Decomposing this interactions shows that whereas there is at least a marginally significant positive relationship between anthropomorphicity and AI responsibility in both valence conditions, the relationship is significantly stronger in the negative valence condition ( r = 0.4994, t = 4.237, p < 0.0001) relative to the positive valence condition ( r = 0.2111, t = 1.794, p = 0.0771).
1
—
—