Category Archives: Implicit Association Test

Hey Social Psychologists: Don’t Mess with Suicide!

Ten years ago, the foundations of psychological science were shaking by the realization that the standard scientific method of psychological science is faulty. Since then it has become apparent that many classic findings are not replicable and many widely used measures are invalid; especially in social psychology (Schimmack, 2020).

However, it is not uncommon to read articles in 2021 that ignore the low credibility of published results. There are too many of these pseudo-scientific articles, but some articles matter more than others; at last to me. I do care about suicide and like many people my age, I know people who have committed suicide. I was therefore concerned when I saw a review article that examines suicide from a dual-process perspective.

Automatic and controlled antecedents of suicidal ideation and action: A dual-process conceptualization of suicidality.

My main concern about this article is that dual-process models in social cognition are based on implicit priming studies with low replicability and implicit measures with low validity (Schimmack, 2021a, 2021b). It is therefore unclear how dual-process models can help us to understand and prevent suicides.

After reading the article, it is clear that the authors make many false statements and present questionable studies that have never been replicated as if they produce a solid body of empirical evidence.

Introduction of the Article

The introduction cites outdated studies that have either not been replicated or produced replication failures.

“Our position is that even these integrative models omit a fundamental and well-established
dynamic of the human mind: that complex human behavior is the result of an interplay between relatively automatic and relatively controlled modes of thought (e.g., Sherman et al., 2014). From basic processes of impression formation (e.g., Fiske et al., 1999) to romantic relationships (e.g., McNulty & Olson, 2015) and intergroup relations (e.g., Devine, 1989), dual-process frameworks that incorporate automatic and controlled cognition have provided a more complete understanding of a broad array of social phenomena.

This is simply not true. For example, there is no evidence that we implicitly love our partners when we consciously hate them or vice versa, and there is no evidence that prejudice occurs outside of awareness.

Automatic cognitions can be characterized as unintentional (i.e., inescapably activated), uncontrollable (i.e., difficult to stop), efficient in operation (i.e., requiring few cognitive resources), and/or unconscious (Bargh, 1994) and are typically captured with implicit measures.

This statement ignores many articles that have criticized the assumption that implicit measures measure implicit constructs. Even the proponent of the most widely used implicit measure have walked back this assumption (Greenwald & Banaji, 2017).

The authors then make the claim that implicit measures of suicide have incremental predictive validity of suicidal behavior.

For example, automatic associations between the self and death predict suicidal ideation and action beyond traditional explicit (i.e., verbal) responses (Glenn et al., 2017).

This claim has been made repeatedly by proponents of implicit measures, so I meta-analyzed the small set of studies that tested this prediction (Schimmack, 2021). Some of these studies produced non-significant results and the literature showed evidence that questionable research practices were used to produce significant results. Overall, the evidence is inconclusive. It is therefore incorrect to point to a single study as if there is clear evidence that implicit measures of suicidality are valid.

Further statements are also based on outdated research and a single reference.

Research on threat has consistently shown that people preferentially process dangers to physical harm by prioritizing attention, response, and recall regarding threats (e.g., Öhman &
Mineka, 2001)
.”

There have been many proposals about stimuli that attract attention, and threatening stimuli are by no means the only attention grabbing stimuli. Sexual stimuli also attract attention and in general arousal rather than valence or threat is a better predictor of attention (Schimmack, 2005).

It is also not clear how threatening stimuli are relevant for suicide which is related to depression rather than anxiety disorders.

The introduction of implicit measures totally disregards the controversy about the validity of implicit measures or the fact that different implicit measures of the same construct show low convergent validity.

Much has been written about implicit measures (for reviews, see De Houwer et al., 2009; Fazio & Olson, 2003; March et al., 2020; Nosek et al., 2011; Olson & Fazio, 2009), but for the present purposes, it is important to note the consensus that implicit measures index the automatic properties of attitudes.

More relevant are claims that implicit measures have been successfully used to understand a variety of clinical topics.

The application of a dual-process framework has consequently improved explanation and prediction in a number of areas involving mental health, including addiction (Wiers & Stacy, 2006), anxiety (Teachman et al., 2012), and sexual assault (Widman & Olson, 2013). Much of this work incorporates advances in implicit measurement in clinical domains (Roefs et al., 2011).

The authors then make the common mistake to conflate self-deception and other-deception. The notion of implicit motives that can influence behavior without awareness implies self-deception. An alternative rational for the use of implicit measures is that they are better measures of consciously accessible thoughts and feelings that individuals are hiding from others. Here we do not need to assume a dual-process model. We simply have to assume that self-report measures are easy to fake, whereas implicit measures can reveal the truth because they are difficult to fake. Thus, even incremental predictive validity does not automatically support a dual-process model of suicide. However, this question is only relevant if implicit measures of suicidality show incremental predictive validity, which has not been demonstrated.

Consistent with the idea that such automatic evaluative associations can predict suicidality later, automatic spouse-negative associations predicted increases in suicidal ideation over time across all three studies, even after accounting for their controlled counterparts (McNulty et al., 2019).

Conclusion Section

In the conclusion section, the authors repeat their false claim that implicit measures of suicidality reflect valid variance in implicit suicidality and that they are superior to explicit measures.

As evidence of their impact on suicidality has accumulated, so has the need for incorporating automatic processes into integrative models that address questions surrounding how and under what circumstances automatic processes impact suicidality, as well as how automatic and controlled processes interact in determining suicide-relevant outcomes.”

Implicit measures are better-suited to assess constructs that are more affective
(Kendrick & Olson, 2012), spontaneous (e.g., Phillips & Olson, 2014), and uncontrollable (e.g., Klauer & Teige-Mocigemba, 2007).

As recent work has shown (e.g., Creemers et al., 2012; Franck, De Raedt, Dereu, et al., 2007; Franklin et al., 2016; Glashouwer et al., 2010; Glenn et al., 2017; Hussey et al., 2016; McNulty et al., 2019; Nock et al., 2010; Tucker,Wingate, et al., 2018), the psychology of suicidality requires formal consideration of automatic processes, their proper measurement, and how they relate
to one another and corresponding controlled processes.

We have articulated a number of hypotheses, several already with empirical support, regarding interactions between automatic and controlled processes in predicting suicidal ideation and lethal acts, as well as their combination into an integrated model.

Then they finally mention the measurement problems of implicit measures.

Research utilizing the model should be mindful of specific challenges. First, although the model answers calls to diversify measurement in suicidality research by incorporating implicit measures, such measures are not without their own problems. Reaction time measures often have problematically low reliabilities, and some include confounds (e.g., Olson et al., 2009). Further, implicit and explicit measures can differ in a number of ways, and structural differences between them can artificially deflate their correspondence (Payne et al., 2008). Researchers should be aware of the strengths and weaknesses of implicit measures.

Evaluation of the Evidence

Here I provide a brief summary of the actual results of studies cited in the review article so that readers can make up their own mind about the relevance and credibility of the evidence.

Creemers, D. H., Scholte, R. H., Engels, R. C., Prinstein, M. J., & Wiers, R. W. (2012). Implicit and explicit self-esteem as concurrent predictors of suicidal ideation, depressive symptoms, and loneliness. Journal of Behavior Therapy and Experimental Psychiatry, 43(1), 638–646

Participants: 95 undergraduate students
Implicit Construct / Measure: Implicit self-esteem / Name Latter Task
Dependent Variables: depression, loneliness, suicidal ideation
Results: No significant direct relationship. Interaction between explicit and implicit self-esteem for suicidal ideation only, b = .28.

Franck, E., De Raedt, R., & De Houwer, J. (2007). Implicit but not explicit self-esteem predicts future depressive symptomatology. Behaviour Research and Therapy, 45(10), 2448–2455.

Participants: 28 clinically depressed patients; 67 not-depressed participants.
Implicit Construct / Measure: Implicit self-esteem / Name Latter Task
Dependent Variable: change in depression controlling for T1
Result: However, after controlling for initial symptoms of depression, implicit, t(48) = 2.21, p = .03, b = .25, but not explicit self-esteem, t(48) = 1.26, p = .22, b = .17, proved to be a significant predictor for depressive symptomatology at 6 months follow-up.

Franck, E., De Raedt, R., Dereu, M., & Van den Abbeele, D. (2007). Implicit and explicit self- esteem in currently depressed individuals with and without suicidal ideation. Journal of Behavior Therapy and Experimental Psychiatry, 38(1), 75–85.

Participants: Depressed patients with suicidal ideation (N = 15), depressed patients without suicidal ideation (N = 14) and controls (N = 15)
Implicit Construct / Measure: Implicit self-esteem / IAT
Dependent variable. Group status
Contrast analysis revealed that the currently depressed individuals with suicidal ideation showed a significantly higher implicit self-esteem as compared to the currently depressed individuals without suicidal ideation, t(43) = 3.0, p < 0.01. Furthermore, the non-depressed controls showed a significantly higher implicit self-esteem as compared to the currently depressed individuals without suicidal ideation, t(43) = 3.7, p < 0.001.
[this finding implies that suicidal depressed patients have HIGHER implicit self-esteem than depressed patients who are not suicidal].

Glashouwer,K.A., de Jong,P. J., Penninx, B.W.,Kerkhof,A. J., vanDyck, R., & Ormel, J. (2010). Do automatic self-associations relate to suicidal ideation? Journal of Psychopathology and Behavioral Assessment, 32(3), 428–437.

Participants: General population (N = 2,837)
Implicit Constructs / Measure: Implicit depression, Implicit Anxiety / IAT
Dependent variable: Suicidal Ideation, Suicide Attempt
Results: simple correlations
Depression IAT – Suicidal Ideation, r = .22
Depression IAT – Suicide Attempt, r = .12
Anxiety IAT – Suicide Ideation, r = .18
Anxiety IAT – Suicide Attempt, r = .11
Controlling for Explicit Measures of Depression / Anxiety
Depression IAT – Suicidal Ideation, b = ..024, p = .179
Depression IAT – Suicide Attempt, b = .037, p = .061
Anxiety IAT – Suicide Ideation, b = .024, p = .178
Anxiety IAT – Suicide Attempt, r = ..039, p = .046

Glenn, J. J., Werntz, A. J., Slama, S. J., Steinman, S. A., Teachman, B. A., &
Nock, M. K. (2017). Suicide and self-injury-related implicit cognition: A
large-scale examination and replication. Journal of Abnormal Psychology,
126(2), 199–211.

Participants: Self-selected online sample with high rates of self-harm (> 50%). Ns = 3,115, 3114
Implicit Constructs / Measure
: Self-Harm, Death, Suicide / IAT
Dependent variables: Group differences (non-suicidal self-injury / control; suicide attempt / control)
Results:
Non-suicidal self-injury versus control
Self-injury IAT, d = .81/.97; Death IAT d = .52/.61, Suicide IAT d = .58/.72
Suicide Attempt versus control
Self-injury IAT, d = ..52/.54; Death IAT d = .37/.32, Suicide IAT d = .54/.67
[these results show that self-ratings and IAT scores reflect a common construct;
they do not show discriminant validity; no evidence that they measure distinct
constructs and they do not show incremental predictive validity]

Hussey, I., Barnes-Holmes, D., & Booth, R. (2016). Individuals with current
suicidal ideation demonstrate implicit “fearlessness of death..” Journal of
Behavior Therapy and Experimental Psychiatry, 51, 1–9.

Participants: 23 patients with suicidal ideation and 25 controls (university students)
Implicit Constructs / Measure: Death attitudes (general / personal) / IRAP
Dependent variable: Group difference
Results: No main effects were found for either group (p = .08). Critically, however, a three-way interaction effect was found between group, IRAP type, and trial-type, F(3, 37) = 3.88, p = .01. Specifically, the suicidal ideation group produced a moderate “my death-not-negative” bias (M = .29, SD = .41), whereas the normative group produced a weak “my death-negative” bias (M = -.12, SD = .38, p < .01). This differential performance was of a very large effect size (Hedges’ g = 1.02).
[This study suggests that evaluations of personal death show stronger relationships than generic death]

McNulty, J. K., Olson, M. A., & Joiner, T. E. (2019). Implicit interpersonal evaluations as a risk factor for suicidality: Automatic spousal attitudes predict changes in the probability of suicidal thoughts. Journal of Personality and Social Psychology, 117(5), 978–997

Participants. Integrative analysis of 399 couples from 3 longitudinal study of marriages.
Implicit Construct / Measure: Partner attitudes / evaluative priming task
Dependent variable: Change in suicidal thoughts (yes/no) over time
Result: (preferred scoring method)
without covariates, b = -.69, se = .27, p = .010.
with covariate, b = -.64, se = .29, p = .027

Nock, M. K., Park, J. M., Finn, C. T., Deliberto, T. L., Dour, H. J., & Banaji, M. R. (2010). Measuring the suicidal mind: Implicit cognition predicts suicidal behavior. Psychological Science, 21(4), 511–517.

Participants. 157 patients with mental health problems
Implicit Construct / Measure: death attitudes / IAT
Dependent variable: Prospective Prediction of Suicide
Result: controlling for prior attempts / no explicit covariates
b = 1.85, SE = 0.94, z = 2.03, p = .042

Tucker, R. P., Wingate, L. R., Burkley, M., & Wells, T. T. (2018). Implicit Association with Suicide as Measured by the Suicide Affect Misattribution Procedure (S-AMP) predicts suicide ideation. Suicide and Life-Threatening Behavior, 48(6), 720–731.

