of evidence accumulation modelling. We begin first with a brief overview of the evidence accumulation modelling framework and their past success within the domain of cognitive psychology. We then highlight five ways in which social cognitive research would benefit from an evidence accumulation approach. This includes 1) greater specification of assumptions; 2) unambiguous comparisons across blocked task conditions; 3) quantifying and comparing the magnitude of effects in standardised measures; 4) a novel approach for studying individual differences; and 5) improved reproducibility and accessibility. These points are illustrated using examples from the domain of social attention. Finally, we outline several methodological and practical considerations, which should help researchers use evidence accumulation models productively. Ultimately it will be seen that evidence accumulation modelling offers a well-developed, accessible and commonly understood framework that can reveal inferences about cognition that may otherwise be out of reach in a traditional analysis of accuracy and RT. This approach, therefore, has the potential to substantially revise our understanding of social cognition.
Addressing the damaging effects of implicit stereotypes—spontaneous, awareness‐independent associations between social groups and particular traits—remains a social imperative. These biases have been linked to negative outcomes in settings ranging from the workplace to medical care facilities. However, many techniques found to reduce implicit biases have been shown to yield short‐lived effects. In the present experiment, we assessed the longevity of reduced implicit racial stereotyping resulting from an intensive training technique that focuses on weakening the fundamental attributional processes underlying implicit stereotyping. Specifically, we aimed to strengthen the likelihood of White participants to consider situational attributions for behaviors performed by Black men that might otherwise have been judged to reflect negative African American stereotypes. White participants were randomly assigned to complete either Situational Attribution Training (SAT), a technique comprised of intensive training (480 trials) to “consider the situation” when making judgments about stereotype‐consistent behaviors performed by Black men, or a control task. Implicit stereotyping was assessed 24h later via the Person Categorization Task and found to be reduced for SAT, versus control, participants even after this delay. Implications for future antibias research and practice are considered.
Background:
Implicit alcohol attitudes are considered important in the etiology of drinking, and theory posits reciprocal associations between them. Research testing reciprocal associations between implicit attitudes (using the Implicit Association Task, IAT) and drinking is limited by a failure to consider multiple processes influencing performance on the IAT and to disaggregate within- and between-person effects. The current study addressed these limitations by using a diffusion model to analyze IAT data and Latent Curve Models with Structured Residuals to test reciprocal associations.
Methods:
The sample included 314 emerging adults from the community (52% female; predominantly non-Hispanic Caucasian (76%) or African American (15%)) assessed annually for three years. Differences between IAT conditions in the drift rate parameter of the EZ-diffusion model (vΔ) were used as an alternative to traditional response-time-based indices from the IAT (d-scores). Differences in drift rate have been found to index implicit attitudes effectively.
Results:
Within-person reciprocal associations were supported, but between-person associations were not. Positive implicit alcohol attitudes (vΔ) were prospectively associated with heavy drinking, which was positively associated with subsequent positive implicit alcohol attitudes.
Conclusions:
We found that positive implicit alcohol attitudes and heavy drinking reinforce each other in a negative cascade within individuals. The results highlight the importance of disaggregating within- and between-person prospective effects when testing dual process models and suggest that the diffusion model may be a fruitful approach to enhance the construct validity of IAT assessed implicit attitudes.
Objective:
Correlational research aiming to validate measures and the construct of implicit self-esteem (ISE) has produced heterogeneous results in the past. We argue that this might be caused by two underappreciated obstacles: the situational malleability of and the construct irrelevant variance in conventional ISE measures. In this study, we aim to address these problems.
Methods:
To this end, we applied process and latent state-trait modeling to Implicit Association Test and Name Letter Task data collected on four occasions across six weeks in a preregistered online study (initial N = 360, final N = 302). We investigated the relation of supposed trait ISE parameters with trait explicit self-esteem (ESE) and a set of criteria.
Results:
Results indicated no latent trait correlation among the different supposed indicators of ISE, small latent trait correlations of indicators of ISE and ESE, and little incremental validity of the supposed ISE measures in predicting potential criterion measures over and above ESE.
Conclusions:
These findings align with previous critical evaluations regarding the supposed measures of ISE and the conceptual validity of ISE as an association and call for a more careful terminology in the field.
