Figure 20.

A reexamination of prior results reveals both consistency with our current results and overlooked evidence for the Bar-Tip Limit (BTL) error. Shown side by side for direct comparison are plotted data from: (a) The present study using the DDoG measure; (b) Pentoney and Berger (2016) (PB); (c) Newman and Scholl (2012) Study 5 (NS5). All graphs label correct Bar-Tip Mean response values in green and computed (or inferred, in the case of NS5) Bar-Tip Limit (BTL) response values in yellow. Notably, the PB study used the exact same 9-point rating scale and graph stimulus (of hypothetical chemical freezing temperature data) as the NS5 study (Figure 9 shows the scale and stimulus) but added ratings of four additional temperatures (−15, −10, 10, 15) to NS5's original two (−5, 5). Comparing PB's results (b) to ours (a) suggests that while PB's version of the probability rating scale measure achieved a fairly high degree of accuracy and precision at values near zero, it still suffered from apparently irreducible inaccuracy and/or imprecision as values diverged from zero (see text for further discussion).