Participants. 138 students oversampled for suicidal ideation
Implicit Construct / Measure: suicide attitudes / AMP
Dependent variable: Suicidal Ideation
Result: simple correlation, r = .24
regression controlling for depression, b = .09, se = .04, p = .028

Taken together the reference show a mix of constructs, measures and outcomes, and p-values cluster just below .05. Not one of these p-values is below .005. Moreover, many studies relied on small convenience samples. The most informative study is the study by Glashouwer et al. that examined incremental predictive validity of a depression IAT in a large, population wide, sample. The result was not significant and the effect size was less than r = .1. Thus, the references do not provide compelling evidence for dual-attitude models of depression.

Conclusion

Social psychology have abused the scientific method for decades. Over the past decade, criticism of their practices has become louder, but many social psychologists ignore this criticism and continue to abuse significance testing and to misrepresent these results as if they provide empirical evidence that can inform understanding of human behavior. This article is just another example of the unwillingness of social psychologists to “clean up their act” (Kahneman, 2012). Readers of this article should be warned that the claims made in this article are not scientific. Fortunately, there is a credible research on depression and suicide outside of social psychology.

Did Social Psychologist Really Develop a Love Test?

Until 2011, social psychologists were able to believe that they were actually doing science. They conducted studies, often rigorous experiments with random assignment, analyzed the data and reported results only when they achieved statistical significance, p < .05. This is how they were trained to do science and most of them believed that this is how science works.

However, in 2011 an article by a well-respected social psychologists changed all this. Daryl Bem published an article that showed time-reversed causal processes. Seemingly, people were able to feel the future (Bem, 2011). This article shock the foundations of social psychology because most social psychologists did not believe in paranormal phenomena. Yet, Bem presented evidence for his crazy claim in 8 out of 9 studies. The only study that did not work was with supraliminal stimuli. The other studies used subliminal stimuli, suggesting that only our unconscious self can feel the future.

Over the past decade it has become apparent that Bem and other social psychologists had misused significance testing. They only paid attention to significant results, p < .05, and ignored non-significant results, p > .05. Selective publishing of significant results means that statistical results no longer distinguished between true and false findings. Everything was significant, even time-reversed implicit priming.

Some areas of social psychology have been hit particularly hard by replication failures. Most prominently, implicit priming research has been called out as a poster child of doubt about social psychological results by Nobel Laureate Kahneman. The basic idea of implicit priming is that stimuli outside of participants’ awareness can influence their behavior. Many implicit priming studies have failed to replicate.

Ten years later, we can examine how social psychologists have responded to the growing evidence that many classic findings were obtained with questionable practices (not reporting the failures) and cannot be replicated. Unfortunately, the response is consistent with psychodynamic theories of ego-defense mechanisms and social psychologists’ own theories of motivated reasoning. For the most part, social psychologists have simply ignored the replication failures in the 2010s and continue to treat old articles as if they provide scientific insights into human behavior. For example, Bargh – a leading figure in the implicit priming world – wrote a whole book about implicit priming that does not mention replication failures and presents questionable research as if they were well-established facts (Schimmack, 2017).

Given the questionable status of implicit priming research, it is not surprising that concerns are also growing about measures that were designed to reflect individual differences in implicit cognitions (Schimmack, 2019). The measures often have low reliability (when you test yourself you get different results each time) and show low convergent validity (one measure of your unconscious feelings towards your spouse doesn’t correlate with another measure of your unconscious feelings towards your spouse). It is therefore suspicious, when researchers consistently find results with these measures because measurement error should make it difficult to get significant results all the time.

Implicit Love

In an article from 2019 (i.e., when the replication crisis in social psychology has been well-established), Hicks and McNulty make the following claims about implicit love; that is feelings that are not reflected in self-reports of affection or marital satisfaction.

Their title is based on a classic article by Bargh and Chartrand.

Readers are not informed that the big claims made by Bargh twenty years ago have failed to be supported by empirical evidence. Especially the claim that stimuli often influence behavior without awareness lacks any credible evidence. It is therefore sad to say that social psychologists have moved on from self-deception (they thought they were doing science, but they did not) to other-deception (spreading false information knowing that credible doubts have been raised about this research). Just like it is time to reclaim humility and honesty in American political life, it is important to demand humility and honesty from American social psychologists, who are dominating social psychology.

The empirical question is whether research on implicit love has produced robust and credible results. One advantage for relationship researchers is that a lot of this research was published after Bem (2011). Thus, researchers could have improved their research practices. This could result in two outcomes. Either relationship researchers reported their results more honestly and did report non-significant results when they emerged, or they increased sample sizes to ensure that small effect sizes could produce statistically significant results.

Hicks and McNulty’s (2019) narrative review makes the following claims about implicit love.

1. The frequency of various sexual behaviors was prospectively associated with automatic partner evaluations assessed with an implicit measure but not with self-reported relationship satisfaction. (Hicks, McNulty, Meltzer, & Olson, 2016).

2. Participants with less responsive partners who felt less connected to their partners during conflict-of-interest situations had more negative automatic partner attitudes at a subsequent assessment but not more negative subjective evaluations (Murray, Holmes, & Pinkus, 2010).

3. Pairing the partner with positive affect from other sources (i.e., positive words and pleasant images) can increase the positivity of automatic partner attitudes relative to a control group.

4. The frequency of orgasm during sex was associated with automatic partner attitudes, whereas sexual frequency was associated only with deliberate reports of relationship satisfaction for participants who believed frequent sex was important for relationship health.

5. More positive automatic partner attitudes have been linked to perceiving fewer problems over time (McNulty, Olson, Meltzer, & Shaffer, 2013).

6. More positive automatic partner attitudes have been linked to self-reporting
fewer destructive behaviours (Murray et al., 2015).

7. More positive automatic partner attitudes have been linked to more cooperative relationship behaviors (LeBel & Campbell, 2013)

8. More positive automatic partner attitudes have been linked to displaying attitude-consistent nonverbal communication in conflict discussions (Faure et al., 2018).

9. More positive automatic partner attitudes were associated with a decreased likelihood of dissolution the following year, even after controlling for explicit relationship satisfaction (Lee, Rogge, & Reis, 2010).

10. Newlyweds’ implicit partner evaluations but not explicit satisfaction within the first few months of marriage were more predictive of their satisfaction 4 years later.

11. People with higher motivation to see their relationship in a positive light because of barriers to exiting their relationships (i.e., high levels of relationship investments and poor alternatives) demonstrated a weaker correspondence between their automatic attitudes and their relationship self-reports.

12. People with more negative automatic evaluations are less trusting of their partners when their working memory capacity is limited (Murray et al., 2011).

These claims are followed with the assurance that “these studies provide compelling evidence that automatic partner attitudes do have implications for relationship outcomes” (p. 256).

Should anybody who reads this article or similar claims in the popular media believe them? Have social psychologists improved their methods to produce more credible results over the past decade?

Fortunately, we can answer this question by examining the statistical evidence that was used to support these claims, using the z-curve method. First, all test statistics are converted into z-scores that represent the strength of evidence against the null-hypothesis (i.e., implicit love has no effect or does not exist) in each study. These z-scores are a function of the effect size and the amount of sampling error in a study (signal/noise ratio). Second, the z-scores are plotted as a histogram to show how many of the reported results provide weak or strong evidence against the null-hypothesis. The data are here for full transparency (Implicit.Love.xlsx).

The figure shows the z-curve for the 30 studies that reported usable test results. Most published z-scores are clustered just above the threshold value of 1.96 that corresponds to the .05 criterion to claim a discovery. This clustering is indicative of the use of selecting significant results from a much larger set of analyses that were run and produced non-significant results. The grey curve from z = 0 to 1.96 shows the predicted number of analyses that were not reported. The file drawer ratio implies that for every significant result there were 12 analyses with non-significant results.

Another way to look at the results is to compare the observed discovery rate with the expected discovery rate. The observed discovery rate is simply the percentage of studies that reported a significant result, which is 29 out of 30 or 97%. The estimated discovery rate is the average power of studies to produce a significant result. It is only 8%. This shows that social psychologists still continue to select only successes and do not report or interpret the failures. Moreover, in this small sample of studies, there is considerable uncertainty around the point estimates. The 95%confidence interval for the replication success probability includes 5%, which is not higher than chance. The complementary finding is that the maximum number of false positives is estimated to be 63%, but could be as high as 100%. In other words, the results make it impossible to conclude that even some of these studies produced a credible result.

In short, the entire research on implicit love is bullshit. Ten years ago, social psychologists had the excuse that they did not know better and misused statistics because they were trained the wrong way. This excuse is wearing thin in 2020. They know better, but they continue to report misleading results and write unscientific articles. In psychology, this is called other-deception, in everyday life it is called lying. Don’t trust social psychologists. Doing so is as stupid as believing Donald Trump when he claims that he won the election.

Invalid Claims about the Validity of Implicit Association Tests

Citation:
Schimmack, U. (2021). Invalid Claims About the Validity of Implicit Association Tests by Prisoners of the Implicit Social-Cognition Paradigm. Perspectives on Psychological Science, 16(2), 435–442. https://doi.org/10.1177/1745691621991860

This post has been revised on March 12, 2021 to make it consistent with the published version (https://doi.org/10.1177/1745691621991860) of my response to commentaries by Vianello and Bar-Anan and Kurdi, Ratliff, and Cunningham in response to my target article about the lack of construct validity of IATs (Schimmack, 2021).

Invalid Claims about the Validity of Implicit Association Tests by Prisoners of the Implicit Social-Cognition Paradigm

Abstract
In a prior publication, I used structural equation modeling of multimethod data to examine the construct validity of Implicit Association Tests. The results showed no evidence that IATs measure implicit constructs (e.g., implicit self-esteem, implicit racial bias). This critique of IATs elicited several responses by implicit social-cognition researchers, who tried to defend the validity and usefulness of IATs. I carefully examine these arguments and show that they lack validity. IAT proponents consistently ignore or misrepresent facts that challenge the validity of IATs as measures of individual differences in implicit cognitions. One response suggests that IATs can be useful even if they merely measure the same constructs as self-report measures, but I find no support for the claim that IATs have practically significant incremental predictive validity. In conclusions, IATs are widely used without psychometric evidence of construct or predictive validity.

Keywords
implicit attitudes, Implicit Association Test, validity, prejudice, suicide, mental health

Greenwald and colleagues (1998) introduced Implicit Association Tests (IATs) as a new method to measure individual differences in implicit cognitions. Twenty years later, IATs are widely used for this purpose, but their construct validity has not been established. Even its creator is no longer sure what IATs measure. Whereas Banaji and Greenwald (2013) confidently described
IATs as “a method that gives the clearest window now available into a region of the mind that is inaccessible to question-asking methods” (p. xiii), they now claim that IATs merely measure “the strengths of associations among concepts” (Cvencek et al., 2020, p. 187). This is akin to saying that an old-fashioned thermometer measures the expansion of mercury: It is true, but it has little to do with thermometers’ purpose of measuring temperature.

Fortunately, we do not need Greenwald or Banaji to define the constructs that IATs are supposed to measure. Twenty years of research with IATs makes it clear what researchers believe they are measuring with IATs. A self-esteem IAT is supposed to measure implicit self-esteem (Greenwald & Farnham, 2000). A race IAT is supposed to measure implicit prejudice (Cunningham et al., 2001), and a suicide IAT is supposed to measure implicit suicidal tendencies that can predict suicidal behaviors above and beyond self-reports (Kurdi et al.,
2021). The empirical question is whether IATs are any good at measuring these constructs. I concluded that most IATs are poor measures of their intended constructs (Schimmack, 2021). This conclusion elicited one implicit and two explicit responses.

Implicit Response

The implicit response is to simply ignore criticism and to make invalid claims about the construct validity of IATs (Greenwald & Lai, 2020). For example, a 2020 article coauthored by Nosek, Greenwald, and Banaji (among others) claimed that “available evidence for validity of
IAT measures of self-esteem is limited (Bosson et al., 2000; Greenwald & Farnham, 2000), with some of the strongest evidence coming from empirical tests of the balance-congruity principle” (Cvencek et al., 2020, p. 191). This statement is as valid as Donald Trump’s claim that an honest count of votes would make him the winner of the 2020 election. Over the past 2 decades, several articles have concluded that self-esteem IATs lack validity (Buhrmester et al., 2011; Falk et al., 2015; Walker & Schimmack, 2008). It is unscientific to omit these references from a literature review.

The balance-congruity principle is also not a strong test of the claim that the self-esteem IAT is a valid measure of individual differences in implicit self-esteem. In contrast, the lack of convergent validity with informant ratings and even other implicit measures of
self-esteem provides strong evidence that self-esteem IATs are invalid (Bosson et al., 2000; Falk et al., 2015). Finally, supporting evidence is surprisingly weak. For example, Greenwald and Farnham’s (2000) highly cited article tested predictive validity of the self-esteem IAT with responses to experimentally manipulated successes and failures (n = 94). They did not even report statistical results. Instead, they suggested that even nonsignificant results should be counted as evidence for the validity of the self-esteem IAT:

Although p values for these two effects straddled the p = .05 level that is often treated as a boundary between noteworthy and ignorable results, any inclination to dismiss these findings should be tempered by noting that these two effects agreed with prediction in both direction and shape. (Greenwald & Farnham, 2000, p. 1032)

Twenty years later, this finding has not been replicated, and psychologists have learned to distrust p values that are marginally significant (Benjamin et al., 2018; Schimmack, 2012, 2020). In conclusion, conflict of interest and motivated biases undermine the objectivity of Greenwald and colleagues in evaluations of IATs’ validity.