Implicit self-associations are theorized to be rigidly and excessively negative in affective disorders like depression. Such information processing patterns may be useful as an approach to parsing heterogeneous etiologies, substrates, and treatment outcomes within the broad syndrome of depression. However, there is a lack of sufficient data on the psychometric, neural, and computational substrates of Implicit Association Test (IAT) performance in patient populations. In a treatment-seeking, clinically depressed sample (n = 122), we administered five variants of the IAT-a dominant paradigm used in hundreds of studies of implicit cognition to date-at two repeated sessions (outside and inside a functional MRI scanner). We examined reliability, clinical correlates, and neural and computational substrates of IAT performance. IAT scores showed adequate (.67-.81) split-half reliability and convergent validity with one another and with relevant explicit symptom measures. Test-retest correlations (in vs. outside the functional MRI scanner) were present but modest (.15-.55). In depressed patients, on average, negative implicit self-representations appeared to be weaker or less efficiently processed relative to positive self-representations; elicited greater recruitment of frontoparietal task network regions; and, according to computational modeling of trial-by-trial data, were driven primarily by differential efficiency of information accumulation for negative and positive attributes. Greater degree of discrepancy between implicit and explicit self-worth predicted depression severity. Overall, these IATs show potential utility in understanding heterogeneous substrates of depression but leave substantial room for improvement. The observed clinical, neural, and computational correlates of implicit self-associations offer novel insights into a simple computer-administered task in a clinical population and point toward heterogeneous depression mechanisms and treatment targets. (PsycInfo Database Record (c) 2020 APA, all rights reserved).
Implicit intergroup biases are automatically activated prejudices and stereotypes that may influence judgments of others on the basis of group membership. We review evidence on the measurement of implicit intergroup biases, finding: implicit intergroup biases reflect the personal and the cultural, implicit measures vary in reliability and validity, and implicit measures vary greatly in their prediction of explicit and behavioral outcomes due to theoretical and methodological moderators. We then discuss three challenges to the application of implicit intergroup biases to real‐world problems: (1) a lack of research on social groups of scientific and public interest, (2) developing implicit measures with diagnostic capabilities, and (3) resolving ongoing ambiguities in the relationship between implicit bias and behavior. Making progress on these issues will clarify the role of implicit intergroup biases in perpetuating inequality.
As the attitude concept is integral to many theories across the social sciences the measurement of attitudes has long been an important topic of research. Over the years, a variety of different attitude measurement techniques have been developed, all of which rely on their own theoretical rationale and methodological characteristics. Generally speaking, these testing procedures can be categorized into one of three fundamentally different families: direct, indirect, and physiological. Media psychologists should be aware of the theoretical and methodological differences between these families of attitude measurement, acknowledge their strengths and weaknesses, and deal with the theoretical complexities associated with relying on a particular approach.
The use of Bayesian statistics in psychology has been of recent focus. This paper discusses the effectiveness and usefulness of “Bayesian statistical modeling” in psychological studies by exploring its differences with traditional methodologies used in psychological studies. First, we explain differences between two trends in Bayesian statistics: hypothesis testing using Bayesian statistics and Bayesian statistical modeling. Second, Bayesian estimation and probabilistic programming languages may make it easy for psychologists to use statistical modeling. Third, the three advantages of studies using Bayesian statistical modeling in psychology are demonstrated. These advantages include developing mathematical explanations of behavioral mechanisms, valid estimation of psychological characteristics, and improved transparency and replicability of the data analysis. Fourth, it is argued that Bayesian statistical modeling and traditional psychological methodology could coexist by influencing each other.
In this commentary, we welcome Schimmack’s reanalysis of Bar-Anan and Vianello’s multitrait multimethod (MTMM) data set, and we highlight some limitations of both the original and the secondary analyses. We note that when testing the fit of a confirmatory model to a data set, theoretical justifications for the choices of the measures to include in the model and how to construct the model improve the informational value of the results. We show that making different, theory-driven specification choices leads to different results and conclusions than those reported by Schimmack (this issue, p. ♦♦♦). Therefore, Schimmack’s reanalyses of our data are insufficient to cast doubt on the Implicit Association Test (IAT) as a measure of automatic judgment. We note other reasons why the validation of the IAT is still incomplete but conclude that, currently, the IAT is the best available candidate for measuring automatic judgment at the person level.
This special issue of Cognition and Emotion assembles recent advances in theorizing and empirical research on the automaticity of evaluative learning. Based on a taxonomy of automatic processes in evaluative learning, we distinguish between processes that are involved in translating evaluative experiences into evaluative mental representations (acquisition), and processes that translate these representations into evaluative biases in perception, thought, and action (activation and application). We emphasize that automaticity concerns the operating conditions of these processes (unawareness, unintentionality, uncontrollability, efficiency), not their operating principles, and thus can vary within specific processes (e.g., inferences can occur in either an automatic or non-automatic fashion). We review and discuss contemporary theories and methodological approaches to automatic processes in evaluative learning against the backdrop of our framework, and we highlight the contributions of the papers of this special issue to the question whether and when evaluative changes can occur in an automatic manner.
