Assessing Children’s Implicit Attitudes Using the Affect Misattribution Procedure

Abstract

In the current research, we examined whether the Affect Misattribution Procedure (AMP) could be successfully adapted as an implicit measure of children’s attitudes. We tested this possibility in 3 studies with 5- to 10-year-old children. In Study 1, we found evidence that children misattribute affect elicited by attitudinally positive (e.g., cute animals) and negative (e.g., aggressive animals) primes to neutral stimuli (inkblots). In Study 2, we found that, as expected, children’s responses following flower and insect primes were moderated by gender. Girls (but not boys) were more likely to judge inkblots as pleasant when they followed flower primes. Children in Study 3 showed predicted affect misattribution following happy-face compared with sad-face primes. In addition, children’s responses on this child-friendly AMP predicted their self-reported empathy: The greater children’s spontaneous misattribution of affect following happy and sad primes, the more children reported feeling the joy and pain of others. These studies provide evidence that the AMP can be adapted as an implicit measure of children’s attitudes, and the results of Study 3 offer novel insight into individual differences in children’s affective responses to the emotional expressions of others.

ACKNOWLEDGMENTS

We would like to thank members of the Interpersonal and Social Cognition Lab for their helpful comments and their assistance with data collection. We would also like to thank the Ontario Science Center administration as well as parents and children who very generously provided us with the opportunity to conduct this research.

FUNDING

This research was supported by an Social Sciences and Humanities Research Council Standard Research Grant (SSHRC-SRG) and a Canadian Foundation for Innovation Leaders Opportunity Fund (CFI-LOF) grant awarded to the second author.

Notes

¹ This age categorization is similar to what has previously been used in research examining the development of children’s implicit social cognition (e.g., Baron & Banaji, 2006; Dunham, Baron, & Banaji, 2006).

² Inkblots were used as neutral targets instead of Chinese characters (Payne et al., 2005) to control for potential familiarity with Chinese characters that might occur as a consequence of our racially diverse participants and testing location. To determine target neutrality, seven adults rated 102 inkblots on a 3-point scale (negative [−1], neutral [0], positive [+1]) and a 2-point scale (negative [0], positive [+1]). Scores were summed across the seven raters for each of these two scales. Twenty-two of the inkblots were consistently rated as positive or negative (e.g., the summed scores were equal to or greater than |5| on the 3-point scale or were less than 2 or greater than 5 on the 2-point scale). These inkblots were removed from the pool. A one-sample t test on the summed scores for the remaining 80 inkblots revealed that the mean of the 3-point scale was not significantly different from neutral (M = 0.19, SD = 1.98), t(79) = 0.85, p = .40.

³ For just less than half of the participants (n = 30), the inkblots were presented for 125 ms. For all other participants, the inkblots were presented for 150 ms. The proportion of pleasant responses following positive and negative primes did not differ by target duration, Fs < 0.26, ps > .61.

⁴ Studies with adults have shown that greater positive affect emerges in response to happy as compared with sad faces (e.g., Donges et al., 2012; Rohr et al., 2012). Although the happy/sad ch-IAT has not been validated with children, this measure was included to examine whether children would show a similar pattern of implicit preference on both the ch-AMP and ch-IAT. Because “the IAT is a very flexible task that can be used to assess almost any type of association between pairs of concepts” (Gawronski & De Houwer, 2014, p. 285) and research with this measure has demonstrated robust effects across a variety of domains with both children (see Olson & Dunham, 2010, for a review) and adults (see Gawronski & De Houwer, 2014, for a review), we felt that this was a useful measure for comparison, despite the lack of prior use with children.

⁵ To examine the incremental predictive ability of absolute attitude estimates on the ch-AMP, we conducted similar hierarchical regression analyses where the relative ch-AMP preference score was replaced by adjusted responses following happy (centered) and sad (centered) primes (r = .18, p = .13) in the third step of the model. Including the ch-AMP priming indexes in the third step significantly increased the amount of empathy variance explained above and beyond that explained in the ch-IAT, ΔF(2, 65) = 8.21, p = .001, ΔR² = .19. Responses following happy (β = .34), t = 3.00, p = .004, and sad primes (β = −.38), t = −3.34, p = .001, reliably predicted empathy, whereas the ch-IAT was no longer a significant predictor (β = .20), t = 1.80, p = .08, in the model. Age was a significant predictor in the third step (β = .24), t = 2.13, p = .04, and gender did not significantly contribute to the model (β = .19), t = 1.71, p = .09. Controlling for participant age and gender, children’s automatically activated positivity in response to happy primes and negativity in response to sad primes were significantly related to their self-reported empathy.

Additional information

Funding

This research was supported by an Social Sciences and Humanities Research Council Standard Research Grant (SSHRC-SRG) and a Canadian Foundation for Innovation Leaders Opportunity Fund (CFI-LOF) grant awarded to the second author.