Explicit Response 1

Vianello and Bar-Anan (2021) criticized my structural equation models of their data. They also presented a new model that appeared to show incremental predictive validity for implicit racial bias and implicit political orientation. I thought it would be possible to resolve some of the disagreement in a direct and open communication with the authors because the disagreement
is about modeling of the same data. I was surprised when the authors declined this offer, given that Bar- Anan coauthored an article that praised the virtues of open scientific communication (Nosek & Bar-Anan, 2012). Readers therefore have to reconcile conflicting viewpoints for themselves. To ensure full transparency, I published syntax, outputs, and a detailed discussion
of the different modeling assumptions on OSF at https://osf.io/wsqfb/.

In brief, a comparison of the models shows that mine is more parsimonious and has better fit than their model. Because the model is more parsimonious, better fit cannot be attributed to overfitting of the data. Rather, the model is more consistent with the actual data, which in most sciences is considered a good reason to favor a model. Vianello and Bar-Anan’s model also produced unexplained, surprising results. For example, the race IAT has only a weak positive loading on the IAT method factor, and the political-orientation IAT even has a moderate negative loading. It is not clear how a method can have negative loadings on a method factor,
and Vianello and Bar-Anan provided no explanation for this surprising finding.

The two models also produce different results regarding incremental predictive validity (Table 1). My model shows no incremental predictive validity for implicit factors. It is also surprising that Vianello and Bar-Anan found incremental predictive validity for voting behaviors,
because the explicit and implicit factors correlated (r) at .9. This high correlation leaves little room for variance in implicit political orientation that is distinct from political orientation measured with self-ratings.

In conclusion, Vianello and Bar-Anan failed to challenge my conclusion that implicit and explicit measures measure mostly the same constructs and that low correlations between explicit and implicit measures reflect measurement error rather than some hidden implicit processes.

Explicit Response 2

The second response (Kurdi et al., 2021) is a confusing 7,000-word article that is short of facts, filled with false claims, and requires more fact-checking than a Trump interview.

False fact 1

The authors begin with the surprising statement that my findings are “not at all incompatible with the way that many social cognition researchers have thought about the construct of (implicit) evaluation” (p. 423). This statement is misleading. For 3 decades, social-cognition
researchers have pursued the idea that many social-cognitive processes that guide behavior occur outside of awareness. For example, Nosek et al. (2011) claim “most human cognition occurs outside conscious awareness or conscious control” (p. 152) and go on to claim that IATs “measure something different from self-report” (p. 153). And just last year, Greenwald and Lai
(2020) claimed that “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” (p. 419).

Social psychologists have also been successful in making the term implicit bias a common term in public discussions of social behavior. The second author, Kathy Ratliff, is director of Project Implicit, which “has a mission to develop and deliver methods for investigating and applying phenomena of implicit social cognition, including especially phenomena of implicit bias based on age, race, gender or other factors” (Kurdi et al., 2021, p. 431). It is not clear what this statement means if we do not make a distinction between traditional research on prejudice with self-report measures and the agenda of Project Implicit to study implicit biases with IATs.
In addition, all three authors have published recent articles that allude to IATs as measures of implicit cognitions.

In a highly cited American Psychologist article, Kurdi and coauthors (2019) claim “in addition to dozens of studies that have established construct validity . . . investigators have asked to what extent, and under what conditions, individual differences in implicit attitudes, stereotypes, and identity are associated with variation in behavior toward individuals as a function of their social group membership” (p. 570). The second author coauthored an article with the claim that “Black participants’ implicit attitudes reflected no ingroup/ outgroup preference . . . Black participants’ explicit attitudes reflected an ingroup preference” ( Jiang et al.,
2019). In 2007, Cunningham wrote that the “distinction between automatic and controlled processes now lies at the heart of several of the most influential models of evaluative processing” (Cunningham & Zelazo, 2007, p. 97). And Cunningham coauthored a review article with the claim that “a variety of tasks have been used to reflect implicit psychopathology associations, with the IAT (Greenwald et al., 1998) used most widely” (Teachman
et al., 2019). Finally, many users of IATs assume that they are measuring implicit constructs that are distinct from constructs that are measured with self-ratings. It is therefore a problem for the construct validity of IATs if they lack discriminant validity. At the least, Kurdi et al. fail to explain why anybody should use IATs if they merely measure the same constructs that can be
measured with cheaper self-ratings. In short, the question whether IATs and explicit measures reflect the same constructs or different constructs has theoretical and empirical relevance, and lack of discriminant validity is a problem for many theories of implicit cognitions (but see Cunningham & Zelazo, 2007).

False fact 2

A more serious false claim is that I found “high correlations between relatively indirect (automatic) measures of mental content, as indexed by the IAT, and relatively direct (controlled) measures of mental content, as indexed by a variety of self-report scales” (p. 423). Table 2 shows some of the correlations among implicit and explicit measures in Vianello and Bar-Anan’s data. Only one of these correlations meets the standard criterion of a high correlation (i.e., r = .5; Cohen, 1988). The other correlations are small to moderate. These correlations show at best moderate convergent validity and no evidence of discriminant validity (i.e., higher implicit-implicit than implicit-explicit correlations). Similar results have been reported since the first IATs were created (Bosson et al., 2000). For 20 years, IAT researchers have ignored these low correlations and made grand claims about the validity of IATs. Kurdi et al. are doubling
down on this misinformation by falsely describing these correlations as high.

False fact 3

The third false claim is that “plenty of evidence in favor of dissociations between direct and indirect measures exists” (p. 428). To support this claim, Kurdi et al. cite a meta-analysis of incremental predictive validity (Kurdi et al., 2019). There are several problems with this claim.
First, the meta-analysis corrects only for random measurement error and not systematic measurement error. To the extent that systematic measurement error is present, incremental validity will shrink because explicit and implicit factors are very highly correlated when both sources of error are controlled (Schimmack, 2021). Second, Kurdi et al. fail to mention effect sizes. The meta-analysis suggests that a perfectly reliable IAT would explain about 2% unique variance. However, IATs have only modest reliability. Thus, manifest IAT scores would explain even less unique variance. Finally, even this estimate has to be interpreted with caution because the meta-analysis did not correct for publication bias and included some questionable studies. For example, Phelps et al. (2003) report, among 12 participants, a correlation of .58 between scores on the race IAT and differences in amygdala activation in response to Black and White faces. Assuming 20% valid variance in the IAT scores (Schimmack, 2021), the validation- corrected correlation would be 1.30. In other words, a correlation of .58 is impossible given the low validity of race-IAT scores. It is well known that correlations in functional MRI studies with small samples are not credible (Vul et al., 2009). Moreover, brain activity is not a social behavior. It is therefore unclear why studies like this were included in Kurdi et al.’s (2019) meta-analysis.

Kurdi et al. also used suicides as an important outcome that can be predicted with suicide and death IATs. They cited two articles to support this claim. Fact checking shows that one article reported a statistically significant result (p = .013; Barnes et al., 2017), whereas the other one did not (p > .50; Glenn et al., 2019). I conducted a meta-analysis of all studies that reported incremental predictive validity of suicide or death IATs. The criterion was suicide attempts in the next 3 to 6 months (Table 3). I found eight studies, but six of them came from a single lab (Matthew K. Nock). Nock was also the first one to report a significant result in an extremely underpowered study that included only two suicide attempts (Nock & Banaji, 2007). Five of the eight studies showed a statistically significant result (63%), but the average observed power to achieve significance was only 42%. This discrepancy suggests the presence of publication bias (Schimmack, 2012). Moreover, significant results are all clustered around .05, and none
of the p values meets the stricter criterion of .005 that has been suggested by Nosek and others to claim a discovery (Benjamin et al., 2018). Thus, there is no conclusive evidence to suggest that suicide IATs have incremental predictive validity in the prediction of suicides. This is not surprising because most of the studies were underpowered and unlikely to detect small effects.
Moreover, effect sizes are bound to be small because the convergent validity between suicide and death IATs is low (r = .21; Chiurliza et al., 2018), suggesting that most of the variance in these IATs is measurement error.

In conclusion, 20 years of research with IATs has produced no credible and replicable evidence that IATs have incremental predictive validity over explicit measures. Even if there is some statistically significant incremental predictive validity, the amount of explained
variance may lack practical significance (Kurdi et al., 2019).

False fact 4

Kurdi et al. (2021) object (p. 424) to my claim that “most researchers regard the IAT as a valid measure of enduring attitudes that vary across individuals” (Schimmack, 2021, p. 397). They claim that “the overwhelming theoretical consensus in the community of attitude researchers.
. . is that attitudes emerge from an interaction of persons and situations” (p. 425). It is instructive to compare this surprising claim with Cunningham and Zelazo’s (2007) definition of attitudes as “relatively stable ideas about whether something is good or bad” (p. 97). Kurdi and Banaji (2017) wrote that “differences in implicit attitudes . . . may arise because of multiple components, including relatively stable components [emphasis added]” (p. 286). Rae and Greenwald (2017) stated that it is a “widespread assumption . . . that implicit attitudes are characteristics of people, almost certainly more so than a property of situations” (p. 297).
Greenwald and Lai (2020) stated that test–retest reliability “places an upper limit on correlational tests of construct validity” (p. 425). This statement makes sense only if we assume that the construct to be measured is stable over the retest interval. It is also not clear how it would be ethical to provide individuals with feedback about their IAT scores on the Project Implicit website, if IAT scores were merely a product of the specific situation at the moment they are taking the test. Finally, how can the suicide IAT be a useful predictor of suicide if it cannot not measure some stable dispositions related to suicidal behaviors?

In conclusion, Kurdi et al.’s definition of attitudes is inconsistent with the common definition of attitudes as relatively enduring evaluations. That being said, the more important question is
whether IATs measure stable attitudes or momentary situational effects. Ironically, some of the best evidence comes from Cunningham. Cunningham et al. (2001) repeatedly measured prejudice four times over a 3-month period with multiple measures, including the race IAT. Cunningham et al. (2001) modeled the data with a single trait factor that explained all of the covariation among different measures of racial attitudes. Thus, Cunningham et al. (2001) provided first evidence that most of the valid variance in race IAT scores is perfectly stable over a 3-month period and that person-by-situation interactions had no effect on racial attitudes. There have been few longitudinal studies with IATs since Cunningham et al.’s (2001) seminal study. However, last year, an article examined stability over a 6-year interval (Onyeador et al., 2020). Racial attitudes of more than 3,000 medical students were measured in the first year of medical school, the fourth year of medical school, and the second year of medical residency.
Table 4 shows the correlations for the explicit feeling thermometer and the IAT scores. The first observation is that the Time-1-to-Time-3 correlation for the IAT scores is not smaller than the Time-1-to-Time-2 or the Time-2-to-Time-3 correlations. This pattern shows that a single trait factor can capture the shared variance among the repeated IAT measures. The second observation is that the bold correlations between explicit ratings and IAT scores on the same occasion are only slightly higher than the correlations for different measurement
occasions. This finding shows that there is very little occasion-specific variance in racial attitudes. The third observation is that IAT correlations over time are higher than the corresponding FT-IAT correlations over time. This finding points to IAT-specific method variance that is revealed in studies with multiple implicit measures (Cunningham et al., 2001; Schimmack, 2021). These findings extend Cunningham et al.’s (2001) findings to
a 6-year period and show that most of the valid variance in race IAT scores is stable over long periods of time.

In conclusion, Kurdi et al.’s claims about person-by-situation effects are not supported by evidence.

Conclusion

Like presidential debates, the commentaries and my response present radically different views of reality. In one world, IATs are valid and useful tools that have led to countless new insights into human behavior. In the other world, IATs are noisy measures that add nothing to the information we already get from cheaper self-reports. Readers not well versed in the literature are likely to be confused rather than informed by these conflicting accounts. Although we may expect such vehement disagreement in politics, we should not expect it among scientists.
A common view of scientists is that they are able to resolve disagreement by carefully looking at data and drawing logical conclusions from empirical facts. However, this model of scientists is naive and wrong.

A major source of disagreement among psychologists is that psychology lacks an overarching paradigm; that is, a set of fundamentally shared assumptions and facts. Psychology does not have one paradigm, but many paradigms. The IAT was developed within the implicit social-cognition paradigm that gained influence in the 1990s (Bargh et al., 1996; Greenwald & Banaji, 1995; Nosek et al., 2011). Over the past decade, it has become apparent that the empirical foundations of this paradigm are shaky (Doyen et al., 2012; D. Kahneman quoted in Yong, 2012, Supplemental Material; Schimmack, 2020). It took a long time to see the problems because paradigms are like prisons that make it impossible to see the world from the outside. A key force that prevents researchers within a paradigm from noticing problems is publication bias. Publication bias ensures that studies that are consistent with a paradigm are published, cited, and highlighted in review articles to provide false evidence in support for a paradigm
(Greenwald & Lai, 2020; Kurdi et al., 2021).

Over the past decade, it has become apparent how pervasive these biases have been, especially in social psychology (Schimmack, 2020). The responses to my critique of IATs merely confirms how powerful paradigms and conflicts of interest can be. It is therefore necessary to allocate more resources to validation projects by independent researchers. In addition, validation studies should be preregistered and properly powered, and results need to be published whether they show validity or not. Conducting validation studies of widely used measures could be an important role for the emerging field of meta-psychology that is not focused on new discoveries, but rather on evaluating paradigmatic research from an outsider, meta-perspective (Carlsson et al., 2017). Viewed from this perspective, many IATs that are in use lack credible evidence of construct validity.