In 1998, Greenwald, McGhee, and Schwartz proposed that the Implicit Association Test (IAT) measures individual differences in implicit social cognition. This claim requires evidence of construct validity. I review the evidence and show that there is insufficient evidence for this claim. Most important, I show that few studies were able to test discriminant validity of the IAT as a measure of implicit constructs. I examine discriminant validity in several multimethod studies and find little or no evidence of discriminant validity. I also show that validity of the IAT as a measure of attitudes varies across constructs. Validity of the self-esteem IAT is low, but estimates vary across studies. About 20% of the variance in the race IAT reflects racial preferences. The highest validity is obtained for measuring political orientation with the IAT (64%). Most of this valid variance stems from a distinction between individuals with opposing attitudes, whereas reaction times contribute less than 10% of variance in the prediction of explicit attitude measures. In all domains, explicit measures are more valid than the IAT, but the IAT can be used as a measure of sensitive attitudes to reduce measurement error by using a multimethod measurement model.
In the last 20 years, research on implicit social cognition has established that social judgments and behavior are guided by attitudes and stereotypes of which the actor may lack awareness. Research using the methods of implicit social cognition has produced the concept of implicit bias, which has generated wide attention not only in social, clinical, and developmental psychology, but also in disciplines outside of psychology, including business, law, criminal justice, medicine, education, and political science. Although this rapidly growing body of research offers prospects of useful societal applications, the theory needed to confidently guide those applications remains insufficiently developed. This article describes the methods that have been developed, the findings that have been obtained, and the theoretical questions that remain to be answered.
Expected final online publication date for the Annual Review of Psychology, Volume 71 is January 4, 2020. Please see http://www.annualreviews.org/page/journal/pubdates for revised estimates.
We hypothesized that scores on the Implicit Association Test (IAT) are confounded with a general cognitive ability of how quickly one can process information when the IAT categories seem incongruent compared to when they are congruent. Across four studies, two IATs on irrelevant dimensions (e.g., delicious-happy) were substantially correlated with IATs assessing prejudice and self-esteem, confirming the general skill confound: Those who lack this skill are biased toward higher prejudice and lower self-esteem IAT scores. However, IATs with just two exemplars in each category were much less affected by this confound than were those with more exemplars. The cognitive skill was associated specifically with the difficulty in responding when categories were incongruent and to the difficulty of mentally switching sets in the middle of an IAT.
One pragmatic goal of implicit tools like the Implicit Association Tests (IAT) is to rule out self-presentation and controlled responding. Three experiments examined whether the IAT meets this goal, using Turkish and German groups along with positive and negative traits. Experiment 1 was an Internet study. After completing a naïve IAT pretest, participants were instructed to fake on a posttest in 3 graded conditions that differed in the explicitness of faking instructions. Experiment 2 replicated and extended the approach in the laboratory, including a no-pretest condition. Results demonstrate that participants who intended to fake were successful, provided the experience of a pretest. Experiment 3 ruled out an alternative account of faking in terms of pretest experience. Faking was mostly due to slow-down on compatible trials, but a notable speed-up on incompatible trials also occurred. Faking remained inconspicuous, especially with nonblatant instructions; experts failed to identify faked data sets.
Many experiments in psychology yield both reaction time and accuracy data. However, no off-the-shelf methods yet exist for
the statistical analysis of such data. One particularly successful model has been the diffusion process, but using it is difficult
in practice because of numerical, statistical, and software problems. We present a general method for performing diffusion
model analyses on experimental data. By implementing design matrices, a wide range of across-condition restrictions can be
imposed on model parameters, in a flexible way. It becomes possible to fit models with parameters regressed onto predictors.
Moreover, data analytical tools are discussed that can be used to handle various types of outliers and contaminants. We briefly
present an easy-touse software tool that helps perform diffusion model analyses.
Flexible control of action requires the ability to disengage from previous goals or task sets. The authors tested the hypothesis that disengagement during intentional shifts between task sets is accompanied by inhibition of the previous task set ("backward inhibition"). As an expression of backward inhibition the authors predicted increased response times when shifting to a task set that had to be abandoned recently and, thus, suffers residual inhibition. The critical backward inhibition effect on the level of abstractly defined perceptual task sets was obtained across 6 different experiments. In addition, it was shown that backward inhibition can be differentiated from negative priming (Experiment 2), that it is tied to top-down sequential control (Experiment 3), that it can account at least partially for "residual shift costs" in set-shifting experiments (Experiment 4), and that it occurs even in the context of preplanned sequences of task sets (Experiment 5).
