Measuring implicit attitudes: A positive framing bias flaw in the Implicit Relational Assessment Procedure (IRAP).

Abstract

How can implicit attitudes best be measured? The Implicit Relational Assessment Procedure (IRAP), unlike the Implicit Association Test (IAT), claims to measure absolute, not just relative, implicit attitudes. In the IRAP, participants make congruent (Fat Person-Active:False; Fat Person-Unhealthy: True) or incongruent (Fat Person-Active: True; Fat PersonUnhealthy: False) responses in different blocks of trials. IRAP experiments have reported positive or neutral implicit attitudes (e.g., neutral attitudes towards fat people) in cases where negative attitudes are normally found on explicit or other implicit measures. It was hypothesized that these results might reflect a Positive Framing Bias (PFB) that occurs when participants complete the IRAP. Implicit attitudes towards categories with varying prior associations (nonwords, social systems, flowers and insects, thin and fat people) were measured. Three conditions (standard, positive framing, and negative framing) were used to measure whether framing influenced estimates of implicit attitudes. It was found that IRAP scores were influenced by how the task was framed to the participants, that the framing effect was modulated by the strength of prior stimulus associations and that a default PFB led to an overestimation of positive implicit attitudes when measured by the IRAP. Overall, the findings question the validity of the IRAP as a tool for the measurement of absolute implicit attitudes. A new tool (Simple Implicit Procedure: SIP) for measuring absolute, not just relative, implicit attitudes is proposed. Implicit attitudes are automatic evaluations that occur outside conscious awareness and are measured without requiring respondents to introspect on their feelings. Explicit attitudes in contrast are the result of deliberate introspection and controlled evaluative judgment (Greenwald & Banaji, 1995). One reason for measuring implicit attitudes is that participants may use self-presentation tactics or respond in a socially desirable manner on explicit self-reports to avoid being perceived as prejudiced. Implicit measures can also be useful in areas where participants might be unwilling to reveal personal psychological attributes or are unaware of these psychological attributes (for a review of implicit attitudes and the tools used to measure them see Gawronski & De Houwer, 2014).The current gold standard method for assessing implicit attitudes is the Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, 1998) and this tool has been increasingly used in clinically relevant areas (see https://implicit.harvard.edu/implicit/pimh/).The IAT has shown promise in predicting self-harm (Randall, Rowe, Dong, Nock, & Colman 2013), social anxiety disorders (Teachman & Allen, 2007) and suicidal ideation (Harrison, Stritzke, Fay, Ellison, & Hudaib, 2014). The current study questions the validity of a recently developed implicit measure, the Implicit Relational Assessment Procedure (IRAP; BarnesHolmes, Barnes-Holmes, Stewart, & Boles, 2010a), which has also been u...

Emotional Faces in Symbolic Relations: A Happiness Superiority Effect Involving the Equivalence Paradigm

“…It seems plausible to consider that the orienting function of the response option true may have been stronger than the orienting function of false ( true may have a stronger orienting function since it frequently serves as a confirmatory, favorable, or positive response in natural language). Indeed, previous research (e.g., O’Shea et al, 2016) has shown that participants may find it easier to respond true to positive stimuli than to press false , if these are the IRAP response options. Based on the current results, it is possible to suggest that a behavioral bias to relate happy faces and true have emerged in the course of the IRAP trial types that presented such a combination (i.e., trial-types 1 and 3).…”

From the IRAP and REC model to a multi-dimensional multi-level framework for analyzing the dynamics of arbitrarily applicable relational responding

“…For example, some of the earliest IRAP publications noted that response times on an IRAP may be influenced by general verbal biases typically found in natural languages, such as the tendency in English to respond "True" more quickly than "False" (Barnes-Holmes, Murphy et al, 2010, p. 62). Indeed, recent evidence indicates that IRAP effects may be influenced by these types of verbal biases and related variables (e.g., Finn et al, 2016;Maloney & BarnesHolmes, 2016;O'Shea, Watson, & Brown, 2016). On balance, the impact of such variables on the predictive validity of the IRAP has yet to be determined, and it is indeed possible that the sensitivity of the IRAP to natural-language biases contributes towards its relatively strong predictive validity (see Vahey et al, 2015).…”

Faking revisited: Exerting strategic control over performance on the Implicit Relational Assessment Procedure

“…Thus, controlling the direction and magnitude of one's IRAP effects seems to be relatively more difficult to achieve when compared with alternative indirect procedures. Critically, however, support for this claim necessitates future research in which the relative sensitivities to manipulation of IRAP, IAT, and AMP effects and other newly introduced indirect procedures (e.g., De Houwer et al, ; O'Shea, Watson, & Brown, ) are directly compared. Surprisingly, past work has tended to focus on the degree to which a single implicit measure can be faked and, in some instances, compared performances on indirect procedures with those on direct procedures.…”