References
*References marked with an asterisk report studies included in
the suicide IAT meta-analysis

Banaji, M. R., & Greenwald, A. G. (2013). Blindspot: Hidden
biases of good people. Delacorte Press.

Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity
of social behavior: Direct effects of trait construct and
stereotype activation on action. Journal of Personality
and Social Psychology, 71(2), 230–244. https://doi.org/
10.1037/0022-3514.71.2.230

*Barnes, S. M., Bahraini, N. H., Forster, J. E., Stearns-Yoder, K. A.,
Hostetter, T. A., Smith, G., Nagamoto, H. T., & Nock,
M. K. (2017). Moving beyond self-report: Implicit associations
about death/ life prospectively predict suicidal
behavior among veterans. Suicide and Life-Threatening
Behavior, 47, 67–77. https://doi.org/10.1111/sltb.12265

Benjamin, D. J., Berger, J. O., Johannesson, M., Nosek, B. A.,
Wagenmakers, E.-J., Berk, R., Bollen, K. A., Brembs, B.,
Brown, L., Camerer, C., Cesarini, D., Chambers, C. D.,
Clyde, M., Cook, T. D., Boeck, P., De, Dienes, Z., Dreber,
A., Easwaran, K., Efferson, C., . . . Johnson, V. E. (2018).
Redefine statistical significance. Nature Human Behaviour,
2, 6–10.

Bosson, J. K., Swann, W. B. Jr., & Pennebaker, J. W. (2000).
Stalking the perfect measure of implicit self-esteem:
The blind men and the elephant revisited? Journal of
Personality and Social Psychology, 79, 631–643. https://
doi.org/10.1037/0022-3514.79.4.631

Buhrmester, M. D., Blanton, H., & Swann, W. B., Jr. (2011).
Implicit self-esteem: Nature, measurement, and a new way
forward. Journal of Personality and Social Psychology,
100(2), 365–385. https://doi.org/10.1037/a0021341

Carlsson, R., Danielsson, H., Heene, M., Ker, Å., Innes, Lakens,
D., Schimmack, U., Schönbrodt, F. D., van Assen, M., &
Weinstein, Y. Inaugural editorial of Meta-Psychology. Meta-
Psychology, 1. https://doi.org/10.15626/MP2017.1001

Chiurliza, B., Hagan, C. R., Rogers, M. L., Podlogar, M. C., Hom,
M. A., Stanley, I. H., & Joiner, T. E. (2018). Implicit measures
of suicide risk in a military sample. Assessment, 25(5),
667–676. https://doi.org/10.1177/1073191116676363
Cohen, J. (1988). Statistical power analysis for the behavioral
sciences (2nd ed.). Erlbaum.

Cunningham, W. A., Preacher, K. J., & Banaji, M. R. (2001).
Implicit attitude measures: Consistency, stability, and
No Evidence for Construct Validity of IAT 441
convergent validity. Psychological Science, 12(2), 163–170
https://doi.org/10.1111/1467-9280.00328

Cunningham, W. A., & Zelazo, P. D. (2007). Attitudes and
evaluations: A social cognitive neuroscience perspective.
Trends in Cognitive Sciences, 11, 97–104. https://
doi.org/10.1016/j.tics.2006.12.005

Cvencek, D., Meltzoff, A. N., Maddox, C. D., Nosek, B. A.,
Rudman, L. A., Devos, T., Dunham, Y., Baron, A. S.,
Steffens, M. C., Lane, K., Horcajo, J., Ashburn Nardo, L.,
Quinby, A., Srivastava, S. B., Schmidt, K., Aidman, E.,
Tang, E., Farnham, S., Mellott, D. S., . . . Greenwald, A. G.
(2020). Meta-analytic use of balanced identity theory to
validate the Implicit Association Test. Personality and
Social Psychology Bulletin, 47(2), 185–200. https://doi
.org/10.1177/0146167220916631

Doyen, S., Klein, O., Pichon, C. L., & Cleeremans, A. (2012).
Behavioral priming: It’s all in the mind, but whose mind?
PLOS ONE, 7(1), Article e29081. https://doi.org/10.1371/
journal.pone.0029081

Falk, C. F., Heine, S. J., Takemura, K., Zhang, C. X., & Hsu,
C. (2015). Are implicit self-esteem measures valid for
assessing individual and cultural differences. Journal of
Personality, 83, 56–68. https://doi.org/10.1111/jopy.12082

*Glenn, C. R., Millner, A. J., Esposito, E. C., Porter, A. C.,
& Nock, M. K. (2019). Implicit identification with death
predicts suicidal thoughts and behaviors in adolescents.
Journal of Clinical Child & Adolescent Psychology, 48,
263–272. https://doi.org/10.1080/15374416.2018.1528548

Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition:
Attitudes, self-esteem, and stereotypes. Psychological
Review, 102(1), 4–27. https://doi.org/10.1037/0033-
295X.102.1.4

Greenwald, A. G., & Farnham, S. D. (2000). Using the Implicit
Association Test to measure self-esteem and self-concept.
Journal of Personality and Social Psychology, 79, 1022–1038
https://doi.org/10.1037/0022-3514.79.6.1022

Greenwald, A. G., & Lai, C. K. (2020). Implicit social cognition.
Annual Review of Psychology, 71, 419–445. https://
doi.org/10.1146/annurev-psych-010419-050837

Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998).
Measuring individual differences in implicit cognition:
The Implicit Association Test. Journal of Personality and
Social Psychology, 74, 1464–1480.

*Harrison, D. P., Stritzke, W. G. K., Fay, N., & Hudaib, A.-R.
(2018). Suicide risk assessment: Trust an implicit probe
or listen to the patient? Psychological Assessment, 30(10),
1317–1329. https://doi.org/10.1037/pas0000577

Jiang, C., Vitiello, C., Axt, J. R., Campbell, J. T., & Ratliff, K. A.
(2019). An examination of ingroup preferences among
people with multiple socially stigmatized identities. Self
and Identity. Advance online publication. https://doi.org/
10.1080/15298868.2019.1657937

Kurdi, B., & Banaji, M. R. (2017). Reports of the death of
the individual difference approach to implicit social cognition
may be greatly exaggerated: A commentary on Payne,
Vuletich, and Lundberg. Psychological Inquiry, 28,
281–287. https://doi.org/10.1080/1047840X.2017.1373555

Kurdi, B., Ratliff, K. A., & Cunningham, W. A. (2021). Can
the Implicit Association Test serve as a valid measure of
automatic cognition? A response to Schimmack (2021).
Perspectives on Psychological Science, 16(2), 422–434.
https://doi.org/10.1177/1745691620904080

Kurdi, B., Seitchik, A. E., Axt, J. R., Carroll, T. J., Karapetyan,
A., Kaushik, N., Tomezsko, D., Greenwald, A. G., &
Banaji, M. R. (2019). Relationship between the Implicit
Association Test and intergroup behavior: A meta-analysis.
American Psychologist, 74(5), 569–586. https://doi.org/
10.1037/amp0000364

*Millner, A. J., Augenstein, T. M., Visser, K. H., Gallagher, K.,
Vergara, G. A., D’Angelo, E. J., & Nock, M. K. (2019). Implicit
cognitions as a behavioral marker of suicide attempts in
adolescents. Archives of Suicide Research, 23(1), 47–63.
https://doi.org/10.1080/13811118.2017.1421488

*Nock, M. K., & Banaji, M. R. (2007). Prediction of suicide ideation
and attempts among adolescents using a brief performance-
based test. Journal of Consulting and Clinical
Psychology, 75(5), 707–715. https://doi.org/10.1037/0022-
006X.75.5.707

*Nock, M. K., Park, J. M., Finn, C. T., Deliberto, T. L.,
Dour, H. J., & Banaji, M. R. (2010). Measuring the suicidal
mind: Implicit cognition predicts suicidal behavior.
Psychological Science, 21(4), 511–517. https://doi
.org/10.1177/0956797610364762

Nosek, B. A., & Bar-Anan, Y. (2012). Scientific utopia: I. Opening
scientific communication. Psychological Inquiry, 23(3),
217–243. https://doi.org/10.1080/1047840X.2012.692215

Nosek, B. A., Hawkins, C. B., & Frazier, R. S. (2011). Implicit
social cognition: From measures to mechanisms. Trends
in Cognitive Sciences, 15(4), 152–159. https://doi.org/
10.1016/j.tics.2011.01.005

Onyeador, I. N., Wittlin, N. M., Burke, S. E., Dovidio, J. F.,
Perry, S. P., Hardeman, R. R., Dyrbye, L. N., Herrin, J.,
Phelan, S. M., & van Ryn, M. (2020). The value of interracial
contact for reducing anti-Black bias among non-Black
physicians: A Cognitive Habits and Growth Evaluation
(CHANGE) study report. Psychological Science, 31(1),
18–30. https://doi.org/10.1177/0956797619879139

Phelps, E. A., Cannistraci, C. J., & Cunningham, W. A. (2003).
Intact performance on an indirect measure of race bias
following amygdala damage. Neuropsychologia, 41(2),
203–208. https://doi.org/10.1016/s0028-3932(02)00150-1

Rae, J. R., & Greenwald, A. G. (2017). Persons or situations?
Individual differences explain variance in aggregated
implicit race attitudes. Psychological Inquiry, 28, 297–300.
https://doi.org/10.1080/1047840X.2017.1373548

*Randall, J. R., Rowe, B. H., Dong, K. A., Nock, M. K., &
Colman, I. (2013). Assessment of self-harm risk using
implicit thoughts. Psychological Assessment, 25(3), 714–721
https://doi.org/10.1037/a0032391

Schimmack, U. (2012). The ironic effect of significant results
on the credibility of multiple-study articles. Psychological
Methods, 17(4), 551–566. https://doi.org/10.1037/a0029487

Schimmack, U. (2020). A meta-psychological perspective on
the decade of replication failures in social psychology.
Canadian Psychology/Psychologie canadienne, 61(4),
364–376. http://doi.org/10.1037/cap0000246

Schimmack, U. (2021). The Implicit Association Test: A method
in search of a construct. Perspectives on Psychological Science, 16(2), 396–414. https://doi.org/10.1177/1745691619863798

Teachman, B. A., Clerkin, E. M., Cunningham, W. A., Dreyer-
Oren, S., & Werntz, A. (2019). Implicit cognition and
psychopathology: Looking back and looking forward.
Annual Review of Clinical Psychology, 15, 123–148.
https://doi.org/10.1146/annurev-clinpsy-050718-095718

*Tello, N., Harika-Germaneau, G., Serra, W., Jaafari, N., &
Chatard, A. (2020). Forecasting a fatal decision: Direct
replication of the predictive validity of the Suicide–
Implicit Association Test. Psychological Science, 31(1),
65–74. https://doi.org/10.1177/0956797619893062

Vianello, M., & Bar-Anan, Y. (2021). Can the Implicit Association
Test measure automatic judgment? The validation continues.
Perspectives on Psychological Science, 16(2), 415–421.
https://doi.org/10.1177/1745691619897960

Vul, E., Harris, C., Winkielman, P., & Pashler, H. (2009).
Puzzlingly high correlations in fMRI studies of emotion,
personality, and social cognition. Perspectives on
Psychological Science, 4(3), 274–290. https://doi.org/10
.1111/j.1745-6924.2009.01125.x

Walker, S. S., & Schimmack, U. (2008). Validity of a happiness
implicit association test as a measure of subjective wellbeing.
Journal of Research in Personality, 42, 490–497.
https://doi.org/10.1016/j.jrp.2007.07.005

Yong, E. (2012 October 12). Nobel laureate challenges
psychologists to clean up their act. Nature. https://doi
.org/10.1038/nature.2012.11535

Psychologists are not immune to the Dunning-Kruger Effect

Background

Bar-Anan and Vianello (2018) published a structural equation model in support of a dual-attitude model that postulates explicit and implicit attitudes towards racial groups, political parties, and the self. I used their data to argue against a dual-attitude model. Vianello and Bar-Anan (2020) wrote a commentary that challenged my conclusions. I was a reviewer of their commentary and pointed out several problems with their new model (Schimmack, 2020). They did not respond to my review and their commentary was published without changes. I wrote a reply to their commentary. In the reply, I merely pointed to my criticism of their new model. Vianello and Bar-Anan wrote a review of my reply, in which they continue to claim that my model is wrong. I invited them to discuss the differences between our models, but they declined. In this blog post, I show that Vianello and Bar-Anan lack insight into the shortcomings of their model, which is consistent with the Dunning-Kruger effect that incompetent individuals lack insight into their own incompetence. On top of this, Vianello and Bar-Anan show willful ignorance by resisting arguments that undermine their motivated belief in dual-attitude models. As I show below, Vianello and Bar-Anan’s model has several unexplained results (e.g, negative loadings on method factors), worse fit than my model, and produces false evidence of incremental predictive validity for the implicit attitude factors.

Introduction

The skill set of psychology researchers is fairly limited. In some areas expertise is needed to create creative experimental setups. In other areas, some expertise in the use of measurement instruments (e.g., EEG) is required. However, for the most part, once data are collected, little expertise is needed. Data are analyzed with simple statistical tools like t-tests, ANOVAs, or multiple regression. These statistical methods are implemented in simple commands and no expertise is required to obtain results from statistics programs like SPSS or R.