There has been significant interest in indirect measures of attitudes like the Implicit Association Test (IAT), presumably because of the possibility of uncovering implicit prejudices. The authors derived a set of qualitative predictions for people's performance in the IAT on the basis of random walk models. These were supported in 3 experiments comparing clearly positive or negative categories to nonwords. They also provided evidence that participants shift their response criterion when doing the IAT. Because of these criterion shifts, a response pattern in the IAT can have multiple causes. Thus, it is not possible to infer a single cause (such as prejudice) from IAT results. A surprising additional result was that nonwords were treated as though they were evaluated more negatively than obviously negative items like insects, suggesting that low familiarity items may generate the pattern of data previously interpreted as evidence for implicit prejudice.
Behavioral scientists have long sought measures of important psychological constructs that avoid response biases and other problems associated with direct reports. Recently, a large number of such indirect, or "implicit," measures have emerged. We review research that has utilized these measures across several domains, including attitudes, self-esteem, and stereotypes, and discuss their predictive validity, their interrelations, and the mechanisms presumably underlying their operation. Special attention is devoted to various priming measures and the Implicit Association Test, largely due to their prevalence in the literature. We also attempt to clarify several unresolved theoretical and empirical issues concerning implicit measures, including the nature of the underlying constructs they purport to measure, the conditions under which they are most likely to relate to explicit measures, the kinds of behavior each measure is likely to predict, their sensitivity to context, and the construct's potential for change.
The effects of aging on decision time were examined in a brightness discrimination experiment with young and older subjects (ages, 60-75 years). Results showed that older subjects were slightly slower than young subjects but just as accurate. Ratcliff's (1978) diffusion model was fit to the data, and it provided a good account of response times, their distributions, and response accuracy. There was a 50-msec slowing of the nondecision components of response time for older subjects relative to young subjects, but response criteria settings and rates of accumulation of evidence from stimuli were roughly equal for the two groups. These results are contrasted with those obtained from letter discrimination and signal-detection-like tasks.
In reporting Implicit Association Test (IAT) results, researchers have most often used scoring conventions described in the first publication of the IAT (A.G. Greenwald, D.E. McGhee, & J.L.K. Schwartz, 1998). Demonstration IATs available on the Internet have produced large data sets that were used in the current article to evaluate alternative scoring procedures. Candidate new algorithms were examined in terms of their (a) correlations with parallel self-report measures, (b) resistance to an artifact associated with speed of responding, (c) internal consistency, (d) sensitivity to known influences on IAT measures, and (e) resistance to known procedural influences. The best-performing measure incorporates data from the IAT's practice trials, uses a metric that is calibrated by each respondent's latency variability, and includes a latency penalty for errors. This new algorithm strongly outperforms the earlier (conventional) procedure.
The effects of aging on response time (RT) are examined in 2 lexical-decision experiments with young and older subjects (age 60-75). The results show that the older subjects were slower than the young subjects, but more accurate. R. Ratcliff s (1978) diffusion model provided a good account of RTs, their distributions, and response accuracy. The fits show an 80-100-ms slowing of the nondecision components of RT for older subjects relative to young subjects and more conservative decision criterion settings for older subjects than for young subjects. The rates of accumulation of evidence were not significantly different for older compared with young subjects (less than 2% and 5% higher for older subjects relative to young subjects in the 2 experiments).
Recently, the role of method-specific variance in the Implicit Association Test (IAT) was examined (McFarland & Crouch, 2002; Mierke & Klauer, 2003). This article presents a new content-unspecific control task for the assessment of task-switching ability within the IAT methodology. Study 1 showed that this task exhibited good internal consistency and stability. Studies 2-4 examined method-specific variance in the IAT and showed that the control task is significantly associated with conventionally scored IAT effects of the IAT-Anxiety. Using the D measures proposed by Greenwald, Nosek, and Banaji (2003), the amount of method-specific variance in the IAT-Anxiety could be reduced. Possible directions for future research are outlined.
The authors argue that implicit measures of social cognition do not reflect only automatic processes but rather the joint contributions of multiple, qualitatively different processes. The quadruple process model proposed and tested in the present article quantitatively disentangles the influences of 4 distinct processes on implicit task performance: the likelihood that automatic bias is activated by a stimulus; that a correct response can be determined; that automatic bias is overcome; and that, in the absence of other information, a guessing bias drives responses. The stochastic and construct validity of the model is confirmed in 5 studies. The model is shown to provide a more nuanced and detailed understanding of the interplay of multiple processes in implicit task performance, including implicit measures of attitudes, prejudice, and stereotyping.
