Left Digit Effect in Atypical Number Line Estimation

• In Event: Poster Session 03
  In Poster Session: PS 03 Topic - Cognitive Processes

Abstract

Number line estimation (NLE) tasks are informative for theories of learning and development, predictive of real-world math outcomes, and particularly useful as teaching tools. Understanding specifically what these tasks are measuring is important for all of these reasons. When designing studies or analyzing and interpreting data, researchers generally assume that target numerals’ magnitudes are what matter, rather than the specific digits that instantiate those magnitudes. For example, it is generally assumed that it does not matter if a participant is asked to estimate “899” or “901,” because these numbers have nearly identical magnitudes. However, recent research suggests that specific digits do matter (Lai, Zax, & Barth, 2018). Both children’s and adults’ NLE placements are heavily influenced by the leftmost digits of target numerals.
The influence of this left digit effect (LDE) has been explored in canonical number line ranges (such as 0-1000 or 0-100), but not in noncanonical number lines (such as 238-1238). However, noncanonical ranges are often commonly used in studies of mathematics knowledge (e.g. Booth & Newton, 2012; Booth, Newton, & Twiss-Garrity, 2014; Chesney & Matthews, 2013; Cohen & Blanc-Goldhammer, 2011; Hurst et al., 2014; Luwel et al., 2018; Matthews & Chesney, 2011; Peeters et al., 2016). Investigations of the LDE in a noncanonical range may also be informative as to the source of this left digit bias: for example, if the LDE appears in canonical but not noncanonical ranges, this might imply that the effect depends on participants’ ability to orient themselves relative to a familiar numerical range.
To assess whether the left digit effect influences placements on a noncanonical line, we plan to conduct a pre-registered study in which adults and children aged 8-11 will be tested on two NLE tasks presented in an online context. There will be two within-subjects conditions based on line type: canonical (0-1000) and noncanonical (238-1238), and two between-subjects conditions based on order: either canonical/noncanonical or noncanonical/canonical. We plan to have a sample of N = 40 usable adults and N=160 usable children (at least 40 participants per age group after exclusions). Overall accuracy will be measured using percent absolute error (PAE) and the LDE will be measured using difference scores of paired numerals falling on either side of “hundreds” boundaries (as in Lai et al., 2018).
We anticipate the completion of data collection and analysis in winter of 2020-2021, with anticipated results available by February 2021. We are primarily interested in whether the left digit effect is present for the noncanonical range as well as the canonical range, and in whether the strength of the effect differs across line types. We predict that our preregistered analyses will reveal a left digit effect in children ages 8-11 and adults for both canonical and noncanonical number lines. We do not have a directional prediction for potential canonical/noncanonical differences.

Authors

• Leah Vaidya, Wesleyan University
  Presenting Author

• Sierra Eisen, Wesleyan University
  Non-Presenting Author

• Lauren Barragan, Wesleyan University
  Non-Presenting Author

• Carolina Montano, Wesleyan University
  Non-Presenting Author

• Sarah Aduke Ohiomah, Wesleyan University
  Non-Presenting Author

• Abby Wolk, Wesleyan University
  Non-Presenting Author

• Andrea Patalano, Wesleyan University
  Non-Presenting Author

• Hilary Barth, Wesleyan University
  Non-Presenting Author