Structural equation modeling is different because researchers have to specify a model that is fitted to the data. With complex data sets, the number of possible models that can be specified increases exponentially and it is not possible to specify all models and to simply pick the model with the best fit. Moreover, there will be many models with similar fit and it requires expertise to pick plausible models. Unfortunately, psychologists receive little formal training in structural equation modeling because graduate training relies heavily on training by supervisors rather than formal training. As most supervisors never received training in structural equation modeling, they cannot teach their graduate student how to perform these analyses. This means that expertise in structural equation modeling varies widely.

An inevitable consequence of wide variation in expertise is that individuals with low expertise have little insight into their limited abilities. This is known as the Dunning-Kruger effect that has been replicated in numerous studies. Even incentives to provide accurate performance estimates do not eliminate the overconfidence of individuals with low levels of expertise (Ehrlinger et al., 2008).

The Dunning-Kruger effect explains Vianello and Bar-Anan’s (2020) response to my article that presents another ill-fitting model that makes little theoretical sense. This overconfidence may also explain why they are unwilling to engage in a discussion of their model with me. They may not realize that my model is superior because they were unable to compare the models or to run more direct comparisons of the models. As their commentary is published in the influential journal Perspectives on Psychological Science and as many readers lack the expertise to evaluate the merits of their criticism, it is necessary to explain clearly why their criticism of my models is invalid and why their new alternative model is flawed.

Reproducing Vianello and Bar-Anan’s Model

I learned the hard way that the best way to fit a structural equation model is to start with small models of parts of the data and then to add variables or other partial models to build a complex model. The reason is that bad fit in smaller models can be easily identified and lead to important model modifications, whereas bad fit in a complex model can have thousands of reasons that are difficult to diagnose. In this particular case, I saw new reason to even fit a complex model for attitudes to political parties, racial groups, and the self. Instead I fitted separate models for each attitude domain. Vianello and Bar-Anan (2020) take issue with this decision.

As for estimating method variance across attitude domains, that is the very logic behind an MTMM design (Campbell & Fiske, 1959; Widaman, 1985): Method variance is shared across measures of different traits that use the same method (e.g., among indirect measures
of automatic racial bias and political preferences). Trait variance is shared across measures of the same trait that use different methods (e.g., among direct and indirect measures of racial attitude). Separating the MTMM matrix into three separate submatrices (one for each
trait), as Schimmack did in his article, misses a main advantage of an MTMM design.

This criticism is based on an outdated notion of validation by means of correlations in a multi-trait-multi-method matrix. In this MTMM tables, every trait is measured with all methods. For example, the Big Five traits are measured with students’ self-ratings, mothers’ ratings, and fathers’ ratings (5 traits x 3 methods). This is not possible for validation studies of explicit and implicit measures because it is assumed that explicit measures measure explicit constructs and implicit measures measure implicit constructs. Thus, it is not possible to fully cross traits and methods. This problem is evident in all models by Bar-Anan and Vianello and myself. Bar-Anan and Vianello make the mistake to assume that using implicit measures for several attitude domains solves this problem, but their assumption that we can use correlations between implicit measures in one domain and implicit measures in another domain to solve this problem is wrong. In fact, it makes matters worse because they fail to model method variance within a single attitude domain properly.

To show this problem, I first constructed measurement models for each attitude domain and then show that combining well-fitting models of three three domains produces a better fitting model than Vianello and Bar-Anan’s model.

Racial Bias

In their revised model, Vianello and Bar-Anan postulate three method factors. One for explicit measures, one for IAT-related measures, and one for the Affective Missatribution Paradigm and the Evaluative Priming Task. It is not possible to estimate a separate method factor for all explicit measures, but it is possible to allow for method factors that are unique to the IAT-related measures and one that is unique to the AMP and EPT. In the first model, I fitted this model to the measures of racial bias. The model appears to have good fit, RMSEA = .013, CFI = 973. In this model, the correlation between the explicit and implicit racial bias factors is r = .80.

However, it would be premature to stop the analysis here because overall fit values in models with many missing values are misleading (Zhang & Savaley, 2019). Even if fit were good, it is good practice to examine the modification indices to see whether some parameters are misspecified.

Inspection of the fit indices shows one very large Modification Index of 146.04 for the residual correlation between the feeling thermometer and the preference ratings. There is a very plausible explanation for this finding. These two measures are very similar and can share method variance. For example, social desirable responding could have the same effect on both ratings. This was the reason why I included only one of the two measures in my model. An alternative is to include both ratings and allow for the correlated residual to model shared method variance.

As predicted by the MI, model fit improved, RMSEA = .006, CFI = .995. Vianello and Bar-Anan (2020) might object that this finding is post-hoc after peeking at the data, while their model is specified theoretically. However, this argument is weak. If they really theoretically predicted that feeling thermometer and direct ratings share no method variance, it is not clear what theory they have in mind. After all, shared rating biases are very common. Moreover, their model also assumes shared method variance between these factors, but it also predicts that this method variance also influences dissimilar measures like the Modern Racism Scale and even ratings of other attitude objects. In short, neither their model nor my models are based on theories, in part because psychologists have ignored to develop and validate measurement theories. Even if it were theoretically predicted that feeling-thermometer and preference ratings do not share method variance, the large MI for this parameter would indicate that this theory is wrong. Thus, the data falsify this prediction. In the modified model, the implicit-explicit correlation increases from .80 to .90, providing even less support for the dual-attitude model.

Further inspection of the MI showed no plausible further improvements of the model. One important finding in this partial model is that there is no evidence of shared method variance between the AMP and EPT, r = -.04. Thus, closer inspection of the correlations among the racial attitude domain suggests two problems for Vianello and Bar-Anan’s model. There is evidence of shared method variance between two explicit measures and there is no evidence of shared method variance between two implicit measures, namely the AMP and EPT.

Next, I built a model for the political orientation domain starting with the specification in Vianello and Bar-Anan’s model. Once more, overall fit appears to be good, RMSEA = .014, CFI = .989. In this model, the correlation between the implicit and explicit factor is r = .9. However, inspection of the MI replicates a residual correlation between feeling thermometer and preference ratings. MI = 91.91. Allowing for this shared method variance improved model fit, RMSEA = .012, CFI = .993, but had little effect on the implicit-explicit correlation, r = .91. In this model, there was some evidence of shared method variance between the AMP and EPT, r = .13.

Next, I put these two well-fitting models together, leaving each model unchanged. The only new question is how measures of racial bias should be related to measures of political orientation. It is common to allow trait factors to correlate freely. This is also what Vianello and Bar-Anan did and I followed this common practices. Thus, there is no theoretical structure imposed on the trait correlations. I did not specify any additional relations for the method factors. If such relationships exist, this should lead to low fit. Model fit seemed to be good, RMSEA = .009, CFI = .982. The biggest MI was observed for the loading of the Modern Racism Scale (MRS) on the explicit political orientation factor, MI = 197.69. This is consistent with the item content of the MRS that combines racism with conservative politics (e.g., being against affirmative action). For that reason, I included the MRS in my measurement model of political orientation (Schimmack, 2020).

Vianello and Bar-Anan (2020) criticize my use of the MRS. “For instance, Schimmack chose to omit one of the indirect measures—the SPF—from the models, to include the Modern Racism Scale (McConahay, 1983) as an indicator of political evaluation, and to omit the thermometer scales from two of his models. We assume that Schimmack had good practical or theoretical reasons for his modelling decisions; unfortunately, however, he did not include those reasons.” If they had inspected the MI, they would have seen that my decision to use the MRS as a different method to measure political orientation was justified by the data as well as by the item-content of the scale.

After allowing for this theoretically expected relationship, model fit improves, chi2(df = 231) = 506.93, RMSEA = .007, CFI = .990. Next I examined whether the IAT method factor for racial bias is related to the IAT method factor for political orientation. Adding this relationship did not improve fit, chi2(230) = 506.65 = RMSEA = .007, CFI = .990. More important, the correlation was not significant, r = -.06. This is a problem for Vianello and Bar-Anan’s model that assumes the two method factors are identical. To test this hypothesis, I fitted a model with a single IAT method factor. This model had worse fit, chi2(231) = 526.99, RMSEA = .007, CFI = .989. Thus, there is no evidence for a general IAT method factor.

I next explored the possibility of a method factor for the explicit measures. I had identified shared method variance for the feeling thermometer and preference ratings for racial bias and for political orientation. I now modeled this shared method variance with method factors and let the two method factors correlate with each other. The addition of a correlation did not improve model fit, chi2(230) = 506.93, RMSEA = .007, CFI = .990 and the correlation between the two explicit method factors was not significant, r = .00. Imposing a single method factor for both attitude domains reduced model fit, chi2(df = 229) = 568.27, RMSEA = .008, CFI = .987.

I also tried to fit a single method factor for the AMP and EPT. The model only converged by constraining two loadings. Then model fit improved slightly, chi2(df = 230) = 501.75, RMSEA = .007, CFI = .990. The problem for Vianello and Bar-Anan is that the better fit was achieved with a negative loading on the method factor. This is inconsistent with the idea that a general method factor inflates correlations across attitude domains.

In sum, there is no evidence that method factors are consistent across the two attitude domains. Therefore I retained the basic model that specified method variance within attitude domains. I then added the three criterion variables to the model. As in Vianello and Bar-Anan’s model, contact was regressed on the explicit and implicit racial bias factor and previous voting and intention to vote were regressed on the explicit and implicit political orientation factors. The residuals were allowed to correlate freely, as in Vianello and Bar-Anan’s model.

Overall model fit decreased slightly for CFI, chi2(df = 297) = 668.61, RMSEA = .007, CFI = .988. MI suggested an additional relationship between the explicit political orientation factor and racial contact. Modifying the model accordingly improved fit slightly, chi2(df = 296) = 660.59, RMSEA = .007, CFI = .988. There were no additional MI involving the two voting measures.

Results were different from Vianello and Bar-Anan’s results. They reported that the implicit factors had incremental predictive validity for all three criterion measures.

In contrast, the model I am developing here shows no incremental predictive validity for the implicit factors.

It is important to note that I create the measurement model before I examined predictive validity. After the measurement model was created, criterion variables were added and the data determined the pattern of results. It is unclear how Vianello and Bar-Anan developed a measurement model with non-existing method factors that produced the desired outcome of significant incremental validity.

To try to reproduce their full result, I also added self-esteem measures to the model. To do so, I first created a measurement model for the self-esteem measures. The basic measurement model had poor fit, chi2(df = 58) = 434.49, RMSEA = .019, CFI = .885. Once more, the MI suggested that feeling-thermometer and preference ratings shared method variance. Allowing for this residual correlation increased model fit, chi2(df = 57) = 165.77, RMSEA = .010, CFI = .967. Another MI suggested a loading of the speeded task on the implicit factor, MI = 54.59. Allowing for this loading further improved model fit, chi2(df = 56) = 110.01, RMSEA = .007, CFI = .983. The crucial correlation between the explicit and implicit factor was r = .36. The correlation in Vianello and Bar-Anan’s model was r = .30.

I then added the self-esteem model to the model with the other two attitude domains, chi2(df = 695) = 1309.59, RMSEA = .006, CFI = .982. Next I added correlations of the IAT method factor for self-esteem with the two other IAT-method factors. This improved model fit, chi2(df = 693) = 1274.59, RMSEA = .006, CFI = .983. The reason was a significant correlation between the IAT method factors for self-esteem and racial bias. I offered an explanation for this finding in my article. Most White respondents associate self with good and White with good. If some respondents are better able to control their automatic tendencies, they will show less pro-self and pro-White biases. In contrast, Vianello and Bar-Anan have no theoretical explanation for a shared method factor across attitude domains. There was no significant correlation between IAT method factors for self-esteem and political orientation. The reason is that political orientation has more balanced automatic tendencies so that method variance does not favor one direction over the other.

This model had better fit with fewer parameters than Vianello and Bar-Anan’s model, chi2(df = 679) = 1719.39, RMSEA = .008, CFI = .970. The critical results of predictive validity remained unchanged.

I also fitted Vianello and Bar-Anan’s model and added four parameters that I identified as missing from their model: (a) the loading of the MRS on the explicit political orientation factor and (b) the correlations between feeling-thermometer and preference ratings for each domain. Making these adjustments improved model fit considerably, chi2(df = 675) = 1235.59, RMSEA = .006, CFI = .984. This modest adjustment altered the pattern of results for the prediction of the three criterion variables. Unlike Vianello and Bar-Anan’s model, the implicit factors no longer predicted any of the three criterion variables.

Conclusion

My interaction with Vianello and Bar-Anan are symptomatic of social psychologists misapplication of the scientific method. Rather than using data to test theories, data are being abused to confirm pre-existing beliefs. This confirmation bias goes against philosophies of science that have demonstrated the need to subject theories to strong tests and to allow data to falsify theories. Verificationism is so ingrained in social psychology that Vianello and Bar-Anan ended up with a model that showed significant incremental predictive validity for all three criterion measures in their model, when this model made several questionable assumptions. They may object that I am biased in the opposite direction, but I presented clear justifications for modeling decisions and my model fits better than their model. In my 2020 article, I showed that Bar-Anan also co-authored another article that exaggerated evidence of predictive validity that disappeared when I reanalyzed the data (Greenwald, Smith, Sriram, Bar-Anan, & Nosek, 2009). Ten years later, social psychologists claim that they have improved their research methods, but Vianello and Bar-Anan’s commentary in 2020 shows that social psychologists have a long way to go. If social psychologists want to (re)gain trust, they need to be willing to discard cherished theories that are not supported by data.