Two experiments investigated adult age differences in episodic and semantic long-term memory tasks, as a test of the hypothesis of specific age-related decline in context memory. Older adults were slower and exhibited lower episodic accuracy than younger adults. Fits of the diffusion model (R. Ratcliff, 1978) revealed age-related increases in non-decisional reaction time for both episodic and semantic retrieval. In Experiment 2, an age difference in boundary separation also indicated an age-related increase in conservative criterion setting. For episodic old-new recognition (Experiment 1) and source memory (Experiment 2), there was an age-related decrease in the quality of decision-driving information (drift rate). As predicted by the context-memory deficit hypothesis, there was no corresponding age-related decline in semantic drift rate.
In the present article, a flexible and fast computer program, called fast-dm, for diffusion model data analysis is introduced. Fast-dm is free software that can be downloaded from the authors' websites. The program allows estimating all parameters of Ratcliff's (1978) diffusion model from the empirical response time distributions of any binary classification task. Fast-dm is easy to use: it reads input data from simple text files, while program settings are specified by commands in a control file. With fast-dm, complex models can be fitted, where some parameters may vary between experimental conditions, while other parameters are constrained to be equal across conditions. Detailed directions for use of fast-dm are presented, as well as results from three short simulation studies exemplifying the utility of fast-dm.
A goodness‐of‐fit test is constructed using sample quantiles for a p.d.f. F(x, θ) (where θ is to be estimated) which is continuous and strictly increasing at the quantiles of interest. It is shown to be asymptotically equivalent, under alternatives “close” to the null, to the usual goodness‐of‐fit test based on grouped observations, the advantage being that we may avoid the difficulties of choosing cell boundaries for grouping.
The present article reviews evidence for the malleability of automatic stereotypes and prejudice. In contrast to assumptions that such responses are fixed and ines- capable, it is shown that automatic stereotypes and prejudice are influenced by, (a) self- and social motives, (b) specific strategies, (c) the perceiver's focus of atten- tion, and (d) the configuration of stimulus cues. In addition, group members' indi- vidual characteristics are shown to influence the extent to which (global) stereo- types and prejudice are automatically activated. This evidence has significant implications for conceptions of automaticity, models of stereotyping and prejudice, and attitude representation. The review concludes with the description of an initial model of early social information processing. Given a thimbleful of facts we rush to make general- izations as large as a tub. … Life is short, and the de- mands upon us for practical adjustments so great, that we cannot let our ignorance detain us in our daily transactions. (Allport, 1954, p. 9)
This article presents the rationale and procedures for conducting a process analysis in evaluation research. Such an analysis attempts to identify the process that mediates the effects of some treatment, by estimating the parameters of a causal chain between the treatment and some outcome variable. Two different procedures for estimating mediation are discussed. In addition we present procedures for examining whether a treatment exerts its effects, in part, by altering the mediating process that produces the outcome. Finally, the benefits of process analysis in evaluation research are underlined.
The diffusion model for two-choice real-time decisions is applied to four psychophysical tasks. The model reveals how stimulus information guides decisions and shows how the information is processed through time to yield sometimes correct and sometimes incorrect decisions. Rapid two-choice decisions yield multiple empirical measures: response times for correct and error responses, the probabilities of correct and error responses, and a variety of interactions between accuracy and response time that depend on instructions and task difficulty. The diffusion model can explain all these aspects of the data for the four experiments we present. The model correctly accounts for error response times, something previous models have failed to do. Variability within the decision process explains how errors are made, and variability across trials correctly predicts when errors are faster than correct responses and when they are slower.
Develops a theory of memory retrieval and shows that it applies over a range of experimental paradigms. Access to memory traces is viewed in terms of a resonance metaphor. The probe item evokes the search set on the basis of probe–memory item relatedness, just as a ringing tuning fork evokes sympathetic vibrations in other tuning forks. Evidence is accumulated in parallel from each probe–memory item comparison, and each comparison is modeled by a continuous random walk process. In item recognition, the decision process is self-terminating on matching comparisons and exhaustive on nonmatching comparisons. The mathematical model produces predictions about accuracy, mean reaction time, error latency, and reaction time distributions that are in good accord with data from 2 experiments conducted with 6 undergraduates. The theory is applied to 4 item recognition paradigms (Sternberg, prememorized list, study–test, and continuous) and to speed–accuracy paradigms; results are found to provide a basis for comparison of these paradigms. It is noted that neural network models can be interfaced to the retrieval theory with little difficulty and that semantic memory models may benefit from such a retrieval scheme. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
It is argued that a model of goal-independent spreading activation in a social or semantic knowledge structure is insufficient to explain implicit association effects in the IAT (Greenwald, McGhee, & Schwartz, 1998). An alternative account is proposed, which attributes IAT effects to differential costs for switching between task sets. Two experiments were conduced to test this account. In Experiment 1, specific task-set switching cost was a function of IAT condition: switching between tasks was associated with significantly more cost in the incompatible IAT phase. In a second experiment the magnitude of the IAT effect was reduced when task-set reconfiguration was possible in advance of or simultaneously with the upcoming stimulus. The results are discussed with respect to recently suggested accounts of the effect.
