Reliability of the left digit effect in number line estimation [Author Accepted Manuscript]

The left digit effect in number line estimation refers to the phenomenon whereby leftmost digits’ magnitudes have an outsized effect on placements. A common demonstration of the effect is that numerals with similar magnitudes but different leftmost digits (e.g., 199 and 202) are estimated to be farther apart on a number line than is warranted. Although the phenomenon is robust, it is not known whether the measure of the left digit effect is reliable for use as an individual difference measure. To address this question, we assessed its reliability in two studies using online samples. In Study 1, adults (N = 145) performed the number line estimation task twice, approximately one month apart. A large left digit effect was found at each time point, but inter-item (αs ≈ .36) and test-retest (r = .36) reliability were low. In Study 2 (with N = 136 adults), we made small changes in the number and order of trials towards increasing reliability. The revised measure had good reliability (αs ≈ .71; r = .77). The findings suggest that there are stable individual differences in the left digit effect in number line estimation, with the revised measure developed here being an improved tool for assessing and studying these differences. This work was supported by NSF DRL-1920445 and benefited from NSF DRL-1561214, both to HB and ALP. reviewed acceptedVersion