References

Bar-Anan, Y., & Vianello, M. (2018). A multi-method multi-trait test of the dual-attitude perspective. Journal of Experimental Psychology: General, 147(8), 1264–1272. https://doi.org/10.1037/xge0000383

Ehrlinger, J., Johnson, K., Banner, M., Dunning, D., & Kruger, J. (2008). Why the unskilled are unaware: Further explorations of (absent) self-insight among the incompetent. Organizational Behavior and Human Decision Processes, 105(1), 98–121. https://doi.org/10.1016/j.obhdp.2007.05.002

Greenwald, A. G., Smith, C. T., Sriram, N., Bar-Anan, Y., & Nosek, B. A. (2009). Implicit race attitudes predicted vote in the 2008 U.S. Presidential election. Analyses of Social Issues and Public Policy (ASAP), 9(1), 241–253. https://doi.org/10.1111/j.1530-2415.2009.01195.x

Schimmack U. The Implicit Association Test: A Method in Search of a Construct. Perspectives on Psychological Science. October 2019. doi:10.1177/1745691619863798

Vianello M, Bar-Anan Y. Can the Implicit Association Test Measure Automatic Judgment? The Validation Continues. Perspectives on Psychological Science. February 2020. doi:10.1177/1745691619897960

Zhang, X. & Savalei, V. (2020) Examining the effect of missing data on RMSEA and CFI under normal theory full-information maximum likelihood, Structural Equation Modeling: A Multidisciplinary Journal, 27:2, 219-239, DOI: 10.1080/10705511.2019.1642111

Racial Bias as a Trait

Prejudice is an important topic in psychology that can be examined from various perspectives. Nevertheless, prejudice research is typically studied by social psychologists. As a result, research has focused on social cognitive processes that are activated in response to racial stimuli (e.g., pictures of African Americans) and experimental manipulations of the situation (e.g., race of experimenter). Other research has focused on cognitive processes that can lead to the formation of racial bias (e.g., the minimal group paradigm). Sometimes this work has been based on a model of prejudice that assumes racial bias is a common attribute of all people (Devine, 1989) and that individuals only differ in their willingness or ability to act on their racial biases.

An alternative view is that racial biases vary across individuals and are shaped by experiences with out-group members. The most prominent theory is contact theory, which postulates that contact with out-group members reduces racial bias. In social psychology, individual differences in racial biases are typically called attitudes, where attitudes are broad dispositions to respond to a class of attitude objects in a consistent manner. For example, individuals with positive attitudes towards African Americans are more likely to have positive thoughts, feelings, and behaviors in interactions with African Americans.

The notion of attitudes as general dispositions shows that attitudes play the same role in social psychology that traits play in personality psychology. For example, extraversion is a general disposition to have more positive thoughts, feelings, and to engage more in social interactions. One important research question in personality psychology are the causes of variation in personality. Why are some people more extraverted than others? A related question is how stable personality traits are. If the causes of extraversion are environmental factors, extraversion should change when the environment changes. If the causes of extraversion are within the person (e.g., early childhood experiences, genetic differences), extraversion should be stable. Thus, the stability of personality traits over time is an empirical question that can only be answered in longitudinal studies that measure personality traits repeatedly. A meta-analysis shows that the Big Five personality traits are highly stable over time (Anusic & Schimmack, 2016).

In comparison, the stability of attitudes has received relatively little attention in social psychology because stable individual differences are often neglected in social cognitive models of attitudes. This is unfortunate because the origins of racial bias are important to the understanding of racial bias and to design interventions that help individuals to reduce their racial biases.

How stable are racial biases?

The lack of data has not stopped social psychologists from speculating about the stability of racial biases. “It’s not as malleable as mood and not as reliable as a personality trait. It’s in between the two–a blend of both a trait and a state characteristic” (Nosek in Azar, 2008). In 2019, Nosek was less certain about the stability of racial biases. “One is does that mean we have have some degree of trait variance because there is some stability over time and what is the rest? Is the rest error or is it state variance in some way, right. Some variation that is meaningful variation that is sensitive to the context of measurement. Surely it is some of both, but we don’t know how much” (The Psychology Podcast, 2019).

Other social psychologists have made stronger claims about the stability of racial bias. Payne argued that racial bias is a state because implicit bias measures show higher internal consistency than retest correlations (Payne, 2017). However, the comparison of internal consistency and retest correlations is problematic because situational factors may simply produce situation-specific measurement errors rather than reflecting real changes in the underlying trait; a problem that is well recognized in personality psychology. To examine this question more thoroughly, it is necessary to obtain multiple retests and decompose the variances into trait, state, and error variances (Anusic & Schimmack, 2016). Even this approach cannot distinguish between state variance and systematic measurement error, which requires multi-method data (Schimmack, 2019).

A Longitudinal Multi-Method Study of Racial Bias

A recent article reported the results of an impressive longitudinal study of racial bias with over 3,000 medical students who completed measures of racial bias and inter-group contact three times over a period of six year (first year of medical school, fourth year of medical school, 2nd year of residency) (Onyeador et al., 2019). I used the openly shared data to fit a multi-method state-trait-error model to the data (https://osf.io/78cqx/).

The model integrates several theoretical assumptions that are consistent with previous research (Schimmack, 2019). First, the model assumes that explicit ratings of racial bias (feeling thermometer) and implicit measures of racial bias (Implicit Association Test) are complementary measures of individual differences in racial bias. Second, the model assumes that one source of variance in racial bias is a stable trait. Third, the model assumes that racial bias differs across racial groups, in that Black individuals have more favorable attitudes towards Black people than members from other groups. Fourth, the model assumes that contact is negatively correlated with racial bias without making a strong causal assumption about the direction of this relationship. The model also assumes that Black individuals have more contact with Black individuals and that contact partially explains why Black individuals have less racial biases.

The new hypotheses that could be explored with these data concerned the presence of state variance in racial bias. First, state variance should produce correlations between the occasion specific variances of the two methods. That is, after statistically removing trait variance, residual state variance in feeling thermometer scores should be correlated with residual variances in IAT scores. For example, as medical students interact more with Black staff and patients in residency, their racial biases could change and this would produce changes in explicit ratings and in IAT scores. Second, state variance is expected to be somewhat stable over shorter time intervals because environments tend to be stable over shorter time intervals.

The model in Figure 1 met standard criteria of model fit, CFI = .997, RMSEA = .016.

Describing the model from left to right, race (0 = Black, 1 = White) has the expected relationship with quantity of contact (quant1) in year 1 (reflecting everyday interactions with Black individuals) and with the racial bias (att) factor. In addition, more contact is related to less pro-White bias (-.28). The attitude factor is a stronger predictor of the explicit trait factor (.78; ft; White feeling-thermometer – Black feeling-thermometer) than on the implicit trait factor (.60, iat). The influence of the explicit trait factor on measures on the three occasions (.58-.63) suggests that about one-third of the variance in these measures is trait variance. The same is true for individual IATs (.59-.62). The effect of the attitude factor on individual IATs (.60 * .60 = .36; .36^2 = .13 suggests that less than 20% of the variance in an individual IAT reflects racial bias. This estimate is consistent with the results from multi-method studies (Schimmack, 2019). However, these results suggests that the amount of valid trait variance can increase up to 36%, by aggregating scores of several IATs. In sum, these results provide first evidence that racial bias is stable over a period of six years and that both explicit ratings and implicit ratings capture trait variance in racial bias.

Turning to the bottom part of the model, there is weak evidence to suggest that residual variances (that are not trait variance) in explicit and implicit ratings are correlated. Although the correlation of r = .06 at time 1 is statistically significant, the correlations at time 2 (r = .03) and time 3 (r = .00) are not. This finding suggests that most of the residual variance is method specific measurement error rather than state-variance in racial bias. There is some evidence that the explicit ratings capture more than occasion-specific measurement error because state variance at time 1 predicts state variance at time 2 (r = .25) and from time 2 to time 3 (r = .20). This is not the case for the IAT scores. Finally, contact with Black medical staff at time 2 is a weak, but significant predictor of explicit measures of racial bias at time 2 and time 3, but it does not predict IAT scores at time 2 and 3. These findings do not support the hypothesis that changes in racial bias measures reflect real changes in racial biases.

The results are consistent with the only other multi-method longitudinal study of racial bias that covered only a brief period of three months. In this study, even implicit measures showed no convergent validity for the state (non-trait) variance on the same occasion (Cunningham, Preacher, & Banaji, 1995).

Conclusion

Examining predictors of individual differences in racial bias is important to understand the origins of racial biases and to develop interventions that help individuals to reduce their racial biases. Examining the stability of racial bias in longitudinal studies shows that these biases are stable dispositions and there is little evidence that they change with changing life-experiences. One explanation is that only close contact may be able to shift attitudes and that few people have close relationships with outgroup members. Thus stable environments may contribute to stability in racial bias.

Given the trait-like nature of racial bias, interventions that target attitudes and general dispositions may be relatively ineffective, as Onyeador et al.’s (2019) article suggested. Thus, it may be more effective to target and assess actual behaviors in diversity training. Expecting diversity training to change general dispositions may be misguided and lead to false conclusions about the effectiveness of diversity training programs.

Anti-Black Bias on the IAT predicts Pro-Black Bias in Behavior

Over 20 years ago, Anthony Greenwald and colleagues introduced the Implicit Association Test (IAT) as a measure of individual differences in implicit bias (Greenwald et al., 1998). The assumption underlying the IAT is that individuals can harbour unconscious, automatic, hidden, or implicit racial biases. These implicit biases are distinct from explicit bias. Somebody could be consciously unbiased, while their unconscious is prejudice. Theoretically, the opposite would also be possible, but taking IAT scores at face value, the unconscious is more prejudice than conscious reports of attitudes imply. It is also assumed that these implicit attitudes can influence behavior in ways that bypass conscious control of behavior. As a result, implicit bias in attitudes leads to implicit bias in behavior.

The problem with this simple model of implicit bias is that it lacks scientific support. In a recent review of validation studies, I found no scientific evidence that the IAT measures hidden or implicit biases outside of people’s awareness (Schimmack, 2019a). Rather, it seems to be a messy measure of consciously accessible attitudes.

Another contentious issue is the predictive validity of IAT scores. It is commonly implied that IAT scores predict bias in actual behavior. This prediction is so straightforward that the IAT is routinely used in implicit bias training (e.g., at my university) with the assumption that individuals who show bias on the IAT are likely to show anti-Black bias in actual behavior.

Even though the link between IAT scores and actual behavior is crucial for the use of the IAT in implicit bias training, this important question has been examined in relatively few studies and many of these studies had serious methodological limitations (Schimmack, 20199b).

To make things even more confusing, a couple of papers even suggested that White individuals’ unconscious is not always biased against Black people: “An unintentional, robust, and replicable Pro-Black bias in social judgment (Axt, Ebersole, & Nosek, 2016; Axt, 2017).

I used the open data of these two articles to examine more closely the relationship between scores on the attitude measures (the Brief Implicit Association Test & a direct explicit rating on a 7-point scale) and performance on a task where participants had to accept or reject 60 applicants into an academic honor society. Along with pictures of applicants, participants were provided with information about academic performance. These data were analyzed with signal-detection theory to obtain a measure of bias. Pro-White bias would be reflected in a lower admission standard for White applicants than for Black applicants. However, despite pro-White attitudes, participants showed a pro-Black bias in their admissions to the honor society.

Figure 1 shows the results for the Brief IAT. The blue lines show are the coordinates with 0 scores (no bias) on both tasks. The decreasing red line shows the linear relationship between BIAT scores on the x-axis and bias in admission decisions on the y-axis. The decreasing trend shows that, as expected, respondents with more pro-White bias on the BIAT are less likely to accept Black applicants. However, the picture also shows that participants with no bias on the BIAT have a bias to select more Black than White applicants. Most important, the vertical red line shows behavior of participants with the average performance on the BIAT. Even though these participants are considered to have a moderate pro-White bias, they show a pro-Black bias in their acceptance rates. Thus, there is no evidence that IAT scores are a predictor of discriminatory behavior. In fact, even the most extreme IAT scores fail to identify participants who discriminate against Black applicants.

A similar picture emerges for the explicit ratings of racial attitudes.

The next analysis examine convergent and predictive validity of the BIAT in a latent variable model (Schimmack, 2019). In this model, the BIAT and the explicit measure are treated as complementary measures of a single attitude for two reasons. First, multi-method studies fail to show that the IAT and explicit measures tap different attitudes (Schimmack, 2019a). Second, it is impossible to model systematic method variance in the BIAT in studies that use only a single implicit measure of attitudes.

The model also includes a group variable that distinguishes the convenience samples in Axt et al.’s studies (2016) and the sample of educators in Axt (2017). The grouping variable is coded with 1 for educators and 0 for the comparison samples.

The model meets standard criteria of model fit, CFI = .996, RMSEA = .002.

Figure 3 shows the y-standardized results so that relationships with the group variable can be interpreted as Cohen’s d effect sizes. The results show a notable difference (d = -59) in attitudes between the two samples with less pro-White attitudes for educators. In addition, educators have a small bias to favor Black applicants in their acceptance decisions (d = .19).