Tests of the validity of probability models can be based on classical chi-squared statistics for discrete random variables. For continuous random variables, chi-squared-based methods are available as well as procedures using order statistics and log-likelihood ratio tests, among others. Classical approaches to model selection are also based on model-fit tests. A related approach is based on tests between hypotheses that correspond to the different families of models. Other approaches are the Bayesian approach, the discrepancy-based method to select models with the goal to minimize the expected value of a given discrepancy criterion, and selection based on the minimum description length principle.
It is desired to determine which of several alternative models adequately describe the data. The properties of a combined distribution containing the component models as special cases are investigated. Using this distribution, statistics are developed for testing for departures from one model in the direction of another and for testing the hypothesis that all models fit the data equally well. The relationship with other procedures is investigated. Examples are given of the use of the method, especially when there are two component models belonging to separate parametric families.
The stability of the Implicit Association Test for assessing anxiety (IAT-Anxiety) is lower than its internal consistency, indicating that the IAT-Anxiety measures both stable and occasion-specific variance. This suggests that the IAT-Anxiety may be not only a valid measure of trait anxiety but also one of state anxiety. To test this assumption, two studies were conducted in which state anxiety was experimentally induced by a public speaking task. However, both studies showed that the IAT-Anxiety score did not change when a state of anxiety was induced. Thus, it seems that occasion-specific factors other than variations in state anxiety lead to occasion-specific variance in the IAT-Anxiety score. Implications for the indirect assessment of personality dispositions with the IAT are discussed. Copyright © 2004 John Wiley & Sons, Ltd.
Comments on D. E. Meyer and colleagues' (see record 1988-28535-001) new technique for examining the time course of information processing, which is a variant of the response signal procedure: On some trials Ss are presented with a signal that requires them to respond, whereas on other trials they respond normally. The accuracy of guesses based on partial information can be determined by using the data from the regular trials and a simple race model to remove the contribution of fast-finishing regular trials from signal trial data. This analysis shows that the accuracy of guesses is relatively low and is either approximately constant or grows slowly over the time course of retrieval. Myers et al argue that this pattern of results rules out most continuous models of information processing. But the analyses presented in the present article show that this pattern is consistent with several stochastic RT models: the simple random walk, the runs, and the continuous diffusion models. The diffusion model is assessed with data from a new experiment using the study–test recognition memory procedure. Fitting the diffusion model to the data from regular trials fixes all parameters of the model except one (the signal encoding and decision parameter). With this one free parameter, the model predicts the observed guessing accuracy. It is concluded that the results obtained from Meyer and colleagues' new technique give qualitative support to some stochastic models and quantitative support to the continuous diffusion model. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
The interpretation of the reaction time difference between positive and negative responses in two-choice matching tasks has been the subject of recent controversy. Proctor and his colleagues hold that the difference represents a difference in processing between same and different judgments, whereas Ratcliff and Hacker argued that the difference can be accounted for in terms of criteria settings. In this article, the way in which several models of choice reaction time and matching can account for this reaction time difference is examined and one particular model, the diffusion model of Ratcliff (1981), is fitted to the data from three published experiments. The results of these fits provide a clear interpretation of the reaction time difference in terms of criteria settings. It is concluded that interpretation of such positive-negative reaction time differences in the absence of a specific model is hazardous at best.
An implicit association test (IAT) measures differential association of 2 target concepts with an attribute. The 2 concepts appear in a 2-choice task (2-choice task (e.g., flower vs. insect names), and the attribute in a 2nd task (e.g., pleasant vs. unpleasant words for an evaluation attribute). When instructions oblige highly associated categories (e.g., flower + pleasant) to share a response key, performance is faster than when less associated categories (e.g., insect & pleasant) share a key. This performance difference implicitly measures differential association of the 2 concepts with the attribute. In 3 experiments, the IAT was sensitive to (a) near-universal evaluative differences (e.g., flower vs. insect), (b) expected individual differences in evaluative associations (Japanese + pleasant vs. Korean + pleasant for Japanese vs. Korean subjects), and (c) consciously disavowed evaluative differences (Black + pleasant vs. White + pleasant for self-described unprejudiced White subjects).
