Comment on Løhre & Teigen (2016). “There is a 60% probability, but I am 70% certain: communicative consequences of external and internal expressions of uncertainty”. Thinking & Reasoning

Click to increase image sizeClick to decrease image size

Acknowledgments

We thank Carsten Erner, Tim Rakow, David Tannenbaum, and Karl Halvor Teigen for helpful discussions relevant to this Comment.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ The unpublished studies of Fox and Malle (1997) were cited by Fox and Irwin (1998). Related studies by Fox et al. (2011) were cited by Fox and Ülkümen (2011).

² We note that the internal-external distinction is logically independent from the epistemic-aleatory distinction, as both epistemic and aleatory uncertainty could, in principle, be attributed to internal or external sources (see Ülkümen et al., 2016, p.1281), though in most contexts we expect these dimensions to be empirically related in the way characterized in Table 1, columns 3 and 4. Likewise, the subjective-objective distinction is logically independent from the internal-external distinction, though we expect them to be empirically related in the way characterized in Table 1, columns 5 and 6 (see Ülkümen et al., 2016, footnote 3).

³ This said, we would argue that people sometimes indicate both epistemic and aleatory uncertainty verbally—as in “I'm pretty sure there is a high probability of rain today” which could indicate lack of confidence in the adequacy if one's model of the world or one's memory of a weather forecast.

⁴ We developed the singular/distributional reasoning measure specifically for this study. While this construct is a familiar one for judgement researchers, it can be challenging to communicate to participants. Thus, we tried to use colloquial language in expressing evaluation criteria and we asked participants to evaluate these dimensions in a single, familiar context that we thought could potentially lend itself to either form of reasoning (rain tomorrow).

⁵ We also ran a fully saturated model, where we regressed subjectivity/objectivity ratings on the term dummy, voice dummy, replicate dummy, and all two-way and three-way interaction terms. We find a reliable two-way interaction effect between term and voice (b = .38, p < .001, contrast value = 0.341), indicating that the effect of objective versus subjective voice is stronger for likelihood terms (“probability”, “chance”) than for confidence terms (“confident”, “sure”). More importantly, we observe a reliable average marginal effect for voice (dy/dx = .99, p < .001) and replicate (dy/dx = .15, p < .001), but not for term (dy/dx = –.07, p = .267).

⁶ We also ran a fully saturated model where we regressed ratings of singular/distributional reasoning on the term dummy, voice dummy, replicate dummy, and all two-way and three-way interaction terms. In this model, the only significant effect was the effect of term (dy/dx = 0.13, p = .037).

Additional information

Funding

This material is based upon work supported by the National Science Foundation under grant number SES-1427469. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.