The model also shows that racial attitudes influence acceptance decisions with a moderate effect size, r = -.398. Finally, the model shows that the BIAT and the single-item explicit rating have modest validity as measures of racial attitudes, r = .392, .429, respectively. The results for the BIAT are consistent with other estimates that a single IAT has no more than 20% (.392^2 = 15%) valid variance. Thus, the results here are entirely consistent with the view that explicit and implicit measures tap a single attitude and that there is no need to postulate hidden, unconscious attitudes that can have an independent influence on behavior.

Based on their results, Axt et al. (2016) caution readers that the relationship between attitudes and behaviors is more complex than the common narrative of implicit bias assumes.

The authors “suggest that the prevailing emphasis on pro-White biases in judgment and behavior in the existing literature would improve by refining the theoretical understanding of under what conditions behavior favoring dominant or minority groups will occur.” (p. 33).

Implications

For two decades, the developers of the IAT have argued that the IAT measures a distinct type of attitudes that reside in individuals’ unconscious and can influence behavior in ways that bypass conscious control. As a result, even individuals who aim to be unbiased might exhibit prejudice in their behavior. Moreover, the finding that the majority of White people show a pro-White bias in their IAT scores was used to explain why discrimination and prejudice persist. This narrative is at the core of implicit bias training.

The problem with this story is that it is not supported by scientific evidence. First, there is no evidence that IAT scores reflect some form of unconscious or implicit bias. Rather, IAT scores seem to tap the same cognitive and affective processes that influence explicit ratings. Second, there is no evidence that processes that influence IAT scores can bypass conscious control of behavior. Third, there is no evidence that a pro-White bias in attitudes automatically produces a pro-White bias in actual behaviors. Not even Freud assumed that unconscious processes would have this effect on behavior. In fact, he postulated that various defense mechanisms may prevent individuals from acting on their undesirable impulses. Thus, the prediction that attitudes are sufficient to predict behavior is too simplistic.

Axt et al. (2016= speculate that “bias correction can occur automatically and without awareness” (p. 32). While this is an intriguing hypothesis, there is little evidence for such smart automatic control processes. This model also implies that it is impossible to predict actual behaviors from attitudes because correction processes can alter the influence of attitudes on behavior. This implies that only studies of actual behavior can reveal the ability of IAT scores to predict actual behavior. For example, only studies of actual behavior can demonstrate whether police officers with pro-White IAT scores show racial bias in the use of force. The problem is that 20 years of IAT research have uncovered no robust evidence that IAT scores actually predict important real-world behaviors (Schimmack, 2019b).

In conclusion, the results of Axt’s studies suggest that the use of the IAT in implicit bias training needs to be reconsidered. Not only are test scores highly variable and often provide false information about individuals’ attitudes; they also do not predict actual behavior of discrimination. It is wrong to assume that individuals who show a pro-White bias on the IAT are bound to act on these attitudes and discriminate against Black people or other minorities. Therefore, the focus on attitudes in implicit bias training may be misguided. It may be more productive to focus on factors that do influence actual behaviors and to provide individuals with clear guidelines that help them to act in accordance with these norms. The belief that this is not sufficient is based on an unsupported model of unconscious forces that can bypass awareness.

This conclusion is not totally new. In 2008, Blanton criticized the use of the IAT in applied settings (IAT: Fad or fabulous?)

“There’s not a single study showing that above and below that cutoff people differ in any way based on that score,” says Blanton.

And Brian Nosek agreed.

Guilty as charged, says the University of Virginia’s Brian Nosek, PhD, an IAT developer.

However, this admission of guilt has not changed behavior. Nosek and other IAT proponents continue to support Project Implicit that provided millions of visitors with false information about their attitudes or mental health issues based on a test with poor psychometric properties. A true admission of guilt would be to stop this unscientific and unethical practice.

References

Axt, J.R. (2017). An unintentional pro-Black bias in judgement among educators. British Journal of Educational Psychology, 87, 408-421.

Axt, J.R., Ebersole, C.R. & Nosek, B.A. (2016). An unintentional, robust, and replicable pro-Black bias in social judgment. Social Cognition34, 1-39.

Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998). Measuring individual differences in implicit cognition: The Implicit Association Test. Journal of Personality and Social Psychology, 74, 1464–1480.

Schimmack, U. (2019). The Implicit Association Test: A Method in Search of a construct. Perspectives on Psychological Sciencehttps://doi.org/10.1177/1745691619863798

Schimmack, U. (2019). The race IAT: A Case Study of The Validity Crisis in Psychology.
https://replicationindex.com/2019/02/06/the-race-iat-a-case-study-of-the-validity-crisis-in-psychology/

Open Communication about the invalidity of the race IAT

In the old days, most scientific communication occured behind closed doors, when reviewers provide anonymous peer-reviews that determine the fate of manuscripts. In the old days, rejected manuscripts would not be able to contribute to scientific communications because nobody would know about them.

All of this has changed with the birth of open science. Now authors can share manuscripts on pre-print servers and researchers can discuss merits of these manuscripts on social media. The benefit of this open scientific communication is that more people can join in and contribute to the communication.

Yoav Bar-Anan co-authored an article with Brian Nosek titled “Scientific Utopia: I. Opening Scientific Communication.” In this spirit of openness, I would like to have an open scientific communication with Yoav and his co-author Michelangelo Vianello about their 2018 article “A Multi-Method Multi-Trait Test of the Dual-Attitude Perspective

I have criticized their model in an in press article in Perspectives of Psychological Science (Schimmack, 2019). In a commentary, Yoav and Michelangelo argue that their model is “compatible with the logic of an MTMM investigation (Campbell & Fiske, 1959). They argue that it is important to have multiple traits to identify method variance in a matrix with multiple measures of multiple traits. They then propose that I lost the ability to identify method variance by examining one attitude (i.e., race, self-esteem, political orientation) at a time. They then point out that I did not include all measures and included the Modern Racism Scale as an indicator of political orientation to note that I did not provide a reason for these choices. While this is true, Yoav and Michelangelo had access to the data and could have tested whether these choices made any differences. They do not. This is obvious for the modern racism scale that can be eliminated from the measurement model without any changes in the overall model.

To cut to the chase, the main source of disagreement is the modelling of method variance in the multi-trait-multi-method data set. The issue is clear when we examine the original model published in Bar-Anan and Vianello (2018).

In this model, method variance in IATs and related tasks like the Brief IAT is modelled with the INDIRECT METHOD factor. The model assumes that all of the method variance that is present in implicit measures is shared across attitude domains and across all implicit measures. The only way for this model to allow for different amounts of method variance in different implicit measures is by assigning different loadings to the various methods. Moreover, the loadings provide information about the nature of the shared variance and the amount of method variance in the various methods. Although this is valuable and important information, the authors never discuss this information and its implications.

Many of these loadings are very small. For example, the loading of the race IAT and the brief race IAT are .11 and .02. In other words, the correlation between these two measures is inflated by .11 * .02 = .0022 points. This means that the correlation of r = .52 between these two measures is r = .5178 after we remove the influence of method variance.

It makes absolutely no sense to accuse me of separating the models, when there is no evidence of implicit method variance that is shared across attitudes. The remaining parameter estimates are not affected if a factor with low loadings is removed from a model.

Here I show that examining one attitude at a time produces exactly the same results as the full model. I focus on the most controversial IAT; the race IAT. After all, there is general agreement that there is little evidence of discriminant validity for political orientation (r = .91, in the Figure above), and there is little evidence for any validity in the self-esteem IAT based on several other investigations of this topic with a multi-method approach (Bosson et al., 2000; Falk et al., 2015).

Model 1 is based on Yoav and Michelangelo’s model that assumes that there is practically no method variance in IAT-variants. Thus, we can fit a simple dual-attitude model to the data. In this model, contact is regressed onto implicit and explicit attitude factors to see the unique contribution of the two factors without making causal assumptions. The model has acceptable fit, CFI = .952, RMSEA = .013.

The correlation between the two factors is .66, while it is r = .69 in the full model in Figure 1. The loading of the race IAT on the implicit factor is .66, while it is .62 in the full model in Figure 1. Thus, as expected based on the low loadings on the IMPLICIT METHOD factor, the results are no different when the model is fitted only to the measure of racial attitudes.

Model 2 makes the assumption that IAT-variants share method variance. Adding the method factor to the model increased model fit, CFI = .973, RMSEA = .010. As the models are nested, it is also possible to compare model fit with a chi-square test. With five degrees of freedom difference, chi-square changed from 167. 19 to 112.32. Thus, the model comparison favours the model with a method factor.

The main difference between the models is that there the evidence is less supportive of a dual attitude model and that the amount of valid variance in the race IAT decreases from .66^2 = 43% to r = .47^2 = 22%.

In sum, the 2018 article made strong claims about the race IAT. These claims were based on a model that implied that there is no systematic measurement error in IAT scores. I showed that this assumption is false and that a model with a method factor for IATs and IAT-variants fits the data better than a model without such a factor. It also makes no theoretical sense to postulate that there is no systematic method variance in IATs, when several previous studies have demonstrated that attitudes are only one source of variance in IAT scores (Klauer, Voss, Schmitz, & Teige-Mocigemba, 2007).

How is it possible that the race IAT and other IATs are widely used in psychological research and on public websites to provide individuals with false feedback about their hidden attitudes without any evidence of its validity as an individual difference measure of hidden attitudes that influence behaviour outside of awareness?

The answer is that most of these studies assumed that the IAT is valid rather than testing its validity. Another reason is that psychological research is focused on providing evidence that confirms theories rather than subjecting theories to empirical tests that they may fail. Finally, psychologists ignore effect sizes. As a result, the finding that IAT scores have incremental predictive validity of less than 4% variance in a criterion is celebrated as evidence for the validity of IATs, but even this small estimate is based on underpowered studies and may shrink in replication studies (cf. Kurdi et al., 2019).

It is understandable that proponents of the IAT respond with defiant defensiveness to my critique of the IAT. However, I am not the first to question the validity of the IAT, but these criticisms were ignored. At least Banaji and Greenwald recognized in 2013 that they do “not have the luxury of believing that what appears true and valid now will always appear so” (p. xv). It is time to face the facts. It may be painful to accept that the IAT is not what it was promised to be 21 years ago, but that is what the current evidence suggests. There is nothing wrong with my models and their interpretation, and it is time to tell visitors of the Project Implicit website that they should not attach any meaning to their IAT scores. A more productive way to counter my criticism of the IAT would be to conduct a proper validation study with multiple methods and validation criteria that are predicted to be uniquely related to IAT scores in a preregistered study.

References

Bosson, J. K., Swann, W. B., Jr., & Pennebaker, J. W. (2000). Stalking the perfect measure of implicit self-esteem: The blind men and the elephant revisited? Journal of Personality and Social Psychology, 79, 631–643.

Falk, C. F., Heine, S. J., Takemura, K., Zhang, C. X., & Hsu, C. (2015). Are implicit self-esteem measures valid for assessing individual and cultural differences. Journal of Personality, 83, 56–68. doi:10.1111/jopy.12082

Klauer, K. C., Voss, A., Schmitz, F., & Teige-Mocigemba, S. (2007). Process components of the Implicit Association Test: A diffusion-model analysis. Journal of Personality and Social Psychology, 93, 353–368.

Kurdi, B., Seitchik, A. E., Axt, J. R., Carroll, T. J., Karapetyan, A., Kaushik, N., . . . Banaji, M. R. (2019). Relationship between the Implicit Association Test and intergroup behavior: A meta-analysis. American Psychologist, 74, 569–586.

Brain Nosek explains the IAT

I spent 20 minutes, actually more than 20 minutes because I had to rewind to transcribe, listening to a recent podcast in which Brain Nosek was asked some questions about the IAT and implicit bias training (The Psychology Podcast, August 1, 2019).

Scott Barry Kaufman: How do you see the IAT now and how did you see it when you started work on Project Implicit? How discrepant are these stats of mind?

Brian Nosek: I hope I have learned a lot from all the research that we have done on it over the years. In the big picture I have the same view that I have had since we did the first set of studies. It is a great tool for research purposes and we have been able to learn a lot about the tool itself and about human behavior and interaction with the tool and a lot about the psychology of things that are [gap] occur with less control AND less awareness than just asking people how they feel about topics. So that has been and continues to be a very productive research area for trying to understand better how humans work.

And then the main concern that we had at onset and that is actually a lot of the discussion of even creating the website is the same anticipated some of the concerns and overuses that happened with the IAT in the present and that is the natural – I don’t know if natural is the right word – the common desire that people have for simple solutions and thinking well a measure is a direct indicator of something that we care about and it shouldn’t have any error in measurement and it should be applicable to lots and lots of situations.  And thus lots of potential of misuse of the IAT despite it being a very productive research tool and education too.  I like the experience of doing it and delivering to an audience and the discussion it provokes; what is it that it means, what does it mean about me, what does it mean about the world; those are really productive intellectual discussions and debates.  But the risk part the overapplication of the IAT for selection processes. We should use this. We should [?] use this for deciding who gets a job or not; we should [?] use this who is on a jury or not. Those are the kind of real-world applications of it as a measure that go far beyond its validity.  And so this isn‘t exact answering your question because even at the very beginning when we launched the website we said explicitly it should not be used for these purposes and I still believe this to be true. What has changed over time is the refinement of where it is we understand the evidence base against some of the major questions. And what is amazing about it is that there has been so much research and we still don’t have a great handle on really big questions relating to the IAT and measures like it.  So this is just part of [unclear]  how hard it is to actually make progress in the study of human behavior.   