Two connectionist frameworks, GRAIN (J. L. McClelland, 1993) and brain-state-in-a-box (J. A. Anderson, 1991), and R. Ratcliff's (1978) diffusion model were evaluated using data from a signal detection task. Dependent variables included response probabilities, reaction times for correct and error responses, and shapes of reaction-time distributions. The diffusion model accounted for all aspects of the data, including error reaction times that had previously been a problem for all response-time models. The connectionist models accounted for many aspects of the data adequately, but each failed to a greater or lesser degree in important ways except for one model that was similar to the diffusion model. The findings advance the development of the diffusion model and show that the long tradition of reaction-time research and theory is a fertile domain for development and testing of connectionist assumptions about how decisions are generated over time.
The Implicit Association Test (IAT) examines the differential association of two object categories (e.g. flower and insect) with attribute categories (e.g. pleasant and unpleasant). When items from congruent categories (e.g. flower + pleasant) share a response key, performance is faster and more accurate than when items from incongruent categories (e.g. insect + pleasant) share a key. Performing incongruent word classification engages inhibitory processes to overcome the prepotent tendency to map emotionally congruent items to the same response key. Using fMRI on subjects undergoing the IAT, we show that the left dorsolateral prefrontal cortex, and to a lesser extent the anterior cingulate cortex, mediate inhibitory processes where manipulation of word association is required.
A brightness discrimination experiment was performed to examine how subjects decide whether a patch of pixels is "bright" or "dark," and stimulus duration, brightness, and speed versus accuracy instructions were manipulated. The diffusion model (Ratcliff, 1978) was fit to the data, and it accounted for all the dependent variables: mean correct and error response times, the shapes of response time distributions for correct and error responses, and accuracy values. Speed-accuracy manipulations affected only boundary separation (response criteria settings) in the model. Drift rate (the rate of accumulation of evidence) in the diffusion model, which represents stimulus quality, increased as a function of stimulus duration and stimulus brightness but asymptoted as stimulus duration increased from 100 to 150 msec. To address the argument that the diffusion model can fit any pattern of data, simulated patterns of plausible data are presented that the model cannot fit.
Three methods for fitting the diffusion model (Ratcliff, 1978) to experimental data are examined. Sets of simulated data were generated with known parameter values, and from fits of the model, we found that the maximum likelihood method was better than the chi-square and weighted least squares methods by criteria of bias in the parameters relative to the parameter values used to generate the data and standard deviations in the parameter estimates. The standard deviations in the parameter values can be used as measures of the variability in parameter estimates from fits to experimental data. We introduced contaminant reaction times and variability into the other components of processing besides the decision process and found that the maximum likelihood and chi-square methods failed, sometimes dramatically. But the weighted least squares method was robust to these two factors. We then present results from modifications of the maximum likelihood and chi-square methods, in which these factors are explicitly modeled, and show that the parameter values of the diffusion model are recovered well. We argue that explicit modeling is an important method for addressing contaminants and variability in nondecision processes and that it can be applied in any theoretical approach to modeling reaction time.
The diffusion model for 2-choice decisions (R. Ratcliff, 1978) was applied to data from lexical decision experiments in which word frequency, proportion of high- versus low-frequency words, and type of nonword were manipulated. The model gave a good account of all of the dependent variables--accuracy, correct and error response times, and their distributions--and provided a description of how the component processes involved in the lexical decision task were affected by experimental variables. All of the variables investigated affected the rate at which information was accumulated from the stimuli--called drift rate in the model. The different drift rates observed for the various classes of stimuli can all be explained by a 2-dimensional signal-detection representation of stimulus information. The authors discuss how this representation and the diffusion model's decision process might be integrated with current models of lexical access.
Based on a task-set switching account of the Implicit Association Test (IAT), the authors predict a specific pattern of aftereffects as a consequence of working through IAT blocks. In Study 1, performance in an evaluative decision task, but not in a color-naming task, was decreased after working through the incompatible rather than compatible block of a flower-insect IAT. In Study 2, response latencies in an evaluative rating task, but not in a color-rating task, were analogously affected, whereas the ratings themselves were not a function of the compatibility of prior IAT blocks. The aftereffects demonstrate reactivity of the IAT; they bear on the mechanisms underlying the IAT and on compatibility-order effects.
The diffusion model (Ratcliff, 1978) allows for the statistical separation of different components of a speeded binary decision process (decision threshold, bias, information uptake, and motor response). These components are represented by different parameters of the model. Two experiments were conducted to test the interpretational validity of the parameters. Using a color discrimination task, we investigated whether experimental manipulations of specific aspects of the decision process had specific effects on the corresponding parameters in a diffusion model data analysis (see Ratcliff, 2002; Ratcliff & Rouder, 1998; Ratcliff, Thapar, & McKoon, 2001, 2003). In support of the model, we found that (1) decision thresholds were higher when we induced accuracy motivation, (2) drift rates (i.e., information uptake) were lower when stimuli were harder to discriminate, (3) the motor components were increased when a more difficult form of response was required, and (4) the process was biased toward rewarded responses.
Theoretically, low correlations between implicit and explicit measures can be due to (a) motivational biases in explicit self reports, (b) lack of introspective access to implicitly assessed representations, (c) factors influencing the retrieval of information from memory, (d) method-related characteristics of the two measures, or (e) complete independence of the underlying constructs. The present study addressed these questions from a meta-analytic perspective, investigating the correlation between the Implicit Association Test (IAT) and explicit self-report measures. Based on a sample of 126 studies, the mean effect size was .24, with approximately half of the variability across correlations attributable to moderator variables. Correlations systematically increased as a function of (a) increasing spontaneity of self-reports and (b) increasing conceptual correspondence between measures. These results suggest that implicit and explicit measures are generally related but that higher order inferences and lack of conceptual correspondence can reduce the influence of automatic associations on explicit self-reports.
The EZ-diffusion model for two-choice response time tasks takes mean response time, the variance of response time, and response accuracy as inputs. The model transforms these data via three simple equations to produce unique values for the quality of information, response conservativeness, and nondecision time. This transformation of observed data in terms of unobserved variables addresses the speed-accuracy trade-off and allows an unambiguous quantification of performance differences in two-choice response time tasks. The EZ-diffusion model can be applied to data-sparse situations to facilitate individual subject analysis. We studied the performance of the EZ-diffusion model in terms of parameter recovery and robustness against misspecification by using Monte Carlo simulations. The EZ model was also applied to a real-world data set.
The Implicit Association Test (IAT; A. G. Greenwald, D. E. McGhee, & J. L. K. Schwartz, 1998) can be used to assess interindividual differences in the strength of associative links between representational structures such as attitude objects and evaluations. Four experiments are reported that explore the extent of method-specific variance in the IAT. The most important findings are that conventionally scored IAT effects contain reliable interindividual differences that are method specific but independent of the measures' content, and that IAT effects can be obtained in the absence of a preexisting association between the response categories. Several techniques to decrease the impact of method-specific variance are evaluated. The best results were obtained with the D measures recently proposed by A. G. Greenwald, B. A. Nosek, and M. R. Banaji (2003).
Measuring task-switching ability in the Implicit Association Test. Experimental Psychol-ogy The malleability of automatic stereotypes and prejudice
- M D Back
- S C Schmukle
- B Klauer Egloff
- Teige- Rblair
Back, M. D., Schmukle, S. C., & Egloff, B. (2005). Measuring task-switching ability in the Implicit Association Test. Experimental Psychol-ogy, 52, 167–179. 366 KLAUER, VOSS, SCHMITZ, AND TEIGE-MOCIGEMBA rBlair, I. V. (2002). The malleability of automatic stereotypes and prejudice. Personality and Social Psychology Review, 6, 242–261
The Implicit Association Test at age 7: A methodological and conceptual review Social psychology and the unconscious: The automa-ticity of higher mental processes
- B A Nosek
- A G Greenwald
- M R Banaji
Nosek, B. A., Greenwald, A. G., & Banaji, M. R. (2006). The Implicit Association Test at age 7: A methodological and conceptual review. In J. A. Bargh (Ed.), Social psychology and the unconscious: The automa-ticity of higher mental processes (pp. 265–292). London: Psychology Press.
Box-and-whisker plot. In MathWorld—A Wol-fram Web resource
- E W Weisstein
Weisstein, E. W. (1999). Box-and-whisker plot. In MathWorld—A Wol-fram Web resource. Retrieved, November 10, 2005, from http:// mathworld.wolfram.com/Box-and-WhiskerPlot.html (Appendix follows)
Box-and-whisker plot
- E W Weisstein
Weisstein, E. W. (1999). Box-and-whisker plot. In MathWorld-A Wolfram Web resource. Retrieved, November 10, 2005, from http://
mathworld.wolfram.com/Box-and-WhiskerPlot.html
(Appendix follows)