Scott Barry Kaufman:  Let’s talk shop for a second [my translation; enough with the BS]. My dissertation at Yale a couple of year after years was looking at the question are there individual differences in implicit cognition.  And the idea was to ask this question because from a trait perspective I felt that was a huge gap in the literature. There was so much research on the reliability and validity of IQ tests for instance, but I wanted to ask the question if we adapt some of these implicit cognition measures from the social psychological experimental literature for an individual differences paradigm you know are they reliable and stable differences. And I have a whole appendix of failed experiments – by the way, you should tell how to publish that some day but we’ll get to that in a second, but so much of my dissertation, I am putting failed in quotes because you know I mean that was useful information … it was virtually impossible to capture reliable individual differences that cohered over time but I did find one that did and I published that as a serial reaction time task, but anyway, before we completely lose my audience which is a general audience I just want to say that I am trying to link this because for me one of the things that I am most wary about with the IAT is like – and this might be more of a feature than a bug – but it may be capturing at this given moment in time when a person is taking the test it is capturing a lot of the societal norms and influences are on that person’s associations but not capturing so much an intrinsic sort of stable individual differences variable. So I just wanted to throw that out and see what your current thoughts on that are.

Brian Nosek:   Yeah, it is clear that it is not trait like in the same way that a measure like the Big Five for personality is trait-like.  It does show stability over time, but much more weakly than that.  Across a variety of topics you might see a test-retest correlation for the IAT measuring the same construct of about .5  The curiosity for this is;  I guess it is a few curiosities. One is does that mean we have have some degree of trait variance because there is some stability over time and what is the rest? Is the rest error or is it state variance in some way, right. Some variation that is meaningful variation that is sensitive to the context of measurement. Surely it is some of both, but we don’t know how much. And there isn’t yet a real good insight on where the prediction components of the IAT are and how it anticipates behavior, right.  If we could separate in a real reliable way the trait part, the state part, and the error part, than we should be able to uniquely predict different type of things between the trait, the state, and the trait components. Another twist which is very interesting that is totally understudied in my view is the variations in which it is state or trait like seems to vary by the topic you are investigating. When you do a Democrat – Republican IAT, to what extent do people favor one over the other, the correlation with self-report is very strong and the stability over time is stronger than when you measure Black-White or some of the other types of topics. So there is also something about the attitude construct itself that you are assessing that is not as much measurement based but that is interacting with the measure that is anticipating the extent to which it is trait or state like. So these are all interesting things that if I had time to study them would be the problems I would be studying, but I had to leave that aside

Scott Barry Kaufman: You touch on a really interesting point about this. How would you measure the outcome of this two-day or week- training thing? It seems that would not be a very good thing to then go back to the IAT and see a difference between the IAT, IAT pre and IAT-post, doesn’t seem like the best outcome you know you’d want, I mean you ….

Brian Nosek I mean you could just change the IAT and that would be the end of it. But, of course, if that doesn’t actually shift behavior then what was the point?

Scott Barry Kaufman:  to what extent are we making advances in demonstrating that there are these implicit influences on explicit behavior that are outside of our value system? Where are we at right now? 

[Uli, coughs, Bargh, elderly priming]

Brian Nosek: Yeah, that is a good question. I cannot really comment on the micro-aggression literature. I don’t follow that as a distinct literature, but on the general point I think it is the big picture story is pretty clear with evidence which is we do things with automaticity, we do things that are counterproductive to our interests all the time, and sometimes we recognize we are doing it, sometimes we don’t, but a lot of time it is not controllable.  But that is a very big picture, very global, very non-specific point.

If you want to find out what 21 years of research on the IAT have shown, you can read my paper (Schimmack, in press, PoPS). In short,

  • most of the variance in the race IAT (Black-White) is random and systematic measurement error.
  • Up to a quarter of the variance reflects racial attitudes that are also reflected in self-report measures of racial attitudes; most clearly in direct ratings of feelings towards Blacks and Whites.
  • there is little evidence that any of the variance in IAT scores reflects some implicit attitudes that are outside of people’s awareness
  • there is no reliable evidence that IAT scores predict discriminatory behavior in the real world
  • visitors of Project Implicit are given invalid feedback that they may hold unconscious biases and are not properly informed about the poor psychometric properties of the test.
  • Founders of Project Implicit have not disclosed how much money they make from speaking engagements related to Project Implicit, royalties from the book “Blindspot,” and do not declare conflict of interest in IAT-related publications.
  • It is not without irony that educators on implicit bias may fail to realize that they have an implicit bias in reading the literature and to dismiss criticism.

The Implicit Association Test: A Measure in Search of a Construct (in press, PoPS)

Here is a link to the manuscript, data, and MPLUS scripts for reproducibility. https://osf.io/mu7e6/

ABSTRACT

Greenwald et al. (1998) proposed that the 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 multi-method studies and find no or weak evidence for 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% valid variance).  Most of this valid variance stems from a distinction between individuals with opposing attitudes, while 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 multi-method measurement model.

Keywords:  Personality, Individual Differences, Social Cognition, Measurement, Construct Validity, Convergent Validity, Discriminant Validity, Structural Equation Modeling

HIGHLIGHTS

Despite its popularity, relatively little is known about the construct validity of the IAT.

As Cronbach (1989) pointed out, construct validation is better examined by independent experts than by authors of a test because “colleagues are especially able to refine the interpretation, as they compensate for blind spots and capitalize on their own distinctive experience” (p. 163).

It is of utmost importance to determine how much of the variance in IAT scores is valid variance and how much of the variance is due to measurement error, especially when IAT scores are used to provide individualized feedback.

There is also no consensus in the literature whether the IAT measures something different from explicit measures.

In conclusion, while there is general consensus to make a distinction between explicit measures and implicit measures, it is not clear what the IAT measures

To complicate matters further, the validity of the IAT may vary across attitude objects. After all the IAT is a method, just like Likert scales are a method, and it is impossible to say that a method is valid (Cronbach, 1971).

At present, relatively little is known about the contribution of these three parameters to observed correlations in hundreds of mono-method studies.

A Critical Review of Greenwald et al.’s (1998) Original Article

In conclusion, the seminal IAT article introduced the IAT as a measure of implicit constructs that cannot be measured with explicit measures, but it did not really test this dual-attitude model.

Construct Validity in 2007

In conclusion, the 2007 review of construct validity revealed major psychometric challenges for the construct validity of the IAT, which explains why some researchers have concluded that the IAT cannot be used to measure individual differences (Payne et al., 2017).  It also revealed that most studies were mono-method studies that could not examine convergent and discriminant validity

Cunningham, Preacher and Banaji (2001)

Another noteworthy finding is that a single factor accounted for correlations among all measures on the same occasion and across measurement occasions. This finding shows that there were no true changes in racial attitudes over the course of this two-month study.  This finding is important because Cunningham et al.’s (2001) study is often cited as evidence that implicit attitudes are highly unstable and malleable (e.g., Payne et al., 2017). This interpretation is based on the failure to distinguish random measurement error and true change in the construct that is being measured (Anusic & Schimmack, 2016).  While Cunningham et al.’s (2001) results suggest that the IAT is a highly unreliable measure, the results also suggest that the racial attitudes that are measured with the race IAT are highly stable over periods of weeks or months. 

Bar-Anan & Vianello, 2018

this large study of construct validity also provides little evidence for the original claim that the IAT measures a new construct that cannot be measured with explicit measures, and confirms the estimate from Cunningham et al. (2001) that about 20% of the variance in IAT scores reflects variance in racial attitudes.

Greenwald et al. (2009)

“When entered after the self-report measures, the two implicit measures incrementally explained 2.1% of vote intention variance, p=.001, and when political conservativism was also included in the model, “the pair of implicit measures incrementally predicted only 0.6% of voting intention variance, p = .05.”  (Greenwald et al., 2009, p. 247).

I tried to reproduce these results with the published correlation matrix and failed to do so. I contacted Anthony Greenwald, who provided the raw data, but I was unable to recreate the sample size of N = 1,057. Instead I obtained a similar sample size of N = 1,035.  Performing the analysis on this sample also produced non-significant results (IAT: b = -.003, se = .044, t = .070, p = .944; AMP: b = -.014, se = .042, t = 0.344, p = .731).  Thus, there is no evidence for incremental predictive validity in this study.

Axt (2018)

With N = 540,723 respondents, sampling error is very small, σ = .002, and parameter estimates can be interpreted as true scores in the population of Project Implicit visitors.  A comparison of the factor loadings shows that explicit ratings are more valid than IAT scores. The factor loading of the race IAT on the attitude factor once more suggests that about 20% of the variance in IAT scores reflects racial attitudes

Falk, Heine, Zhang, and Hsu (2015)

Most important, the self-esteem IAT and the other implicit measures have low and non-significant loadings on the self-esteem factor. 

Bar-Anan & Vianello (2018)

Thus, low validity contributes considerably to low observed correlations between IAT scores and explicit self-esteem measures.

Bar-Anan & Vianello (2018) – Political Orientation

More important, the factor loading of the IAT on the implicit factor is much higher than for self-esteem or racial attitudes, suggesting over 50% of the variance in political orientation IAT scores is valid variance, π = .79, σ = .016.  The loading of the self-report on the explicit ratings was also higher, π = .90, σ = .010

Variation of Implicit – Explicit Correlations Across Domains

This suggests that the IAT is good in classifying individuals into opposing groups, but it has low validity of individual differences in the strength of attitudes.

What Do IATs Measure?

The present results suggest that measurement error alone is often sufficient to explain these low correlations.  Thus, there is little empirical support for the claim that the IAT measures implicit attitudes that are not accessible to introspection and that cannot be measured with self-report measures. 

For 21 years the lack of discriminant validity has been overlooked because psychologists often fail to take measurement error into account and do not clearly distinguish between measures and constructs.

In the future, researchers need to be more careful when they make claims about constructs based on a single measure like the IAT because measurement error can produce misleading results.

Researchers should avoid terms like implicit attitude or implicit preferences that make claims about constructs simply because attitudes were measured with an implicit measure

Recently, Greenwald and Banaji (2017) also expressed concerns about their earlier assumption that IAT scores reflect unconscious processes.  “Even though the present authors find themselves occasionally lapsing to use implicit and explicit as if they had conceptual meaning, they strongly endorse the empirical understanding of the implicit– explicit distinction” (p. 862).

How Well Does the IAT Measure What it Measures?

Studies with the IAT can be divided into applied studies (A-studies) and basic studies (B-studies).  B-studies employ the IAT to study basic psychological processes.  In contrast, A-studies use the IAT as a measure of individual differences. Whereas B-studies contribute to the understanding of the IAT, A-studies require that IAT scores have construct validity.  Thus, B-studies should provide quantitative information about the psychometric properties for researchers who are conducting A-studies. Unfortunately, 21 years of B-studies have failed to do so. For example, after an exhaustive review of the IAT literature, de Houwer et al. (2009) conclude that “IAT effects are reliable enough to be used as a measure of individual differences” (p. 363).  This conclusion is not helpful for the use of the IAT in A-studies because (a) no quantitative information about reliability is given, and (b) reliability is necessary but not sufficient for validity.  Height can be measured reliably, but it is not a valid measure of happiness. 

This article provides the first quantitative information about validity of three IATs.  The evidence suggests that the self-esteem IAT has no clear evidence of construct validity (Falk et al., 2015).  The race-IAT has about 20% valid variance and even less valid variance in studies that focus on attitudes of members from a single group.  The political orientation IAT has over 40% valid variance, but most of this variance is explained by group-differences and overlaps with explicit measures of political orientation.  Although validity of the IAT needs to be examined on a case by case basis, the results suggest that the IAT has limited utility as a measurement method in A-studies.  It is either invalid or the construct can be measured more easily with direct ratings.

Implications for the Use of IAT scores in Personality Assessment

I suggest to replace the reliability coefficient with the validity coefficient.  For example, if we assume that 20% of the variance in scores on the race IAT is valid variance, the 95%CI for IAT scores from Project Implicit (Axt, 2018), using the D-scoring method, with a mean of .30 and a standard deviation of.46 ranges from -.51 to 1.11. Thus, participants who score at the mean level could have an extreme pro-White bias (Cohen’s d = 1.11/.46 = 2.41), but also an extreme pro-Black Bias (Cohen’s d = -.51/.46 = -1.10).  Thus, it seems problematic to provide individuals with feedback that their IAT score may reveal something about their attitudes that is more valid than their beliefs. 

Conclusion

Social psychologists have always distrusted self-report, especially for the measurement of sensitive topics like prejudice.  Many attempts were made to measure attitudes and other constructs with indirect methods.  The IAT was a major breakthrough because it has relatively high reliability compared to other methods.  Thus, creating the IAT was a major achievement that should not be underestimated because the IAT lacks construct validity as a measure of implicit constructs. Even creating an indirect measure of attitudes is a formidable feat. However, in the early 1990s, social psychologists were enthralled by work in cognitive psychology that demonstrated unconscious or uncontrollable processes (Greenwald & Banaji, 1995). Implicit measures were based on this work and it seemed reasonable to assume that they might provide a window into the unconscious (Banaji & Greenwald, 2013). However, the processes that are involved in the measurement of attitudes with implicit measures are not the personality characteristics that are being measured.  There is nothing implicit about being a Republican or Democrat, gay or straight, or having low self-esteem.  Conflating implicit processes in the measurement of attitudes with implicit personality constructs has created a lot of confusion. It is time to end this confusion. The IAT is an implicit measure of attitudes with varying validity.  It is not a window into people’s unconscious feelings, cognitions, or attitudes.